From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 01 17:25:37 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xXl8-0004Rg-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 01 Jun 2000 17:23:50 +0100 Received: from [212.250.233.16] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xXl6-0004RX-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 01 Jun 2000 17:23:49 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id HAA12619; Thu, 1 Jun 2000 07:37:51 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id HAA00733; Thu, 1 Jun 2000 07:34:13 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006010634.HAA00733@flint.arm.linux.org.uk> Subject: Re: Patch #1 for CL7500FE support To: jgg@ualberta.ca (Jason Gunthorpe) Date: Thu, 1 Jun 2000 07:34:13 +0100 (BST) Cc: ee96162@ee.iitm.ernet.in (Lavu Sridhar), sami@hmec.co.kr (Jungjun Kim), linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Jason Gunthorpe" at May 31, 2000 05:48:51 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jason Gunthorpe writes: > To make this work I decided to entirely rewrite the head.S, it seemed to > make some assumptions that are not valid on this chip. In particular you > cannot read back the MMU control register, you get 0. head.S does not expect that you can read back the MMU control register. Which head.S are you mentioning here? The one in arch/arm/kernel, or arch/arm/boot/compressed? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 01 18:07:32 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xYQU-00059t-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 01 Jun 2000 18:06:34 +0100 Received: from [199.185.220.103] (helo=priv-edtnes03-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xYQT-00059k-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 01 Jun 2000 18:06:34 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes03-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000601180415.BRIY656.priv-edtnes03-hme0.telusplanet.net@wakko.deltatee.com>; Thu, 1 Jun 2000 12:04:15 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12xZKa-0001oN-00 (Debian); Thu, 1 Jun 2000 12:04:32 -0600 Date: Thu, 1 Jun 2000 12:04:31 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Russell King - ARM Linux Admin cc: Lavu Sridhar , Jungjun Kim , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Patch #1 for CL7500FE support In-Reply-To: <200006010634.HAA00733@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 1 Jun 2000, Russell King - ARM Linux Admin wrote: > head.S does not expect that you can read back the MMU control register. > Which head.S are you mentioning here? The one in arch/arm/kernel, or > arch/arm/boot/compressed? The one in boot/compressed. The one in the main kernel is OK. This bit: mrc p15, 0, r0, c1, c0 orr r0, r0, #0x1000 @ I-cache enable #ifndef DEBUG orr r0, r0, #0x003d @ Write buffer, mmu #endif mcr p15, 0, r0, c1, c0 The first instruction will return 0 on this chip. Though that whole routine might never be run.. I will revisit that before I make a proper patch. It may be that all it needs is the code to turn off the mmu inserted at the top: ldr r3, .MyDummyStart; @ Real start address mov r0,#((1<<5) | (1<<4) | 0) @ 32 bit Mode, MMU off mcr p15,0,r1,c5,c0 @ Flush TLBs mcr p15, 0, r0, c1, c0; @ Load the new control settings mov pc,r3; @ Jump to the real code Part of my initial investigation was warped due to that silly ARM Debugger software, still have to revisit that. :< I haven't looked at the main kernel head.S too closely, it worked on the first try! Russell, Can you suggest how to properly patch for device addresses? For instance the LAN chip on this board is at 0x80090300 and there are only two serial ports (serial driver seems to find 4, two are bogus looking). What is the best solution here? Use a kernel command line [haven't got that working yet :>] or extend the probe range of the drivers with an #ifdef? Thanks Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 01 20:09:30 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xaKy-0007Kw-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 01 Jun 2000 20:09:00 +0100 Received: from [194.168.180.25] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xaKw-0007Ki-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 01 Jun 2000 20:08:58 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id VAA15228; Thu, 1 Jun 2000 21:06:36 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id VAA00742; Thu, 1 Jun 2000 21:06:11 +0100 Message-Id: <200006012006.VAA00742@raistlin.arm.linux.org.uk> Subject: Re: Patch #1 for CL7500FE support To: jgg@ualberta.ca (Jason Gunthorpe) Date: Thu, 1 Jun 2000 21:06:10 +0100 (BST) Cc: ee96162@ee.iitm.ernet.in (Lavu Sridhar), sami@hmec.co.kr (Jungjun Kim), linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Jason Gunthorpe" at Jun 01, 2000 12:04:31 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jason Gunthorpe writes: > The one in boot/compressed. The one in the main kernel is OK. This bit: Ah ha. That bit is only ever reached if we are running on a StrongARM processor. The architecture of the code is not ready, and will probably not be ready for 2.4.0real, so the intention is that it is existing processors only (not ARM6 or ARM7). > I haven't looked at the main kernel head.S too closely, it worked on the > first try! Good! It's designed carefully to contain around zero dependencies on processor type. > Russell, Can you suggest how to properly patch for device addresses? For > instance the LAN chip on this board is at 0x80090300 and there are only > two serial ports (serial driver seems to find 4, two are bogus looking). > What is the best solution here? Use a kernel command line [haven't got > that working yet :>] or extend the probe range of the drivers with an > #ifdef? Serial ports aren't too much of a problem - there is an architecture- dependent include file in include/asm-arm/arch-cl7500/serial.h which contains them. They look like they're there for a reason, so I'd be tempted to leave them there, or play around with the flags until they do get correctly detected/not detected. It may be wise to check what setserial -bav reports for each device once you get usermode up and running. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 02 01:52:14 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xffu-00042m-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 02 Jun 2000 01:50:58 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xfft-00042f-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 02 Jun 2000 01:50:57 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000602014839.XLMW625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com> for ; Thu, 1 Jun 2000 19:48:39 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12xga1-0002Sx-00 (Debian); Thu, 1 Jun 2000 19:48:57 -0600 Date: Thu, 1 Jun 2000 19:48:57 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Patch #1 for CL7500FE support In-Reply-To: <200006012006.VAA00742@raistlin.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 1 Jun 2000, Russell King - ARM Linux Admin wrote: > Ah ha. That bit is only ever reached if we are running on a StrongARM > processor. The architecture of the code is not ready, and will probably > not be ready for 2.4.0real, so the intention is that it is existing > processors only (not ARM6 or ARM7). Ah OK. That means if I stick that code I pasted at the top of the loader it should work just fine, probably will work on all the 32 bit CPUs at least. I'll try that tomorrow. > Good! It's designed carefully to contain around zero dependencies on > processor type. Yeah, that was a dream, I was very happy I didn't have to debug deeper! > They look like they're there for a reason, so I'd be tempted to leave > them there, or play around with the flags until they do get correctly > detected/not detected. I'll look the hardware address up in the guide I have and see what is there... If anything it should be the ''standard'' alternate serial address on the ISA bus I'd think. Other CL7500FE people: I spent all day working on the ethernet chip. My board for some odd reason had a corrupt serial eeprom, I had to rewrite it (will post code..) and the cs8900 driver is coded to expect some kind of weird Card-Specific eeprom data :< I had just finished a patch around that.. All that is left is to figure out what the kernel IRQ is for the pin the chip is on and see if DMA can be made to operate. I'll mail this patch to Andrew Morton and ask his advice I suppose. Thanks, Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 02 15:32:55 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xsTl-0007qX-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 02 Jun 2000 15:31:17 +0100 Received: from [198.232.147.16] (helo=mail01-oak.pilot.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xsTk-0007qQ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 02 Jun 2000 15:31:16 +0100 Received: from Altera.COM (mail.altera.com [137.57.1.1]) by mail01-oak.pilot.net with ESMTP id IAA28662 for ; Fri, 2 Jun 2000 08:28:57 -0700 (PDT) Received: from sj-gw01.altera.com by Altera.COM (8.8.8+Sun/SMI-4.1) id IAA25039; Fri, 2 Jun 2000 08:30:44 -0700 (PDT) Received: by sj-gw01.altera.com with Internet Mail Service (5.5.2650.21) id ; Fri, 2 Jun 2000 08:30:21 -0700 Message-ID: <82B1591A280FD411B91200D0B75B484A388632@sis-uk-msg01.altera.com> From: Clive Davies To: linux-arm-kernel@lists.arm.linux.org.uk Subject: memory settings Date: Fri, 2 Jun 2000 08:27:15 -0700 X-Mailer: Internet Mail Service (5.5.2650.21) Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Can anyone tell me what the #defines TASK_SIZE, PAGE_OFFSET and PHYS_OFFSET in include/asm/arch/memory.h refer to and how they are determined for different memory maps? If could tell me where to look for this information that would be even better. TIA Clive unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 02 19:10:48 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xvrn-0003rJ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 02 Jun 2000 19:08:19 +0100 Received: from [194.168.180.103] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xvrl-0003rC-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 02 Jun 2000 19:08:18 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id SAA20267; Fri, 2 Jun 2000 18:21:16 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id SAA01894; Fri, 2 Jun 2000 18:20:44 +0100 Message-Id: <200006021720.SAA01894@raistlin.arm.linux.org.uk> Subject: Re: memory settings To: cdavies@altera.com (Clive Davies) Date: Fri, 2 Jun 2000 18:20:43 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <82B1591A280FD411B91200D0B75B484A388632@sis-uk-msg01.altera.com> from "Clive Davies" at Jun 02, 2000 08:27:15 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Clive Davies writes: > Can anyone tell me what the #defines TASK_SIZE, PAGE_OFFSET and PHYS_OFFSET > in include/asm/arch/memory.h refer to and how they are determined for > different memory maps? If could tell me where to look for this information > that would be even better. Virtual memory map 0xffffffff +--------+ | I/O | | | | Kernel | 0xc0000000 +--------+ ------------ PAGE_OFFSET | | - TASK_SIZE ^ | | ^ | | | | | | | | | | | | | | User | | | | | | | | | | | | | | | | | | | | | | | 0x00000000 +--------+ - - TASK_SIZE defines the top of user memory. PAGE_OFFSET defines the virtual address at which the kernels view of memory starts. PHYS_OFFSET is the physical start of memory. The 3GB/1GB split between user and kernel space is the best situation. What I tend to do when porting a new architecture is allocate RAM from 0xc0000000 upwards, and IO from 0xffffffff downwards. Note that large areas of dynamically allocated IO memory (eg, PCI memory space) should not be mapped at boot time, but should be mapped on an as-needed basis using the ioremap/iounmap calls, and accessed using the {read[bwl],write[bwl]} calls. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 02 22:54:35 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12xzOV-0007x1-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 02 Jun 2000 22:54:19 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12xzOU-0007wv-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 02 Jun 2000 22:54:18 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000602225201.BDWC625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com> for ; Fri, 2 Jun 2000 16:52:01 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12y0Id-0003kI-00 (Debian); Fri, 2 Jun 2000 16:52:19 -0600 Date: Fri, 2 Jun 2000 16:52:19 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Ooops? Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Still working on this Cirrus board, I have it running semi-usably now, it has been up in user land doing busy things for two hours now. But I've got some scary messages that I have no clue about: apt-get: unhandled page fault at pc=0x40199004, lr=0x40198fa8 (bad address=0x0405655c, code 2) apt-get: unhandled page fault at pc=0x40199004, lr=0x40198fa8 (bad address=0x04077bac, code 2) install-info (352): undefined instruction: pc=02048fa4 And this oops [did I decode it right?] ksymoops 2.3.4 on armv3l 2.3.99-pre8-rmk2. Options used -V (default) -K (specified) -L (specified) -O (specified) -m /boot/System.map-2.3.99-pre8-rmk2 (default) Code: <1>Unable to handle kernel paging request at virtual address 0208485c Warning (Oops_code): trailing garbage ignored on Code: line Text: 'Code: <1>Unable to handle kernel paging request at virtual address 0208485c' Garbage: 'Unable to handle kernel paging request at virtual address 0208485c' Using defaults from ksymoops -t elf32-littlearm -a arm Warning (Oops_code_values): Code looks like message, not hex digits. No disassembly attempted. *pgd = 10f03811, *pmd = 10f03811, *pte = 00000000, *ppte = 00000000 Internal error: Oops: 2 CPU: 0 pc : [] lr : [] sp : c0551f70 ip : c0551f2c fp : c0551f8c r10: e104c09c r9 : 401282d0 r8 : c0551ff4 Warning (Oops_set_i370_regs): garbage 'r9 : 401282d0 r8 : c0551ff4' at end of i370 register line ignored r7 : 00000003 r6 : c0105f10 r5 : 02084864 r4 : fffffffe r3 : c012e760 r2 : c012e768 r1 : c012e768 r0 : c0105ef8 Flags: NzCv IRQs off FIQs on Mode SVC_32 Segment user Control: EE133F10 Table: EB00313A DAC: EA00000B Process ntpdate (pid: 1831, stackpage=c0551000) Stack: c0551f60: c00243d8 c001bc50 a0000093 ffffffff e104c09c 4010e388 02084864 0000017f c0551f80: c0551fb4 c0551f90 c001c2a0 c001bb30 e104c09c 4010e388 00000000 0000017f c0551fa0: 00000001 4012a324 00000000 c0551fb8 c00167cc c001c28c 4010e388 00000000 c0551fc0: 00000001 4010e38c 4010e38c 4010e388 00000000 00000001 00000001 401282d0 c0551fe0: 4012a324 bffffc3c 00000001 bffffc1c 4011a3b4 4011a530 60000010 ffffffff Backtrace: Function entered at [] from [] r7 = 0000017F r6 = 02084864 r5 = 4010E388 r4 = E104C09C Function entered at [] from [] Code: e59f0044 e3540000 (e7951104) e2844001 01a00006 Error (Oops_bfd_perror): set_section_contents Bad value >>PC; c001bc50 <===== >>r10; e104c09c Trace; c001bb20 Trace; c001c2a0 Trace; c001c27c Trace; c00167cc <__dabt_usr+2c/40> 3 warnings and 1 error issued. Results may not be reliable. ----------------------------- After the oops the machine seems to be fine, it is still running at least. This oops looks really odd, like I am running ksymoops wrong?? Does anyone have some suggestions on what these messages all could mean? Also.. The only way I was able to get the kernel to be stable was to replace the 16meg SIMM shipped with the board with a 32Meg SIMM. The kernel still thinks there is 16M of ram, but with the 16M part it oops all over the place. I only have two guesses here, 1) There is an off-by-one type error in the memory sizing code and the kernel is going past the end of ram/ 2) My SIMM is bad 3) The memory settings (EDO/FPM/speed) are wrong, though both SIMMs are 60ns .. Thanks, Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 04 18:52:53 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12yeXL-0006BI-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 04 Jun 2000 18:50:11 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96162) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12yeXH-0006Ae-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 04 Jun 2000 18:50:08 +0100 Received: from localhost (ee96162@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id XAA10548; Sun, 4 Jun 2000 23:22:07 +0530 Date: Sun, 4 Jun 2000 23:22:07 +0530 (IST) From: Lavu Sridhar Reply-To: ee96162@ee.iitm.ernet.in To: Jason Gunthorpe cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, I have downloaded the kernel image (from the site u mentioned) to the Cirrus 7500FE board using angelboot and ttyS0. After downloading I do not see any error messages on the screen(connected to the VGA port on the Cirrus Logic Board) Lavu Sridhar PS : Cirrus board's COM1 is connected to COM1 of host Cirrus board's COM2 is unconnected Cirrus board's VGA is connected to a PC monitor Regards, Lavu Sridhar --------------------------------------------------------------------- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 04 21:03:37 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12ygc1-0007ik-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 04 Jun 2000 21:03:09 +0100 Received: from [199.185.220.104] (helo=priv-edtnes04-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12ygc0-0007ib-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 04 Jun 2000 21:03:08 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes04-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000604200305.NSVV4208.priv-edtnes04-hme0.telusplanet.net@wakko.deltatee.com>; Sun, 4 Jun 2000 14:03:05 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12ygcH-0008Mf-00 (Debian); Sun, 4 Jun 2000 14:03:25 -0600 Date: Sun, 4 Jun 2000 14:03:25 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Lavu Sridhar cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Sun, 4 Jun 2000, Lavu Sridhar wrote: > I have downloaded the kernel image (from the site u mentioned) to > the Cirrus 7500FE board using angelboot and ttyS0. After downloading I do > not see any error messages on the screen(connected to the VGA port on the > Cirrus Logic Board) Nobody has written a console driver yet, so the video outputs will not show anything. After downloading the image you have to attach a terminal program to the serial port. Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 05 10:24:36 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12yt73-0008AB-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 05 Jun 2000 10:24:01 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96209) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12yt6y-00089B-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 05 Jun 2000 10:23:58 +0100 Received: from localhost (ee96209@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id OAA18130; Mon, 5 Jun 2000 14:56:14 +0530 Date: Mon, 5 Jun 2000 14:56:14 +0530 (IST) From: Vikas Datta To: Jason Gunthorpe cc: Lavu Sridhar , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, > Nobody has written a console driver yet, so the video outputs will not > show anything. After downloading the image you have to attach a terminal > program to the serial port. The connections I have used for bootlader and the serial console are the serial port (COM1) on the board and the host. After downloading the image (using angelboot) I have used minicom in order to see the kernel messages. Minicom does not show anything either. Can you please explain the set-up (connection and terminal program) you have used. Thanks, Regards Vikas Datta unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 05 13:26:52 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12yvwP-00027m-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 05 Jun 2000 13:25:13 +0100 Received: from [195.108.189.102] (helo=white.plant.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12yvwO-00027c-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 05 Jun 2000 13:25:12 +0100 Received: from [194.134.11.47] by white.plant.nl (NTMail 4.30.0013/NT3720.00.67541ba1) with ESMTP id iontfaaa for ; Mon, 5 Jun 2000 14:21:03 +0200 Message-ID: <393B9B77.3A109E6A@maycom.nl> Date: Mon, 05 Jun 2000 14:22:15 +0200 From: Daven Baptist X-Mailer: Mozilla 4.7 [en] (X11; I; Linux 2.2.13 i686) X-Accept-Language: en MIME-Version: 1.0 To: linux-arm-kernel@lists.arm.linux.org.uk Subject: module problems Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk can someone help me, i wrote/copied a extreem simpel module #define MODULE #include int init_module() { printk("hello,world -this is the kernel speaking\n"); return 0; } void cleanup_module() { printk("Short is the life of a kernel module\n"); } i compiled it and put it on a ramdisk but as soon as i load it it tells me i'm not using the richt kernel ?? [root@Linux /]$insmod hello.o hello.o: kernel-module version mismatch hello.o was compiled for kernel version 2.2.13-rmk2-np14 while this kernel is version 2.3.99-pre8-rmk1-np5. [root@Linux /]$insmod hello.o -f Warning:hello,world -this is the kernel speaking kernel-module version mismatch hello.o was compiled for kernel version 2.2.13-rmk2-np14 while this kernel is version 2.3.99-pre8-rmk1-np5 [root@Linux /]$rmmod hello Short is the life of a kernel module but as far as i know i'm using 2.3.99~~~~ how can i solve this 'problem' ?? Daven unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 05 20:29:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12z2Xm-0008QB-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 05 Jun 2000 20:28:14 +0100 Received: from [130.161.42.1] (helo=duteinh.et.tudelft.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12z2Xl-0008Q4-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 05 Jun 2000 20:28:13 +0100 Received: from kalman.et.tudelft.nl by duteinh.et.tudelft.nl (8.9.3/BB-2.0) id VAA06757; Mon, 5 Jun 2000 21:28:17 +0200 (MET DST) Received: (from erik@localhost) by kalman.et.tudelft.nl (8.8.8/8.8.8/Debian/GNU) id VAA31863; Mon, 5 Jun 2000 21:28:16 +0200 Message-Id: <200006051928.VAA31863@kalman.et.tudelft.nl> Subject: Re: module problems In-Reply-To: <393B9B77.3A109E6A@maycom.nl> from Daven Baptist at "Jun 5, 0 02:22:15 pm" To: daven@maycom.nl (Daven Baptist) Date: Mon, 5 Jun 2000 21:28:16 +0200 (CEST) Cc: linux-arm-kernel@lists.arm.linux.org.uk From: J.A.K.Mouw@its.tudelft.nl (Erik Mouw) Organization: Eric Conspiracy Secret Labs X-Eric-Conspiracy: There is no conspiracy! X-Mailer: ELM [version 2.4ME+ PL39 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Daven Baptist wrote: > can someone help me, i wrote/copied a extreem > [snip] > i compiled it and put it on a ramdisk > but as soon as i load it it tells me i'm not using > the richt kernel ?? > > [root@Linux /]$insmod hello.o > hello.o: kernel-module version mismatch > hello.o was compiled for kernel version 2.2.13-rmk2-np14 [...] > but as far as i know i'm using 2.3.99~~~~ > how can i solve this 'problem' ?? My guess is that you're using my cross compiler toolchain, which is actually compiled with linux-2.2.13-rmk2-np14. What's happening is that your kernel is indeed linux-2.3.99-something, but that you're compiling against the linux-2.2.13 headers. You can solve the problem in two ways: - Remove the original kernel headers and make a softlink to the current kernel headers: cd /data/lart/cross/arm-linux/include mv asm asm-orig mv linux linux-orig ln -s /path/to/recent/kernel/linux/include/linux linux ln -s /path/to/recent/kernel/linux/include/asm asm - Compile with different compiler flags: arm-linux-gcc -Wall -O2 -D__KERNEL__ -DMODULE -nostdinc \ -I/path/to/recent/kernel/linux -c hello.c -o hello.o Erik -- J.A.K. (Erik) Mouw, Information and Communication Theory Group, Department of Electrical Engineering, Faculty of Information Technology and Systems, Delft University of Technology, PO BOX 5031, 2600 GA Delft, The Netherlands Phone: +31-15-2783635 Fax: +31-15-2781843 Email: J.A.K.Mouw@its.tudelft.nl WWW: http://www-ict.its.tudelft.nl/~erik/ unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 06 02:20:03 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12z81I-00054b-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 06 Jun 2000 02:19:04 +0100 Received: from [199.185.220.103] (helo=priv-edtnes03-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12z81H-00054U-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 06 Jun 2000 02:19:03 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes03-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000606011906.QULH656.priv-edtnes03-hme0.telusplanet.net@wakko.deltatee.com>; Mon, 5 Jun 2000 19:19:06 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12z81f-0002H3-00 (Debian); Mon, 5 Jun 2000 19:19:27 -0600 Date: Mon, 5 Jun 2000 19:19:27 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Vikas Datta cc: Lavu Sridhar , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Mon, 5 Jun 2000, Vikas Datta wrote: > The connections I have used for bootlader and the serial console are the > serial port (COM1) on the board and the host. After downloading the image > (using angelboot) I have used minicom in order to see the kernel messages. > Minicom does not show anything either. Can you please explain the set-up > (connection and terminal program) you have used. Hum, that sounds all right. can you watch the leds in the upper left corner they will go all 8 on when then kernel is booting. Minicom should work, if you have the serial settings right. The best way to test minicom is to fire it up and then hit the rest button, you should see an Angel boot message and some garbage (packet marker) characters. 9600,N81 are the settings, probably want to disable flow control too. Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 06 02:36:10 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12z8HS-0005IX-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 06 Jun 2000 02:35:46 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12z8HR-0005IR-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 06 Jun 2000 02:35:45 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000606013546.JOHI625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com>; Mon, 5 Jun 2000 19:35:46 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12z8Hn-0002JJ-00 (Debian); Mon, 5 Jun 2000 19:36:07 -0600 Date: Mon, 5 Jun 2000 19:36:07 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com Reply-To: Jason Gunthorpe To: linux-arm-kernel@lists.arm.linux.org.uk cc: Russell King - ARM Linux Admin Subject: Patch for cs89x0 and CL7500FE Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi, Enclosed is my patch for the ethernet driver. It enables it to support the Cirrus eval board(s?) and I think tidies up a few related things inside the driver. It does not yet support DMA operation, I will work on that next week I hope. A note: My board came with a corrupt EEPROM, so even this driver will not work on such a board. If you have a board like this the driver will tell you so and refuse to load. Russell, I think this is the first patch I have that would be suitable for inclusion in the kernel. If you think it is good I will forward it to Andrew (the cs89x0 driver guy) and get his thoughts, you might want to put in your patch set.. I tried for 2 hours today to get head.S to work with very little luck, so the booting patches will have to wait. On the plus side, I just got a EP7209 based eval board (http://www.cirrus.com/design/products/overview/index.cfm?ProductID=109). I really need linux working on that so my efforts will be focused there for the next while. The board comes with a Linux 2.2.1 port from some odd company (Royal Linux?), but it doesn't seem to work.. :| Thanks, Jason --- linux2/drivers/net/cs89x0.c Tue May 2 13:40:59 2000 +++ linux/drivers/net/cs89x0.c Mon Jun 5 10:45:33 2000 @@ -128,10 +128,26 @@ #include "cs89x0.h" -/* First, a few definitions that the brave might change. */ -/* A zero-terminated list of I/O addresses to be probed. */ +/* First, a few definitions that the brave might change. + A zero-terminated list of I/O addresses to be probed. Some special flags.. + Addr & 1 = Read back the address port, look for signature and reset + the page window before probing + Addr & 3 = Reset the page window and probe + The CLPS eval board has the Cirrus chip at 0x80090300, in ARM IO space, + but it is possible that a Cirrus board could be plugged into the ISA + slots. */ +/* The cs8900 has 4 IRQ pins, software selectable. cs8900_irq_map maps + them to system IRQ numbers. This mapping is card specific and is set to + the configuration of the Cirrus Eval board for this chip. */ +#ifdef CONFIG_ARCH_CLPS7500 +static unsigned int netcard_portlist[] __initdata = + { 0x80090303, 0x300, 0x320, 0x340, 0x360, 0x200, 0x220, 0x240, 0x260, 0x280, 0x2a0, 0x2c0, 0x2e0, 0}; +static unsigned int cs8900_irq_map[] = {12,0,0,0}; +#else static unsigned int netcard_portlist[] __initdata = { 0x300, 0x320, 0x340, 0x360, 0x200, 0x220, 0x240, 0x260, 0x280, 0x2a0, 0x2c0, 0x2e0, 0}; +static unsigned int cs8900_irq_map[] = {10,11,12,5}; +#endif #if DEBUGGING static unsigned int net_debug = DEBUGGING; @@ -362,15 +378,17 @@ /* if they give us an odd I/O address, then do ONE write to the address port, to get it back to address zero, where we - expect to find the EISA signature word. */ + expect to find the EISA signature word. An IO with a base of 0x3 + will skip the test for the ADD_PORT. */ if (ioaddr & 1) { - ioaddr &= ~1; - if ((inw(ioaddr + ADD_PORT) & ADD_MASK) != ADD_SIG) - return ENODEV; + if ((ioaddr & 2) != 2) + if ((inw((ioaddr & ~3)+ ADD_PORT) & ADD_MASK) != ADD_SIG) + return ENODEV; + ioaddr &= ~3; outw(PP_ChipID, ioaddr + ADD_PORT); } - if (inw(ioaddr + DATA_PORT) != CHIP_EISA_ID_SIG) + if (inw(ioaddr + DATA_PORT) != CHIP_EISA_ID_SIG) { retval = ENODEV; goto out1; @@ -403,15 +421,77 @@ dev->base_addr); reset_chip(dev); - - /* First check to see if an EEPROM is attached*/ + + /* Here we read the current configuration of the chip. If there + is no Extended EEPROM then the idea is to not disturb the chip + configuration, it should have been correctly setup by automatic + EEPROM read on reset. So, if the chip says it read the EEPROM + the driver will always do *something* instead of complain that + adapter_cnf is 0. */ + if ((readreg(dev, PP_SelfST) & (EEPROM_OK | EEPROM_PRESENT)) == + (EEPROM_OK|EEPROM_PRESENT)) { + /* Load the MAC. */ + for (i=0; i < ETH_ALEN/2; i++) { + unsigned int Addr; + Addr = readreg(dev, PP_IA+i*2); + dev->dev_addr[i*2] = Addr & 0xFF; + dev->dev_addr[i*2+1] = Addr >> 8; + } + + /* Load the Adapter Configuration. + Note: Barring any more specific information from some + other source (ie EEPROM+Schematics), we would not know + how to operate a 10Base2 interface on the AUI port. + However, since we do read the status of HCB1 and use + settings that always result in calls to control_dc_dc(dev,0) + a BNC interface should work if the enable pin + (dc/dc converter) is on HCB1. It will be called AUI + however. */ + + lp->adapter_cnf = 0; + i = readreg(dev, PP_LineCTL); + /* Preserve the setting of the HCB1 pin. */ + if ((i & (HCB1 | HCB1_ENBL)) == (HCB1 | HCB1_ENBL)) + lp->adapter_cnf |= A_CNF_DC_DC_POLARITY; + /* Save the sqelch bit */ + if ((i & LOW_RX_SQUELCH) == LOW_RX_SQUELCH) + lp->adapter_cnf |= A_CNF_EXTND_10B_2 | A_CNF_LOW_RX_SQUELCH; + /* Check if the card is in 10Base-t only mode */ + if ((i & (AUI_ONLY | AUTO_AUI_10BASET)) == 0) + lp->adapter_cnf |= A_CNF_10B_T | A_CNF_MEDIA_10B_T; + /* Check if the card is in AUI only mode */ + if ((i & (AUI_ONLY | AUTO_AUI_10BASET)) == AUI_ONLY) + lp->adapter_cnf |= A_CNF_AUI | A_CNF_MEDIA_AUI; + /* Check if the card is in Auto mode. */ + if ((i & (AUI_ONLY | AUTO_AUI_10BASET)) == AUTO_AUI_10BASET) + lp->adapter_cnf |= A_CNF_AUI | A_CNF_10B_T | + A_CNF_MEDIA_AUI | A_CNF_MEDIA_10B_T | A_CNF_MEDIA_AUTO; + + /* IRQ. Other chips already probe, see below. */ + if (lp->chip_type == CS8900) + lp->isa_config = readreg(dev, PP_CS8900_ISAINT) & INT_NO_MASK; + + printk( "[Cirrus EEPROM] "); + } + + printk("\n"); + + /* First check to see if an EEPROM is attached. */ if ((readreg(dev, PP_SelfST) & EEPROM_PRESENT) == 0) - printk(KERN_WARNING "\ncs89x0: No EEPROM, relying on command line....\n"); + printk(KERN_WARNING "cs89x0: No EEPROM, relying on command line....\n"); else if (get_eeprom_data(dev, START_EEPROM_DATA,CHKSUM_LEN,eeprom_buff) < 0) { printk(KERN_WARNING "\ncs89x0: EEPROM read failed, relying on command line.\n"); } else if (get_eeprom_cksum(START_EEPROM_DATA,CHKSUM_LEN,eeprom_buff) < 0) { - printk(KERN_WARNING "\ncs89x0: EEPROM checksum bad, relying on command line\n"); + /* Check if the chip was able to read its own configuration starting + at 0 in the EEPROM*/ + if ((readreg(dev, PP_SelfST) & (EEPROM_OK | EEPROM_PRESENT)) != + (EEPROM_OK|EEPROM_PRESENT)) + printk(KERN_WARNING "cs89x0: Extended EEPROM checksum bad and no Cirrus EEPROM, relying on command line\n"); + } else { + /* This reads an extended EEPROM that is not documented + in the CS8900 datasheet. */ + /* get transmission control word but keep the autonegotiation bits */ if (!lp->auto_neg_cnf) lp->auto_neg_cnf = eeprom_buff[AUTO_NEG_CNF_OFFSET/2]; /* Store adapter configuration */ @@ -464,17 +544,12 @@ } else { i = lp->isa_config & INT_NO_MASK; if (lp->chip_type == CS8900) { - /* the table that follows is dependent upon how you wired up your cs8900 - * in your system. The table is the same as the cs8900 engineering demo - * board. irq_map also depends on the contents of the table. Also see - * write_irq, which is the reverse mapping of the table below. */ - switch(i) { - case 0: i = 10; break; - case 1: i = 11; break; - case 2: i = 12; break; - case 3: i = 5; break; - default: printk("\ncs89x0: bug: isa_config is %d\n", i); - } + /* Translate the IRQ using the IRQ mapping table. */ + if (i > sizeof(cs8900_irq_map)/sizeof(cs8900_irq_map[0])) + printk("\ncs89x0: bug: isa_config is %d\n", i); + else + i = cs8900_irq_map[i]; + lp->irq_map = CS8900_IRQ_MAP; /* fixed IRQ map for CS8900 */ } else { int irq_map_buff[IRQ_MAP_LEN/2]; @@ -490,7 +565,7 @@ dev->irq = i; } - printk(", IRQ %d", dev->irq); + printk(" IRQ %d", dev->irq); #if ALLOW_DMA if (lp->use_dma) @@ -741,7 +816,9 @@ struct net_local *lp = (struct net_local *)dev->priv; unsigned int selfcontrol; int timenow = jiffies; - /* control the DC to DC convertor in the SelfControl register. */ + /* control the DC to DC convertor in the SelfControl register. + Note: This is hooked up to a general purpose pin, might not + always be a DC to DC convertor. */ selfcontrol = HCB1_ENBL; /* Enable the HCB1 bit as an output */ if (((lp->adapter_cnf & A_CNF_DC_DC_POLARITY) != 0) ^ on_not_off) @@ -916,13 +993,13 @@ int i; if (chip_type == CS8900) { - switch(irq) { - case 10: i = 0; break; - case 11: i = 1; break; - case 12: i = 2; break; - case 5: i = 3; break; - default: i = 3; break; - } + /* Search the mapping table for the corrisponding IRQ pin. */ + for (i = 0; i != sizeof(cs8900_irq_map)/sizeof(cs8900_irq_map[0]); i++) + if (cs8900_irq_map[i] == irq) + break; + /* Not found */ + if (i == sizeof(cs8900_irq_map)/sizeof(cs8900_irq_map[0])) + i = 3; writereg(dev, PP_CS8900_ISAINT, i); } else { writereg(dev, PP_CS8920_ISAINT, irq); unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 06 02:39:52 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12z8L9-0005Nn-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 06 Jun 2000 02:39:35 +0100 Received: from [199.185.220.104] (helo=priv-edtnes04-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12z8L8-0005Ng-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 06 Jun 2000 02:39:34 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes04-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000606013937.SMOS4208.priv-edtnes04-hme0.telusplanet.net@wakko.deltatee.com>; Mon, 5 Jun 2000 19:39:37 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12z8LX-0002KJ-00 (Debian); Mon, 5 Jun 2000 19:39:59 -0600 Date: Mon, 5 Jun 2000 19:39:59 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com Reply-To: Jason Gunthorpe To: Russell King - ARM Linux Admin cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Quick patch for cl7500 loader Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi Russell, This is an isolated patch that fixes up the ucompress.h for cl7500. Whoever wrote the original was using com2 and had the wrong baud divisor (weird). I also made the serial port base easially changeable.. Thanks, Jason --- linux2/include/asm-arm/arch-cl7500/uncompress.h Tue Nov 23 23:23:11 1999 +++ linux/include/asm-arm/arch-cl7500/uncompress.h Mon Jun 5 19:35:09 2000 @@ -5,11 +5,12 @@ */ #define BASE 0x03010000 +#define SERBASE (BASE + (0x3f8 << 2)) static __inline__ void putc(char c) { - while (!(*((volatile unsigned int *)(BASE + 0xbf4)) & 0x20)); - *((volatile unsigned int *)(BASE + 0xbe0)) = c; + while (!(*((volatile unsigned int *)(SERBASE + 0x14)) & 0x20)); + *((volatile unsigned int *)(SERBASE)) = c; } /* @@ -27,13 +28,13 @@ static __inline__ void arch_decomp_setup(void) { - int baud = 3686400 / (9600 * 16); + int baud = 3686400 / (9600 * 32); - *((volatile unsigned int *)(BASE + 0xBEC)) = 0x80; - *((volatile unsigned int *)(BASE + 0xBE0)) = baud & 0xff; - *((volatile unsigned int *)(BASE + 0xBE4)) = (baud & 0xff00) >> 8; - *((volatile unsigned int *)(BASE + 0xBEC)) = 3; /* 8 bits */ - *((volatile unsigned int *)(BASE + 0xBF0)) = 3; /* DTR, RTS */ + *((volatile unsigned int *)(SERBASE + 0xC)) = 0x80; + *((volatile unsigned int *)(SERBASE + 0x0)) = baud & 0xff; + *((volatile unsigned int *)(SERBASE + 0x4)) = (baud & 0xff00) >> 8; + *((volatile unsigned int *)(SERBASE + 0xC)) = 3; /* 8 bits */ + *((volatile unsigned int *)(SERBASE + 0x10)) = 3; /* DTR, RTS */ } /* unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 06 07:25:01 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zCmi-0000Py-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 06 Jun 2000 07:24:20 +0100 Received: from [210.59.228.108] (helo=relay5.url.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 12zCmg-0000PT-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 06 Jun 2000 07:24:19 +0100 Received: from ms3.url.com.tw ([210.59.228.13]) by AccSMTP/NT 2.5 (210.59.228.108) [210.59.228.108]; Tue, 6 Jun 2000 14:38:56 +0800 Received: from pads3 ([140.114.79.13]) by AccSMTP/NT 2.5 (210.59.228.36) [210.59.228.36]; Tue, 6 Jun 2000 14:24:22 +0800 Message-ID: <000801bfcf7f$e4bab220$0d4f728c@cs.nthu.edu.tw> From: "Kevin Cheng" To: Subject: how to modify "t-bare" ? Date: Tue, 6 Jun 2000 14:24:37 +0800 MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0005_01BFCFC2.F20B5FE0" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2314.1300 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk This is a multi-part message in MIME format. ------=_NextPart_000_0005_01BFCFC2.F20B5FE0 Content-Type: text/plain; charset="big5" Content-Transfer-Encoding: quoted-printable Hi there, I want to build a cross compiler for "arm-coff" format, and I followed Chris Rutter's page (http://www.inkvine.fluff.org/~chris/arm-tools.html) on Building the GNU toolchain for ARM targets. There are always error message: ./libgcc2.c:41: stdlib.h: No such file or directory ./libgcc2.c:42: unistd.h: No such file or directory make[3]: *** [libgcc2.a] Error 1 I employed the "-Dinhibit_libc" hack in "t-linux", still error. And I checked the Makefile, I fonnd that it merged the "t-bare" file, not "t-linux". Is anyone can tell me how to modify the "t-bare" file? Thank for any opinion. Kevinmailto:beast@ms3.url.com.tw ------=_NextPart_000_0005_01BFCFC2.F20B5FE0 Content-Type: text/html; charset="big5" Content-Transfer-Encoding: quoted-printable
Hi there,
I want to = build a cross compiler for=20 "arm-coff" format,
and I followed Chris Rutter's page
(http://www.in= kvine.fluff.org/~chris/arm-tools.html)
on=20 Building the GNU toolchain for ARM targets. There are
always error=20 message:
 
      ./libgcc2.c:41:=20 stdlib.h: No such file or directory
     =20 ./libgcc2.c:42: unistd.h: No such file or=20 directory
      make[3]: *** [libgcc2.a] = Error=20 1
 
I employed the = "-Dinhibit_libc" hack in "t-linux",=20 still
error. And I checked the Makefile, I fonnd that it = merged
the=20 "t-bare" file, not "t-linux". Is anyone can tell me
how to modify the = "t-bare" file?
Thank for any opinion.
 
Kevinmailto:beast@ms3.url.com.tw ------=_NextPart_000_0005_01BFCFC2.F20B5FE0-- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 06 15:29:35 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zKLQ-0006yt-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 06 Jun 2000 15:28:40 +0100 Received: from [202.141.26.182] (helo=agasthya.ee.iitm.ernet.in ident=ee96209) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zKLL-0006yM-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 06 Jun 2000 15:28:37 +0100 Received: from localhost (ee96209@localhost) by agasthya.ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id UAA21280; Tue, 6 Jun 2000 20:03:06 +0530 X-Authentication-Warning: agasthya.ee.iitm.ernet.in: ee96209 owned process doing -bs Date: Tue, 6 Jun 2000 20:03:06 +0530 (IST) From: Vikas Datta To: Jason Gunthorpe cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, I have tried replacing the SIMM with a good one. It still shows the same error. My vmlinux is 1461K and zImage is 533K. I have tried debugging the uncompressing function. The misc.c calls inflate.c in /linux/lib. I placed various debugpoints using error() and finally found that the program is getting stuck at the huft_build function in /linux/lib/inflate.c The program returns an error in DEBG("huft 3") saying : bad input; more codes than bits ! I have used binutils 2.95.2 for building the kernel. I would be trying the hexdump later and would compare the first 512 bytes of the zImage loaded with the zImage on the host PC. any different ideas? Thanks, Regards, Vikas On Mon, 5 Jun 2000, Jason Gunthorpe wrote: > > On Tue, 6 Jun 2000, Vikas Datta wrote: > > > Thank you for the suggestions. I have now got the minicom working. > > While downloading the zImage using the angelbootloader at the address > > location 0x208000, i get the leds glowing. One led then goes off and the > > message on minicom from the cirrus COM2 port is : >>CL-PS7500FE Loader, Verison: Jun 3 2000 08:54:09 >>Loading without relocation.. >>Output_Start: 10008000 >>free_mem_ptr: 1028C008 >>free_mem_ptr_end: 1029C008Uncompressing Linux... >>invalid compressed format (err=2) >>-- System halted > > Hmm. In all my testing I never once got *this* message. > > It can only mean one of two things, I think. > a) The original file, or the serial download caused file corruption > b) Your ram is acting flaky like mine was. > > I suppose there is a faint possibility of memory corruption, or running > off the end of ram or something.. Make certain your kernel vmlinux is < > 2m. (Mine was 1.5meg vmlinux, 500k zImage) > > Otherwise I would suggest you pull the SIMM and pop in another, see if > that works better. Maybe write a little memory tester too.. SIMMS have to > be or 16meg or better and I am using 60ns parts, EDO I think. > > It may also be memory timings. You can adjust those with some chip > specific registers. > > > Is there a way to debug the program and see the assembly-level > > instructions of the zImage file? > > No, at this point you should be looking at using puth's and puts's to > debug the decompressor stages. Look to see if the output pointer goes past > the load address for the startup (Kernel > 2Meg) or if the data it is > decompressing is valid. If you can show that there is a 1 bit error in the > comrpessed data then it would definately point to funny RAM. If you show > that the compressed data is totally hosed then some pointer is going out > of bounds some how. > > Maybe print out a hex dump of 512 bytes starting from the front and > 512bytes starting from the CRC failing block. That would give some idea > what is going on. > > > when i try using ../boot/compressed/vmlinux, i get no response on the > > minicom and none of the leds glow either. > > That is an Elf Image, you cannot use it until you use objcopy to make it > into a flat memory dump (the zImage file). > > Jason > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 13:47:53 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zfCy-0000LY-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 13:45:20 +0100 Received: from [195.134.130.75] (helo=pip5.mpl.ch) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zfCy-0000Fj-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 13:45:20 +0100 Received: from elsoft.ch (pcdm.elsoft.ch [195.134.130.116]) by pip5.mpl.ch (8.9.3/8.9.3) with ESMTP id KAA04458 for ; Wed, 7 Jun 2000 10:11:35 +0200 Message-ID: <393E0372.639ADD6E@elsoft.ch> Date: Wed, 07 Jun 2000 10:10:26 +0200 From: David =?iso-8859-1?Q?M=FCller?= (ELSOFT AG) Organization: ELSOFT AG X-Mailer: Mozilla 4.73 [en] (WinNT; U) X-Accept-Language: de-CH,de,en MIME-Version: 1.0 To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Console on serial port on EBSA285 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello I would like to have the console I/O redirected to the serial port of my "VGA less" EBSA285. I have the kernel (2.3.39 + patch-2.3.39-rmk1) configured for "Console on serial port" and have tried the "console=" kernel parameter, but without success. What am i missing? Dave unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 14:19:15 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zfjO-00012a-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 14:18:50 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zfj0-000116-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 14:18:28 +0100 Received: from localhost (ee96209@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id SAA16600; Wed, 7 Jun 2000 18:40:00 +0530 Date: Wed, 7 Jun 2000 18:40:00 +0530 (IST) From: Vikas Datta To: Jason Gunthorpe cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, On Tue, 6 Jun 2000, Jason Gunthorpe wrote: > Try to increase the stack size maybe? In the vmlinux.lds.in file. > I tried increasing the stack size....but no change.... I then wrote a small code in misc.c that adds up all 32-bit ARM words in the kernel from the DRAM and displays their sum....similar to the EEPROM checksum! The checksum is added in the zImage and checked seperately. Care is taken that both the programs add the same ARM words. I get zImage Feedback from DRAM 7065f500 7064f50e trial one with entry point in bank 2 bc45e689 bc44e687 trial two with entry point in bank 0 Atleast two nibbles seem to get corrupted during the download to DRAM. Is it because angelboot is doing something wrong ? Or is my inflate.c compiled using binutils 2.95.2 for the kernel 2.3.99-pre8 incorrect. I have also independently checked the DRAM using a memory test program running over SDT. When i compared the hexdump of the whole image, i find that errors occur inconsistently at different places each time.... sometimes in the middle of the kernel and sometimes towards the end. However the kernel always gets stuck in the huft_build function in /linux/lib/inflate.c What could be going wrong.....? There does seem to be file corruption somewhere..... Thanks, Regards, Vikas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 15:07:23 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zgTz-0001sd-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 15:06:59 +0100 Received: from [199.185.220.112] (helo=priv-edtnes12-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zgTy-0001sX-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 15:06:59 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes12-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000607150440.JRHA24907.priv-edtnes12-hme0.telusplanet.net@wakko.deltatee.com>; Wed, 7 Jun 2000 09:04:40 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12zhOB-0005Rx-00 (Debian); Wed, 7 Jun 2000 09:05:03 -0600 Date: Wed, 7 Jun 2000 09:05:03 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Vikas Datta cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, 7 Jun 2000, Vikas Datta wrote: > that errors occur inconsistently at different places each time.... > sometimes in the middle of the kernel and sometimes towards the end. Can you try running checksum multiple times? See if it stays the same over each run. AFAIK the angel protocol has per block checksums so data should not be getting corrupted while it is in transit. However.. it may be that the angelboot program does not understand a CRC failure reply, try a lower baud rate maybe? Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 16:28:22 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zhkJ-0003D3-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 16:27:55 +0100 Received: from [216.163.180.10] (helo=c0mailgw08.prontomail.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zhkI-0003CF-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 16:27:54 +0100 Received: from web104 (216.163.176.204) by c0mailgw08.prontomail.com (NPlex 4.5.049) id 3925E52E000C11FB for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 7 Jun 2000 09:24:51 -0700 X-Version: canoe 6.2.2329.0 From: "Dan" Message-Id: <6F21EC0BC8C34D115A040005B8275D56@dmcdon.canoemail.com> Date: Wed, 7 Jun 2000 10:24:18 -0600 X-Priority: Normal Content-Type: text/plain; charset=iso-8859-1 To: linux-arm-kernel@lists.arm.linux.org.uk Reply-To: dmcdon@canoemail.com Subject: Floating Point Emulation X-Mailer: Web Based Pronto Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello: We have recently ported Linux to an SA1110-based board for a portable application. With little experience with the ARM processor and none with Linux, we were prepared for quite the battle. However, much to our surprise it was amazingly easy. I just about fell off my chair when we saw the console login come up and we typed root. Cool! Now to the task at hand. The Linux Kernel uses Floating Point Emulation coupled via the ARM's undefined instruction vector. The development environment -- gcc and the libs -- relies on the use of floating point instructions being emitted by the compiler. My problem is that I have a small C++ stand-alone application that I want to use to test new boards. I would like to build the application using the arm-linux development environment. I do not want to spawn a second set of libraries compatible with the -msoft-float flag. Can the Floating Point Emulation used by the Linux Kernel be used without the kernel ? The code at the undefined instruction vector passes a data area from the current task into the FPEmulator. Is this just a scratch area or is it used to pass data to the Emulator ? Any advice would be appreciated. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 16:41:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zhxV-0003Re-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 16:41:33 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zhxU-0003RV-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 16:41:32 +0100 Date: Wed, 7 Jun 2000 12:38:38 -0400 (EDT) From: Nicolas Pitre To: Dan cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation In-Reply-To: <6F21EC0BC8C34D115A040005B8275D56@dmcdon.canoemail.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, 7 Jun 2000, Dan wrote: > Hello: > > We have recently ported Linux to an SA1110-based > board for a portable application. > > With little experience with the ARM processor and > none with Linux, we were prepared for quite the > battle. However, much to our surprise it was amazingly > easy. I just about fell off my chair when we saw > the console login come up and we typed root. Cool! ;-) > My problem is that I have a small C++ stand-alone > application that I want to use to test new boards. > I would like to build the application using the > arm-linux development environment. I do not want to > spawn a second set of libraries compatible with the > -msoft-float flag. If you can arrange for not using floating point operations in your code i.e. using fixed point math techniques instead, then you may just ignore the whole issue. Do you really need floating point for your testing app anyway? Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 16:52:25 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zi7j-0003wa-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 16:52:07 +0100 Received: from [63.192.220.206] (helo=hermes.mvista.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zi7Y-0003vW-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 16:51:58 +0100 Received: from mvista.com (IDENT:mhatle@gemini.mvista.com [10.0.0.101]) by hermes.mvista.com (8.9.3/8.9.3) with ESMTP id JAA19148; Wed, 7 Jun 2000 09:48:54 -0700 Message-ID: <393E7CAD.BF286751@mvista.com> Date: Wed, 07 Jun 2000 09:47:41 -0700 From: Mark Hatle Organization: MontaVista Software, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: dmcdon@canoemail.com CC: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <6F21EC0BC8C34D115A040005B8275D56@dmcdon.canoemail.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk What timing.. :) I'm fighting the same thing. I am attempting to turn OFF the floating point emulation in the kernel and remove the need for the second library. All of the other platforms that I am used to use the built in gcc library for floating point emulation. I am currently working on patching gcc (2.95.2) to do soft-float the way other platforms do it. (If you look in the gcc source in gcc/config/arm you'll see a but of "t-*" files. all of the files EXCEPT the linux one use the built-in soft-floating point. So it should be possible.) FYI, I have yet to run the tests on the floating point, so what I have now may not work the way I think it does! If you would like a copy of my patches email me off the list and I'll go into exactly what I'm doing. --Mark Dan wrote: > > Hello: > > We have recently ported Linux to an SA1110-based > board for a portable application. > > With little experience with the ARM processor and > none with Linux, we were prepared for quite the > battle. However, much to our surprise it was amazingly > easy. I just about fell off my chair when we saw > the console login come up and we typed root. Cool! > > Now to the task at hand. > > The Linux Kernel uses Floating Point Emulation coupled > via the ARM's undefined instruction vector. The development > environment -- gcc and the libs -- relies on the use > of floating point instructions being emitted by the > compiler. > > My problem is that I have a small C++ stand-alone > application that I want to use to test new boards. > I would like to build the application using the > arm-linux development environment. I do not want to > spawn a second set of libraries compatible with the > -msoft-float flag. > > Can the Floating Point Emulation used by the Linux > Kernel be used without the kernel ? > > The code at the undefined instruction vector passes > a data area from the current task into the FPEmulator. > > Is this just a scratch area or is it used to pass > data to the Emulator ? > > Any advice would be appreciated. > > -------------------------------- > Bye! > Dan McDonald > The Late Night Software Shop > Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 17:09:30 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12ziOH-0004EZ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 17:09:13 +0100 Received: from [216.163.180.10] (helo=c0mailgw08.prontomail.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12ziOG-0004EE-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 17:09:13 +0100 Received: from web104 (216.163.176.204) by c0mailgw08.prontomail.com (NPlex 4.5.049) id 3925E52E000C1960; Wed, 7 Jun 2000 10:06:11 -0700 X-Version: canoe 6.2.2329.0 From: "Dan" Message-Id: <47E1EC0BC8C34D115A040005B8275D56@dmcdon.canoemail.com> Date: Wed, 7 Jun 2000 11:05:37 -0600 X-Priority: Normal Content-Type: text/plain; charset=iso-8859-1 To: linux-arm-kernel@lists.arm.linux.org.uk Reply-To: dmcdon@canoemail.com Subject: Re: Floating Point Emulation CC: nico@cam.org X-Mailer: Web Based Pronto Mime-Version: 1.0 Content-Transfer-Encoding: quoted-printable Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, 7 Jun 2000, Nicolas wrote: > If you can arrange for not using floating point operations in your code > i.e. using fixed point math techniques instead, then you may just ignore > the whole issue. =A0Do you really need floating point for your testing app > anyway? Hi Nicolas: You are probably right. It is probably easier to remove the floating point calculations then it is to get the FPEmulation working. It should also make the application smaller. Perhaps this particular issue is best left for another day when floating point cannot be readily bypassed. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 17:36:18 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zioK-0004dx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 17:36:08 +0100 Received: from [216.163.180.10] (helo=c0mailgw10.prontomail.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zioJ-0004dN-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 17:36:07 +0100 Received: from web104 (216.163.176.204) by c0mailgw10.prontomail.com (NPlex 4.5.049) id 392EC8FB0008BC1F; Wed, 7 Jun 2000 10:32:45 -0700 X-Version: canoe 6.2.2329.0 From: "Dan" Message-Id: <5252EC0BC8C34D115A040005B8275D56@dmcdon.canoemail.com> Date: Wed, 7 Jun 2000 11:32:13 -0600 X-Priority: Normal Content-Type: text/plain; charset=iso-8859-1 To: fray@mvista.com Reply-To: dmcdon@canoemail.com Subject: Re: Floating Point Emulation CC: linux-arm-kernel@lists.arm.linux.org.uk X-Mailer: Web Based Pronto Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Mark Hatle wrote: > > What timing.. :) > [snip] > > I am currently working on patching gcc (2.95.2) to do soft-float the way > other platforms do it. (If you look in the gcc source in gcc/config/arm > you'll see a but of "t-*" files. all of the files EXCEPT the linux one > use the built-in soft-floating point. So it should be possible.) FYI, > I have yet to run the tests on the floating point, so what I have now > may not work the way I think it does! > > If you would like a copy of my patches email me off the list and I'll go > into exactly what I'm doing. > [snip] Hi Mark: Thanks for your response. I don't know if I am brave enough to do battle with gcc and the Linux libraries. I am a wimp in such matters - I tend not to fight the native development environments unless absolutely necessary. I find if I manipulate the environments too much it turns into a support headache making it difficult to pick up new releases, etc. It's the "when in Rome, do as the Romans do" philosophy. :-) I always wondered where those darn flags were. Thanks. Good luck. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 18:57:12 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zk4X-0005wB-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 18:56:57 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zk4V-0005vz-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 18:56:56 +0100 From: Message-Id: To: sa1100-linux@pa.dec.com Subject: Problems writing back dirty data from the SA-110 data cache Date: Wed, 7 Jun 2000 11:03:20 -0700 Mime-Version: 1.0 Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk We are trying to port linux/drivers/scsi/advansys.c to a StrongArm platform running Linux 2.2.13. We are experiencing cache coherency problems between the SA-110 and a PCI master (AdvanSys SCSI chip). We have attempted to implement the cache write back and invalidate functions in linux/arch/arm/mm/proc-sa110.s to achieve cache coherency, but every attempt has resulted in memory corruption and memory violations. To help diagnose the problem, we simplified our strategy to rely primarily on the "_sa110_flush_cache_all" function which does a 16K read from the Footbridge SA-110 Cache Flush area followed by an SA-110 Coprocessor "Drain Write Buffer" instruction. Attaching an analyzer to the processor bus and the PCI bus we can see the following sequence for a particular command the driver queues to the SCSI chip. (Note: the command structures are cache line aligned and are exactly three cache lines in length.) 1. The driver calls "_sa110_flush_cache_all" to write back and invalidate the command structure from the cache. During the 16K read loop, two of the three cache lines belonging to the command structure are successfully written to the processor bus. The third cache line is not written even though it was modified by the processor. 2. The driver updates command queue linkage and calls "_sa110_cache_wback_area" to write back the modified linkage. This function also includes an SA-110 Coprocessor "Drain Write Buffer" instruction. 3. The processor performs an unbuffered write to the SCSI chip to inform the chip of the new command. 4. The SCSI chip reads the command structure from processor memory. 5. The third cache line is finally written to the processor bus (75 us after step 3, 14 us after step 4) the next time "_sa110_flush_cache_all" is called (apparently by the kernel). 6. The SCSI chip returns the corrupted command structure to the driver. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 22:17:00 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12znBj-00010M-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 22:16:35 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96162) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12znBg-0000zm-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 22:16:33 +0100 Received: from localhost (ee96162@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id DAA02177; Thu, 8 Jun 2000 03:41:55 +0530 Date: Thu, 8 Jun 2000 03:41:55 +0530 (IST) From: Lavu Sridhar Reply-To: ee96162@ee.iitm.ernet.in To: Jason Gunthorpe cc: Vikas Datta , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk The checksum seems fine. I have also checked the hardware side, it seems fine. Tried with lower baud rates also. The error message of comes from the function DEBG("huft3 ") in lib/inflate.c which has the following comment line: "bad input: more codes than bits" I want to know if an uncompressed linux image using the '1.5MB vmlinux' can be made and downloaded into the DRAM. Regards, Lavu Sridhar --------------------------------------------------------------------- On Wed, 7 Jun 2000, Jason Gunthorpe wrote: >Can you try running checksum multiple times? See if it stays the same over >each run. AFAIK the angel protocol has per block checksums so data should >not be getting corrupted while it is in transit. However.. it may be that >the angelboot program does not understand a CRC failure reply, try a lower >baud rate maybe? > >Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 07 23:58:19 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zolx-0002XK-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 07 Jun 2000 23:58:05 +0100 Received: from [132.233.247.11] (helo=thalia.fm.intel.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zolv-0002XD-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 07 Jun 2000 23:58:04 +0100 Received: from SMTP (fmsmsxvs01-1.fm.intel.com [132.233.42.201]) by thalia.fm.intel.com (8.9.1a+p1/8.9.1/d: relay.m4,v 1.29 2000/05/27 05:56:37 nzand Exp $) with SMTP id XAA21884 for ; Wed, 7 Jun 2000 23:56:37 GMT Received: from fmsmsx27.FM.INTEL.COM ([132.233.48.27]) by 132.233.48.201 (Norton AntiVirus for Internet Email Gateways 1.0) ; Wed, 07 Jun 2000 23:55:46 0000 (GMT) Received: by fmsmsx27.fm.intel.com with Internet Mail Service (5.5.2448.0) id ; Wed, 7 Jun 2000 16:55:45 -0700 Message-ID: <3D33CF40366DD111AC4100A0C96B22AC0870B498@fmsmsx34.fm.intel.com> From: "Naik, Uday" To: linux-arm-kernel@lists.arm.linux.org.uk Subject: ramdisk file system - memory leak ?? Date: Wed, 7 Jun 2000 16:55:41 -0700 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi, On my IXP1200 system (little endian), I am using a ram disk that I got from netwinder.org. When the system starts up it has approx 33% disk space. Every time I run a utility, some ram disk space is gone and is never freed. I am doing ftps to this disk and everytime I put a file in, space is used up. But deleting the file does not free up space. Eventually I get a disk full error. Before I dig into this problem, anybody has any clues for me. regards Uday unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 05:12:06 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12ztfT-0006SK-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 05:11:43 +0100 Received: from [199.185.220.112] (helo=priv-edtnes12-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12ztfS-0006SE-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 05:11:42 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes12-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000608050924.NADG24907.priv-edtnes12-hme0.telusplanet.net@wakko.deltatee.com>; Wed, 7 Jun 2000 23:09:24 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 12zuZg-0006vi-00 (Debian); Wed, 7 Jun 2000 23:09:48 -0600 Date: Wed, 7 Jun 2000 23:09:48 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Lavu Sridhar cc: Vikas Datta , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, Lavu Sridhar wrote: > The error message of comes from the function DEBG("huft3 ") in > lib/inflate.c which has the following comment line: > "bad input: more codes than bits" I think you need to determine if the data is corrupted in memory before compression or during decompression. > I want to know if an uncompressed linux image using the '1.5MB vmlinux' > can be made and downloaded into the DRAM. I don't think that is wise.. if you cannot even get a compressed image into ram OK then you will probably have problems with uncompressed images as well. Likely random kernel oops :< I've been using this board and another using angel and the zImages for two weeks now, never once has anything like this happened.. I just had another thought.. What if there is a bug in your compiler that is producing funny code? I have placed the image I am using up at http://amber.deltatee.com/~jgg/cl-ps7500fe/zImage I used these options to download it: base 0x208000 entry 0x208000 device /dev/ttyS1 options "9600 8N1" baud 115200 Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 08:28:16 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zwjN-0000we-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 08:27:57 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96209) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zwjD-0000wI-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 08:27:48 +0100 Received: from localhost (ee96209@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id NAA22968; Thu, 8 Jun 2000 13:53:06 +0530 Date: Thu, 8 Jun 2000 13:53:06 +0530 (IST) From: Vikas Datta To: Jason Gunthorpe cc: Lavu Sridhar , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, Your idea worked.......the compiler was faulty and was indeed producing funny code. With the zImage on your site, my board now boots upto the kernel panic...........it looks really great !! We were using the arm-linux cross Tool Chain. The Tool Chain is made up of: binutils-2.9.5.0.22, gcc-2.95.2 - C, C++, f77, CHILL, java and objc. glibc-2.1.2. with the International crypt library. Which one are you using? Thanks, Regards, Vikas On Wed, 7 Jun 2000, Jason Gunthorpe wrote: > I just had another thought.. What if there is a bug in your compiler that > is producing funny code? unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 09:49:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zy0B-00020N-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 09:49:23 +0100 Received: from [130.161.40.91] (helo=thorgal.et.tudelft.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zy0B-00020H-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 09:49:23 +0100 Received: from [130.161.40.82] (tvs082.et.tudelft.nl [130.161.40.82]) by thorgal.et.tudelft.nl (8.8.7/8.8.7) with ESMTP id LAA06486; Thu, 8 Jun 2000 11:47:06 +0200 Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 8 Jun 2000 11:46:57 +0200 To: linux-arm-kernel@lists.arm.linux.org.uk From: "J.D. Bakker" Subject: Flushing I/D caches from userspace Cc: lart@lart.tudelft.nl Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi all, Is there any way to do cache flushes from userspace under ARMLinux ? I'm developing a fixed point FFT library, and I want to do timings with both hot and cold cache. The only way I can think of is to walk an array the size of the D cache, and write a dummy function that fills the I cache, but that seems a tad hackish. Any thoughts ? JDB. [and yes, the library will be [L]GPL'd once it works properly] -- LART. 250 MIPS under one Watt. Free hardware design files. http://www.lart.tudelft.nl/ unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 10:38:55 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zylt-0002e8-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 10:38:42 +0100 Received: from [62.6.164.226] (helo=mail.linuxgrrls.org ident=postfix) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zylt-0002e1-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 10:38:41 +0100 Received: from localhost (localhost [127.0.0.1]) by mail.linuxgrrls.org (Postfix) with ESMTP id 388847480D; Thu, 8 Jun 2000 11:34:35 +0100 (BST) Date: Thu, 8 Jun 2000 11:34:35 +0100 (BST) From: Kira Brown To: "J.D. Bakker" Cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Subject: Re: Flushing I/D caches from userspace In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, J.D. Bakker wrote: > Hi all, > > Is there any way to do cache flushes from userspace under ARMLinux ? I'm The way we used to do it under RISC OS was to write out enough data to disparate pages that we could be sure that the cache had been flushed. It's not a very smart way of doing things but hey... kira. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 10:50:43 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zyxN-0002oX-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 10:50:33 +0100 Received: from [130.161.40.91] (helo=thorgal.et.tudelft.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zyxM-0002oL-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 10:50:32 +0100 Received: from [130.161.40.82] (tvs082.et.tudelft.nl [130.161.40.82]) by thorgal.et.tudelft.nl (8.8.7/8.8.7) with ESMTP id MAA06715; Thu, 8 Jun 2000 12:48:02 +0200 Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 8 Jun 2000 12:47:42 +0200 To: Kira Brown From: "J.D. Bakker" Subject: Re: Flushing I/D caches from userspace Cc: "J.D. Bakker" , linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk At 12:34 +0200 08-06-2000, Kira Brown wrote: >On Thu, 8 Jun 2000, J.D. Bakker wrote: >> Is there any way to do cache flushes from userspace under ARMLinux ? I'm > >The way we used to do it under RISC OS was to write out enough data to >disparate pages that we could be sure that the cache had been flushed. > >It's not a very smart way of doing things but hey... On a processor with an SA-1 core this won't work as the caches are allocate-on-read. I thought this was the norm for ARM ? JDB. -- "There is a style of design I call "wishful thinking engineering." It starts with something like "pigs can fly if you feed them enough beans" and develops utopian plans such as like having everyone commute to work riding on personal pigs, and along the way ignores minor details such as the consequent rain of the non-gaseous byproducts." (Vernon Schryver in n.a.n-a.e) unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 10:53:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 12zz0L-0002rQ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 10:53:37 +0100 Received: from [62.6.164.226] (helo=mail.linuxgrrls.org ident=postfix) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 12zz0K-0002rJ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 10:53:36 +0100 Received: from localhost (localhost [127.0.0.1]) by mail.linuxgrrls.org (Postfix) with ESMTP id 6DF957480D; Thu, 8 Jun 2000 11:49:30 +0100 (BST) Date: Thu, 8 Jun 2000 11:49:30 +0100 (BST) From: Kira Brown To: "J.D. Bakker" Cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Subject: Re: Flushing I/D caches from userspace In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, J.D. Bakker wrote: > At 12:34 +0200 08-06-2000, Kira Brown wrote: > >On Thu, 8 Jun 2000, J.D. Bakker wrote: > >> Is there any way to do cache flushes from userspace under ARMLinux ? I'm > > > >The way we used to do it under RISC OS was to write out enough data to > >disparate pages that we could be sure that the cache had been flushed. > > > >It's not a very smart way of doing things but hey... > > On a processor with an SA-1 core this won't work as the caches are > allocate-on-read. I thought this was the norm for ARM ? That was what we did to flush the write buffer- if we didn't do that regularly, we couldn't cache the video RAM without artifacts. kira. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 15:33:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1303MW-0006iB-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 15:32:48 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1303MV-0006i4-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 15:32:47 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000608153029.UADQ625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com>; Thu, 8 Jun 2000 09:30:29 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 1304Gi-0007io-00 (Debian); Thu, 8 Jun 2000 09:30:53 -0600 Date: Thu, 8 Jun 2000 09:30:52 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Vikas Datta cc: Lavu Sridhar , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: CL7500FE Kernel Image In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, Vikas Datta wrote: > Your idea worked.......the compiler was faulty and was indeed producing > funny code. With the zImage on your site, my board now boots upto the > kernel panic...........it looks really great !! Hm.. That doesn't mean it is a compiler problem, but it sure is something to do with how your image is being built.. > We were using the arm-linux cross Tool Chain. The Tool Chain is made up of: > > binutils-2.9.5.0.22, > gcc-2.95.2 - C, C++, f77, CHILL, java and objc. > glibc-2.1.2. with the International crypt library. > > Which one are you using? I hand compiled my own, binutils 2.9.5.0.41, gcc 2.95.2+arm-patch+debian-patch, glibc-2.1.3 You can grab it and try my build if you want, same place as the zImage, unpack the tarball into /opt (so that /opt/arm-linux/bin is created) Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 08 23:02:27 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130AM7-0004Mu-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 08 Jun 2000 23:00:51 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130AM6-0004Mm-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 08 Jun 2000 23:00:50 +0100 Received: from ims.crl.dec.com (ims.crl.dec.com [16.11.0.11]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id SAA07431; Thu, 8 Jun 2000 18:58:31 -0400 (EDT) Received: by ims.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Thu, 8 Jun 2000 19:03:17 -0400 Message-ID: From: George France To: "'dmcdon@canoemail.com'" , fray@mvista.com Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Floating Point Emulation Date: Thu, 8 Jun 2000 18:59:17 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Why are you not using -msoft-float? --George -----Original Message----- From: Dan [mailto:dmcdon@canoemail.com] Sent: Wednesday, June 07, 2000 1:32 PM To: fray@mvista.com Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation Mark Hatle wrote: > > What timing.. :) > [snip] > > I am currently working on patching gcc (2.95.2) to do soft-float the way > other platforms do it. (If you look in the gcc source in gcc/config/arm > you'll see a but of "t-*" files. all of the files EXCEPT the linux one > use the built-in soft-floating point. So it should be possible.) FYI, > I have yet to run the tests on the floating point, so what I have now > may not work the way I think it does! > > If you would like a copy of my patches email me off the list and I'll go > into exactly what I'm doing. > [snip] Hi Mark: Thanks for your response. I don't know if I am brave enough to do battle with gcc and the Linux libraries. I am a wimp in such matters - I tend not to fight the native development environments unless absolutely necessary. I find if I manipulate the environments too much it turns into a support headache making it difficult to pick up new releases, etc. It's the "when in Rome, do as the Romans do" philosophy. :-) I always wondered where those darn flags were. Thanks. Good luck. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 00:25:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130BeZ-0005Mo-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 00:23:59 +0100 Received: from [63.192.220.206] (helo=hermes.mvista.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130BeY-0005MT-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 00:23:58 +0100 Received: from mvista.com (IDENT:mhatle@gemini.mvista.com [10.0.0.101]) by hermes.mvista.com (8.9.3/8.9.3) with ESMTP id RAA13923; Thu, 8 Jun 2000 17:21:10 -0700 Message-ID: <3940382B.32E9B3A9@mvista.com> Date: Thu, 08 Jun 2000 17:19:55 -0700 From: Mark Hatle Organization: MontaVista Software, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: George France CC: "'dmcdon@canoemail.com'" , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I can answer for myself at least. The way -msoft-float is setup for the netwinders isn't the way I want to do things. The regular -msoft-float uses an external libfloat library, having to use an external library for floating point is not the correct solution IMHO, especially when gcc provides the functionality internally. I believe Dan didn't want to deal with external libraries. FYI, all of the other arm platforms use gcc's built in floating point emulation except for the linux target. I have a patch that does the same for linux. --Mark George France wrote: > > Why are you not using -msoft-float? > > --George > > -----Original Message----- > From: Dan [mailto:dmcdon@canoemail.com] > Sent: Wednesday, June 07, 2000 1:32 PM > To: fray@mvista.com > Cc: linux-arm-kernel@lists.arm.linux.org.uk > Subject: Re: Floating Point Emulation > > Mark Hatle wrote: > > > > What timing.. :) > > > [snip] > > > > I am currently working on patching gcc (2.95.2) to do soft-float > the way > > other platforms do it. (If you look in the gcc source in > gcc/config/arm > > you'll see a but of "t-*" files. all of the files EXCEPT the linux > one > > use the built-in soft-floating point. So it should be possible.) > FYI, > > I have yet to run the tests on the floating point, so what I have > now > > may not work the way I think it does! > > > > If you would like a copy of my patches email me off the list and > I'll go > > into exactly what I'm doing. > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 00:48:00 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130C0D-0005cm-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 00:46:21 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130C0C-0005cZ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 00:46:20 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id UAA11584; Thu, 8 Jun 2000 20:42:09 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Thu, 8 Jun 2000 20:42:09 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: "J.D. Bakker" cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Subject: Re: Flushing I/D caches from userspace In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, J.D. Bakker wrote: > Hi all, > > Is there any way to do cache flushes from userspace under ARMLinux ? Isn't a syscall for that? Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 05:57:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130Gqj-0000jm-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 05:56:53 +0100 Received: from [129.94.242.29] (helo=note.orchestra.cse.unsw.EDU.AU) by www.linux.org.uk with smtp (Exim 3.13 #1) id 130Gqh-0000jd-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 05:56:52 +0100 Received: From paulaner With LocalMail ; Fri, 9 Jun 2000 15:53:02 +1000 From: Adam 'WeirdArms' Wiggins To: "J.D. Bakker" Date: Fri, 9 Jun 2000 15:53:02 +1000 (EST) X-Sender: awiggins@paulaner.disy.cse.unsw.EDU.AU cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Subject: Re: Flushing I/D caches from userspace In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 8 Jun 2000, J.D. Bakker wrote: > Hi all, > > Is there any way to do cache flushes from userspace under ARMLinux ? I'm > developing a fixed point FFT library, and I want to do timings with both > hot and cold cache. The only way I can think of is to walk an array the > size of the D cache, and write a dummy function that fills the I cache, but > that seems a tad hackish. Allowing user space to flush the caches is can lead to real problems. Maybe with a /dev/ to allow priviledged users but in general you don't want any old user to do it. > > Any thoughts ? > > JDB. > [and yes, the library will be [L]GPL'd once it works properly] > > -- > LART. 250 MIPS under one Watt. Free hardware design files. > http://www.lart.tudelft.nl/ > > > -- > To unsubscribe from this list: send the line "unsubscribe lart" in > the body of a message to majordomo@lart.tudelft.nl > Please read the LART FAQ at http://www.lart.tudelft.nl/faq.php3 > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 07:15:33 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130I4I-0001p0-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 07:14:58 +0100 Received: from [194.64.24.3] (helo=pink.ma.choin.net ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130I4G-0001ot-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 07:14:56 +0100 Received: from RR ([195.180.132.160]) by pink.ma.choin.net (8.9.3/8.9.3) with SMTP id JAA58691 for ; Fri, 9 Jun 2000 09:12:04 +0200 (CEST) (envelope-from rrascher@ids-imaging.de) Received: by RR with Microsoft Mail id <01BFD1F2.83A03A30@RR>; Fri, 9 Jun 2000 09:10:11 +0200 Message-ID: <01BFD1F2.83A03A30@RR> From: Ralf Rascher To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Invalid Assembler Instruction Date: Fri, 9 Jun 2000 09:09:59 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hallo, I try to compile a Linux Kernel 2_3_99-pre8 with patch-2_3_99-prek8-rmk1 and diff-2_3_99-prek8-rmk1-np5 for an Intel Assabet Board. I use binutils-2_10-1rc2 and gcc-2_95_2-1nw6 on an i686 Debian Linux PC. But I get an error when compiling the kernel. # entry-common.S: Assembler message # entry-common.S:38: Error: can only set flag field with immediate value I found out that this is the command: msr cpsr_c, #I_BIT | MODE_SVC which is several time in that file. Is that command right, or how can I change the command, or the assembler (if the command is right) to work? Thanks, Ralf unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 09:41:26 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130KLX-0003l5-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 09:40:55 +0100 Received: from [130.161.40.91] (helo=thorgal.et.tudelft.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130KLW-0003kw-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 09:40:54 +0100 Received: from [130.161.40.82] (tvs082.et.tudelft.nl [130.161.40.82]) by thorgal.et.tudelft.nl (8.8.7/8.8.7) with ESMTP id LAA10761; Fri, 9 Jun 2000 11:38:33 +0200 Message-Id: In-Reply-To: References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Fri, 9 Jun 2000 11:38:15 +0200 To: "Adam 'WeirdArms' Wiggins" From: "J.D. Bakker" Subject: Re: Flushing I/D caches from userspace Cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk At 07:53 +0200 09-06-2000, Adam 'WeirdArms' Wiggins wrote: > Allowing user space to flush the caches is can lead to real >problems. Such as ? Considering that a context switch to a process with a different VM context causes a cache flush anyway, I can't imagine what harm it would do. JDB. -- Jan-Derk Bakker, bakker@mmc.et.tudelft.nl The lazy man's proverb: 'There's no business like slow business !' unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 13:58:38 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130OLX-0007AJ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 13:57:11 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96209) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130OLK-00079S-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 13:57:07 +0100 Received: from localhost (ee96209@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id TAA06168 for ; Fri, 9 Jun 2000 19:17:40 +0530 Date: Fri, 9 Jun 2000 19:17:40 +0530 (IST) From: Vikas Datta To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Serial options for a console Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, How do i give the serial options (for a serial console) in the kernel so that the default serial port is COMM2 (/dev/ttyS0) and not COMM1. Which file has to be modified for this. I am making the kernel for a cl-ps7500fe target. Thanks in advance, Vikas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 15:31:12 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130Pnd-0008Tj-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 15:30:17 +0100 Received: from [207.245.35.202] (helo=kei.netwinder.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130Pnc-0008TZ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 15:30:16 +0100 Received: from netwinder.org ([10.1.19.121]) by kei.netwinder.org (8.9.3/8.9.3) with ESMTP id LAA03056; Fri, 9 Jun 2000 11:27:54 -0400 Message-ID: <39410BDD.C09A0232@netwinder.org> Date: Fri, 09 Jun 2000 11:23:09 -0400 From: Scott Bambrough Organization: Rebel.com X-Mailer: Mozilla 4.7 [en] (Win95; U) X-Accept-Language: en MIME-Version: 1.0 To: Ralf Rascher CC: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Re: Invalid Assembler Instruction References: <01BFD1F2.83A03A30@RR> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Ralf Rascher wrote: > > # entry-common.S: Assembler message > # entry-common.S:38: Error: can only set flag field with immediate value > > I found out that this is the command: > > msr cpsr_c, #I_BIT | MODE_SVC > The assembler is correct. You can only use the immediate form with cpsr_f. This is a change I recently put in the 2.10 branch of gas, to make the mainline and the 2.10 assembler behave in a similar manner. While this form seems to work, according to the ARM architechure manual the results are unpredictable. You have to change all the instructions of this type in the kernel code, which I'm told is quite a chore. Scott -- Scott Bambrough - Software Engineer REBEL.COM http://www.rebel.com NetWinder http://www.netwinder.org unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 17:42:25 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130Ror-00020c-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 17:39:41 +0100 Received: from [195.44.63.19] (helo=wpk-smtp-relay.cwci.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130Roq-000204-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 17:39:40 +0100 Received: from moo ([195.44.217.189]) by wpk-smtp-relay.cwci.net with Microsoft SMTPSVC(5.5.1877.387.38); Fri, 9 Jun 2000 18:36:42 +0100 Date: Fri, 09 Jun 2000 18:35:37 +0100 From: Peter Naulls To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Serial options for a console Message-ID: <5c2decc49%peter@erble.freeserve.co.uk> In-Reply-To: X-Mailer: Messenger v1.40f for RISC OS MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Posting-Agent: RISC OS Newsbase 0.60m User-Agent: POPstar/2.02 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk In message Vikas Datta wrote: > How do i give the serial options (for a serial console) in the kernel so > that the default serial port is COMM2 (/dev/ttyS0) and not COMM1. Which > file has to be modified for this. I am making the kernel for a cl-ps7500fe > target. Why don't you swap the device numbers on /dev/ttyS0 and /dev/ttyS1? -- ------------------------------------------------------------------------ Peter Naulls - peter@erble.freeserve.co.uk Homepage - http://free.prohosting.com/~chocky/ Java for RISC OS and ARM - http://free.prohosting.com/~chocky/java/ Debian Linux on RiscPCs - http://www.erble.freeserve.co.uk/ ------------------------------------------------------------------------ unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 09 19:15:48 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130TJA-0003RD-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 09 Jun 2000 19:15:04 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130TJ7-0003R3-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 09 Jun 2000 19:15:02 +0100 Received: from server01.asi (akron-80-176.ar51.net [208.225.80.176]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id PAA31697; Fri, 9 Jun 2000 15:12:28 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Fri, 9 Jun 2000 15:13:06 -0400 Message-ID: From: Jeff Sutherland To: "Linux ARM kernel mailing list (E-mail)" Cc: "John G. Dorsey (E-mail)" , "Nicolas Pitre (E-mail)" Subject: A word of warning... Date: Fri, 9 Jun 2000 15:13:04 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk A word of warning to those attempting to reduce the size of their occupied disk space: I've just spent 3 days chasing down a problem with PCMCIA network drivers on SA1110, finally figured it out. How many of you are using 'arm-linux-strip' to reduce the size of your object modules? Well, I got burned using '--strip-unneeded' on a module. Funny, the stripped module would run on a 2.3.99pre* kernel, but building the exact same source code and using the same strip parameters caused it to not work in any of the 2.4.0-test1-X-X kernels. ???? The 'busybox' version of insmod would load it fine. lsmod would show it loaded. Its symbols showed up in /proc/ksyms. BUT, the initialization code at the end of pcnet_cs WOULD NOT BE EXECUTED. Obviously --strip-unneeded took out something that WAS needed. Other stripped modules (8390.o) seem to load and work fine. I'll have to experiment with some of the other 'strip' options to see which ones work, if any. I wonder what changed in the kernel to cause this behaviour? BTW it's probably something in pcnet_cs.c anyway, not the kernel- you know, one of those latent bugs that finally got exposed. At least I'll be able to sleep tonight... Regards, //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 10 01:35:57 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130ZEm-000838-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 10 Jun 2000 01:34:56 +0100 Received: from [129.94.242.29] (helo=note.orchestra.cse.unsw.EDU.AU) by www.linux.org.uk with smtp (Exim 3.13 #1) id 130ZEk-00082y-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 10 Jun 2000 01:34:55 +0100 Received: From paulaner With LocalMail ; Sat, 10 Jun 2000 11:31:12 +1000 From: Adam 'WeirdArms' Wiggins To: "J.D. Bakker" Date: Sat, 10 Jun 2000 11:31:12 +1000 (EST) X-Sender: awiggins@paulaner.disy.cse.unsw.EDU.AU cc: linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl Subject: Re: Flushing I/D caches from userspace In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Fri, 9 Jun 2000, J.D. Bakker wrote: > At 07:53 +0200 09-06-2000, Adam 'WeirdArms' Wiggins wrote: > > Allowing user space to flush the caches is can lead to real > >problems. > > Such as ? Considering that a context switch to a process with a different > VM context causes a cache flush anyway, I can't imagine what harm it would > do. Good point. I was going off the idea that the caches were not flushed during a context switch... Simplest thing would be a syscall then. Cheers Adam unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 10 13:23:01 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 130kHP-00075v-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 10 Jun 2000 13:22:23 +0100 Received: from [195.180.235.152] (helo=storm.local ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 130kHN-00075o-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 10 Jun 2000 13:22:22 +0100 Received: by munich.netsurf.de via sendmail from stdin id (Debian Smail3.2.0.102) for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 10 Jun 2000 00:41:44 +0200 (CEST) Date: Sat, 10 Jun 2000 00:41:44 +0200 From: Andreas Bombe To: =?iso-8859-1?Q?David_M=FCller?= Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Console on serial port on EBSA285 Message-ID: <20000610004144.B4306@storm.local> References: <393E0372.639ADD6E@elsoft.ch> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit User-Agent: Mutt/1.1.12i In-Reply-To: <393E0372.639ADD6E@elsoft.ch>; from d.mueller@elsoft.ch on Wed, Jun 07, 2000 at 10:10:26AM +0200 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, Jun 07, 2000 at 10:10:26AM +0200, David Müller wrote: > Hello > > I would like to have the console I/O redirected to > the serial port of my "VGA less" EBSA285. > > I have the kernel (2.3.39 + patch-2.3.39-rmk1) configured > for "Console on serial port" and have tried the "console=" > kernel parameter, but without success. Seems like you are trying to use the standard serial driver. During configure, say 'Y' to "Non-standard serial port support" and then enable the 21825 serial port support and the accompanied console on 21825 port option. -- Andreas E. Bombe DSA key 0x04880A44 http://home.pages.de/~andreas.bombe/ http://linux1394.sourceforge.net/ unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 07:42:56 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311RF-0001mj-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 07:41:41 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311RD-0001mc-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 07:41:40 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id IAA15478; Sun, 11 Jun 2000 08:09:01 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id IAA01578; Sun, 11 Jun 2000 08:05:04 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110705.IAA01578@flint.arm.linux.org.uk> Subject: Re: Console on serial port on EBSA285 To: d.mueller@elsoft.ch (David M|ller (ELSOFT AG)) Date: Sun, 11 Jun 2000 08:05:04 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <393E0372.639ADD6E@elsoft.ch> from "David M|ller (ELSOFT AG)" at Jun 07, 2000 10:10:26 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk David M|ller (ELSOFT AG) writes: > I have the kernel (2.3.39 + patch-2.3.39-rmk1) configured > for "Console on serial port" and have tried the "console=" > kernel parameter, but without success. Note that the standard serial port is a 16550-type device, not the DC21285 on-board serial port. To use the on-board port, you need to select the CONFIG_SERIAL_NONSTANDARD, CONFIG_SERIAL_21285 and CONFIG_SERIAL_21285_CONSOLE options. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 07:42:56 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311RT-0001mx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 07:41:55 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311RR-0001mc-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 07:41:54 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id IAA15508; Sun, 11 Jun 2000 08:18:31 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id IAA01624; Sun, 11 Jun 2000 08:14:35 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110714.IAA01624@flint.arm.linux.org.uk> Subject: Re: Invalid Assembler Instruction To: rrascher@ids-imaging.de (Ralf Rascher) Date: Sun, 11 Jun 2000 08:14:35 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk ('linux-arm-kernel@lists.arm.linux.org.uk') In-Reply-To: <01BFD1F2.83A03A30@RR> from "Ralf Rascher" at Jun 09, 2000 09:09:59 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Ralf Rascher writes: > msr cpsr_c, #I_BIT | MODE_SVC > > which is several time in that file. Is that command right, or how can I change > the command, or the assembler (if the command is right) to work? It works correctly on all known (by Linux at least) ARM chips. However, it is marked as "undefined" in the ARM architecture reference manual, and recently binutils has started to implement this as an illegal combination of operands. I believe that the last kernel I released last Saturday should have these all fixed. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 07:47:17 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311Wb-0001qn-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 07:47:13 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311WZ-0001qf-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 07:47:11 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id IAA15638; Sun, 11 Jun 2000 08:44:26 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id IAA01687; Sun, 11 Jun 2000 08:40:29 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110740.IAA01687@flint.arm.linux.org.uk> Subject: Re: Floating Point Emulation To: fray@mvista.com (Mark Hatle) Date: Sun, 11 Jun 2000 08:40:29 +0100 (BST) Cc: france@crl.dec.com (George France), dmcdon@canoemail.com ('dmcdon@canoemail.com'), linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <3940382B.32E9B3A9@mvista.com> from "Mark Hatle" at Jun 08, 2000 05:19:55 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Mark Hatle writes: > The regular -msoft-float uses an external libfloat library, having to > use an external library for floating point is not the correct solution > IMHO, especially when gcc provides the functionality internally. Please explain what you mean more clearly. AFAICR, GCC does not provide any floating point emulation. Without -msoft-float, GCC will emit the hardware instructions for floating point operations, and therefore you will need some floating point emulation code. There are two possibilities: 1. Build with -msoft-float and use libfloat 2. Build without -msoft-float and use either the Acorn FPE or NWFPE. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 08:04:27 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311nE-00024a-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 08:04:24 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311nC-00024R-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 08:04:23 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id JAA15682; Sun, 11 Jun 2000 09:01:58 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id IAA01743; Sun, 11 Jun 2000 08:58:02 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110758.IAA01743@flint.arm.linux.org.uk> Subject: Re: Flushing I/D caches from userspace To: nico@cam.org (Nicolas Pitre) Date: Sun, 11 Jun 2000 08:58:02 +0100 (BST) Cc: bakker@thorgal.et.tudelft.nl (J.D. Bakker), linux-arm-kernel@lists.arm.linux.org.uk, lart@lart.tudelft.nl In-Reply-To: from "Nicolas Pitre" at Jun 08, 2000 08:42:09 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Nicolas Pitre writes: > Isn't a syscall for that? Yes there is. Number 0x9f0002: r0 = start r1 = length r2 = flags bit 0 - flush I cache as well. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 08:09:25 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311s2-00029K-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 08:09:22 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311s0-00029D-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 08:09:21 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id JAA15721; Sun, 11 Jun 2000 09:06:28 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id JAA01779; Sun, 11 Jun 2000 09:02:31 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110802.JAA01779@flint.arm.linux.org.uk> Subject: Re: Serial options for a console To: ee96209@ee.iitm.ernet.in (Vikas Datta) Date: Sun, 11 Jun 2000 09:02:31 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Vikas Datta" at Jun 09, 2000 07:17:40 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Vikas Datta writes: > How do i give the serial options (for a serial console) in the kernel so > that the default serial port is COMM2 (/dev/ttyS0) and not COMM1. Which > file has to be modified for this. I am making the kernel for a cl-ps7500fe > target. Sounds like you're going about this the wrong way. You can use: console=ttyS1,9600n8 to use the second serial port. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 08:16:52 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1311zE-0002F0-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 08:16:48 +0100 Received: from [212.250.233.52] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1311zC-0002Eu-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 08:16:47 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id JAA15754; Sun, 11 Jun 2000 09:14:32 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id JAA01819; Sun, 11 Jun 2000 09:10:35 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006110810.JAA01819@flint.arm.linux.org.uk> Subject: Re: ramdisk file system - memory leak ?? To: uday.naik@intel.com (Naik, Uday) Date: Sun, 11 Jun 2000 09:10:35 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <3D33CF40366DD111AC4100A0C96B22AC0870B498@fmsmsx34.fm.intel.com> from "Naik, Uday" at Jun 07, 2000 04:55:41 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Naik, Uday writes: > On my IXP1200 system (little endian), I am using > a ram disk that I got from netwinder.org. When the > system starts up it has approx 33% disk space. Every > time I run a utility, some ram disk space is gone and is > never freed. I am doing ftps to this disk and everytime > I put a file in, space is used up. But deleting the file > does not free up space. Eventually I get a disk full error. Some questions: 1. Are you quitting the ftp client between the download and the df? 2. Is something else using the file? 3. What kernel version are you using? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 21:26:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131EII-0001qN-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 21:25:18 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131EIH-0001qD-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 21:25:17 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000611212256.GKHA625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com> for ; Sun, 11 Jun 2000 15:22:56 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 131FCV-00085I-00 (Debian); Sun, 11 Jun 2000 15:23:23 -0600 Date: Sun, 11 Jun 2000 15:23:23 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Serial options for a console In-Reply-To: <200006110802.JAA01779@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Sun, 11 Jun 2000, Russell King - ARM Linux Admin wrote: > Sounds like you're going about this the wrong way. > > You can use: > > console=ttyS1,9600n8 > > to use the second serial port. What would be nice is if that kernel command line CONFIG option were moved into the main configuration so it could always be used (I think it is hidden under a netwinder option?). Right now I haven't spent any time working on finding/writing a more fancy boot loader that could do command line options. Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 21:30:36 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131ENF-0001uY-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 21:30:25 +0100 Received: from [212.250.233.6] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131END-0001uO-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 21:30:24 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id WAA18642; Sun, 11 Jun 2000 22:28:06 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id WAA08737; Sun, 11 Jun 2000 22:27:26 +0100 Message-Id: <200006112127.WAA08737@raistlin.arm.linux.org.uk> Subject: Re: Serial options for a console To: jgg@ualberta.ca (Jason Gunthorpe) Date: Sun, 11 Jun 2000 22:27:26 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Jason Gunthorpe" at Jun 11, 2000 03:23:23 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jason Gunthorpe writes: > What would be nice is if that kernel command line CONFIG option were moved > into the main configuration so it could always be used (I think it is > hidden under a netwinder option?). Right now I haven't spent any time > working on finding/writing a more fancy boot loader that could do command > line options. It only appears for architectures that want to use it. If you have an architecture that needs it, add the relevent test to the relevent if statement in arch/arm/config.in. It is incorrect for it to appear for all architectures, since it is not used on all architectures. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 11 21:46:52 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131Ed6-00026X-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 11 Jun 2000 21:46:48 +0100 Received: from [199.185.220.111] (helo=priv-edtnes11-hme0.telusplanet.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131Ed4-00026P-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 11 Jun 2000 21:46:47 +0100 Received: from wakko.deltatee.com ([209.115.196.25]) by priv-edtnes11-hme0.telusplanet.net (InterMail vM.4.01.02.11 201-229-116-111) with ESMTP id <20000611214434.GMFU625.priv-edtnes11-hme0.telusplanet.net@wakko.deltatee.com>; Sun, 11 Jun 2000 15:44:34 -0600 Received: from localhost (wakko.deltatee.com) [127.0.0.1] (jgg) by wakko.deltatee.com with smtp (Exim 2.11 #1) id 131FXR-00087x-00 (Debian); Sun, 11 Jun 2000 15:45:01 -0600 Date: Sun, 11 Jun 2000 15:45:01 -0600 (MDT) From: Jason Gunthorpe X-Sender: jgg@wakko.deltatee.com To: Russell King - ARM Linux Admin cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Serial options for a console In-Reply-To: <200006112127.WAA08737@raistlin.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Sun, 11 Jun 2000, Russell King - ARM Linux Admin wrote: > It only appears for architectures that want to use it. If you have an > architecture that needs it, add the relevent test to the relevent if > statement in arch/arm/config.in. Thanks, I think I will do that. > It is incorrect for it to appear for all architectures, since it is not > used on all architectures. Hmm. I see how it works.. Wasn't how I thought. Jason unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 17:14:02 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131WlC-00010d-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 17:08:22 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131WlB-00010V-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 17:08:21 +0100 Received: from ims.crl.dec.com (ims.crl.dec.com [16.11.0.11]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id MAA14174; Mon, 12 Jun 2000 12:07:24 -0400 (EDT) Received: by ims.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Mon, 12 Jun 2000 12:12:13 -0400 Message-ID: From: George France To: "'Russell King - ARM Linux Admin'" , fray@mvista.com Cc: dmcdon@canoemail.com, linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Floating Point Emulation Date: Mon, 12 Jun 2000 12:07:23 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I also do not understand. Maximum verbosity please. --George George France, france@crl.dec.com Cambridge Research Laboratory, Compaq Computer Corporation One Cambridgecenter MS: CRL Cambridge, MA 02142 USA -----Original Message----- From: Russell King - ARM Linux Admin [mailto:linux@arm.linux.org.uk] Sent: Sunday, June 11, 2000 3:40 AM To: fray@mvista.com Cc: france@crl.dec.com; dmcdon@canoemail.com; linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation Mark Hatle writes: > The regular -msoft-float uses an external libfloat library, having to > use an external library for floating point is not the correct solution > IMHO, especially when gcc provides the functionality internally. Please explain what you mean more clearly. AFAICR, GCC does not provide any floating point emulation. Without -msoft-float, GCC will emit the hardware instructions for floating point operations, and therefore you will need some floating point emulation code. There are two possibilities: 1. Build with -msoft-float and use libfloat 2. Build without -msoft-float and use either the Acorn FPE or NWFPE. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 17:37:44 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131XBr-0001dg-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 17:35:55 +0100 Received: from [216.163.180.10] (helo=c0mailgw09.prontomail.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131XBq-0001d7-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 17:35:55 +0100 Received: from web104 (216.163.176.204) by c0mailgw09.prontomail.com (NPlex 4.5.049) id 390DEC9A0010EC15; Mon, 12 Jun 2000 09:35:11 -0700 X-Version: canoe 6.2.2329.0 From: "Dan" Message-Id: Date: Mon, 12 Jun 2000 10:34:07 -0600 X-Priority: Normal Content-Type: text/plain; charset=iso-8859-1 To: linux-arm-kernel@lists.arm.linux.org.uk Reply-To: dmcdon@canoemail.com Subject: RE: Floating Point Emulation CC: fray@mvista.com X-Mailer: Web Based Pronto Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello folks: I guess I started this little confusion. When I was trying to build a stand-alone application that used floating point calculations. My problem was that my stand-alone application did not import the floating point emulation using the undefined instruction vector. I tried re-compiling with -msoft-float and linking in libfloat; however, libgcc.a had been compiled without -msoft-float. The GCC manual says EVERYTHING should be compiled one way or the other. I was reluctant to start re-compiling libraries, etc. After all I am a newbie at this GCC and Linux stuff. I changed the application to use some fixed-point tricks and my problem mostly went away. As one recommended fix Mark Hatle recommended using the floating point that is now built into GCC. That is, you should no longer have to use floating point emulation or libfloat. He is working on some patches that will allow the kernel to be compiled without the NetWinder Emulation and without libfloat. Apparently this is the way other targets compile Linux. I went to the GCC site and was looking through their FAQ. http://www.objsw.com/CrossGCC/FAQ-6.html This site does say that libgcc supports floating point for most embedded targets. The FAQ is a little vague though. I am sure that Mark can shed more light on this than I can. It may be an opportunity to eliminate the floating point emulator from the ARM kernel if the compiler can take care of it. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 18:10:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131Xi7-0002CO-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 18:09:15 +0100 Received: from [63.192.220.206] (helo=hermes.mvista.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131Xi6-0002Bw-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 18:09:14 +0100 Received: from mvista.com (IDENT:mhatle@gemini.mvista.com [10.0.0.101]) by hermes.mvista.com (8.9.3/8.9.3) with ESMTP id KAA00827; Mon, 12 Jun 2000 10:08:37 -0700 Message-ID: <394518B7.CD4DF743@mvista.com> Date: Mon, 12 Jun 2000 10:07:03 -0700 From: Mark Hatle Organization: MontaVista Software, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: Russell King - ARM Linux Admin CC: George France , "'dmcdon@canoemail.com'" , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006110740.IAA01687@flint.arm.linux.org.uk> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Russell King - ARM Linux Admin wrote: > > Mark Hatle writes: > > The regular -msoft-float uses an external libfloat library, having to > > use an external library for floating point is not the correct solution > > IMHO, especially when gcc provides the functionality internally. > > Please explain what you mean more clearly. AFAICR, GCC does not > provide any floating point emulation. Without -msoft-float, GCC > will emit the hardware instructions for floating point operations, > and therefore you will need some floating point emulation code. > > There are two possibilities: > > 1. Build with -msoft-float and use libfloat > 2. Build without -msoft-float and use either the Acorn FPE or NWFPE. (Hopefully this will answer Russell's, George's and Dan's questions all at once.) Gcc (2.95.2 at least) does in fact have an internal floating point library. This library is used by other soft floating point targets, such as PowerPC. The system compiles in two source files to enable the built in soft floating point. fp-bit.c and dp-bit.c (FYI they are actually the same source file .../gcc/config/fp-bit.c, but the t-makefile copies it twice and adds a couple of #defines at the top of the file.) My understanding is that the gcc even with the built in floating point uses the method shown in http://www.netwinder.org/~scottb/notes/FP-Notes-6.html. The difference is that we don't need the external libfloat anymore. (i.e. things compiled with -mhard-float and -msoft-float are still not compatable, and there isn't a way to allow them to be.) FYI, I have a kernel on my netwinder that does not have NWFPE compile in, and I do not have libfloat. Everything I've compiled so far (bash, ldconfig, util-linux) has worked correctly with my patches. If you are interested in how I did the patch you can look at the gcc source in .../gcc/config/arm and then t-arm-elf is a generic config file, and t-linux is the one for linux. The t-arm-elf has the pieces in it to activate the gcc built-in soft floating point. Other arm targets seem to do the same. From what I can tell (quickly) it looks like the embedded style targets use the built in, and the bsd and linux target use an external library. I'll wrap up by changing the "two possibilities" above: 1) Use -msoft-float and use gcc's built-ins (with my patch) 2) Use -msoft-float and libfloat (default) 3) Use (-mhard-float and use) the compiled in NWFPE. Keep in mind that you can't link soft-float and hard-float items together. Questions? Hopefully I've answered them, but please ask more! --Mark unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 19:26:32 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131Yug-0003VV-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 19:26:18 +0100 Received: from [128.100.10.203] (helo=pyxis.lrtw.org ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131Yuf-0003VO-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 19:26:17 +0100 Received: from localhost (tangf@localhost) by pyxis.lrtw.org (8.9.3/8.9.3/Debian 8.9.3-6) with SMTP id OAA30264; Mon, 12 Jun 2000 14:26:15 -0400 Date: Mon, 12 Jun 2000 14:26:15 -0400 (EDT) From: Felix Tang X-Sender: tangf@pyxis.lrtw.org To: George France cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Text on soft-float by Scott Bambrough... In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk http://www.netwinder.org/~scottb/notes/FP-Notes-all.html Give it a read, I think it will help clear up some things. Felix On Mon, 12 Jun 2000, George France wrote: > I also do not understand. Maximum verbosity please. > > --George > > George France, france@crl.dec.com > Cambridge Research Laboratory, Compaq Computer Corporation > One Cambridgecenter MS: CRL > Cambridge, MA 02142 USA > > > > -----Original Message----- > From: Russell King - ARM Linux Admin [mailto:linux@arm.linux.org.uk] > Sent: Sunday, June 11, 2000 3:40 AM > To: fray@mvista.com > Cc: france@crl.dec.com; dmcdon@canoemail.com; > linux-arm-kernel@lists.arm.linux.org.uk > Subject: Re: Floating Point Emulation > > > Mark Hatle writes: > > The regular -msoft-float uses an external libfloat library, having to > > use an external library for floating point is not the correct solution > > IMHO, especially when gcc provides the functionality internally. > > Please explain what you mean more clearly. AFAICR, GCC does not > provide any floating point emulation. Without -msoft-float, GCC > will emit the hardware instructions for floating point operations, > and therefore you will need some floating point emulation code. > > There are two possibilities: > > 1. Build with -msoft-float and use libfloat > 2. Build without -msoft-float and use either the Acorn FPE or NWFPE. > _____ > |_____| ------------------------------------------------- ---+---+- > | | Russell King linux@arm.linux.org.uk --- --- > | | | | http://www.arm.linux.org.uk/ / / | > | +-+-+ --- -+- > / | THE developer of ARM Linux |+| /|\ > / | | | --- | > +-+-+ ------------------------------------------------- /\\\ | > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 19:40:21 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131Z85-0003ii-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 19:40:09 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131Z84-0003ia-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 19:40:08 +0100 Received: from crlwebmail.crl.dec.com (crlwebmail.crl.dec.com [16.11.0.2]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id OAA17383; Mon, 12 Jun 2000 14:40:06 -0400 (EDT) Received: by crlwebmail.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Mon, 12 Jun 2000 14:44:42 -0400 Message-ID: From: George France To: "'dmcdon@canoemail.com'" , linux-arm-kernel@lists.arm.linux.org.uk Cc: fray@mvista.com Subject: RE: Floating Point Emulation Date: Mon, 12 Jun 2000 14:40:02 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello Dan; I think you are assuming that people only wish to use gcc. If you are using a different compiler, you still need the floating point emulations in the kernel. --George George France, france@crl.dec.com Cambridge Research Laboratory, Compaq Computer Corporation One Cambridgecenter MS: CRL Cambridge, MA 02142 USA -----Original Message----- From: Dan [mailto:dmcdon@canoemail.com] Sent: Monday, June 12, 2000 12:34 PM To: linux-arm-kernel@lists.arm.linux.org.uk Cc: fray@mvista.com Subject: RE: Floating Point Emulation Hello folks: I guess I started this little confusion. When I was trying to build a stand-alone application that used floating point calculations. My problem was that my stand-alone application did not import the floating point emulation using the undefined instruction vector. I tried re-compiling with -msoft-float and linking in libfloat; however, libgcc.a had been compiled without -msoft-float. The GCC manual says EVERYTHING should be compiled one way or the other. I was reluctant to start re-compiling libraries, etc. After all I am a newbie at this GCC and Linux stuff. I changed the application to use some fixed-point tricks and my problem mostly went away. As one recommended fix Mark Hatle recommended using the floating point that is now built into GCC. That is, you should no longer have to use floating point emulation or libfloat. He is working on some patches that will allow the kernel to be compiled without the NetWinder Emulation and without libfloat. Apparently this is the way other targets compile Linux. I went to the GCC site and was looking through their FAQ. http://www.objsw.com/CrossGCC/FAQ-6.html This site does say that libgcc supports floating point for most embedded targets. The FAQ is a little vague though. I am sure that Mark can shed more light on this than I can. It may be an opportunity to eliminate the floating point emulator from the ARM kernel if the compiler can take care of it. -------------------------------- Bye! Dan McDonald The Late Night Software Shop Sign up today for your Free E-mail at: http://www.canoe.ca/CanoeMail unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 19:53:55 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131ZLH-0003wM-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 19:53:47 +0100 Received: from [207.245.35.202] (helo=kei.netwinder.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131ZLG-0003wD-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 19:53:46 +0100 Received: from netwinder.org ([10.1.19.121]) by kei.netwinder.org (8.9.3/8.9.3) with ESMTP id OAA32471; Mon, 12 Jun 2000 14:53:22 -0400 Message-ID: <39453095.9BF4C900@netwinder.org> Date: Mon, 12 Jun 2000 14:48:54 -0400 From: Scott Bambrough Organization: Rebel.com X-Mailer: Mozilla 4.7 [en] (Win95; U) X-Accept-Language: en MIME-Version: 1.0 To: Mark Hatle CC: Russell King - ARM Linux Admin , George France , "'dmcdon@canoemail.com'" , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006110740.IAA01687@flint.arm.linux.org.uk> <394518B7.CD4DF743@mvista.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk > GCC 2.95 does in fact have an internal floating point > library. This library is used by other soft floating point targets, > such as PowerPC. The system compiles in two source files to enable the > built in soft floating point. fp-bit.c and dp-bit.c Mark is correct here. The file fp-bit.c is a relatively new addition that resulted from the gcc2 code merges. It offers the same set of entry points as libfloat. However the version on the 2.95 branch has several bugs, and should be brought up to the same level as the mainline AFAIK. AFAICT, this work is meant to be used on the mainline, not on the branch. There is no reason it cannot be used in 2.95, it just wasn't tested to my knowledge. This code wasn't available when we first did libfloat (that would be gcc 2.81 days). > FYI, I have a kernel on my netwinder that does not have NWFPE compile > in, and I do not have libfloat. Everything I've compiled so far (bash, > ldconfig, util-linux) has worked correctly with my patches. I would expect this to be the case. A better test would be the glibc math test suite or the UCB tests from netlib. Do the tests gcc.c-torture/execute/ieee/fp-cmp-[1-4].c pass in the gcc testsuite? > If you are interested in how I did the patch you can look at the gcc > source in .../gcc/config/arm and then t-arm-elf is a generic config > file, and t-linux is the one for linux. The t-arm-elf has the pieces in > it to activate the gcc built-in soft floating point. Other arm targets > seem to do the same. From what I can tell (quickly) it looks like the > embedded style targets use the built in, and the bsd and linux target > use an external library. I have a patch similar to this for the gcc mainline to fix the fp-cmp-4 test suite failure seen on Linux. Question is do I patch the specs file as well? I see what happens after more testing. Scott -- Scott Bambrough - Software Engineer REBEL.COM http://www.rebel.com NetWinder http://www.netwinder.org unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 20:45:58 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131a9Y-0004ps-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 20:45:44 +0100 Received: from [158.152.220.239] (helo=tazenda.demon.co.uk ident=mail) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131a9R-0004pf-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 20:45:37 +0100 Received: from kings-cross.london.uk.eu.org [::ffff:192.168.2.83] by tazenda.demon.co.uk with esmtp (Exim 3.11 #1 (Debian)) id 131a8D-0005NO-00; Mon, 12 Jun 2000 20:44:21 +0100 Received: from localhost ([::ffff:127.0.0.1] helo=kings-cross.london.uk.eu.org ident=phil) by kings-cross.london.uk.eu.org with esmtp (Exim 3.12 #1) id 131a8B-0001Yh-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 20:44:19 +0100 X-Mailer: exmh version 2.1.1 10/15/1999 (debian) To: linux-arm-kernel@lists.arm.linux.org.uk Subject: ptrace patch Mime-Version: 1.0 Content-Type: multipart/signed; boundary="==_Exmh_435057350P"; micalg=pgp-sha1; protocol="application/pgp-signature" Content-Transfer-Encoding: 7bit Date: Mon, 12 Jun 2000 20:44:18 +0100 From: Philip Blundell Message-Id: Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk --==_Exmh_435057350P Content-Type: text/plain; charset=us-ascii Any comments on this patch? It's been in CVS at netwinder.org for some time. I think it's necessary for correct single-stepping. p. --- linux/arch/arm/kernel/ptrace.c Sun Apr 30 11:15:42 2000 +++ linux/arch/arm/kernel/ptrace.c Thu Mar 9 08:03:29 2000 @@ -15,6 +15,7 @@ #include #include #include +#include /* * does not yet catch signals sent when the child dies. @@ -92,7 +93,6 @@ static unsigned long ptrace_getrn (struc else val = get_stack_long (child, reg); -printk ("r%02d=%08lX ", reg, val); return val; } @@ -105,12 +105,10 @@ static unsigned long ptrace_getaluop2 (s int shift; int type; -printk ("op2="); if (insn & 1 << 25) { val = insn & 255; shift = (insn >> 8) & 15; type = 3; -printk ("(imm)"); } else { val = get_stack_long (child, insn & 15); @@ -120,9 +118,7 @@ printk ("(imm)"); shift = (insn >> 7) & 31; type = (insn >> 5) & 3; -printk ("(r%02ld)", insn & 15); } -printk ("sh%dx%d", type, shift); switch (type) { case 0: val <<= shift; break; case 1: val >>= shift; break; @@ -133,7 +129,6 @@ printk ("sh%dx%d", type, shift); val = (val >> shift) | (val << (32 - shift)); break; } -printk ("=%08lX ", val); return val; } @@ -150,7 +145,6 @@ static unsigned long ptrace_getldrop2 (s shift = (insn >> 7) & 31; type = (insn >> 5) & 3; -printk ("op2=r%02ldsh%dx%d", insn & 15, shift, type); switch (type) { case 0: val <<= shift; break; case 1: val >>= shift; break; @@ -161,7 +155,6 @@ printk ("op2=r%02ldsh%dx%d", insn & 15, val = (val >> shift) | (val << (32 - shift)); break; } -printk ("=%08lX ", val); return val; } @@ -170,13 +163,11 @@ get_branch_address(struct task_struct *c { unsigned long alt = 0; -printk(KERN_DEBUG "ptrace_set_bpt: insn=%08lX pc=%08lX ", insn, pc); switch (insn & 0x0e100000) { case 0x00000000: case 0x00100000: case 0x02000000: case 0x02100000: /* data processing */ - printk ("data "); switch (insn & 0x01e0f000) { case 0x0000f000: alt = ptrace_getrn(child, insn) & ptrace_getaluop2(child, insn); @@ -220,9 +211,8 @@ printk(KERN_DEBUG "ptrace_set_bpt: insn= } break; - case 0x04100000: /* ldr */ + case 0x06100000: /* ldr */ if ((insn & 0xf000) == 0xf000) { -printk ("ldr "); alt = ptrace_getrn(child, insn); if (insn & 1 << 24) { if (insn & 1 << 23) @@ -237,9 +227,8 @@ printk ("ldr "); } break; - case 0x06100000: /* ldr imm */ + case 0x04100000: /* ldr imm */ if ((insn & 0xf000) == 0xf000) { -printk ("ldrimm "); alt = ptrace_getrn(child, insn); if (insn & 1 << 24) { if (insn & 1 << 23) @@ -258,7 +247,6 @@ printk ("ldrimm "); if (insn & (1 << 15)) { unsigned long base; int nr_regs; -printk ("ldm "); if (insn & (1 << 23)) { nr_regs = insn & 65535; @@ -291,7 +279,6 @@ printk ("ldm "); case 0x0a000000: case 0x0a100000: { /* bl or b */ signed long displ; -printk ("b/bl "); /* It's a branch/branch link: instead of trying to * figure out whether the branch will be taken or not, * we'll put a breakpoint at either location. This is @@ -306,7 +293,6 @@ printk ("b/bl "); } break; } -printk ("=%08lX\n", alt); return alt; } @@ -332,6 +318,8 @@ add_breakpoint(struct task_struct *child return res; } +#define predicate(x) (x & 0xf0000000) + int ptrace_set_bpt (struct task_struct *child) { struct debug_info *dbg = &child->thread.debug; @@ -346,22 +334,20 @@ int ptrace_set_bpt (struct task_struct * dbg->nsaved = 0; - res = add_breakpoint(child, dbg, pc + 4); + alt = get_branch_address(child, pc, insn); + if (alt) + res = add_breakpoint(child, dbg, alt); - if (res == 0) { - alt = get_branch_address(child, pc, insn); - if (alt) - res = add_breakpoint(child, dbg, alt); - } + if (!res && (!alt || (predicate(insn) != 0xe0000000))) + res = add_breakpoint(child, dbg, pc + 4); } return res; } -/* Ensure no single-step breakpoint is pending. Returns non-zero - * value if child was being single-stepped. +/* Ensure no single-step breakpoint is pending. */ -int ptrace_cancel_bpt (struct task_struct *child) +void ptrace_do_cancel_bpt (struct task_struct *child) { struct debug_info *dbg = &child->thread.debug; unsigned long tmp; @@ -380,8 +366,6 @@ int ptrace_cancel_bpt (struct task_struc printk(KERN_ERR "ptrace_cancel_bpt: weirdness\n"); write_long(child, dbg->bp[i].address, dbg->bp[i].insn); } - - return nsaved != 0; } asmlinkage int sys_ptrace(long request, long pid, long addr, long data) @@ -492,7 +476,7 @@ asmlinkage int sys_ptrace(long request, child->exit_code = data; wake_up_process (child); /* make sure single-step breakpoint is gone. */ - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; @@ -506,7 +490,7 @@ asmlinkage int sys_ptrace(long request, wake_up_process (child); child->exit_code = SIGKILL; /* make sure single-step breakpoint is gone. */ - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; @@ -575,7 +559,7 @@ asmlinkage int sys_ptrace(long request, child->p_pptr = child->p_opptr; SET_LINKS(child); /* make sure single-step breakpoint is gone. */ - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; --- linux/arch/arm/kernel/signal.c Sun Jun 4 17:07:26 2000 +++ linux/arch/arm/kernel/signal.c Sun Jun 4 17:40:15 2000 @@ -22,6 +22,7 @@ #include #include #include +#include #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) @@ -31,8 +32,6 @@ asmlinkage int sys_wait4(pid_t pid, unsigned long * stat_addr, int options, unsigned long *ru); asmlinkage int do_signal(sigset_t *oldset, struct pt_regs * regs, int syscall); -extern int ptrace_cancel_bpt (struct task_struct *); -extern int ptrace_set_bpt (struct task_struct *); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) { --- linux/include/asm-arm/breakpoint.h Thu Jan 1 01:00:00 1970 +++ linux/include/asm-arm/breakpoint.h Fri Aug 6 19:09:02 1999 @@ -0,0 +1,24 @@ +#ifndef __ASM_BREAKPOINT_H +#define __ASM_BREAKPOINT_H + +extern void ptrace_do_cancel_bpt (struct task_struct *); +extern int ptrace_set_bpt (struct task_struct *); + +static inline int +ptrace_query_bpt (struct task_struct *child) +{ + return (child->thread.debug.nsaved != 0); +} + +static inline int +ptrace_cancel_bpt (struct task_struct *child) +{ + int r; + if (r = ptrace_query_bpt (child), r == 0) + return 0; + + ptrace_do_cancel_bpt (child); + return r; +} + +#endif --- linux/arch/arm/kernel/traps.c Sat May 27 19:53:37 2000 +++ linux/arch/arm/kernel/traps.c Mon Jun 12 20:41:56 2000 @@ -24,6 +24,7 @@ #include #include #include +#include extern void c_backtrace (unsigned long fp, int pmode); extern int ptrace_cancel_bpt (struct task_struct *); --==_Exmh_435057350P Content-Type: application/pgp-signature -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Content-Type: text/plain; charset=us-ascii Any comments on this patch? It's been in CVS at netwinder.org for some time. I think it's necessary for correct single-stepping. p. - --- linux/arch/arm/kernel/ptrace.c Sun Apr 30 11:15:42 2000 +++ linux/arch/arm/kernel/ptrace.c Thu Mar 9 08:03:29 2000 @@ -15,6 +15,7 @@ #include #include #include +#include /* * does not yet catch signals sent when the child dies. @@ -92,7 +93,6 @@ static unsigned long ptrace_getrn (struc else val = get_stack_long (child, reg); - -printk ("r%02d=%08lX ", reg, val); return val; } @@ -105,12 +105,10 @@ static unsigned long ptrace_getaluop2 (s int shift; int type; - -printk ("op2="); if (insn & 1 << 25) { val = insn & 255; shift = (insn >> 8) & 15; type = 3; - -printk ("(imm)"); } else { val = get_stack_long (child, insn & 15); @@ -120,9 +118,7 @@ printk ("(imm)"); shift = (insn >> 7) & 31; type = (insn >> 5) & 3; - -printk ("(r%02ld)", insn & 15); } - -printk ("sh%dx%d", type, shift); switch (type) { case 0: val <<= shift; break; case 1: val >>= shift; break; @@ -133,7 +129,6 @@ printk ("sh%dx%d", type, shift); val = (val >> shift) | (val << (32 - shift)); break; } - -printk ("=%08lX ", val); return val; } @@ -150,7 +145,6 @@ static unsigned long ptrace_getldrop2 (s shift = (insn >> 7) & 31; type = (insn >> 5) & 3; - -printk ("op2=r%02ldsh%dx%d", insn & 15, shift, type); switch (type) { case 0: val <<= shift; break; case 1: val >>= shift; break; @@ -161,7 +155,6 @@ printk ("op2=r%02ldsh%dx%d", insn & 15, val = (val >> shift) | (val << (32 - shift)); break; } - -printk ("=%08lX ", val); return val; } @@ -170,13 +163,11 @@ get_branch_address(struct task_struct *c { unsigned long alt = 0; - -printk(KERN_DEBUG "ptrace_set_bpt: insn=%08lX pc=%08lX ", insn, pc); switch (insn & 0x0e100000) { case 0x00000000: case 0x00100000: case 0x02000000: case 0x02100000: /* data processing */ - - printk ("data "); switch (insn & 0x01e0f000) { case 0x0000f000: alt = ptrace_getrn(child, insn) & ptrace_getaluop2(child, insn); @@ -220,9 +211,8 @@ printk(KERN_DEBUG "ptrace_set_bpt: insn= } break; - - case 0x04100000: /* ldr */ + case 0x06100000: /* ldr */ if ((insn & 0xf000) == 0xf000) { - -printk ("ldr "); alt = ptrace_getrn(child, insn); if (insn & 1 << 24) { if (insn & 1 << 23) @@ -237,9 +227,8 @@ printk ("ldr "); } break; - - case 0x06100000: /* ldr imm */ + case 0x04100000: /* ldr imm */ if ((insn & 0xf000) == 0xf000) { - -printk ("ldrimm "); alt = ptrace_getrn(child, insn); if (insn & 1 << 24) { if (insn & 1 << 23) @@ -258,7 +247,6 @@ printk ("ldrimm "); if (insn & (1 << 15)) { unsigned long base; int nr_regs; - -printk ("ldm "); if (insn & (1 << 23)) { nr_regs = insn & 65535; @@ -291,7 +279,6 @@ printk ("ldm "); case 0x0a000000: case 0x0a100000: { /* bl or b */ signed long displ; - -printk ("b/bl "); /* It's a branch/branch link: instead of trying to * figure out whether the branch will be taken or not, * we'll put a breakpoint at either location. This is @@ -306,7 +293,6 @@ printk ("b/bl "); } break; } - -printk ("=%08lX\n", alt); return alt; } @@ -332,6 +318,8 @@ add_breakpoint(struct task_struct *child return res; } +#define predicate(x) (x & 0xf0000000) + int ptrace_set_bpt (struct task_struct *child) { struct debug_info *dbg = &child->thread.debug; @@ -346,22 +334,20 @@ int ptrace_set_bpt (struct task_struct * dbg->nsaved = 0; - - res = add_breakpoint(child, dbg, pc + 4); + alt = get_branch_address(child, pc, insn); + if (alt) + res = add_breakpoint(child, dbg, alt); - - if (res == 0) { - - alt = get_branch_address(child, pc, insn); - - if (alt) - - res = add_breakpoint(child, dbg, alt); - - } + if (!res && (!alt || (predicate(insn) != 0xe0000000))) + res = add_breakpoint(child, dbg, pc + 4); } return res; } - -/* Ensure no single-step breakpoint is pending. Returns non-zero - - * value if child was being single-stepped. +/* Ensure no single-step breakpoint is pending. */ - -int ptrace_cancel_bpt (struct task_struct *child) +void ptrace_do_cancel_bpt (struct task_struct *child) { struct debug_info *dbg = &child->thread.debug; unsigned long tmp; @@ -380,8 +366,6 @@ int ptrace_cancel_bpt (struct task_struc printk(KERN_ERR "ptrace_cancel_bpt: weirdness\n"); write_long(child, dbg->bp[i].address, dbg->bp[i].insn); } - - - - return nsaved != 0; } asmlinkage int sys_ptrace(long request, long pid, long addr, long data) @@ -492,7 +476,7 @@ asmlinkage int sys_ptrace(long request, child->exit_code = data; wake_up_process (child); /* make sure single-step breakpoint is gone. */ - - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; @@ -506,7 +490,7 @@ asmlinkage int sys_ptrace(long request, wake_up_process (child); child->exit_code = SIGKILL; /* make sure single-step breakpoint is gone. */ - - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; @@ -575,7 +559,7 @@ asmlinkage int sys_ptrace(long request, child->p_pptr = child->p_opptr; SET_LINKS(child); /* make sure single-step breakpoint is gone. */ - - ptrace_cancel_bpt (child); + ptrace_do_cancel_bpt (child); ret = 0; goto out; - --- linux/arch/arm/kernel/signal.c Sun Jun 4 17:07:26 2000 +++ linux/arch/arm/kernel/signal.c Sun Jun 4 17:40:15 2000 @@ -22,6 +22,7 @@ #include #include #include +#include #define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP))) @@ -31,8 +32,6 @@ asmlinkage int sys_wait4(pid_t pid, unsigned long * stat_addr, int options, unsigned long *ru); asmlinkage int do_signal(sigset_t *oldset, struct pt_regs * regs, int syscall); - -extern int ptrace_cancel_bpt (struct task_struct *); - -extern int ptrace_set_bpt (struct task_struct *); int copy_siginfo_to_user(siginfo_t *to, siginfo_t *from) { - --- linux/include/asm-arm/breakpoint.h Thu Jan 1 01:00:00 1970 +++ linux/include/asm-arm/breakpoint.h Fri Aug 6 19:09:02 1999 @@ -0,0 +1,24 @@ +#ifndef __ASM_BREAKPOINT_H +#define __ASM_BREAKPOINT_H + +extern void ptrace_do_cancel_bpt (struct task_struct *); +extern int ptrace_set_bpt (struct task_struct *); + +static inline int +ptrace_query_bpt (struct task_struct *child) +{ + return (child->thread.debug.nsaved != 0); +} + +static inline int +ptrace_cancel_bpt (struct task_struct *child) +{ + int r; + if (r = ptrace_query_bpt (child), r == 0) + return 0; + + ptrace_do_cancel_bpt (child); + return r; +} + +#endif - --- linux/arch/arm/kernel/traps.c Sat May 27 19:53:37 2000 +++ linux/arch/arm/kernel/traps.c Mon Jun 12 20:41:56 2000 @@ -24,6 +24,7 @@ #include #include #include +#include extern void c_backtrace (unsigned long fp, int pmode); extern int ptrace_cancel_bpt (struct task_struct *); -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.0 (GNU/Linux) Comment: Exmh version 2.1.1 10/15/1999 (debian) iD8DBQE5RT2SVTLPJe9CT30RAhfqAJ4lY78Nma4zX1KCVq5Wu2ZB2ILSVACfYDSc bQ6F4mKhtpOd0u5Np2aGrAI= =sh8K -----END PGP SIGNATURE----- --==_Exmh_435057350P-- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 21:07:53 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131aUc-00058F-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 21:07:30 +0100 Received: from [63.192.220.206] (helo=hermes.mvista.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131aUF-00057N-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 21:07:07 +0100 Received: from mvista.com (IDENT:mhatle@gemini.mvista.com [10.0.0.101]) by hermes.mvista.com (8.9.3/8.9.3) with ESMTP id NAA04098; Mon, 12 Jun 2000 13:06:03 -0700 Message-ID: <3945424D.A0BB948F@mvista.com> Date: Mon, 12 Jun 2000 13:04:29 -0700 From: Mark Hatle Organization: MontaVista Software, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: Scott Bambrough CC: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006110740.IAA01687@flint.arm.linux.org.uk> <394518B7.CD4DF743@mvista.com> <39453095.9BF4C900@netwinder.org> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Scott Bambrough wrote: > > > GCC 2.95 does in fact have an internal floating point > > library. This library is used by other soft floating point targets, > > such as PowerPC. The system compiles in two source files to enable the > > built in soft floating point. fp-bit.c and dp-bit.c > > Mark is correct here. The file fp-bit.c is a relatively new addition that > resulted from the gcc2 code merges. It offers the same set of entry points as > libfloat. However the version on the 2.95 branch has several bugs, and should > be brought up to the same level as the mainline AFAIK. AFAICT, this work is > meant to be used on the mainline, not on the branch. There is no reason it > cannot be used in 2.95, it just wasn't tested to my knowledge. This code wasn't > available when we first did libfloat (that would be gcc 2.81 days). > Interesting, for our PPC work we have a bunch of items that are backported from the mainline to the 2.95.2. I will ask if the floating point is included in that work. If it is, then I may already have a patch for that. > > FYI, I have a kernel on my netwinder that does not have NWFPE compile > > in, and I do not have libfloat. Everything I've compiled so far (bash, > > ldconfig, util-linux) has worked correctly with my patches. > > I would expect this to be the case. A better test would be the glibc math test > suite or the UCB tests from netlib. Do the tests > gcc.c-torture/execute/ieee/fp-cmp-[1-4].c pass in the gcc testsuite? > I have not run any of the gcc testsuites yet. Mostly I have written small test programs that do the basic + - * / operations and the correct results are returned. I will be testing with the gcc and glibc tests "soon", but I'm not sure yet when I will have the opportunity to do it. > > If you are interested in how I did the patch you can look at the gcc > > source in .../gcc/config/arm and then t-arm-elf is a generic config > > file, and t-linux is the one for linux. The t-arm-elf has the pieces in > > it to activate the gcc built-in soft floating point. Other arm targets > > seem to do the same. From what I can tell (quickly) it looks like the > > embedded style targets use the built in, and the bsd and linux target > > use an external library. > > I have a patch similar to this for the gcc mainline to fix the fp-cmp-4 test > suite failure seen on Linux. Question is do I patch the specs file as well? I > see what happens after more testing. > What in the specs file were you considering changing? On the PPC side, we've had a bit of internal discussion on various things we can do to better support soft-float. One of which is to change the %{msoft-float: -D_SOFT_FLOAT} to %{!mhard-float: -D_SOFT_FLOAT}. (The arm define is different, but I think you can see what I'm getting at.) Also my patch activates the --without-fp on the config, and thats what triggers the SPEC file change. --Mark unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 21:43:36 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131b3B-0005iW-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 21:43:13 +0100 Received: from [212.250.233.13] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131b39-0005iP-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 21:43:11 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id VAA01568; Mon, 12 Jun 2000 21:42:50 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id VAA10820; Mon, 12 Jun 2000 21:42:04 +0100 Message-Id: <200006122042.VAA10820@raistlin.arm.linux.org.uk> Subject: Re: ptrace patch To: Philip.Blundell@pobox.com (Philip Blundell) Date: Mon, 12 Jun 2000 21:42:03 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Philip Blundell" at Jun 12, 2000 08:44:18 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Do you really have to send messages that are twice as long as they have to be? If this is how GPG works, then it ought to be consigned to the dustbin. Please fix your mailer/GPG configuration. Philip Blundell writes: > I think it's necessary for correct single-stepping. Seems to correct a problem with the ldr/ldr immediate stuff, but other than that its just performance enhancements. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 21:51:54 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131bBR-0005tF-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 21:51:45 +0100 Received: from [158.152.220.239] (helo=tazenda.demon.co.uk ident=mail) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131bBP-0005sB-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 21:51:43 +0100 Received: from kings-cross.london.uk.eu.org [::ffff:192.168.2.83] by tazenda.demon.co.uk with esmtp (Exim 3.11 #1 (Debian)) id 131bDa-0005YD-00; Mon, 12 Jun 2000 21:53:58 +0100 Received: from localhost ([::ffff:127.0.0.1] helo=kings-cross.london.uk.eu.org ident=phil) by kings-cross.london.uk.eu.org with esmtp (Exim 3.12 #1) id 131bDY-0001n1-00; Mon, 12 Jun 2000 21:53:56 +0100 X-Mailer: exmh version 2.1.1 10/15/1999 (debian) To: Russell King - ARM Linux Admin Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: ptrace patch In-Reply-To: Message from Russell King - ARM Linux Admin of "Mon, 12 Jun 2000 21:42:03 BST." <200006122042.VAA10820@raistlin.arm.linux.org.uk> References: <200006122042.VAA10820@raistlin.arm.linux.org.uk> Mime-Version: 1.0 Content-Type: multipart/signed; boundary="==_Exmh_466238737P"; micalg=pgp-sha1; protocol="application/pgp-signature" Content-Transfer-Encoding: 7bit Date: Mon, 12 Jun 2000 21:53:55 +0100 From: Philip Blundell Message-Id: Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk --==_Exmh_466238737P Content-Type: text/plain; charset=us-ascii >Do you really have to send messages that are twice as long as they >have to be? If this is how GPG works, then it ought to be consigned >to the dustbin. No, just my misconfiguration. Sorry. >Seems to correct a problem with the ldr/ldr immediate stuff, but other >than that its just performance enhancements. This bit is important too: - res = add_breakpoint(child, dbg, pc + 4); + alt = get_branch_address(child, pc, insn); + if (alt) + res = add_breakpoint(child, dbg, alt); - if (res == 0) { - alt = get_branch_address(child, pc, insn); - if (alt) - res = add_breakpoint(child, dbg, alt); - } + if (!res && (!alt || (predicate(insn) != 0xe0000000))) + res = add_breakpoint(child, dbg, pc + 4); } p. --==_Exmh_466238737P Content-Type: application/pgp-signature -----BEGIN PGP MESSAGE----- Version: GnuPG v1.0.0 (GNU/Linux) Comment: Exmh version 2.1.1 10/15/1999 (debian) owGVU79rFEEUziUK3sKiKQUlL4XHHdm7i0FBc1wCUQkW8QeJjU2Y3X23M2Z3ZpmZ 9W41QbAIlimsxVL/CNHCTtDCxl4trGwtRH2z55HSuOwyO2++733fe4858Gdq07N3 lj6c+752c6/27ps9udjJTNJZunTxfAdHGb98bePzFSUtStveKnNcBosj281TJmQP Is60QdsvTJuZSAjf872VqwpKVYBGlqYlcHYfwSowKGPI0BiWoAHLmQWm6WQoIgRm IFUycavlWFKSCS3EVYDrAwoLA/RyNYT1W+swVHrHBA4tQVhQRcLtGA+RkkYkEmNK QxGCQFwYGwrZcf5uqADu0R6yEjJhCD0QSaGZFUp2ADaV1mUFXNlEzIxLGlEMIzIM uVZhihkMheVV5jTWXfpAZBnGglkEY4vBIICwIFeE0M4FZ3JcsrBmLJ6jHiidMUnV o+RuzajHplLecsWGVJarOcuVtky66tSyO23X65pa2AcWx9shtXknV0LaZsRFGgcQ h0kAeQQLcKHV872Fep2lltAJWkKTEN8mImUwE0YeBSCkkQR/88ARxACaRGpV7H+L OWjvrzNHrQh9WGzBwyr2PwYq/KG+2x3ZAIH3/In/ecdqNOjHqe/uQjPXNKGIRtSs tGCeLI5wcfy0Wket9rC14OR8L69m5ntPVmeO19x9mdyn2enm3NTz/V+NM/t3D+Ze rf04UaqNT6/ffv059az79PfZ6Mvp0ctTH98/9o7lj140bv8B =zc6+ -----END PGP MESSAGE----- --==_Exmh_466238737P-- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 21:54:17 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131bDp-0005vl-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 21:54:13 +0100 Received: from [207.245.35.202] (helo=kei.netwinder.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131bDo-0005vf-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 21:54:12 +0100 Received: from netwinder.org ([10.1.19.121]) by kei.netwinder.org (8.9.3/8.9.3) with ESMTP id QAA01023; Mon, 12 Jun 2000 16:54:07 -0400 Message-ID: <39454CE3.B21ED2E1@netwinder.org> Date: Mon, 12 Jun 2000 16:49:39 -0400 From: Scott Bambrough Organization: Rebel.com X-Mailer: Mozilla 4.7 [en] (Win95; U) X-Accept-Language: en MIME-Version: 1.0 To: Mark Hatle CC: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006110740.IAA01687@flint.arm.linux.org.uk> <394518B7.CD4DF743@mvista.com> <39453095.9BF4C900@netwinder.org> <3945424D.A0BB948F@mvista.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Mark Hatle wrote: > > I have not run any of the gcc testsuites yet. Mostly I have written > small test programs that do the basic + - * / operations and the correct > results are returned. I will be testing with the gcc and glibc tests > "soon", but I'm not sure yet when I will have the opportunity to do it. I'd be interested in the results when you get the chance. > What in the specs file were you considering changing? On the PPC side, > we've had a bit of internal discussion on various things we can do to > better support soft-float. One of which is to change the %{msoft-float: > -D_SOFT_FLOAT} to %{!mhard-float: -D_SOFT_FLOAT}. (The arm define is > different, but I think you can see what I'm getting at.) Also my patch > activates the --without-fp on the config, and thats what triggers the > SPEC file change. I don't know whether that would fly on the ARM. I suspect to conform to the ATPCS we will need a version where soft, hard and no floating point options would be available. Such a change as you describe would preclude that. As for what I would change in the specs file, the linux-elf version automatically links in libfloat. If we have the soft-float entry points in libgcc, is this really necessary and is it likely to cause problems, link warnings, etc. I don't know how many people actually use these routines as well, so I'm not convinced they work correctly. It looks like someone has set up the code to handle the ARM doubles, and little endian/big endian troubles. The code makes liberal use of structs, and that always makes me nervous, given the alignment constraints on structs. Everything may be fine, but to date, nothing with ARM floating point has been easy. Just about every package that handles floats/doubles seems to require a measure of porting on the ARM. Why should this be any different? I need to be convinced it is ok. This requires testing. Scott -- Scott Bambrough - Software Engineer REBEL.COM http://www.rebel.com NetWinder http://www.netwinder.org unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 21:59:51 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131bJF-00063P-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 21:59:49 +0100 Received: from [212.250.233.13] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131bJD-000635-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 21:59:48 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id VAA01665; Mon, 12 Jun 2000 21:59:26 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id VAA11581; Mon, 12 Jun 2000 21:58:39 +0100 Message-Id: <200006122058.VAA11581@raistlin.arm.linux.org.uk> Subject: Re: Floating Point Emulation To: scottb@netwinder.org (Scott Bambrough) Date: Mon, 12 Jun 2000 21:58:38 +0100 (BST) Cc: fray@mvista.com (Mark Hatle), linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <39454CE3.B21ED2E1@netwinder.org> from "Scott Bambrough" at Jun 12, 2000 04:49:39 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Scott Bambrough writes: > Everything may be fine, but to date, nothing with ARM floating point has been > easy. Just about every package that handles floats/doubles seems to require a > measure of porting on the ARM. Why should this be any different? I need to be > convinced it is ok. This requires testing. Care to enlighten us a little more with a list of "typical" problems with the floats/doubles? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 22:18:11 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131bas-0006Pg-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 22:18:02 +0100 Received: from [207.245.35.202] (helo=kei.netwinder.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131bar-0006PY-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 22:18:01 +0100 Received: from netwinder.org ([10.1.19.121]) by kei.netwinder.org (8.9.3/8.9.3) with ESMTP id RAA01250; Mon, 12 Jun 2000 17:17:57 -0400 Message-ID: <39455279.985A860C@netwinder.org> Date: Mon, 12 Jun 2000 17:13:29 -0400 From: Scott Bambrough Organization: Rebel.com X-Mailer: Mozilla 4.7 [en] (Win95; U) X-Accept-Language: en MIME-Version: 1.0 To: Russell King - ARM Linux Admin CC: Mark Hatle , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006122058.VAA11581@raistlin.arm.linux.org.uk> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Russell King - ARM Linux Admin wrote: > > Care to enlighten us a little more with a list of "typical" problems with > the floats/doubles? Typically most software is written for little endian Intel architechures. In the stuff I have ported, the biggest trouble is with the ARM double format. The next biggest problem is with masks for testing for things like NaN's, masking out the exponent etc. Well designed code seems to have a define for the masks they use, other stuff uses numeric literals, which have to be hunted down individually. I've never had to deal with little/big endian issues myself. The only other thing that I would actually check in this code is the structures used to represent floats. People who work on x86 platforms expect packed structs; they never consider alignment issues. All this being said, I've only had a cursory look at fp-bit.c. It may suffer from none of these potential problems. I only wanted to point out the possible pitfalls garnered from my experiences. Scott -- Scott Bambrough - Software Engineer REBEL.COM http://www.rebel.com NetWinder http://www.netwinder.org unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 12 22:35:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131bry-0006iD-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 12 Jun 2000 22:35:42 +0100 Received: from [63.192.220.206] (helo=hermes.mvista.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131brx-0006hd-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 12 Jun 2000 22:35:41 +0100 Received: from mvista.com (IDENT:mhatle@gemini.mvista.com [10.0.0.101]) by hermes.mvista.com (8.9.3/8.9.3) with ESMTP id OAA05685; Mon, 12 Jun 2000 14:34:39 -0700 Message-ID: <39455711.BA7072A7@mvista.com> Date: Mon, 12 Jun 2000 14:33:05 -0700 From: Mark Hatle Organization: MontaVista Software, Inc. X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: Scott Bambrough CC: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Floating Point Emulation References: <200006110740.IAA01687@flint.arm.linux.org.uk> <394518B7.CD4DF743@mvista.com> <39453095.9BF4C900@netwinder.org> <3945424D.A0BB948F@mvista.com> <39454CE3.B21ED2E1@netwinder.org> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Scott Bambrough wrote: > > Mark Hatle wrote: > > What in the specs file were you considering changing? On the PPC side, > > we've had a bit of internal discussion on various things we can do to > > better support soft-float. One of which is to change the %{msoft-float: > > -D_SOFT_FLOAT} to %{!mhard-float: -D_SOFT_FLOAT}. (The arm define is > > different, but I think you can see what I'm getting at.) Also my patch > > activates the --without-fp on the config, and thats what triggers the > > SPEC file change. > > I don't know whether that would fly on the ARM. I suspect to conform to the > ATPCS we will need a version where soft, hard and no floating point options > would be available. Such a change as you describe would preclude that. As for > what I would change in the specs file, the linux-elf version automatically links > in libfloat. If we have the soft-float entry points in libgcc, is this really > necessary and is it likely to cause problems, link warnings, etc. I think I stated that poorly. What I ment to say was that if during the config of gcc you passed in --without-fp, the spec files default would change as I noted above. If you did not pass in a --without-fp, then the default would be used. I understand a version for soft and hard, but what is a no floating point version? On the PowerPC the default version is "hard" while a version compiled into the "nof" directory (MULTILIB in the t- makefiles) is for the soft-float. THe -lfloat in the current spec file should be removed. TO use the internal floating point. If it's not removed linking problems occur. > I don't know how many people actually use these routines as well, so I'm not > convinced they work correctly. It looks like someone has set up the code to > handle the ARM doubles, and little endian/big endian troubles. The code makes > liberal use of structs, and that always makes me nervous, given the alignment > constraints on structs. I honestly don't know anything about the alignment on ARM. Certain PPC cores are required to be 8-byte aligned, things like the stack must be 16-byte aligned. Most of these cores do not have FPUs, and this soft floating point has been used successfully on them. > ... This requires testing. I agree 100%! --Mark unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 13 16:39:24 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131sm9-0004Qj-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 13 Jun 2000 16:38:49 +0100 Received: from [198.232.147.16] (helo=mail01-oak.pilot.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131sm8-0004Qa-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 13 Jun 2000 16:38:48 +0100 Received: from Altera.COM (mail.altera.com [137.57.1.1]) by mail01-oak.pilot.net with ESMTP id IAA22919 for ; Tue, 13 Jun 2000 08:38:48 -0700 (PDT) Received: from sj-gw01.altera.com by Altera.COM (8.8.8+Sun/SMI-4.1) id IAA14566; Tue, 13 Jun 2000 08:40:30 -0700 (PDT) Received: by sj-gw01.altera.com with Internet Mail Service (5.5.2650.21) id ; Tue, 13 Jun 2000 08:40:13 -0700 Message-ID: <82B1591A280FD411B91200D0B75B484A38864C@sis-uk-msg01.altera.com> From: Clive Davies To: linux-arm-kernel@lists.arm.linux.org.uk Subject: ARM720 problem? Date: Tue, 13 Jun 2000 08:37:03 -0700 X-Mailer: Internet Mail Service (5.5.2650.21) Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk The following appeared in the anouncement of the 2.4.0-test1-ac7-rmk1 path release. Does anyone know if this was also a problem in 2.3.99-pre6-rmk1? If so, any chance of a couple of pointers to the bits of code I need to change (I don't really understand the description below, I'm afraid. TIA Clive >Things to watch out for: > >* The ARM720 data abort code needs fixing up wrt the register > optimisations (r1 -> r3, r2 -> r1) unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 13 17:34:08 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131tbb-0005Rx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 13 Jun 2000 17:31:59 +0100 Received: from [202.141.26.181] (helo=ee.iitm.ernet.in ident=ee96162) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131tbX-0005RO-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 13 Jun 2000 17:31:57 +0100 Received: from localhost (ee96162@localhost) by ee.iitm.ernet.in (8.9.3/8.9.3) with ESMTP id WAA18587; Tue, 13 Jun 2000 22:03:20 +0530 Date: Tue, 13 Jun 2000 22:03:20 +0530 (IST) From: Lavu Sridhar Reply-To: ee96162@ee.iitm.ernet.in To: Jason Gunthorpe cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: telnet on cirrus board Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I have installed the telnet package (debian) and am able to telnet from the Cirrus Board to the outside. However, i keep getting this error message on the screen: nfs_refresh_inode: inode number mismatch expected (0x303/0x6410e), got (0x303/0x41079) I do not understand why it is coming and how to rectify this problem. I will try to install the telnetd package and see if I am able to avoid the 'respawning too fast' problem and telnet to the Cirrus Board. Regards, Lavu Sridhar --------------------------------------------------------------------- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 13 18:30:01 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131uUJ-0006ZJ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 13 Jun 2000 18:28:31 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131uUH-0006Z9-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 13 Jun 2000 18:28:29 +0100 Received: from CBRAKE (akron-80-155.ar51.net [208.225.80.155]) by ice.ar51.net (8.9.3/8.8.7) with SMTP id NAA15051; Tue, 13 Jun 2000 13:28:29 -0400 From: "Cliff Brake" To: , "'Clive Davies'" Subject: RE: ARM720 problem? Date: Tue, 13 Jun 2000 13:27:23 -0400 Message-ID: <001501bfd55c$a3499800$1c00a8c0@CBRAKE> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook CWS, Build 9.0.2416 (9.0.2910.0) Importance: Normal In-Reply-To: <82B1591A280FD411B91200D0B75B484A38864C@sis-uk-msg01.altera.com> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk >The following appeared in the anouncement of the >2.4.0-test1-ac7-rmk1 path >release. Does anyone know if this was also a problem in >2.3.99-pre6-rmk1? If >so, any chance of a couple of pointers to the bits of code I >need to change >(I don't really understand the description below, I'm afraid. > >TIA >Clive > >>Things to watch out for: >> >>* The ARM720 data abort code needs fixing up wrt the register >> optimisations (r1 -> r3, r2 -> r1) Yes this does apply to pre6 also. I think the following patch implements the required fix. I do not have my ARM720 system completely up yet as it is having some problems running init after it loads it from the ramdisk -- baddataabort() errors -- so there might still be some problems. Cliff *********************************** --- linux/arch/arm/mm/proc-arm720.S Thu Jun 1 15:47:59 2000 +++ linux-2.4.0-test1-ac6-rmk1-cpb2_hack/arch/arm/mm/proc-arm720.S Fri Jun 9 08:11:04 2000 @@ -121,8 +121,9 @@ addeq r7, r0, r7, lsl #2 @ Do correction (signed) Ldata_saver7: str r7, [sp, r5, lsr #14] @ Put register Ldata_simple: mrc p15, 0, r0, c6, c0, 0 @ get FAR - mrc p15, 0, r1, c5, c0, 0 @ get FSR - and r1, r1, #15 + mrc p15, 0, r3, c5, c0, 0 @ get FSR + and r3, r3, #15 + mov r1, r2 mov pc, lr ENTRY(cpu_arm720_data_abort) unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 13 18:31:28 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131uX6-0006eb-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 13 Jun 2000 18:31:24 +0100 Received: from [194.168.180.16] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131uX4-0006eO-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 13 Jun 2000 18:31:23 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id SAA01371; Tue, 13 Jun 2000 18:31:02 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id SAA02555; Tue, 13 Jun 2000 18:30:07 +0100 Message-Id: <200006131730.SAA02555@raistlin.arm.linux.org.uk> Subject: Re: telnet on cirrus board To: ee96162@ee.iitm.ernet.in Date: Tue, 13 Jun 2000 18:30:07 +0100 (BST) Cc: jgg@ualberta.ca (Jason Gunthorpe), linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Lavu Sridhar" at Jun 13, 2000 10:03:20 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Lavu Sridhar writes: > I have installed the telnet package (debian) and am able to telnet from > the Cirrus Board to the outside. However, i keep getting this error > message on the screen: > > nfs_refresh_inode: inode number mismatch > expected (0x303/0x6410e), got (0x303/0x41079) Heh, same problem I'm seeing. Let me guess - you're running a Linux knfsd server? The problem comes when you boot a machine which has no way to obtain the current time. In this case, it always boots up with the same time, which means that the NFS requests are always seeded with the same transaction id (or xid). Unfortunately, the Linux knfsd server keeps a persistent cache of the last 1024 replies, and only uses the client address, xid and procedure to validate their use. The fix is lovely - when you reboot the client, make sure you reboot the NFS server at the same time! If you've got knfsd compiled as a module, you can shut down the knfsd processes, unload the module, reload it, and restart the knfsd processes to achieve the same effect. Sick, isn't it? ;( _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 13 18:33:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 131uZF-0006j4-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 13 Jun 2000 18:33:37 +0100 Received: from [194.168.180.16] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 131uZD-0006ih-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 13 Jun 2000 18:33:36 +0100 Received: from raistlin.arm.linux.org.uk (root@raistlin [192.168.0.3]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id SAA01384; Tue, 13 Jun 2000 18:32:50 +0100 From: Russell King - ARM Linux Admin Received: (from linux@localhost) by raistlin.arm.linux.org.uk (8.7.4/8.7.3) id SAA02569; Tue, 13 Jun 2000 18:31:56 +0100 Message-Id: <200006131731.SAA02569@raistlin.arm.linux.org.uk> Subject: Re: ARM720 problem? To: cbrake@accelent.com (Cliff Brake) Date: Tue, 13 Jun 2000 18:31:55 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk, cdavies@altera.com ('Clive Davies') In-Reply-To: <001501bfd55c$a3499800$1c00a8c0@CBRAKE> from "Cliff Brake" at Jun 13, 2000 01:27:23 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Cliff Brake writes: > Yes this does apply to pre6 also. I think the following patch > implements the required fix. I do not have my ARM720 system completely > up yet as it is having some problems running init after it loads it from > the ramdisk -- baddataabort() errors -- so there might still be some > problems. Right area, probably works, but not what I was expecting. It's fixed in the lastest and greatest 2.4.0test1-ac14-rmk1. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 07:16:54 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1326S5-0000H4-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 07:15:01 +0100 Received: from [211.44.193.250] (helo=www.palmpalm.com ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1326S4-0000Gs-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 07:15:00 +0100 Received: from hmjung ([211.44.193.210]) by www.palmpalm.com (8.9.3/8.8.7) with SMTP id PAA25829 for ; Wed, 14 Jun 2000 15:14:20 +0900 Message-ID: <005001bfd5c8$05b09fa0$d2c12cd3@co.kr> From: "Hyun-Mo Chung" To: Subject: Use of __KERNEL_SYSCALLS__ ? Date: Wed, 14 Jun 2000 15:16:05 +0900 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6600 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi, I'm porting USB Hub driver from 2.3.99 to rmk series of 2.2.2. The driver calls kernel_thread(), which never causes any trouble in 2.3.99, but does in 2.2.2. The problem is, in 2.2.2, that the macro definition of kernel_thread() is enclosed by #ifdef __KERNEL_SYSCALLS__ #endif pair in include/asm-arm/unistd.h and so a reference to kernel_thread() is not resolved. I have no idea what the __KERNEL_SYSCALLS__ is for and whether #defining it in the Hub driver before #including unistd.h is a proper solution to the problem. (Yes! I'm a newbie to the Kernel World. :-) Thanks in advance. Hyun-Mo unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 13:30:51 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132CHM-000542-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 13:28:20 +0100 Received: from [210.230.34.106] (helo=so-fs.musicpod.co.jp) by www.linux.org.uk with smtp (Exim 3.13 #1) id 132CHL-00053v-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 13:28:19 +0100 Received: (qmail 12286 invoked from network); 14 Jun 2000 12:28:19 -0000 Received: from bak.vasara.co.jp (imai@192.168.20.13) by so-fs.musicpod.co.jp with SMTP; 14 Jun 2000 12:28:19 -0000 Date: Wed, 14 Jun 2000 21:28:19 +0900 Message-ID: From: Kunihiko IMAI To: linux-arm-kernel@lists.arm.linux.org.uk Subject: USB driver (was: Use of __KERNEL_SYSCALLS__ ?) In-Reply-To: In your message of "Wed, 14 Jun 2000 15:16:05 +0900" <005001bfd5c8$05b09fa0$d2c12cd3@co.kr> References: <005001bfd5c8$05b09fa0$d2c12cd3@co.kr> User-Agent: Wanderlust/1.1.1 (Purple Rain) WEMI/1.13.7 (Shimada) CLIME/1.13.6 (=?ISO-2022-JP?B?GyRCQ2YlTj4xGyhC?=) MULE XEmacs/21.1 (patch 6) (Big Bend) (i386-vine-linux) MIME-Version: 1.0 (generated by WEMI 1.13.7 - "Shimada") Content-Type: text/plain; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, At Wed, 14 Jun 2000 15:16:05 +0900, Hyun-Mo Chung wrote: > I'm porting USB Hub driver from 2.3.99 to rmk series of 2.2.2. What is the CPU or architecture? I am interested in. Because I am also porting USB OHCI driver for SA1111 USB interface. I had suffered from cache coherency problem. It worked with 'nocache' kernel command line option , but did not work without the option. But, just now, it has worked with memory cache enabled. Somewhat buggy at high layer, but it seems to work at OHCI layer. I'm very very happy now :-) Thanks. _._. __._ _.. . .___ ._. _____ _... ._ _._ _.._. .____ .__. ... . _._ Kunihiko IMAI unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 16:39:58 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132FEn-0007eW-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 16:37:53 +0100 Received: from [139.175.250.58] (helo=mail2000.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 132FEl-0007dr-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 16:37:53 +0100 Received: By OpenMail Mailer;Wed, 14 Jun 2000 23:45:41 +0800 (CST) From: "Chien-Min Lee" Reply-To: cm_lee@mail2000.com.tw Subject: Re: USB driver Message-ID: <960997541.42975.cm_lee@mail2000.com.tw> To: "imai" Cc: "linux-arm-kernel" Date: Wed, 14 Jun 2000 23:45:41 +0800 (CST) MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 8bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi: >What is the CPU or architecture? I am interested in. >Because I am also porting USB OHCI driver for SA1111 USB >interface. Do you plan to open release this OHCI layer driver for SA1111 ?? BTW, which kernel version do you use?? Have a nice day! kirk ----== Mailed via Openfind ==----- http://mail2000.com.tw/ ´£¨Ñ§K¶Oªº e-mail «H½cªA°È unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 16:57:32 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132FVz-0007uE-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 16:55:39 +0100 Received: from [212.19.67.123] (helo=gate.tao-group.com ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 132FVy-0007u7-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 16:55:38 +0100 Received: from mithril.tao.co.uk (mithril.tao.co.uk [172.16.100.96]) by gate.tao-group.com (8.9.3/8.9.3) with ESMTP id QAA30358 for ; Wed, 14 Jun 2000 16:55:41 +0100 Received: from mithril.tao.co.uk (IDENT:ach@mithril [127.0.0.1]) by mithril.tao.co.uk (8.9.3/8.8.7) with ESMTP id QAA06427 for ; Wed, 14 Jun 2000 16:56:15 +0100 Message-Id: <200006141556.QAA06427@mithril.tao.co.uk> To: linux-arm-kernel@lists.arm.linux.org.uk Subject: GCC/Binutils versions Date: Wed, 14 Jun 2000 16:56:15 +0100 From: Andy Hayward Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Russell, Phil, Is there a list giving the versions of binutils/gcc needed to build the various ARM linux-2.2.x kernels? For the linux-7110 we're currently using binutils-2.9.1.0.15/egcs-1.1b and my naive attempt to use binutils-2.9.5/gcc-2.95.2 failed miserably. ach unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 19:50:11 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132IAL-0001kE-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 19:45:29 +0100 Received: from [194.168.180.65] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 132IAI-0001k8-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 19:45:27 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id TAA04321; Wed, 14 Jun 2000 19:30:28 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id TAA09879; Wed, 14 Jun 2000 19:29:32 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006141829.TAA09879@flint.arm.linux.org.uk> Subject: Re: Use of __KERNEL_SYSCALLS__ ? To: hmjung@palmpalm.com (Hyun-Mo Chung) Date: Wed, 14 Jun 2000 19:29:32 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <005001bfd5c8$05b09fa0$d2c12cd3@co.kr> from "Hyun-Mo Chung" at Jun 14, 2000 03:16:05 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hyun-Mo Chung writes: > I'm porting USB Hub driver from 2.3.99 to rmk series of 2.2.2. > The driver calls kernel_thread(), which never causes any trouble in 2.3.99, > but does in 2.2.2. Why do you use a known buggy version of the 2.2 series of kernels? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 14 22:12:26 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132KRf-00049s-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 14 Jun 2000 22:11:31 +0100 Received: from [24.229.50.7] (helo=cyberiansoftware.com) by www.linux.org.uk with smtp (Exim 3.13 #1) id 132KRd-00049i-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 14 Jun 2000 22:11:29 +0100 Received: (qmail 4331 invoked from network); 14 Jun 2000 21:11:28 -0000 Received: from other.cyberiansoftware.com (HELO cyberiansoftware.com) (192.168.1.3) by cyberiansoftware.com with SMTP; 14 Jun 2000 21:11:28 -0000 Message-ID: <3947F51D.5F7BA50A@cyberiansoftware.com> Date: Wed, 14 Jun 2000 17:11:57 -0400 From: Tom Walsh X-Mailer: Mozilla 4.73 [en] (X11; U; Linux 2.2.12-20 i686) X-Accept-Language: en MIME-Version: 1.0 To: linux-arm-kernel Subject: Getting setup: CS7500 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, The Three Stooges at NuHorizons finally messed up and delivered my Cirrus Logic EDB-7500 eval board, but, they still had a final zinger to lay on me, no Linux development stuff at all on the CDROM. I had spoken with an engineer over at Cirrus and he told me that there would be a copy on the CDROM, well, somebody goofed. Any suggestions, webpages, etc. that I can start reffering to so I can get a GNUC development environment setup, and snag a copy of the current ARM720 kernel sources? TIA TomW -- Tom Walsh - WN3L - Embedded Systems Consultant 'www.openhardware.net', 'www.cyberiansoftware.com' "Windows? No thanks, I have work to do..." unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 15 01:45:37 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132Nks-0006xg-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 15 Jun 2000 01:43:34 +0100 Received: from [211.44.193.250] (helo=www.palmpalm.com ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 132Nkr-0006xX-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 15 Jun 2000 01:43:33 +0100 Received: from hmjung ([211.44.193.210]) by www.palmpalm.com (8.9.3/8.8.7) with SMTP id JAA31238 for ; Thu, 15 Jun 2000 09:42:48 +0900 Message-ID: <005e01bfd662$e3072aa0$d2c12cd3@co.kr> From: "Hyun-Mo Chung" To: References: <005001bfd5c8$05b09fa0$d2c12cd3@co.kr> Subject: Re: USB driver (was: Use of __KERNEL_SYSCALLS__ ?) Date: Thu, 15 Jun 2000 09:44:39 +0900 MIME-Version: 1.0 Content-Type: text/plain; charset="Windows-1252" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6600 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk > What is the CPU or architecture? I am interested in. > Because I am also porting USB OHCI driver for SA1111 USB interface. I'm doing exactly the same work as what you do. > I had suffered from cache coherency problem. > It worked with 'nocache' kernel command line option , but did not work > without the option. I'm in the earliest stage of the work and have not yet gone through such a problem. > But, just now, it has worked with memory cache enabled. > Somewhat buggy at high layer, but it seems to work at OHCI layer. > I'm very very happy now :-) I'm very very envious of you :-) Could you tell me which kernel version you are using ? In my humble opinion, as for OHCI layer, there seems nothing to be done for SA1111 because it is strictly compilant with the OHCI specification. Hyun-Mo unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 15 16:23:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132bTl-00062g-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 15 Jun 2000 16:22:49 +0100 Received: from [195.108.189.102] (helo=white.plant.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 132bTk-00062a-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 15 Jun 2000 16:22:48 +0100 Received: from [194.134.11.51] by white.plant.nl (NTMail 4.30.0013/NT3720.00.67541ba1) with ESMTP id tghxfaaa for ; Thu, 15 Jun 2000 17:22:11 +0200 Message-ID: <3948F404.FC6F8FA6@maycom.nl> Date: Thu, 15 Jun 2000 17:19:32 +0200 From: Daven Baptist X-Mailer: Mozilla 4.7 [en] (X11; I; Linux 2.2.13 i686) X-Accept-Language: en MIME-Version: 1.0 CC: "Linux ARM kernel mailing list (E-mail)" Subject: a device driver problem References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk hello, i hope some one can help me with a device driver. i build a device driver to drive a custom display, but how can i pipe a char in to the device driver? i like to use somthing like: echo 12345> /dev/mlcd0 can some one point me in the richt direction ? Daven unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 15 17:41:09 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132chB-0007GJ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 15 Jun 2000 17:40:45 +0100 Received: from [128.100.10.203] (helo=pyxis.lrtw.org ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 132chA-0007G9-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 15 Jun 2000 17:40:44 +0100 Received: from localhost (tangf@localhost) by pyxis.lrtw.org (8.9.3/8.9.3/Debian 8.9.3-6) with SMTP id MAA26596; Thu, 15 Jun 2000 12:40:38 -0400 Date: Thu, 15 Jun 2000 12:40:38 -0400 (EDT) From: Felix Tang X-Sender: tangf@pyxis.lrtw.org To: Daven Baptist cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: a device driver problem In-Reply-To: <3948F404.FC6F8FA6@maycom.nl> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Try looking in this doc... http://www.linuxdoc.org/LDP/lkmpg/mpg.html Felix On Thu, 15 Jun 2000, Daven Baptist wrote: > hello, > > i hope some one can help me with a device driver. > > i build a device driver to drive a custom display, > but how can i pipe a char in to the device driver? > > i like to use somthing like: > > echo 12345> /dev/mlcd0 > > can some one point me in the richt direction ? > > Daven > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 16 08:28:58 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132qX7-0002lY-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 16 Jun 2000 08:27:17 +0100 Received: from [210.230.34.106] (helo=so-fs.musicpod.co.jp) by www.linux.org.uk with smtp (Exim 3.13 #1) id 132qX6-0002lS-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 16 Jun 2000 08:27:16 +0100 Received: (qmail 10813 invoked from network); 16 Jun 2000 07:27:13 -0000 Received: from bak.vasara.co.jp (imai@192.168.20.13) by so-fs.musicpod.co.jp with SMTP; 16 Jun 2000 07:27:13 -0000 Date: Fri, 16 Jun 2000 16:27:13 +0900 Message-ID: From: Kunihiko IMAI To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: USB driver In-Reply-To: In your message of "Wed, 14 Jun 2000 23:45:41 +0800 (CST)" <960997541.42975.cm_lee@mail2000.com.tw> References: <960997541.42975.cm_lee@mail2000.com.tw> User-Agent: Wanderlust/1.1.1 (Purple Rain) WEMI/1.13.7 (Shimada) CLIME/1.13.6 (=?ISO-2022-JP?B?GyRCQ2YlTj4xGyhC?=) MULE XEmacs/21.1 (patch 6) (Big Bend) (i386-vine-linux) MIME-Version: 1.0 (generated by WEMI 1.13.7 - "Shimada") Content-Type: text/plain; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, At Wed, 14 Jun 2000 23:45:41 +0800 (CST), Chien-Min Lee wrote: > Do you plan to open release this OHCI layer driver > for SA1111 ?? I can openly release the patch. But the solution of avoiding cache problem is too silly or stupid. That is, call flush_cache_all() many time. It needs more blush-up, I think. But if you want it, I can send it. And, I got a mail personally. He wrote that he made a patch for SA110, 21285 footbridge with a PCI connected USB controller. It has a set of original memory allocation routines. I tried it, but it failed at memory allocation layer. > BTW, which kernel version do you use?? I use a linux-2.3.99-pre3-rmk1-np2 source tree. > Have a nice day! Have a nice hack! _._. __._ _.. . .___ ._. _____ _... ._ _._ _.._. .____ .__. ... . _._ Kunihiko IMAI unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 16 14:57:22 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 132wc0-0007V0-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 16 Jun 2000 14:56:44 +0100 Received: from [139.175.250.58] (helo=mail2000.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 132wbz-0007UY-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 16 Jun 2000 14:56:44 +0100 Received: By OpenMail Mailer;Fri, 16 Jun 2000 21:56:03 +0800 (CST) From: "Chien-Min Lee" Reply-To: cm_lee@mail2000.com.tw Subject: Re: USB driver Message-ID: <961163763.10837.cm_lee@mail2000.com.tw> To: "imai" Cc: "linux-arm-kernel" Date: Fri, 16 Jun 2000 21:56:03 +0800 (CST) MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 8bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi: >I can openly release the patch. But the solution of avoiding cache >problem is too silly or stupid. That is, call flush_cache_all() many >time. It needs more blush-up, I think. >But if you want it, I can send it. Really!! Yes, I want to try it. Thanks in advance!! Have a nice day! kirk ----== Mailed via Openfind ==----- http://mail2000.com.tw/ ´£¨Ñ§K¶Oªº e-mail «H½cªA°È unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 16 21:03:26 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1332K0-00041c-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 16 Jun 2000 21:02:32 +0100 Received: from [194.168.180.28] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1332Jo-00040w-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 16 Jun 2000 21:02:27 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id UAA04599; Fri, 16 Jun 2000 20:26:28 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id UAA22743; Fri, 16 Jun 2000 20:25:29 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006161925.UAA22743@flint.arm.linux.org.uk> Subject: Re: USB driver To: imai@vasara.co.jp (Kunihiko IMAI) Date: Fri, 16 Jun 2000 20:25:29 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: from "Kunihiko IMAI" at Jun 16, 2000 04:27:13 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Kunihiko IMAI writes: > I can openly release the patch. But the solution of avoiding cache > problem is too silly or stupid. That is, call flush_cache_all() many > time. It needs more blush-up, I think. You'd be a lot better off implementing this under late 2.3 or 2.4 kernels because they support uncached memory for this type of device communication. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 16 22:04:31 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1333Hn-0004wN-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 16 Jun 2000 22:04:19 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1333Hm-0004wE-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 16 Jun 2000 22:04:18 +0100 Received: from highwind.crl.dec.com (highwind.crl.dec.com [16.11.0.177]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id RAA14756; Fri, 16 Jun 2000 17:04:04 -0400 (EDT) Received: (from jamey@localhost) by highwind.crl.dec.com (8.9.3/RWD-8.8.8) id RAA06343; Fri, 16 Jun 2000 17:04:04 -0400 Date: Fri, 16 Jun 2000 17:04:04 -0400 Message-Id: <200006162104.RAA06343@highwind.crl.dec.com> X-Authentication-Warning: highwind.crl.dec.com: jamey set sender to jamey@highwind.crl.dec.com using -f From: James Hicks To: linux-arm-kernel@lists.arm.linux.org.uk Cc: alan@lxorguk.ukuu.org.uk Subject: Re: USB driver Reply-to: jamey@crl.dec.com Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Here is a patch to usb-ohci.[ch] that fixes the cache coherence problem using pci_alloc_consistent. We have tested this on the Compaq Personal Server (SA110/footbridge/Opti USB OHCI) with 2.4.0-test1-ac7-rmk1. Thanks to my summer intern, Brandon Sanders, for testing the latest version we have working. We haven't tested it against x86 yet, but Brandon is working on it. There still seems to be a problem when unplugging devices -- causing the kernel to crash. I assume this doesn't happen on x86. -Jamey Common subdirectories: linux-2.4.0-test1-ac7/drivers/usb/serial and linux-2.4.0-test1-ac7-crl1/drivers/usb/serial diff -u linux-2.4.0-test1-ac7/drivers/usb/usb-ohci.c linux-2.4.0-test1-ac7-crl1/drivers/usb/usb-ohci.c --- linux-2.4.0-test1-ac7/drivers/usb/usb-ohci.c Fri Jun 16 13:29:47 2000 +++ linux-2.4.0-test1-ac7-crl1/drivers/usb/usb-ohci.c Fri Jun 16 16:54:42 2000 @@ -70,6 +70,135 @@ static DECLARE_WAIT_QUEUE_HEAD (op_wakeup); static LIST_HEAD (ohci_hcd_list); static spinlock_t usb_ed_lock = SPIN_LOCK_UNLOCKED; +static spinlock_t usb_td_lock = SPIN_LOCK_UNLOCKED; + + + +#ifdef CONFIG_ARCH_PERSONAL_SERVER +/*-------------------------------------------------------------------------* + * Code to check addresses are in range + *-------------------------------------------------------------------------*/ + +static int __verify_virt(volatile void *virt, int line) +{ + unsigned long addr = (unsigned long)virt; + int ok = ((addr & 0xF0000000) == 0xC0000000); + if (!ok) + printk(KERN_CRIT __FILE__ ":%d: Expected virtual address, got %p\n", line, virt); + return ok; +} +static int __verify_dma(dma_addr_t dma, int line) +{ + unsigned long addr = (unsigned long)dma; + int ok = ((addr & 0xF0000000) == 0xE0000000); + if (!ok) + printk(KERN_CRIT __FILE__ ":%d: Expected DMA address, got %#08x\n", line, dma); + return ok; +} +#define verify_virt(virt) __verify_virt((virt), __LINE__) +#define verify_dma(dma) __verify_dma((dma), __LINE__) + +#else /* !CONFIG_ARCH_PERSONAL_SERVER, address range macros are empty */ + +#define verify_virt(virt) +#define verify_dma(dma) + +#endif /* !CONFIG_ARCH_PERSONAL_SERVER */ + +/*-------------------------------------------------------------------------* + * Consistent allocation functions + *-------------------------------------------------------------------------*/ + +static int ohci_consistent_storage_create(struct pci_dev *pdev, struct ohci_consistent_storage *ocsp) +{ + int num_eds = NUM_EDS * 256; + int num_tds = 2048; + int i; + char *storage = NULL; + if (ocsp == NULL) + return -EINVAL; + storage = ocsp->virt = pci_alloc_consistent(pdev, + sizeof(ed_t) * num_eds + sizeof(td_t) * num_tds, + &ocsp->dma_addr); + if (storage == NULL) + return -ENOMEM; + ocsp->edlist = NULL; + for (i = 0; i < num_eds; i++) { + ed_t *ed = (ed_t *)(storage + i * sizeof(ed_t)); + *(ed_t **)ed = ocsp->edlist; + ocsp->edlist = ed; + ed->state = ED_NEW; + } + ocsp->tdlist = NULL; + for (i = 0; i < num_tds; i++) { + td_t *td = (td_t *)(storage + i * sizeof(td_t)); + *(td_t **)td = ocsp->tdlist; + ocsp->tdlist = td; + } + + ocsp->virt_to_dma = (long)ocsp->dma_addr - (long)ocsp->virt; + return 0; +} + +static ed_t *ohci_alloc_consistent_ed(struct ohci *ohci) +{ + int flags = 0; + ed_t *ed; + if (!in_interrupt()) spin_lock_irqsave (&usb_ed_lock, flags); + ed = ohci->consistent_storage.edlist; + ohci->consistent_storage.edlist = *(ed_t **)ed; + if (!in_interrupt()) spin_unlock_irqrestore (&usb_ed_lock, flags); + if (0) printk(KERN_CRIT "alloc_consistent_ed: ed=%p\n", ed); + return ed; +} + +static void ohci_free_consistent_ed(struct ohci *ohci, ed_t *ed) +{ + int flags = 0; + if (!in_interrupt()) spin_lock_irqsave (&usb_ed_lock, flags); + *(ed_t **)ed = ohci->consistent_storage.edlist; + ohci->consistent_storage.edlist = ed; + if (!in_interrupt()) spin_unlock_irqrestore (&usb_ed_lock, flags); +} + +static td_t *ohci_alloc_consistent_td(struct ohci *ohci) +{ + int flags = 0; + td_t *td; + if (!in_interrupt()) spin_lock_irqsave (&usb_td_lock, flags); + td = ohci->consistent_storage.tdlist; + ohci->consistent_storage.tdlist = *(td_t **)td; + if (!in_interrupt()) spin_unlock_irqrestore (&usb_td_lock, flags); + verify_virt(td); + if (0) printk(KERN_CRIT "alloc_consistent_td: td=%p\n", td); + return td; +} + +static void ohci_free_consistent_td(struct ohci *ohci, td_t *td) +{ + int flags = 0; + verify_virt(td); + if (!in_interrupt()) spin_lock_irqsave (&usb_td_lock, flags); + *(td_t **)td = ohci->consistent_storage.tdlist; + ohci->consistent_storage.tdlist = td; + if (!in_interrupt()) spin_unlock_irqrestore (&usb_td_lock, flags); +} + +static dma_addr_t ohci_virt_to_bus(struct ohci *ohci, volatile void *obp) +{ + dma_addr_t dma = (long)(obp + ohci->consistent_storage.virt_to_dma); + verify_virt(obp); + verify_dma(dma); + return dma; +} + +static void *ohci_bus_to_virt(struct ohci *ohci, dma_addr_t dma_addr) +{ + void *virt = (void *)(dma_addr - ohci->consistent_storage.virt_to_dma); + verify_virt(virt); + verify_dma(dma_addr); + return virt; +} /*-------------------------------------------------------------------------* * URB support functions @@ -79,6 +208,7 @@ static void urb_rm_priv (urb_t * urb) { + struct ohci *ohci = (ohci_t *) urb->dev->bus->hcpriv; urb_priv_t * urb_priv = urb->hcpriv; int i; @@ -86,7 +216,7 @@ for (i = 0; i < urb_priv->length; i++) { if (urb_priv->td [i]) { - OHCI_FREE (urb_priv->td [i]); + OHCI_FREE_TD (ohci, urb_priv->td [i]); } } kfree (urb->hcpriv); @@ -150,12 +280,12 @@ __u32 * ed_p; for (i= 0; i < 32; i++) { j = 5; - ed_p = &(ohci->hcca.int_table [i]); + ed_p = &(ohci->hcca->int_table [i]); if (*ed_p == 0) continue; printk (KERN_DEBUG __FILE__ ": %s branch int %2d(%2x):", str, i, i); while (*ed_p != 0 && j--) { - ed_t *ed = (ed_t *) bus_to_virt(le32_to_cpup(ed_p)); + ed_t *ed = (ed_t *) ohci_bus_to_virt(ohci, le32_to_cpup(ed_p)); printk (" ed: %4x;", ed->hwINFO); ed_p = &ed->hwNextED; } @@ -313,7 +443,7 @@ ohci_dump_status (controller); if (verbose) ep_print_int_eds (controller, "hcca"); - dbg ("hcca frame #%04x", controller->hcca.frame_no); + dbg ("hcca frame #%04x", controller->hcca->frame_no); ohci_dump_roothub (controller, 1); } @@ -472,7 +602,7 @@ /* allocate the TDs */ for (i = 0; i < size; i++) { - OHCI_ALLOC (urb_priv->td[i], sizeof (td_t)); + OHCI_ALLOC_TD (ohci, urb_priv->td[i]); if (!urb_priv->td[i]) { usb_dec_dev_use (urb->dev); urb_rm_priv (urb); @@ -490,7 +620,7 @@ /* for ISOC transfers calculate start frame index */ if (urb->transfer_flags & USB_ISO_ASAP) { urb->start_frame = ((ed->state == ED_OPER)? (ed->last_iso + 1): - (le16_to_cpu (ohci->hcca.frame_no) + 10)) & 0xffff; + (le16_to_cpu (ohci->hcca->frame_no) + 10)) & 0xffff; } urb->status = USB_ST_URB_PENDING; urb->actual_length = 0; @@ -574,6 +704,8 @@ static int sohci_alloc_dev (struct usb_device *usb_dev) { + int i; + struct ohci *ohci = usb_dev->bus->hcpriv; struct ohci_device * dev; dev = kmalloc (sizeof (*dev), GFP_KERNEL); @@ -582,6 +714,12 @@ memset (dev, 0, sizeof (*dev)); + for (i = 0; i < NUM_EDS; i++) { + dev->ed[i] = ohci_alloc_consistent_ed(ohci); + if (dev->ed[i] == NULL) + return -ENOMEM; + } + usb_dev->hcpriv = dev; return 0; @@ -607,13 +745,14 @@ /* delete all TDs of all EDs */ spin_lock_irqsave (&usb_ed_lock, flags); for(i = 0; i < NUM_EDS; i++) { - ed = &(dev->ed[i]); + ed = dev->ed[i]; if (ed->state != ED_NEW) { if (ed->state == ED_OPER) ep_unlink (ohci, ed); ep_rm_ed (usb_dev, ed); ed->state = ED_DEL; cnt++; } + ohci_free_consistent_ed(ohci, ed); } spin_unlock_irqrestore (&usb_ed_lock, flags); @@ -637,7 +776,7 @@ { ohci_t * ohci = usb_dev->bus->hcpriv; - return le16_to_cpu (ohci->hcca.frame_no); + return le16_to_cpu (ohci->hcca->frame_no); } /*-------------------------------------------------------------------------*/ @@ -718,9 +857,9 @@ case CTRL: ed->hwNextED = 0; if (ohci->ed_controltail == NULL) { - writel (virt_to_bus (ed), &ohci->regs->ed_controlhead); + writel (ohci_virt_to_bus (ohci, ed), &ohci->regs->ed_controlhead); } else { - ohci->ed_controltail->hwNextED = cpu_to_le32 (virt_to_bus (ed)); + ohci->ed_controltail->hwNextED = cpu_to_le32 (ohci_virt_to_bus (ohci, ed)); } ed->ed_prev = ohci->ed_controltail; ohci->ed_controltail = edi; @@ -729,9 +868,9 @@ case BULK: ed->hwNextED = 0; if (ohci->ed_bulktail == NULL) { - writel (virt_to_bus (ed), &ohci->regs->ed_bulkhead); + writel (ohci_virt_to_bus (ohci, ed), &ohci->regs->ed_bulkhead); } else { - ohci->ed_bulktail->hwNextED = cpu_to_le32 (virt_to_bus (ed)); + ohci->ed_bulktail->hwNextED = cpu_to_le32 (ohci_virt_to_bus (ohci, ed)); } ed->ed_prev = ohci->ed_bulktail; ohci->ed_bulktail = edi; @@ -746,12 +885,12 @@ for (i = 0; i < ep_rev (6, interval); i += inter) { inter = 1; - for (ed_p = &(ohci->hcca.int_table[ep_rev (5, i) + int_branch]); - (*ed_p != 0) && (((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->int_interval >= interval); - ed_p = &(((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->hwNextED)) - inter = ep_rev (6, ((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->int_interval); + for (ed_p = &(ohci->hcca->int_table[ep_rev (5, i) + int_branch]); + (*ed_p != 0) && (((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->int_interval >= interval); + ed_p = &(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->hwNextED)) + inter = ep_rev (6, ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->int_interval); ed->hwNextED = *ed_p; - *ed_p = cpu_to_le32 (virt_to_bus (ed)); + *ed_p = cpu_to_le32 (ohci_virt_to_bus (ohci, ed)); } #ifdef DEBUG ep_print_int_eds (ohci, "LINK_INT"); @@ -762,16 +901,16 @@ ed->hwNextED = 0; ed->int_interval = 1; if (ohci->ed_isotail != NULL) { - ohci->ed_isotail->hwNextED = cpu_to_le32 (virt_to_bus (ed)); + ohci->ed_isotail->hwNextED = cpu_to_le32 (ohci_virt_to_bus (ohci, ed)); ed->ed_prev = ohci->ed_isotail; } else { for ( i = 0; i < 32; i += inter) { inter = 1; - for (ed_p = &(ohci->hcca.int_table[ep_rev (5, i)]); + for (ed_p = &(ohci->hcca->int_table[ep_rev (5, i)]); *ed_p != 0; - ed_p = &(((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->hwNextED)) - inter = ep_rev (6, ((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->int_interval); - *ed_p = cpu_to_le32 (virt_to_bus (ed)); + ed_p = &(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->hwNextED)) + inter = ep_rev (6, ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->int_interval); + *ed_p = cpu_to_le32 (ohci_virt_to_bus (ohci, ed)); } ed->ed_prev = NULL; } @@ -810,7 +949,7 @@ if(ohci->ed_controltail == ed) { ohci->ed_controltail = ed->ed_prev; } else { - ((ed_t *) bus_to_virt (le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; + ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; } break; @@ -823,7 +962,7 @@ if (ohci->ed_bulktail == ed) { ohci->ed_bulktail = ed->ed_prev; } else { - ((ed_t *) bus_to_virt (le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; + ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; } break; @@ -832,11 +971,11 @@ interval = ed->int_interval; for (i = 0; i < ep_rev (6, interval); i += inter) { - for (ed_p = &(ohci->hcca.int_table[ep_rev (5, i) + int_branch]), inter = 1; + for (ed_p = &(ohci->hcca->int_table[ep_rev (5, i) + int_branch]), inter = 1; (*ed_p != 0) && (*ed_p != ed->hwNextED); - ed_p = &(((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->hwNextED), - inter = ep_rev (6, ((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->int_interval)) { - if(((ed_t *) bus_to_virt (le32_to_cpup (ed_p))) == ed) { + ed_p = &(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->hwNextED), + inter = ep_rev (6, ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->int_interval)) { + if(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p))) == ed) { *ed_p = ed->hwNextED; break; } @@ -853,18 +992,18 @@ if (ohci->ed_isotail == ed) ohci->ed_isotail = ed->ed_prev; if (ed->hwNextED != 0) - ((ed_t *) bus_to_virt (le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; + ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (&ed->hwNextED)))->ed_prev = ed->ed_prev; if (ed->ed_prev != NULL) { ed->ed_prev->hwNextED = ed->hwNextED; } else { for (i = 0; i < 32; i += inter) { inter = 1; - for (ed_p = &(ohci->hcca.int_table[ep_rev (5, i)]); + for (ed_p = &(ohci->hcca->int_table[ep_rev (5, i)]); *ed_p != 0; - ed_p = &(((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->hwNextED)) { - inter = ep_rev (6, ((ed_t *) bus_to_virt (le32_to_cpup (ed_p)))->int_interval); - if(((ed_t *) bus_to_virt (le32_to_cpup (ed_p))) == ed) { + ed_p = &(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->hwNextED)) { + inter = ep_rev (6, ((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p)))->int_interval); + if(((ed_t *) ohci_bus_to_virt (ohci, le32_to_cpup (ed_p))) == ed) { *ed_p = ed->hwNextED; break; } @@ -899,17 +1038,16 @@ spin_lock (&usb_ed_lock); - ed = ed_ret = &(usb_to_ohci (usb_dev)->ed[(usb_pipeendpoint (pipe) << 1) | - (usb_pipecontrol (pipe)? 0: usb_pipeout (pipe))]); + ed = ed_ret = (usb_to_ohci (usb_dev)->ed[(usb_pipeendpoint (pipe) << 1) | (usb_pipecontrol (pipe)? 0: usb_pipeout (pipe))]); if((ed->state & ED_DEL) || (ed->state & ED_URB_DEL)) return NULL; /* pending delete request */ if (ed->state == ED_NEW) { ed->hwINFO = cpu_to_le32 (OHCI_ED_SKIP); /* skip ed */ - OHCI_ALLOC (td, sizeof (*td)); /* dummy td; end of td list for ed */ + OHCI_ALLOC_TD (ohci, td); /* dummy td; end of td list for ed */ if(!td) return NULL; /* out of memory */ - ed->hwTailP = cpu_to_le32 (virt_to_bus (td)); + ed->hwTailP = cpu_to_le32 (ohci_virt_to_bus (ohci, td)); ed->hwHeadP = ed->hwTailP; ed->state = ED_UNLINK; ed->type = usb_pipetype (pipe); @@ -952,7 +1090,7 @@ writel (OHCI_INTR_SF, &ohci->regs->intrstatus); writel (OHCI_INTR_SF, &ohci->regs->intrenable); /* enable sof interrupt */ - frame = le16_to_cpu (ohci->hcca.frame_no) & 0x1; + frame = le16_to_cpu (ohci->hcca->frame_no) & 0x1; ed->ed_rm_list = ohci->ed_rm_list[frame]; ohci->ed_rm_list[frame] = ed; @@ -972,33 +1110,45 @@ /* prepare a TD */ -static void td_fill (unsigned int info, void * data, int len, urb_t * urb, int index) + static void td_fill (struct ohci *ohci, unsigned int info, dma_addr_t data, int len, urb_t * urb, int index) { volatile td_t * td, * td_pt; urb_priv_t * urb_priv = urb->hcpriv; + if (urb_priv == NULL) return; + if (index >= urb_priv->length) { err("internal OHCI error: TD index > length"); return; } td_pt = urb_priv->td [index]; + if (td_pt == NULL) return; + + /* fill the old dummy TD */ - td = urb_priv->td [index] = (td_t *) bus_to_virt (le32_to_cpup (&urb_priv->ed->hwTailP) & 0xfffffff0); + td = urb_priv->td [index] = (td_t *) ohci_bus_to_virt (ohci, le32_to_cpup (&urb_priv->ed->hwTailP) & 0xfffffff0); + if (td == NULL) return; + td->ed = urb_priv->ed; + + if (td->ed == NULL) return; + td->index = index; td->urb = urb; td->hwINFO = cpu_to_le32 (info); + if ((td->ed->type & 3) == PIPE_ISOCHRONOUS) { td->hwCBP = cpu_to_le32 (((!data || !len)? - 0 : virt_to_bus (data)) & 0xFFFFF000); + 0 : data) & 0xFFFFF000); td->ed->last_iso = info & 0xffff; } else { - td->hwCBP = cpu_to_le32 (((!data || !len)? 0 : virt_to_bus (data))); + td->hwCBP = cpu_to_le32 (((!data || !len)? 0 : data)); } - td->hwBE = cpu_to_le32 ((!data || !len )? 0: virt_to_bus (data + len - 1)); - td->hwNextTD = cpu_to_le32 (virt_to_bus (td_pt)); - td->hwPSW [0] = cpu_to_le16 ((virt_to_bus (data) & 0x0FFF) | 0xE000); + td->hwBE = cpu_to_le32 ((!data || !len )? 0: (data + len - 1)); + td->hwNextTD = cpu_to_le32 (ohci_virt_to_bus (ohci, td_pt)); + td->hwPSW [0] = cpu_to_le16 ((data & 0x0FFF) | 0xE000); + td_pt->hwNextTD = 0; td->ed->hwTailP = td->hwNextTD; @@ -1016,10 +1166,15 @@ void * ctrl = urb->setup_packet; void * data = urb->transfer_buffer; int data_len = urb->transfer_buffer_length; + dma_addr_t dma_data = + pci_map_single(ohci->consistent_storage.pci_dev, + data, data_len, usb_pipeout (urb->pipe) ? PCI_DMA_TODEVICE : PCI_DMA_FROMDEVICE); int cnt = 0; __u32 info = 0; unsigned int toggle = 0; + urb_priv->transfer_buffer_dma_addr = dma_data; + /* OHCI handles the DATA-toggles itself, we just use the USB-toggle bits for reseting */ if(usb_gettoggle(urb->dev, usb_pipeendpoint(urb->pipe), usb_pipeout(urb->pipe))) { toggle = TD_T_TOGGLE; @@ -1035,12 +1190,12 @@ info = usb_pipeout (urb->pipe)? TD_CC | TD_DP_OUT : TD_CC | TD_DP_IN ; while(data_len > 4096) { - td_fill (info | (cnt? TD_T_TOGGLE:toggle), data, 4096, urb, cnt); - data += 4096; data_len -= 4096; cnt++; + td_fill (ohci, info | (cnt? TD_T_TOGGLE:toggle), dma_data, 4096, urb, cnt); + dma_data += 4096; data_len -= 4096; cnt++; } info = usb_pipeout (urb->pipe)? TD_CC | TD_DP_OUT : TD_CC | TD_R | TD_DP_IN ; - td_fill (info | (cnt? TD_T_TOGGLE:toggle), data, data_len, urb, cnt); + td_fill (ohci, info | (cnt? TD_T_TOGGLE:toggle), dma_data, data_len, urb, cnt); cnt++; writel (OHCI_BLF, &ohci->regs->cmdstatus); /* start bulk list */ break; @@ -1048,27 +1203,29 @@ case PIPE_INTERRUPT: info = usb_pipeout (urb->pipe)? TD_CC | TD_DP_OUT | toggle: TD_CC | TD_R | TD_DP_IN | toggle; - td_fill (info, data, data_len, urb, cnt++); + td_fill (ohci, info, dma_data, data_len, urb, cnt++); break; case PIPE_CONTROL: info = TD_CC | TD_DP_SETUP | TD_T_DATA0; - td_fill (info, ctrl, 8, urb, cnt++); + td_fill (ohci, info, pci_map_single(ohci->consistent_storage.pci_dev, ctrl, 8, PCI_DMA_TODEVICE), + 8, urb, cnt++); if (data_len > 0) { info = usb_pipeout (urb->pipe)? TD_CC | TD_R | TD_DP_OUT | TD_T_DATA1 : TD_CC | TD_R | TD_DP_IN | TD_T_DATA1; - td_fill (info, data, data_len, urb, cnt++); + td_fill (ohci, info, dma_data, data_len, urb, cnt++); } info = usb_pipeout (urb->pipe)? TD_CC | TD_DP_IN | TD_T_DATA1: TD_CC | TD_DP_OUT | TD_T_DATA1; - td_fill (info, NULL, 0, urb, cnt++); + td_fill (ohci, info, (dma_addr_t)NULL, 0, urb, cnt++); writel (OHCI_CLF, &ohci->regs->cmdstatus); /* start Control list */ break; case PIPE_ISOCHRONOUS: for (cnt = 0; cnt < urb->number_of_packets; cnt++) { - td_fill (TD_CC|TD_ISO | ((urb->start_frame + cnt) & 0xffff), - (__u8 *) data + urb->iso_frame_desc[cnt].offset, + td_fill (ohci, + TD_CC|TD_ISO | ((urb->start_frame + cnt) & 0xffff), + dma_data + urb->iso_frame_desc[cnt].offset, urb->iso_frame_desc[cnt].length, urb, cnt); } break; @@ -1119,9 +1276,9 @@ ((td->index == 0) || (td->index == urb_priv->length - 1)))) { if (tdBE != 0) { if (td->hwCBP == 0) - urb->actual_length = bus_to_virt (tdBE) - urb->transfer_buffer + 1; + urb->actual_length = tdBE - urb_priv->transfer_buffer_dma_addr + 1; else - urb->actual_length = bus_to_virt (tdCBP) - urb->transfer_buffer; + urb->actual_length = tdCBP - urb_priv->transfer_buffer_dma_addr; } } } @@ -1142,11 +1299,11 @@ spin_lock_irqsave (&usb_ed_lock, flags); - td_list_hc = le32_to_cpup (&ohci->hcca.done_head) & 0xfffffff0; - ohci->hcca.done_head = 0; + td_list_hc = le32_to_cpup (&ohci->hcca->done_head) & 0xfffffff0; + ohci->hcca->done_head = 0; while (td_list_hc) { - td_list = (td_t *) bus_to_virt (td_list_hc); + td_list = (td_t *) ohci_bus_to_virt (ohci, td_list_hc); if (TD_CC_GET (le32_to_cpup (&td_list->hwINFO))) { urb_priv = (urb_priv_t *) td_list->urb->hcpriv; @@ -1190,8 +1347,8 @@ spin_lock_irqsave (&usb_ed_lock, flags); for (ed = ohci->ed_rm_list[frame]; ed != NULL; ed = ed->ed_rm_list) { - tdTailP = bus_to_virt (le32_to_cpup (&ed->hwTailP) & 0xfffffff0); - tdHeadP = bus_to_virt (le32_to_cpup (&ed->hwHeadP) & 0xfffffff0); + tdTailP = ohci_bus_to_virt (ohci, le32_to_cpup (&ed->hwTailP) & 0xfffffff0); + tdHeadP = ohci_bus_to_virt (ohci, le32_to_cpup (&ed->hwHeadP) & 0xfffffff0); edINFO = le32_to_cpup (&ed->hwINFO); td_p = &ed->hwHeadP; @@ -1199,7 +1356,7 @@ urb_t * urb = td->urb; urb_priv_t * urb_priv = td->urb->hcpriv; - td_next = bus_to_virt (le32_to_cpup (&td->hwNextTD) & 0xfffffff0); + td_next = ohci_bus_to_virt (ohci, le32_to_cpup (&td->hwNextTD) & 0xfffffff0); if ((urb_priv->state == URB_DEL) || (ed->state & ED_DEL)) { tdINFO = le32_to_cpup (&td->hwINFO); if (TD_CC_GET (tdINFO) < 0xE) dl_transfer_length (td); @@ -1220,7 +1377,7 @@ } if (ed->state & ED_DEL) { /* set by sohci_free_dev */ struct ohci_device * dev = usb_to_ohci (ohci->dev[edINFO & 0x7F]); - OHCI_FREE (tdTailP); /* free dummy td */ + OHCI_FREE_TD (ohci, tdTailP); /* free dummy td */ ed->hwINFO = cpu_to_le32 (OHCI_ED_SKIP); ed->state = ED_NEW; /* if all eds are removed wake up sohci_free_dev */ @@ -1735,7 +1892,7 @@ writel (0, &ohci->regs->ed_controlhead); writel (0, &ohci->regs->ed_bulkhead); - writel (virt_to_bus (&ohci->hcca), &ohci->regs->hcca); /* a reset clears this */ + writel (ohci->hcca_dma_addr, &ohci->regs->hcca); /* a reset clears this */ fminterval = 0x2edf; writel ((fminterval * 9) / 10, &ohci->regs->periodicstart); @@ -1787,14 +1944,14 @@ struct ohci_regs * regs = ohci->regs; int ints; - if ((ohci->hcca.done_head != 0) && !(le32_to_cpup (&ohci->hcca.done_head) & 0x01)) { + if ((ohci->hcca->done_head != 0) && !(le32_to_cpup (&ohci->hcca->done_head) & 0x01)) { ints = OHCI_INTR_WDH; } else { if ((ints = (readl (®s->intrstatus) & readl (®s->intrenable))) == 0) return; } - // dbg("Interrupt: %x frame: %x", ints, le16_to_cpu (ohci->hcca.frame_no)); + // dbg("Interrupt: %x frame: %x", ints, le16_to_cpu (ohci->hcca->frame_no)); if (ints & OHCI_INTR_UE) { ohci->disabled++; @@ -1814,7 +1971,7 @@ } if (ints & OHCI_INTR_SF) { - unsigned int frame = le16_to_cpu (ohci->hcca.frame_no) & 1; + unsigned int frame = le16_to_cpu (ohci->hcca->frame_no) & 1; writel (OHCI_INTR_SF, ®s->intrdisable); if (ohci->ed_rm_list[!frame] != NULL) { dl_del_list (ohci, !frame); @@ -1835,7 +1992,7 @@ /* for load balancing of the interrupt branches */ for (i = 0; i < NUM_INTS; i++) ohci->ohci_int_load[i] = 0; - for (i = 0; i < NUM_INTS; i++) ohci->hcca.int_table[i] = 0; + for (i = 0; i < NUM_INTS; i++) ohci->hcca->int_table[i] = 0; ohci->ed_rm_list [0] = NULL; ohci->ed_rm_list [1] = NULL; @@ -1848,20 +2005,31 @@ /* allocate OHCI */ -static ohci_t * hc_alloc_ohci (void * mem_base) + static ohci_t * hc_alloc_ohci (struct pci_dev *pdev, void * mem_base) { ohci_t * ohci; struct usb_bus * bus; + int err; ohci = (ohci_t *) __get_free_pages (GFP_KERNEL, 1); if (!ohci) return NULL; memset (ohci, 0, sizeof (ohci_t)); + ohci->hcca = (struct ohci_hcca *)pci_alloc_consistent(pdev, sizeof(struct ohci_hcca), &ohci->hcca_dma_addr); + if (!ohci->hcca) + return NULL; ohci->irq = -1; ohci->regs = mem_base; + hc_reinit_ohci( ohci ); + + if ((err = ohci_consistent_storage_create(pdev, &ohci->consistent_storage))) { + printk(KERN_ERR "hc_alloc_ohci: failed to allocate consistent storage: err=%d\n", err); + return NULL; + } + bus = usb_alloc_bus (&sohci_device_operations); if (!bus) { free_pages ((unsigned long) ohci, 1); @@ -1924,7 +2092,7 @@ (unsigned long) mem_base, bufp); printk(KERN_INFO __FILE__ ": %s\n", dev->name); - ohci = hc_alloc_ohci (mem_base); + ohci = hc_alloc_ohci (dev, mem_base); if (!ohci) { return -ENOMEM; } diff -u linux-2.4.0-test1-ac7/drivers/usb/usb-ohci.h linux-2.4.0-test1-ac7-crl1/drivers/usb/usb-ohci.h --- linux-2.4.0-test1-ac7/drivers/usb/usb-ohci.h Fri Jun 16 13:27:42 2000 +++ linux-2.4.0-test1-ac7-crl1/drivers/usb/usb-ohci.h Fri Jun 16 16:54:42 2000 @@ -374,11 +374,29 @@ __u16 td_cnt; // number of tds already serviced int state; void * wait; + dma_addr_t transfer_buffer_dma_addr; td_t * td[0]; // list pointer to all corresponding TDs associated with this request } urb_priv_t; #define URB_DEL 1 + +/* + * In order to use the pci_alloc_consistent interface to allocate ED's and + * TD's, we need to keep a mapping between the virtual addresses and the + * dma addresses. We do this using a kmem_cache, allowing it a single slab + * only. + */ + +struct ohci_consistent_storage { + char *virt; + dma_addr_t dma_addr; + long virt_to_dma; + ed_t *edlist; /* freelist of eds */ + td_t *tdlist; /* freelist of tds */ + struct pci_dev *pci_dev; +}; + /* * This is the full ohci controller description * @@ -386,9 +404,9 @@ * a subset of what the full implementation needs. (Linus) */ - typedef struct ohci { - struct ohci_hcca hcca; /* hcca */ + struct ohci_hcca *hcca; /* hcca */ + dma_addr_t hcca_dma_addr; int irq; int disabled; /* e.g. got a UE, we're hung */ @@ -410,6 +428,7 @@ struct usb_bus * bus; struct usb_device * dev[128]; struct virt_root_hub rh; + struct ohci_consistent_storage consistent_storage; } ohci_t; @@ -417,7 +436,7 @@ #define NUM_EDS 32 /* num of preallocated endpoint descriptors */ struct ohci_device { - ed_t ed[NUM_EDS]; + ed_t *ed[NUM_EDS]; int ed_cnt; void * wait; }; @@ -432,7 +451,7 @@ static ed_t * ep_add_ed(struct usb_device * usb_dev, unsigned int pipe, int interval, int load); static void ep_rm_ed(struct usb_device * usb_dev, ed_t * ed); /* td */ -static void td_fill(unsigned int info, void * data, int len, urb_t * urb, int index); +static void td_fill(struct ohci *, unsigned int info, dma_addr_t data, int len, urb_t * urb, int index); static void td_submit_urb(urb_t * urb); /* root hub */ static int rh_submit_urb(urb_t * urb); @@ -440,11 +459,11 @@ static int rh_init_int_timer(urb_t * urb); #ifdef DEBUG -#define OHCI_FREE(x) kfree(x); printk("OHCI FREE: %d: %4x\n", -- __ohci_free_cnt, (unsigned int) x) -#define OHCI_ALLOC(x,size) (x) = kmalloc(size, in_interrupt() ? GFP_ATOMIC : GFP_KERNEL); printk("OHCI ALLO: %d: %4x\n", ++ __ohci_free_cnt,(unsigned int) x) +#define OHCI_FREE_TD(ohci,x) ohci_free_consistent_td(ohci, x); printk("OHCI FREE: %d: %4x\n", -- __ohci_free_cnt, (unsigned int) x) +#define OHCI_ALLOC_TD(ohci,x) (x) = ohci_alloc_consistent_td(ohci); printk("OHCI ALLO: %d: %4x\n", ++ __ohci_free_cnt,(unsigned int) x) static int __ohci_free_cnt = 0; #else -#define OHCI_FREE(x) kfree(x) -#define OHCI_ALLOC(x,size) (x) = kmalloc(size, in_interrupt() ? GFP_ATOMIC : GFP_KERNEL) +#define OHCI_FREE_TD(ohci, x) ohci_free_consistent_td(ohci, x) +#define OHCI_ALLOC_TD(ohci, x) (x) = ohci_alloc_consistent_td(ohci) #endif unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 17 10:09:03 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133EY8-0004ec-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 17 Jun 2000 10:05:56 +0100 Received: from [139.175.250.58] (helo=mail2000.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 133EY7-0004eE-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 17 Jun 2000 10:05:56 +0100 Received: By OpenMail Mailer;Sat, 17 Jun 2000 15:39:19 +0800 (CST) From: "Chien-Min Lee" Reply-To: cm_lee@mail2000.com.tw Subject: Driver for USB Device Controller on SA1100/1110 Message-ID: <961227559.58204.cm_lee@mail2000.com.tw> To: "linux-arm-kernel" Date: Sat, 17 Jun 2000 15:39:19 +0800 (CST) MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 8bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Dear All: Is there any driver for SA11x0 USB Device Controller existing ?? Have a nice day! kirk ----== Mailed via Openfind ==----- http://mail2000.com.tw/ ´£¨Ñ§K¶Oªº e-mail «H½cªA°È unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 18 17:58:38 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133iII-0005h7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 18 Jun 2000 17:51:34 +0100 Received: from [24.147.1.143] (helo=chmls05.mediaone.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133iIH-0005gz-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 18 Jun 2000 17:51:33 +0100 Received: from brightstareng.com (h0060082e53bb.ne.mediaone.net [24.91.81.9]) by chmls05.mediaone.net (8.8.7/8.8.7) with ESMTP id MAA08897 for ; Sun, 18 Jun 2000 12:51:26 -0400 (EDT) Message-ID: <394CFDBB.36005BC8@brightstareng.com> Date: Sun, 18 Jun 2000 12:50:03 -0400 From: Stuart Adams Organization: Bright Star Engineering Inc. X-Mailer: Mozilla 4.72 [en]C-MOENE (WinNT; U) X-Accept-Language: en MIME-Version: 1.0 To: linux-arm-kernel@lists.arm.linux.org.uk Subject: problem with 2.4 kernel and save_and_disable_irqs macro Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Just ran into a problem with the 2.4 kernel. The save_and_disable_irqs macro now actually enables FIQ rupts and forces SVC mode. This bit me because I happened to have a FIQ rupt pending during kernel boot and the first call to save_and_disable_irqs (in test_and_clear_bit) enable FIQ rupts and the kernel vectored off into never-never land. Seems like it would be better to go back to just oring 0x80 to the cpsr. -- Stuart unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 18 18:12:00 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133iWb-0005sR-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 18 Jun 2000 18:06:21 +0100 Received: from [194.168.180.65] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133iWa-0005sH-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 18 Jun 2000 18:06:21 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id SAA13233; Sun, 18 Jun 2000 18:06:02 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id SAA07522; Sun, 18 Jun 2000 18:04:46 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006181704.SAA07522@flint.arm.linux.org.uk> Subject: Re: problem with 2.4 kernel and save_and_disable_irqs macro To: sja@brightstareng.com (Stuart Adams) Date: Sun, 18 Jun 2000 18:04:46 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <394CFDBB.36005BC8@brightstareng.com> from "Stuart Adams" at Jun 18, 2000 12:50:03 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Stuart Adams writes: > The save_and_disable_irqs macro now actually enables > FIQ rupts and forces SVC mode. This is something I'd prefer to keep, since its performance-related, and find a different way around the problem. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 08:39:13 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133w5p-0007FV-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 08:35:37 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133w5m-0007F4-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 08:35:35 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id PAA20687 for ; Mon, 19 Jun 2000 15:36:20 +0800 Date: Mon, 19 Jun 2000 15:36:20 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk Subject: UCB1300 touch screen patch Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi, According to Philips Semiconductors' application notes regarding the differences between UCB1300 and UCB1200 [1], ``the touchscreen interface should be programmed to pressure mode first for a duration of 1 SIB frame before setting to position mode for measurement.'' Sample patch below for 2.4.0-test1-ac7-rmk1-np6 (this also fixes a small typo in sa1100_ts.h and has a corrected normalize function). Now Microwindows Minesweeper is running fine on the Assabet. :-) Regards, Tak-Shing Chan [1] http://www.semiconductors.com/acrobat/applicationnotes/AN809_1.pdf --- /usr/src/linux/drivers/char/sa1100_ts.c.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux/drivers/char/sa1100_ts.c Mon Jun 19 11:51:19 2000 @@ -211,6 +211,11 @@ pressure = sa1100_ts_measure_pressure(); if((pressure - ground) > UGL_SA1100_TS_PRESSURE_THRESHOLD) { x += sa1100_ts_measure_x(); +#ifdef CONFIG_SA1100_ASSABET + /* See Philips' AN809 */ + sa1100_ts_regwrite(UGL_UCB1200_TS_CR, + UGL_UCB1200_TS_PRESSURE); +#endif y += sa1100_ts_measure_y(); } else { *in_x = SA1100_TS_NOVALUE; @@ -262,10 +267,13 @@ int sa1100_ts_normalize(int * in_x, int * in_y) { - +#ifdef CONFIG_SA1100_ASSABET + *in_x = (((SA1100_TS_RAW_MAX_X - *in_x)) * SA1100_TS_RES_X) / (SA1100_TS_RAW_MAX_X - SA1100_TS_RAW_MIN_X); + *in_y = (((SA1100_TS_RAW_MAX_Y - *in_y)) * SA1100_TS_RES_Y) / (SA1100_TS_RAW_MAX_Y - SA1100_TS_RAW_MIN_Y); +#else *in_x = (((*in_x - SA1100_TS_RAW_MIN_X)) * SA1100_TS_RES_X) / (SA1100_TS_RAW_MAX_X - SA1100_TS_RAW_MIN_X); *in_y = (((*in_y - SA1100_TS_RAW_MIN_Y)) * SA1100_TS_RES_Y) / (SA1100_TS_RAW_MAX_Y - SA1100_TS_RAW_MIN_Y); - +#endif return 0; } @@ -424,12 +432,21 @@ memset(&_g_settings,0,sizeof(_g_settings)); g_settings = &_g_settings; +#ifdef CONFIG_SA1100_ASSABET + g_settings->m_raw_max_x = 944; + g_settings->m_raw_max_y = 944; + g_settings->m_raw_min_x = 70; + g_settings->m_raw_min_y = 70; + g_settings->m_res_x = 320; + g_settings->m_res_y = 240; +#else g_settings->m_raw_max_x = 885; g_settings->m_raw_max_y = 885; g_settings->m_raw_min_x = 70; g_settings->m_raw_min_y = 70; g_settings->m_res_x = 640; g_settings->m_res_y = 480; +#endif g_settings->m_fudge_x = 1; g_settings->m_fudge_y = 1; g_settings->m_avg_sample = 10; --- /usr/src/linux/drivers/char/sa1100_ts.h.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux/drivers/char/sa1100_ts.h Mon Jun 19 11:09:45 2000 @@ -104,9 +104,9 @@ #define UGL_SA1100_TS_PRESSURE_THRESHOLD (g_settings->m_pressure) #define SA1100_TS_RAW_MAX_X (g_settings->m_raw_max_x) -#define SA1100_TS_RAW_MAX_Y (g_settings->m_raw_max_x) +#define SA1100_TS_RAW_MAX_Y (g_settings->m_raw_max_y) #define SA1100_TS_RAW_MIN_X (g_settings->m_raw_min_x) -#define SA1100_TS_RAW_MIN_Y (g_settings->m_raw_min_x) +#define SA1100_TS_RAW_MIN_Y (g_settings->m_raw_min_y) #define SA1100_TS_RES_X (g_settings->m_res_x) #define SA1100_TS_RES_Y (g_settings->m_res_y) #define SA1100_TS_FUDGE_X (g_settings->m_fudge_x) unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 09:21:53 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133wn8-00082X-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 09:20:22 +0100 Received: from [195.108.189.102] (helo=white.plant.nl) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133wn8-00082Q-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 09:20:22 +0100 Received: from [194.134.11.42] by white.plant.nl (NTMail 4.30.0013/NT3720.00.67541ba1) with ESMTP id eyqyfaaa for ; Mon, 19 Jun 2000 10:19:31 +0200 Message-ID: <394DD6F9.C0908D88@maycom.nl> Date: Mon, 19 Jun 2000 10:16:57 +0200 From: Daven Baptist X-Mailer: Mozilla 4.7 [en] (X11; I; Linux 2.2.13 i686) X-Accept-Language: en MIME-Version: 1.0 CC: linux-arm-kernel , linux-arm@vger.rutgers.edu Subject: uploading device drivers? References: <961227559.58204.cm_lee@mail2000.com.tw> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk can some one tell me if there is a easy/faster way to upload a device driver or file to a running assabet board ? now i add the file to the disk image, then i restart the system (+/- 5 min :( ) and i can access the file. but this takes (tooooo) longe. is there a faster way ? thanks Daven unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 10:16:32 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133xe0-0000I7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 10:15:00 +0100 Received: from [62.161.177.33] (helo=[62.161.177.33]) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133xdz-0000I1-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 10:14:59 +0100 Received: from [192.168.1.13] by [192.168.1.2] with SMTP (QuickMail Pro Server for Mac 2.0); 19 JUN 00 11:16:32 UT From: Benjamin Herrenschmidt To: Subject: Current CVS 2.2.15 vs. 285 debug options Date: Mon, 19 Jun 2000 11:14:08 +0200 Message-Id: <20000619091408.2070@mailhost.mipsys.com> X-Mailer: CTM PowerMail 3.0.3 MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi ! When I enable the low-level debug options (in kernel hacking, the ones that enable the 285 0serial port stuffs in head_armv.S), then head_armv.S fails to compile. I'm getting a weird error about non-const expression in ".if" around the ".ltorg" statement. I'm not expert in GNU assembler and it's macros, so I didn't try to fix that, I was wondering if this could be due to me using an outdated toolchain or if this was actually a bug. Regards, Ben. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 12:06:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133zNj-0001iG-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 12:06:19 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133zNi-0001hp-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 12:06:18 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id TAA22348; Mon, 19 Jun 2000 19:06:49 +0800 Date: Mon, 19 Jun 2000 19:06:49 +0800 (CST) From: Tak-Shing Chan To: Daven Baptist cc: linux-arm-kernel , linux-arm@vger.rutgers.edu Subject: Re: uploading device drivers? In-Reply-To: <394DD6F9.C0908D88@maycom.nl> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Daven, Have you tried modules? Just upload the drivers you need (by ftp or sz) and insmod/rmmod them manually. Tak-Shing Chan On Mon, 19 Jun 2000, Daven Baptist wrote: > can some one tell me if there is a easy/faster way to > upload a device driver or file to a running assabet > board ? > > now i add the file to the disk image, then i restart > the system (+/- 5 min :( ) and i can access the file. > > but this takes (tooooo) longe. is there a faster way ? > > thanks > > Daven > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 12:43:14 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 133zvu-00028s-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 12:41:38 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 133zvs-00028G-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 12:41:37 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id TAA22469 for ; Mon, 19 Jun 2000 19:42:27 +0800 Date: Mon, 19 Jun 2000 19:42:27 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Assabet UDA1341/DMA patch Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi Nicolas, I found that your UDA1341 output is shaking at a steady frequency (it's been there since 2.3.99pre-kernels). Here's the patch: Tak-Shing Chan --- /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/dma-sa1100.c.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/dma-sa1100.c Mon Jun 19 17:33:07 2000 @@ -110,13 +110,13 @@ if( use_bufa ){ DMA->ClrDCSR = DCSR_DONEA|DCSR_STRTA; DMA->DBSA = dma_ptr; - DMA->DBTA = size-1; + DMA->DBTA = size; DMA->SetDCSR = DCSR_STRTA|DCSR_IE|DCSR_RUN; DPRINTK(" with a\n"); }else{ DMA->ClrDCSR = DCSR_DONEB|DCSR_STRTB; DMA->DBSB = dma_ptr; - DMA->DBTB = size-1; + DMA->DBTB = size; DMA->SetDCSR = DCSR_STRTB|DCSR_IE|DCSR_RUN; DPRINTK(" with b\n"); } --- /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/sound/sa1100-uda1341.c.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/sound/sa1100-uda1341.c Mon Jun 19 19:40:09 2000 @@ -272,7 +272,8 @@ #define AUDIO_CHANNELS_DEFAULT 2 #define AUDIO_RATE_DEFAULT 44100 #define AUDIO_NBFRAGS_DEFAULT 8 -#define AUDIO_FRAGSIZE_DEFAULT 8192 +#define AUDIO_FRAGSIZE_DEFAULT 4096 +#define AUDIO_DMA_OVERSHOOT 64 /* @@ -546,7 +547,7 @@ output_stream.dma_spin = 0; } if( s->dma_spin ){ - while( sa1100_start_dma(s->DMA, (void*)FLUSH_BASE_PHYS, MAX_DMA_SIZE) == 0 ) + while( sa1100_start_dma(s->DMA, (void*)FLUSH_BASE_PHYS, MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT) == 0 ) s->dma_spin_ref++; } break; @@ -556,7 +557,7 @@ * Let's try to start DMA on the current buffer. * If no DMA is available then we stop here. */ - chunksize = (b->size < MAX_DMA_SIZE) ? b->size : MAX_DMA_SIZE; + chunksize = (b->size < MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT) ? b->size : MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT; DPRINTK("DMA buf %d (%d) chnk %d\n", s->dma_head_idx, b->ref, chunksize); if( sa1100_start_dma( s->DMA, b->ptr, chunksize ) != 0 ){ DPRINTK("DMA busy\n"); @@ -1091,7 +1092,7 @@ static inline void audio_ssp_init( void ) { /* Setup the uarts */ - PPAR |= PPC_TXD4; + PPAR |= PPAR_SPR; GAFR |= (GPIO_SSP_TXD | GPIO_SSP_RXD | GPIO_SSP_SCLK | GPIO_SSP_SFRM | GPIO_SSP_CLK); GPDR |= (GPIO_SSP_TXD | GPIO_SSP_SCLK | GPIO_SSP_SFRM ); unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 13:43:02 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1340sw-0002sZ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 13:42:38 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1340sv-0002sE-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 13:42:37 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id UAA23438 for ; Mon, 19 Jun 2000 20:43:27 +0800 Date: Mon, 19 Jun 2000 20:43:27 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet UDA1341/DMA patch Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi Nicolas, The last patch is actually a wrong one (my fault), please move PPAR |= PPAR_SPR down three lines (or use the following patch instead). Tak-Shing Chan --- /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/dma-sa1100.c.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/dma-sa1100.c Mon Jun 19 17:33:07 2000 @@ -110,13 +110,13 @@ if( use_bufa ){ DMA->ClrDCSR = DCSR_DONEA|DCSR_STRTA; DMA->DBSA = dma_ptr; - DMA->DBTA = size-1; + DMA->DBTA = size; DMA->SetDCSR = DCSR_STRTA|DCSR_IE|DCSR_RUN; DPRINTK(" with a\n"); }else{ DMA->ClrDCSR = DCSR_DONEB|DCSR_STRTB; DMA->DBSB = dma_ptr; - DMA->DBTB = size-1; + DMA->DBTB = size; DMA->SetDCSR = DCSR_STRTB|DCSR_IE|DCSR_RUN; DPRINTK(" with b\n"); } --- /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/sound/sa1100-uda1341.c.orig Mon Jun 19 10:44:21 2000 +++ /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/sound/sa1100-uda1341.c Mon Jun 19 20:34:06 2000 @@ -272,7 +272,8 @@ #define AUDIO_CHANNELS_DEFAULT 2 #define AUDIO_RATE_DEFAULT 44100 #define AUDIO_NBFRAGS_DEFAULT 8 -#define AUDIO_FRAGSIZE_DEFAULT 8192 +#define AUDIO_FRAGSIZE_DEFAULT 4096 +#define AUDIO_DMA_OVERSHOOT 64 /* @@ -546,7 +547,7 @@ output_stream.dma_spin = 0; } if( s->dma_spin ){ - while( sa1100_start_dma(s->DMA, (void*)FLUSH_BASE_PHYS, MAX_DMA_SIZE) == 0 ) + while( sa1100_start_dma(s->DMA, (void*)FLUSH_BASE_PHYS, MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT) == 0 ) s->dma_spin_ref++; } break; @@ -556,7 +557,7 @@ * Let's try to start DMA on the current buffer. * If no DMA is available then we stop here. */ - chunksize = (b->size < MAX_DMA_SIZE) ? b->size : MAX_DMA_SIZE; + chunksize = (b->size < MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT) ? b->size : MAX_DMA_SIZE-AUDIO_DMA_OVERSHOOT; DPRINTK("DMA buf %d (%d) chnk %d\n", s->dma_head_idx, b->ref, chunksize); if( sa1100_start_dma( s->DMA, b->ptr, chunksize ) != 0 ){ DPRINTK("DMA busy\n"); @@ -1091,11 +1092,11 @@ static inline void audio_ssp_init( void ) { /* Setup the uarts */ - PPAR |= PPC_TXD4; GAFR |= (GPIO_SSP_TXD | GPIO_SSP_RXD | GPIO_SSP_SCLK | GPIO_SSP_SFRM | GPIO_SSP_CLK); GPDR |= (GPIO_SSP_TXD | GPIO_SSP_SCLK | GPIO_SSP_SFRM ); GPDR &= ~(GPIO_SSP_RXD | GPIO_SSP_CLK); + PPAR |= PPAR_SPR; Ser4SSCR0 = SSCR0_SSE + SSCR0_DataSize(16) + SSCR0_TI; Ser4SSCR1 = SSCR1_SClkIactL + SSCR1_SClk1P + SSCR1_ExtClk; unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 13:55:11 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 13414w-00032X-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 13:55:02 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 13414v-00032L-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 13:55:01 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id IAA15084; Mon, 19 Jun 2000 08:54:13 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Mon, 19 Jun 2000 08:54:13 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Daven Baptist cc: linux-arm-kernel , linux-arm@vger.rutgers.edu Subject: Re: uploading device drivers? In-Reply-To: <394DD6F9.C0908D88@maycom.nl> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Set your terminal to 115200 baud and upload the module with zmodem. >From the shell prompt on Assabet: stty speed 115200 [set your terminal to 115200 bauds] rz [send your module file with zmodem] On Mon, 19 Jun 2000, Daven Baptist wrote: > can some one tell me if there is a easy/faster way to > upload a device driver or file to a running assabet > board ? > > now i add the file to the disk image, then i restart > the system (+/- 5 min :( ) and i can access the file. > > but this takes (tooooo) longe. is there a faster way ? > > thanks > > Daven > > unsubscribe: body of `unsubscribe linux-arm' to majordomo@vger.rutgers.edu > ++ Please use linux-arm-kernel@lists.arm.linux.org.uk for ++ > ++ kernel-related discussions. ++ > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 15:03:04 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 13428L-0003tw-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 15:02:37 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 13428K-0003tk-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 15:02:36 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id KAA15128; Mon, 19 Jun 2000 10:01:42 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Mon, 19 Jun 2000 10:01:42 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Tak-Shing Chan cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Mon, 19 Jun 2000, Tak-Shing Chan wrote: > Hi Nicolas, > > I found that your UDA1341 output is shaking at a steady > frequency (it's been there since 2.3.99pre-kernels). Here's > the patch: However you're limitting the DMA buffer to 4096 bytes which might be quite small, especially if the system is under heavy load. The SA1100 manual explictly says that the DMA count register is 13 bit wide for a maximal DMA size of 8192 bytes, which is highly adviseable for audio. However 13 bits can represent a max value of 0x1fff i.e. 8191. Since most registers on the SA1100 are defined as reg = value - 1, I though it was the case for the DMA too, although it is not strictly mentionned for the DMA count reg. What I've noticed though is that even if you specify 0 for the DMA count reg, you'll get at least one datum through the DMA engine which tend to confirm the -1 guess. Anybody can confirm this? Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 19 18:49:31 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1345eg-0007HA-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 19 Jun 2000 18:48:14 +0100 Received: from [139.175.250.58] (helo=mail2000.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 1345ed-0007GY-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 19 Jun 2000 18:48:13 +0100 Received: By OpenMail Mailer;Tue, 20 Jun 2000 01:47:09 +0800 (CST) From: "Chien-Min Lee" Reply-To: cm_lee@mail2000.com.tw Subject: Re: uploading device drivers? Message-ID: <961436827.55873.cm_lee@mail2000.com.tw> To: "nico" Cc: "linux-arm-kernel" , "linux-arm" , "daven" Date: Tue, 20 Jun 2000 01:47:07 +0800 (CST) MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 8bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi: >Set your terminal to 115200 baud and upload the module >with zmodem. >From the shell prompt on Assabet: > stty speed 115200 > [set your terminal to 115200 bauds] > rz > [send your module file with zmodem] I have tried it. However, I can not find the file I uploaded on assabet. I use the 'Send file' option of minicom with 'zmodem' protocol to upload target file. Did I miss something ?? Thanks kirk ----== Mailed via Openfind ==----- http://mail2000.com.tw/ ´£¨Ñ§K¶Oªº e-mail «H½cªA°È unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 04:44:33 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Ex9-0007iK-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 04:43:55 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Ex2-0007ha-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 04:43:53 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id LAA25006; Tue, 20 Jun 2000 11:44:08 +0800 Date: Tue, 20 Jun 2000 11:44:08 +0800 (CST) From: Tak-Shing Chan To: Nicolas Pitre cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Nicolas, I think that involves two things. First, regarding the meaning of DBTA--from repeated listening experiments (from 4096 to 8176) I'm positive that it's value and not value-1. The result of missing samples in digital recordings are just too obvious... [by the way 11.6.1.4 says TCA is in bytes.] Second, isn't the extra datum you detected caused by DMA overshoot (and the value is supposed to be fetched and ignored)? My patch explicitly takes DMA overshoot into account. It seems like it's safe to specify AUDIO_DMA_OVERSHOOT as 16 and AUDIO_FRAGSIZE_DEFAULT as 8176. Any confirmations? IMHO 8176 is big enough for most purposes. Meanwhile 64 and 4096 are ``failsafe'' values but not optimal. Tak-Shing Chan On Mon, 19 Jun 2000, Nicolas Pitre wrote: > However you're limitting the DMA buffer to 4096 bytes which might be quite > small, especially if the system is under heavy load. > > The SA1100 manual explictly says that the DMA count register is 13 bit > wide for a maximal DMA size of 8192 bytes, which is highly adviseable for > audio. However 13 bits can represent a max value of 0x1fff i.e. 8191. > Since most registers on the SA1100 are defined as reg = value - 1, I > though it was the case for the DMA too, although it is not strictly > mentionned for the DMA count reg. What I've noticed though is that even > if you specify 0 for the DMA count reg, you'll get at least one datum > through the DMA engine which tend to confirm the -1 guess. > > Anybody can confirm this? > > > Nicolas > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 05:48:55 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Fxu-00008s-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 05:48:46 +0100 Received: from [210.230.34.106] (helo=so-fs.musicpod.co.jp) by www.linux.org.uk with smtp (Exim 3.13 #1) id 134Fxs-00008d-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 05:48:44 +0100 Received: (qmail 12269 invoked from network); 20 Jun 2000 04:48:43 -0000 Received: from bak.vasara.co.jp (imai@192.168.20.13) by so-fs.musicpod.co.jp with SMTP; 20 Jun 2000 04:48:43 -0000 Date: Tue, 20 Jun 2000 13:48:42 +0900 Message-ID: From: Kunihiko IMAI To: linux-arm-kernel@lists.arm.linux.org.uk Subject: a patch for SA1111 USB interface (was Re: USB driver) In-Reply-To: In your message of "Fri, 16 Jun 2000 17:04:04 -0400" <200006162104.RAA06343@highwind.crl.dec.com> References: <200006162104.RAA06343@highwind.crl.dec.com> User-Agent: Wanderlust/1.1.1 (Purple Rain) WEMI/1.13.7 (Shimada) CLIME/1.13.6 (=?ISO-2022-JP?B?GyRCQ2YlTj4xGyhC?=) MULE XEmacs/21.1 (patch 6) (Big Bend) (i386-vine-linux) MIME-Version: 1.0 (generated by WEMI 1.13.7 - "Shimada") Content-Type: text/plain; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, I've just caught up the latest version of linux kernel. And I've ported James's version of usb-ohci.c driver for SA1111. It's a quite a quick hack, but I believe that there remains only political or formal problems. Note (or known bug): o This patch requires a setup for SA1111 DMA. I'll post it later. o Not consider co-existence with a PCI-connected OHCI USB device. o mis-used bus_to_virt() and virt_to_bus() instead of phys_to_virt() and virt_to_phys(). o CONFIG_SA1100_XP860 is a machine arch. definition. XP860 has a SA1110 and a SA1111. So it is similar to Intel assabet + neponset. I believe that it is easy to port this patch to them. o Too dirty :-) Thanks. _._. __._ _.. . .___ ._. _____ _... ._ _._ _.._. .____ .__. ... . _._ Kunihiko IMAI --- usb-ohci.c.JAMES Mon Jun 19 21:24:21 2000 +++ usb-ohci.c Tue Jun 20 13:44:02 2000 @@ -61,6 +61,21 @@ #include #endif +#ifdef CONFIG_SA1111 + +#define SA1111_OHCI_BASE SA1111_p2v(_SA1111(0x0400)) +#define SA1111_OHCI_EXTENT 512 +#define SA1111_OHCI_IRQ NIRQHCIM + +#ifdef CONFIG_SA1100_XP860 +#define PwrSensePolLow 1 +#define PwrCtrlPolLow 1 +#else +#define PwrSensePolLow 0 +#define PwrCtrlPolLow 0 +#endif + +#endif /* For initializing controller (mask in an HCFS mode too) */ #define OHCI_CONTROL_INIT \ @@ -90,7 +105,11 @@ static int __verify_dma(dma_addr_t dma, int line) { unsigned long addr = (unsigned long)dma; +#ifndef CONFIG_SA1100_XP860 int ok = ((addr & 0xF0000000) == 0xE0000000); +#else + int ok = ((addr & 0xF0000000) == 0xC0000000); +#endif if (!ok) printk(KERN_CRIT __FILE__ ":%d: Expected DMA address, got %#08x\n", line, dma); return ok; @@ -2071,8 +2090,12 @@ usb_deregister_bus (ohci->bus); usb_free_bus (ohci->bus); +#ifndef CONFIG_SA1111 /* unmap the IO address space */ iounmap (ohci->regs); +#else + release_region ( SA1111_OHCI_BASE, SA1111_OHCI_EXTENT ); +#endif free_pages ((unsigned long) ohci, 1); } @@ -2119,9 +2142,16 @@ ohci->irq = irq; hc_start (ohci); +#ifndef CONFIG_SA1111 pmdev = pm_register (PM_PCI_DEV, PM_PCI_ID(dev), handle_pm_event); +#else + /* FIX ME: this is quite dummy */ + pmdev = pm_register (9999, + 9999, + handle_pm_event); +#endif if (pmdev) pmdev->data = ohci; @@ -2140,6 +2170,7 @@ static int hc_start_ohci (struct pci_dev * dev) { +#ifndef CONFIG_SA1111 unsigned long mem_base; if (pci_enable_device(dev) < 0) @@ -2154,6 +2185,10 @@ return -EFAULT; } return hc_found_ohci (dev, dev->irq, (void *) mem_base); +#else + request_region ( SA1111_OHCI_BASE, SA1111_OHCI_EXTENT, "usb-ohci" ); + return hc_found_ohci (dev, SA1111_OHCI_IRQ, SA1111_OHCI_BASE ); +#endif } /*-------------------------------------------------------------------------*/ @@ -2257,6 +2292,7 @@ int ohci_hcd_init (void) { +#ifndef CONFIG_SA1111 int ret = -ENODEV; struct pci_dev * dev = NULL; @@ -2268,6 +2304,20 @@ pmu_register_sleep_notifier (&ohci_sleep_notifier); #endif return ret; +#else + struct pci_dev * dev = NULL; + volatile unsigned long *Reset = SA1111_p2v(_SA1111(0x051c)); + + /* turn the USB clock power on */ + SKPCR |= 0x81; + mdelay(1); + + /* reset USB */ + *Reset = ( PwrSensePolLow << 6 ) + | ( PwrCtrlPolLow << 7 ); + + return hc_start_ohci(dev); +#endif /*CONFIG_SA1111*/ } /*-------------------------------------------------------------------------*/ @@ -2296,6 +2346,10 @@ INIT_LIST_HEAD (&ohci->ohci_hcd_list); hc_release_ohci (ohci); } +#ifdef CONFIG_SA1111 + /* turn the USB clock off */ + SKPCR &= ~0x1; +#endif } #endif //MODULE unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 05:57:55 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134G6e-0000FL-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 05:57:48 +0100 Received: from [210.230.34.106] (helo=so-fs.musicpod.co.jp) by www.linux.org.uk with smtp (Exim 3.13 #1) id 134G6d-0000FD-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 05:57:47 +0100 Received: (qmail 12487 invoked from network); 20 Jun 2000 04:57:47 -0000 Received: from bak.vasara.co.jp (imai@192.168.20.13) by so-fs.musicpod.co.jp with SMTP; 20 Jun 2000 04:57:47 -0000 Date: Tue, 20 Jun 2000 13:57:47 +0900 Message-ID: From: Kunihiko IMAI To: linux-arm-kernel@lists.arm.linux.org.uk Subject: a patch for SA1111 bus master DMA User-Agent: Wanderlust/1.1.1 (Purple Rain) WEMI/1.13.7 (Shimada) CLIME/1.13.6 (=?ISO-2022-JP?B?GyRCQ2YlTj4xGyhC?=) MULE XEmacs/21.1 (patch 6) (Big Bend) (i386-vine-linux) MIME-Version: 1.0 (generated by WEMI 1.13.7 - "Shimada") Content-Type: text/plain; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, I made a patch for SA1111 bus master DMA. But I wonder that this code should be placed here. Or should be defined another macro? Thanks. _._. __._ _.. . .___ ._. _____ _... ._ _._ _.._. .____ .__. ... . _._ Kunihiko IMAI --- linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/hw-sa1100.c.ORIG Mon Jun 19 14:25:36 2000 +++ linux-2.4.0-test1-ac19-rmk1-np1/arch/arm/kernel/hw-sa1100.c Tue Jun 20 13:10:41 2000 @@ -112,6 +112,34 @@ * any other SA-1111 functional blocks must be enabled separately * using the SKPCR. */ + + { + /* + * SA1111 DMA bus master setup + */ + int cas; + + /* SA1111 side */ + switch ( (MDCNFG>>12) & 0x03 ) { + case 0x02: + cas = 0; break; + case 0x03: + cas = 1; break; + default: + cas = 1; break; + } + SMCR = 1 /* 1: memory is SDRAM */ + | ( 1 << 1 ) /* 1:MBGNT is enable */ + | ( ((MDCNFG >> 4) & 0x07) << 2 ) /* row address lines */ + | ( cas << 5 ); /* CAS ratency */ + + /* SA1110 side */ + GPDR |= 1<<21; + GPDR &= ~(1<<22); + GAFR |= ( (1<<21) | (1<<22) ); + + TUCR |= (1<<10); + } } #endif @@ -129,6 +157,10 @@ "hasn't been configured in the kernel\n" ); #endif } + } else if ( machine_is_xp860() ) { +#ifdef CONFIG_SA1111 + sa1111_init(); +#endif } } unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 06:47:28 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134GsS-0000tn-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 06:47:12 +0100 Received: from [210.154.37.2] (helo=funai-tky.co.jp ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134GsQ-0000td-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 06:47:10 +0100 Received: from hiropu ([192.168.1.19]) by funai-tky.co.jp (8.9.1/3.7Wpr7-98092916) with SMTP id OAA10559 for ; Tue, 20 Jun 2000 14:47:06 +0900 Date: Tue, 20 Jun 2000 14:47:10 +0900 From: Hiroshi Ishii To: linux-arm-kernel Subject: Re: UCB1300 touch screen patch In-Reply-To: References: Message-Id: <394F055E197.864AHIROSHI@ns.funai-tky.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver 1.26.03 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Mon, 19 Jun 2000 15:36:20 +0800 (CST) Tak-Shing Chan wrote: > (this also fixes a small typo in sa1100_ts.h and has a corrected > normalize function). Now Microwindows Minesweeper is running > fine on the Assabet. :-) > Hummmm... My board (Assabet) out following message. Unknown output type (FE00) Why.....? ---- Funai Electric R & D System Software Engineer Hiroshi Ishii unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 09:19:12 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134JEt-0000Z7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 09:18:31 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134JEs-0000Yi-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 09:18:30 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id QAA27227; Tue, 20 Jun 2000 16:19:00 +0800 Date: Tue, 20 Jun 2000 16:19:00 +0800 (CST) From: Tak-Shing Chan To: Tak-Shing Chan cc: Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi Nicolas, AUDIO_DMA_OVERSHOOT=2 and AUDIO_FRAGSIZE_DEFAULT=8190 works fine. Another observation: odd fragsizes break things (the last 16-bit sample became 8-bit). It looks like DMA overshoot is not a problem at all because 8190 is the biggest possible even number. Maybe renaming AUDIO_DMA_OVERSHOOT to AUDIO_EVEN_HACK is a better idea. ;) Results so far: * Odd fragsizes with tca=size-1 : failed * Even fragsizes with tca=size-1: failed * Odd fragsizes with tca=size : failed * Even fragsizes with tca=size : passed * Valid fragsize range : 4096-8190 Tak-Shing Chan On Tue, 20 Jun 2000, Tak-Shing Chan wrote: > Nicolas, > > I think that involves two things. First, regarding the > meaning of DBTA--from repeated listening experiments (from 4096 > to 8176) I'm positive that it's value and not value-1. The > result of missing samples in digital recordings are just too > obvious... [by the way 11.6.1.4 says TCA is in bytes.] > > Second, isn't the extra datum you detected caused by DMA > overshoot (and the value is supposed to be fetched and ignored)? > My patch explicitly takes DMA overshoot into account. > > It seems like it's safe to specify AUDIO_DMA_OVERSHOOT as 16 > and AUDIO_FRAGSIZE_DEFAULT as 8176. Any confirmations? IMHO > 8176 is big enough for most purposes. Meanwhile 64 and 4096 are > ``failsafe'' values but not optimal. > > Tak-Shing Chan > > On Mon, 19 Jun 2000, Nicolas Pitre wrote: > >> However you're limitting the DMA buffer to 4096 bytes which might be quite >> small, especially if the system is under heavy load. >> >> The SA1100 manual explictly says that the DMA count register is 13 bit >> wide for a maximal DMA size of 8192 bytes, which is highly adviseable for >> audio. However 13 bits can represent a max value of 0x1fff i.e. 8191. >> Since most registers on the SA1100 are defined as reg = value - 1, I >> though it was the case for the DMA too, although it is not strictly >> mentionned for the DMA count reg. What I've noticed though is that even >> if you specify 0 for the DMA count reg, you'll get at least one datum >> through the DMA engine which tend to confirm the -1 guess. >> >> Anybody can confirm this? >> >> >> Nicolas >> > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 10:03:51 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134JvD-00016E-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 10:02:15 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134JvB-00015d-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 10:02:13 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id RAA27675; Tue, 20 Jun 2000 17:02:06 +0800 Date: Tue, 20 Jun 2000 17:02:06 +0800 (CST) From: Tak-Shing Chan To: Hiroshi Ishii cc: linux-arm-kernel Subject: Re: UCB1300 touch screen patch In-Reply-To: <394F055E197.864AHIROSHI@ns.funai-tky.co.jp> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Beware of the minor device number: /dev/ts : 34 /dev/tsraw : 18 Tak-Shing Chan On Tue, 20 Jun 2000, Hiroshi Ishii wrote: > On Mon, 19 Jun 2000 15:36:20 +0800 (CST) > Tak-Shing Chan wrote: > >> (this also fixes a small typo in sa1100_ts.h and has a corrected >> normalize function). Now Microwindows Minesweeper is running >> fine on the Assabet. :-) >> > > Hummmm... My board (Assabet) out following message. > > Unknown output type (FE00) > > Why.....? > > ---- > Funai Electric R & D > System Software Engineer > Hiroshi Ishii > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 10:31:09 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134KMz-0001Rs-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 10:30:57 +0100 Received: from [210.154.37.2] (helo=funai-tky.co.jp ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134KMw-0001Rl-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 10:30:55 +0100 Received: from hiropu ([192.168.1.19]) by funai-tky.co.jp (8.9.1/3.7Wpr7-98092916) with SMTP id SAA11983 for ; Tue, 20 Jun 2000 18:30:51 +0900 Date: Tue, 20 Jun 2000 18:30:51 +0900 From: Hiroshi Ishii To: linux-arm-kernel Subject: Re: UCB1300 touch screen patch In-Reply-To: References: <394F055E197.864AHIROSHI@ns.funai-tky.co.jp> Message-Id: <394F39CB30B.864BHIROSHI@ns.funai-tky.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver 1.26.03 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Tue, 20 Jun 2000 17:02:06 +0800 (CST) Tak-Shing Chan wrote: > Beware of the minor device number: > > /dev/ts : 34 > /dev/tsraw : 18 > Thanks. Working it cool!!! ---- Funai Electric R & D System Software Engineer Hiroshi Ishii unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 12:35:11 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134MIk-0002wx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 12:34:42 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134MIi-0002wY-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 12:34:40 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id TAA29451; Tue, 20 Jun 2000 19:35:26 +0800 Date: Tue, 20 Jun 2000 19:35:26 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk cc: nico@cam.org Subject: Touchscreen/IRQ problem Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Nicolas, The touchscreen driver works fine for 2.4.0-test1-ac7-rmk1-np6 but not working at all when I use 2.4.0-test1-ac19-rmk1-np1. Are there any big changes regarding IRQs or timers between these two kernels? I've replaced GRER |= (1 << 23) with set_GPIO_IRQ_edge( (1 << 23), GPIO_RISING_EDGE) but that isn't a fix... Tak-Shing Chan unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 15:10:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Oj4-0005BY-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 15:10:02 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Oj2-0005BH-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 15:10:00 +0100 Date: Tue, 20 Jun 2000 10:09:48 -0400 (EDT) From: Nicolas Pitre To: Tak-Shing Chan cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Tue, 20 Jun 2000, Tak-Shing Chan wrote: > Hi Nicolas, > > AUDIO_DMA_OVERSHOOT=2 and AUDIO_FRAGSIZE_DEFAULT=8190 works > fine. Another observation: odd fragsizes break things (the last > 16-bit sample became 8-bit). It looks like DMA overshoot is not That one is easy since audio fragment sizes should always be a power of 2 for buffering efficiency. Sometimes a fragment might not be full, but samples are always 16-bit so no odd-size fragments should happen. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 15:17:06 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Opj-0005LW-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 15:16:55 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Opi-0005LO-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 15:16:54 +0100 Received: from server01.asi (akron-80-160.ar51.net [208.225.80.160]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id JAA06855; Tue, 20 Jun 2000 09:16:11 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Tue, 20 Jun 2000 10:17:31 -0400 Message-ID: From: Jeff Sutherland To: "Nicolas Pitre (E-mail)" , "Linux ARM kernel mailing list (E-mail)" Subject: NFS mounting of root filesystem on Assabet Date: Tue, 20 Jun 2000 10:17:29 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Has anyone successfully NFS mounted a root filesystem for their Assabet at boot time? I'm running down what appears to be a spinlock issue in 2.4.0-test1-ac19-rmk1-np1 in function get_write_access() in fs/openi.c. The simple (non-irq) spinlocks spin_lock() and spin_unlock() are being used here, but the file /dev/console is being redirected to /dev/ttyS0 currently. Execution seems to stall following atomic_inc() near the end of fs/openi.c get_write_acces() in what appears to be a spin deadlock. LED D8 continues to blink for about 10 seconds (D9 is on solid) after reaching this point, after which D8 stops and the kernel croaks. Do we need to possibly use the safer (but slower) write_lock_irqsave() and write_unlock_irqsave() here instead? Some insight please... //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 15:18:16 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Or0-0005Mv-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 15:18:14 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Oqz-0005Mn-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 15:18:13 +0100 Date: Tue, 20 Jun 2000 10:18:03 -0400 (EDT) From: Nicolas Pitre To: Tak-Shing Chan cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Touchscreen/IRQ problem In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Tue, 20 Jun 2000, Tak-Shing Chan wrote: > Nicolas, > > The touchscreen driver works fine for 2.4.0-test1-ac7-rmk1-np6 > but not working at all when I use 2.4.0-test1-ac19-rmk1-np1. Are > there any big changes regarding IRQs or timers between these two > kernels? I've replaced GRER |= (1 << 23) with set_GPIO_IRQ_edge( > (1 << 23), GPIO_RISING_EDGE) but that isn't a fix... You must use set_GPIO_IRQ_edge() before request_irq(), and use IRQ_GPIO23 as the IRQ number argument to request_irq(), not 23 directly. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 19:17:34 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134SUv-0000W7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 19:11:42 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134SUu-0000Vw-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 19:11:40 +0100 Received: from server01.asi (akron-80-157.ar51.net [208.225.80.157]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id NAA22689; Tue, 20 Jun 2000 13:11:03 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Tue, 20 Jun 2000 14:12:24 -0400 Message-ID: From: Jeff Sutherland To: "Linux ARM kernel mailing list (E-mail)" Cc: "Nicolas Pitre (E-mail)" Subject: Technical question: problems with NFS root mounting on SA1110 (As sabet) Date: Tue, 20 Jun 2000 14:12:23 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk re: kernel 2.4.0-test1-ac19-rmk1-np1 Well, I figured out where it's getting into trouble, but I'm not sure how to fix it. First of all, there might be more than one show stopper issue that's keeping this from working. Earlier I modified spinlock handling in fs/namei.c which got me beyond that point, but now it's blowing up at line #661 of fs/open.c. The line: error = f->f_op->open(inode,f); never returns, and the kernel eventually crashes. I haven't figured out how the pointer value of open() gets initialized. Maybe it isn't, and that's the problem. Commenting out this line and sticking a value of -1 in error results in the kernel continuing to boot, and eventually results in a panic message: VFS: Cannot open root device "nfs" or 00:ff Please append a correct "root=" boot option Kernel panic: VFS: Unable to mount root fs on 00:ff which is what I'd expect. The network device is functional, although there's some mucking about in linuxrc that needs to be done (let's hope I don't have a chicken vs. egg problem here...) Any ideas? Regards, //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 20:43:30 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Tud-0001hr-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 20:42:19 +0100 Received: from [194.168.180.70] (helo=caramon.arm.linux.org.uk) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134TuZ-0001hH-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 20:42:18 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id UAA23052; Tue, 20 Jun 2000 20:41:33 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id UAA22229; Tue, 20 Jun 2000 20:40:00 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006201940.UAA22229@flint.arm.linux.org.uk> Subject: Re: NFS mounting of root filesystem on Assabet To: jsutherland@accelent.com (Jeff Sutherland) Date: Tue, 20 Jun 2000 20:40:00 +0100 (BST) Cc: nico@cam.org ("Nicolas Pitre (E-mail)"), linux-arm-kernel@lists.arm.linux.org.uk ("Linux ARM kernel mailing list (E-mail)") In-Reply-To: from "Jeff Sutherland" at Jun 20, 2000 10:17:29 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jeff Sutherland writes: > Has anyone successfully NFS mounted a root filesystem for their Assabet at > boot time? I'm running down what appears to be a spinlock issue in > 2.4.0-test1-ac19-rmk1-np1 in function get_write_access() in fs/openi.c. spinlocks don't do anything on ARM, except the irq versions which enable/disable interrupts. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 20:46:41 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Txb-0001kH-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 20:45:23 +0100 Received: from [194.168.180.70] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134TxZ-0001k8-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 20:45:22 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id UAA23075; Tue, 20 Jun 2000 20:44:49 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id UAA22256; Tue, 20 Jun 2000 20:43:17 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006201943.UAA22256@flint.arm.linux.org.uk> Subject: Re: NFS mounting of root filesystem on Assabet To: jsutherland@accelent.com (Jeff Sutherland) Date: Tue, 20 Jun 2000 20:43:17 +0100 (BST) Cc: nico@cam.org ("Nicolas Pitre (E-mail)"), linux-arm-kernel@lists.arm.linux.org.uk ("Linux ARM kernel mailing list (E-mail)") In-Reply-To: from "Jeff Sutherland" at Jun 20, 2000 10:17:29 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jeff Sutherland writes: > Has anyone successfully NFS mounted a root filesystem for their Assabet at > boot time? Oh, NFS root filesystems do work on these kernels - my EBSA110 is running root-NFS'd ac19 at the moment (will be running ac22 later tonight). _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 20:47:25 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134TzU-0001m9-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 20:47:20 +0100 Received: from [194.168.180.70] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134TzT-0001m0-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 20:47:19 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id UAA23092; Tue, 20 Jun 2000 20:46:20 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id UAA22273; Tue, 20 Jun 2000 20:44:47 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006201944.UAA22273@flint.arm.linux.org.uk> Subject: Re: Technical question: problems with NFS root mounting on SA1110 (As To: jsutherland@accelent.com (Jeff Sutherland) Date: Tue, 20 Jun 2000 20:44:47 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk ("Linux ARM kernel mailing list (E-mail)"), nico@cam.org ("Nicolas Pitre (E-mail)") In-Reply-To: from "Jeff Sutherland" at Jun 20, 2000 02:12:23 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jeff Sutherland writes: > Well, I figured out where it's getting into trouble, but I'm not sure how to > fix it. First of all, there might be more than one show stopper issue > that's keeping this from working. Earlier I modified spinlock handling in > fs/namei.c which got me beyond that point, but now it's blowing up at line > #661 of fs/open.c. The line: > > error = f->f_op->open(inode,f); It sounds very much like a compiler/binutils problem. What versions are you using to compile the kernel? Also, it may be useful to know what file it's trying to open - you can get this from the dentry structure. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 21:08:44 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134UK7-00026v-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 21:08:39 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134UK6-00026j-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 21:08:38 +0100 Received: from server01.asi (akron-80-157.ar51.net [208.225.80.157]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id PAA30706; Tue, 20 Jun 2000 15:07:59 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Tue, 20 Jun 2000 16:09:19 -0400 Message-ID: From: Jeff Sutherland To: "'Russell King - ARM Linux Admin'" Cc: "'\"Linux ARM kernel mailing list (E-mail)\"'" Subject: RE: Technical question: problems with NFS root mounting on SA1110 (As Date: Tue, 20 Jun 2000 16:09:13 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk >> Earlier I modified > spinlock handling in > > fs/namei.c which got me beyond that point, but now it's > blowing up at line > > #661 of fs/open.c. The line: > > > > error = f->f_op->open(inode,f); > > It sounds very much like a compiler/binutils problem. What > versions are you > using to compile the kernel? binutils 2.9.5.0.27, gcc -v says "gcc version 2.95.2 19991024 (release)" I've been watching the CVS on netwinder.org for a later set to try. Our 'pipe' is not big enough here at work to just download the sources willy-nilly... > Also, it may be useful to know what file it's trying to open > - you can get > this from the dentry structure. /dev/ttyS0. I'm still drilling down through all the layers. Currently in tty_open (drivers/char/tty_io.c) and failing on a call to whatever's being pointed at in tty->driver.open (hmm, looks like trouble in rs_open...). Since yours is working, and Nico hinted that it had something to do with serial drivers in the SA1100 stuff, I think I'm getting closer. //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 21:34:07 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Uib-0002TK-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 21:33:57 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Uia-0002TC-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 21:33:56 +0100 Date: Tue, 20 Jun 2000 16:33:48 -0400 (EDT) From: Nicolas Pitre To: Russell King - ARM Linux Admin cc: Jeff Sutherland , "Linux ARM kernel mailing list (E-mail)" Subject: Re: NFS mounting of root filesystem on Assabet In-Reply-To: <200006201943.UAA22256@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Tue, 20 Jun 2000, Russell King - ARM Linux Admin wrote: > Jeff Sutherland writes: > > Has anyone successfully NFS mounted a root filesystem for their Assabet at > > boot time? > > Oh, NFS root filesystems do work on these kernels - my EBSA110 is running > root-NFS'd ac19 at the moment (will be running ac22 later tonight). The problem with Assabet is the ethernet device is PCMCIA. So the initrd must contain the /linuxrc + PCMCIA tools to make a network card actually apear before the real root FS i.e. NFS is mounted. Thus the real problem is executing /linuxrc, not NFS root. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 21:40:36 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134Uox-0002ZS-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 21:40:31 +0100 Received: from [194.168.180.70] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134Uov-0002ZJ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 21:40:30 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id VAA23341; Tue, 20 Jun 2000 21:40:28 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id VAA22569; Tue, 20 Jun 2000 21:38:55 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006202038.VAA22569@flint.arm.linux.org.uk> Subject: Re: Technical question: problems with NFS root mounting on SA1110 To: jsutherland@accelent.com (Jeff Sutherland) Date: Tue, 20 Jun 2000 21:38:55 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk ("'\"Linux ARM kernel mailing list (E-mail)\"'") In-Reply-To: from "Jeff Sutherland" at Jun 20, 2000 04:09:13 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jeff Sutherland writes: > binutils 2.9.5.0.27 > gcc -v says "gcc version 2.95.2 19991024 (release)" Hmm, I'm using both 2.9.5.0.16 and 2.9.5.0.32 to build the 2.4 kernels, and my-own-special-brew-of-gcc and gcc 2.95.1 19990816. > /dev/ttyS0. I'm still drilling down through all the layers. Currently in > tty_open (drivers/char/tty_io.c) and failing on a call to whatever's being > pointed at in tty->driver.open (hmm, looks like trouble in rs_open...). > Since yours is working, and Nico hinted that it had something to do with > serial drivers in the SA1100 stuff, I think I'm getting closer. Sounds like its the problem. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 20 21:56:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134V4P-0002n6-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 20 Jun 2000 21:56:29 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134V4O-0002n0-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 20 Jun 2000 21:56:28 +0100 Received: from server01.asi (akron-80-157.ar51.net [208.225.80.157]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id PAA01700; Tue, 20 Jun 2000 15:55:49 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Tue, 20 Jun 2000 16:57:02 -0400 Message-ID: From: Jeff Sutherland To: "'Nicolas Pitre'" , "'Russell King - ARM Linux Admin'" Cc: "'Linux ARM kernel mailing list (E-mail)'" Subject: RE: NFS mounting of root filesystem on Assabet Date: Tue, 20 Jun 2000 16:56:59 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk > > Oh, NFS root filesystems do work on these kernels - my > EBSA110 is running > > root-NFS'd ac19 at the moment (will be running ac22 later tonight). > > The problem with Assabet is the ethernet device is PCMCIA. > So the initrd > must contain the /linuxrc + PCMCIA tools to make a network > card actually > apear before the real root FS i.e. NFS is mounted. Thus the > real problem > is executing /linuxrc, not NFS root. > > > Nicolas Right. Step one is to uncompress and mount the ramdisk as root. So far, so good. The error comes next when attempting to reopen the console on ttyS0. If I ignore the error for the moment, the next thing that happens is the open of linuxrc (which it finds), then it runs linuxrc which finishes bringing up the network, then the root file system is unmounted, the nfs file system is mounted as /, and (optionally) the ramdisk is remounted as /initrd. Then I should get a login prompt, but of course I get a kernel panic because opening of pseudo device /dev/nfs I've disabled. Something in drivers/char/serial.c rs_open is getting hung up, hope to figure it out real soon now... //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 03:21:58 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134a8p-00073s-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 03:21:23 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134a8n-00073V-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 03:21:22 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id KAA32247; Wed, 21 Jun 2000 10:21:53 +0800 Date: Wed, 21 Jun 2000 10:21:53 +0800 (CST) From: Tak-Shing Chan To: Nicolas Pitre cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Touchscreen/IRQ problem In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Nicolas, Thanks for your hints. By the way, should I use IRQ_GPIO23 or IRQ_GPIO_UCB1300_IRQ? I know they're the same but I wonder if you've any guildlines as to when to use which... Tak-Shing Chan On Tue, 20 Jun 2000, Nicolas Pitre wrote: > You must use set_GPIO_IRQ_edge() before request_irq(), and use IRQ_GPIO23 > as the IRQ number argument to request_irq(), not 23 directly. > > > Nicolas > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 05:07:43 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134bmv-0008Gc-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 05:06:53 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134bmt-0008GD-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 05:06:51 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id MAA01095 for ; Wed, 21 Jun 2000 12:07:35 +0800 Date: Wed, 21 Jun 2000 12:07:35 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Assabet LCD patch again Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi, Better LCD settings for Assabet. This patch also fixes the PBS bug while switching consoles. Tak-Shing Chan --- /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/video/sa1100fb.c.orig Mon Jun 19 10:44:13 2000 +++ /usr/src/linux-2.4.0-test1-ac19-rmk1-np1/drivers/video/sa1100fb.c Wed Jun 21 12:03:18 2000 @@ -849,15 +849,15 @@ LCCR0_LDM + LCCR0_BAM + LCCR0_ERM + LCCR0_Act + LCCR0_LtlEnd + LCCR0_DMADel(0); lcd_shadow.lccr1 = - LCCR1_DisWdth(var->xres) + LCCR1_HorSnchWdth(4) + - LCCR1_BegLnDel(30) + LCCR1_EndLnDel(30); + LCCR1_DisWdth(var->xres) + LCCR1_HorSnchWdth(6) + + LCCR1_BegLnDel(61) + LCCR1_EndLnDel(9); lcd_shadow.lccr2 = - LCCR2_DisHght(var->yres) + LCCR2_VrtSnchWdth(1) + - LCCR2_BegFrmDel(0) + LCCR2_EndFrmDel(0); + LCCR2_DisHght(var->yres) + LCCR2_VrtSnchWdth(2) + + LCCR2_BegFrmDel(3) + LCCR2_EndFrmDel(0); lcd_shadow.lccr3 = LCCR3_OutEnH + LCCR3_PixFlEdg + LCCR3_VrtSnchH + LCCR3_HorSnchH + LCCR3_ACBsCntOff + - LCCR3_ACBsDiv(2) + LCCR3_PixClkDiv(28); + LCCR3_ACBsDiv(2) + LCCR3_PixClkDiv(38); /* Set board control register to handle new color depth */ sa1100fb_assabet_set_truecolor(var->bits_per_pixel >= 16); @@ -1138,6 +1138,7 @@ DPRINTK("blank=%d info->modename=%s\n", blank, info->modename); if (blank) { + if (current_par.visual != FB_VISUAL_TRUECOLOR) for (i = 0; i < current_par.palette_size; i++) sa1100fb_palette_write(i, sa1100fb_palette_encode(i, 0, 0, 0, 0)); sa1100fb_disable_lcd_controller(); @@ -1176,6 +1177,9 @@ fb_display[con].var.activate = FB_ACTIVATE_NOW; DPRINTK("fb_display[%d].var.activate=%x\n", con, fb_display[con].var.activate); sa1100fb_set_var(&fb_display[con].var, con, info); + current_par.v_palette_base[0] = (current_par.v_palette_base[0] & + 0xcfff) | SA1100_PALETTE_MODE_VAL(current_par.bits_per_pixel); + return 0; } unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 11:30:14 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134hkr-0002JO-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 11:29:09 +0100 Received: from [194.64.24.3] (helo=pink.ma.choin.net ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134hkq-0002JH-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 11:29:08 +0100 Received: from RR ([195.180.132.131]) by pink.ma.choin.net (8.9.3/8.9.3) with SMTP id MAA67388 for ; Wed, 21 Jun 2000 12:28:12 +0200 (CEST) (envelope-from rrascher@ids-imaging.de) Received: by RR with Microsoft Mail id <01BFDB7B.EE8B7340@RR>; Wed, 21 Jun 2000 12:26:31 +0200 Message-ID: <01BFDB7B.EE8B7340@RR> From: Ralf Rascher To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Compilig Kernel 2.4.0-test1-ac22-rmk1 Date: Wed, 21 Jun 2000 12:26:10 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, I wanted to compile the kernel 2.4.0-test1-ac22 with the = patch_2_4_0-test1-ac22-rmk1 for an assabet board. I don=B4t want to use the LCD so I choose the = options "Standart/generic (dump) serial support". But I got an error messages: serial.c:300: `STD_COM4_FLAGS' undeclared here (not in a function) serial.c:300: initializer elemment is not constant serial.c:300: (near initialization for `rs_tabel[3].flags') serial.c:300: parse error before `EXTA_SERIAL_PORT_DEFNS' I also want to use the PCMCIA Netcard on the assabet so I enabled PCMCIA/CardBuss support and PCMCIA network device support. But I got an error message: drivers/pcmcia/pcmcia.o: In function `cis_readable': drivers/pcmcia/pcmcia.o(.text+0x4e20): undefined reference to = `__arch_ioremap' drivers/pcmcia/pcmcia.o(.text+0x4e20): relocation truncated to fit: = R_ARM_PC24 __arch_ioremap drivers/pcmcia/pcmcia.o(.text+0x4e98): undefined reference to = `__arch_ioremap' drivers/pcmcia/pcmcia.o(.text+0x4e98): relocation truncated to fit: = R_ARM_PC24 __arch_ioremap=20 And the file SA1100 was missing in the /include/asm/arch/ directory, so I copied it from an older kernelversion. So how can I enable PCMCIA = and the in/output dump to the serial port. I use gcc version 2.95.2 20000516 and = binutils 2.10. Thanks, Ralf unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 13:58:20 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134k4t-00040e-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 13:57:59 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134k4s-00040V-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 13:57:58 +0100 Received: from server01.asi (akron-80-186.ar51.net [208.225.80.186]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id IAA15424; Wed, 21 Jun 2000 08:57:38 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Wed, 21 Jun 2000 08:58:19 -0400 Message-ID: From: Jeff Sutherland To: "'Ralf Rascher'" , "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: RE: Compilig Kernel 2.4.0-test1-ac22-rmk1 Date: Wed, 21 Jun 2000 08:58:18 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk > Hello, >=20 > I wanted to compile the kernel 2.4.0-test1-ac22 with the=20 > patch_2_4_0-test1-ac22-rmk1 > for an assabet board. I don=B4t want to use the LCD so I choose=20 > the options > "Standart/generic (dump) serial support". But I got an error = messages: >=20 > serial.c:300: `STD_COM4_FLAGS' undeclared here (not in a function) > serial.c:300: initializer elemment is not constant > serial.c:300: (near initialization for `rs_tabel[3].flags') > serial.c:300: parse error before `EXTA_SERIAL_PORT_DEFNS' >=20 > I also want to use the PCMCIA Netcard on the assabet so I enabled > PCMCIA/CardBuss support and PCMCIA network device support. > But I got an error message: >=20 > drivers/pcmcia/pcmcia.o: In function `cis_readable': > drivers/pcmcia/pcmcia.o(.text+0x4e20): undefined reference to=20 > `__arch_ioremap' > drivers/pcmcia/pcmcia.o(.text+0x4e20): relocation truncated=20 > to fit: R_ARM_PC24 __arch_ioremap > drivers/pcmcia/pcmcia.o(.text+0x4e98): undefined reference to=20 > `__arch_ioremap' > drivers/pcmcia/pcmcia.o(.text+0x4e98): relocation truncated=20 > to fit: R_ARM_PC24 __arch_ioremap=20 >=20 > And the file SA1100 was missing in the /include/asm/arch/ directory, > so I copied it from an older kernelversion. So how can I=20 > enable PCMCIA and the > in/output dump to the serial port. I use gcc version 2.95.2=20 > 20000516 and binutils > 2.10. >=20 > Thanks, >=20 > Ralf You are missing a patch to support the Assabet. Use -ac19, not -ac22, = then apply Nicolas Pitre's patch -np1 from ftp.netwinder.org/users/n/nico. = After applying all the patches, simply 'make assabet_config', then make menuconfig, accept all defaults (but check for pc card support), make = dep, make zImage, and that's it. The default Assabet config is to not use = the LCD display, but put the console out serial port 1. Attach the = Neponset board, turn on Neponset support in config, and use its port3 instead if = you wish. //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 15:13:34 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134lEn-00053i-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 15:12:17 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134lEm-00053Y-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 15:12:16 +0100 Date: Wed, 21 Jun 2000 10:12:07 -0400 (EDT) From: Nicolas Pitre To: Tak-Shing Chan cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Touchscreen/IRQ problem In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, 21 Jun 2000, Tak-Shing Chan wrote: > Nicolas, > > Thanks for your hints. By the way, should I use IRQ_GPIO23 > or IRQ_GPIO_UCB1300_IRQ? I know they're the same but I wonder > if you've any guildlines as to when to use which... Of course IRQ_GPIO_UCB1300_IRQ is better since it's self-explanatory. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 21:36:35 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134rBB-0002Qe-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 21:32:57 +0100 Received: from [194.168.180.4] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134rB9-0002QW-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 21:32:56 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id VAA27595; Wed, 21 Jun 2000 21:00:58 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id UAA26501; Wed, 21 Jun 2000 20:59:19 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006211959.UAA26501@flint.arm.linux.org.uk> Subject: Re: Compilig Kernel 2.4.0-test1-ac22-rmk1 To: rrascher@ids-imaging.de (Ralf Rascher) Date: Wed, 21 Jun 2000 20:59:19 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk ('linux-arm-kernel@lists.arm.linux.org.uk') In-Reply-To: <01BFDB7B.EE8B7340@RR> from "Ralf Rascher" at Jun 21, 2000 12:26:10 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by caramon.arm.linux.org.uk id VAA27595 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Ralf Rascher writes: > I wanted to compile the kernel 2.4.0-test1-ac22 with the patch_2_4_0-te= st1-ac22-rmk1 > for an assabet board. I don=B4t want to use the LCD so I choose the opt= ions > "Standart/generic (dump) serial support". But I got an error messages: This is a 16550-type serial port, which is not what you have. > I also want to use the PCMCIA Netcard on the assabet so I enabled > PCMCIA/CardBuss support and PCMCIA network device support. I don't believe that the kernels PCMCIA support has been ported to the assabet yet. > And the file SA1100 was missing in the /include/asm/arch/ directory, > so I copied it from an older kernelversion. So how can I enable PCMCIA = and the > in/output dump to the serial port. I use gcc version 2.95.2 20000516 an= d binutils > 2.10. Ah, you need to get Nicos patch as well then. _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 21 21:48:26 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134rPp-0002bO-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 21 Jun 2000 21:48:05 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 134rPo-0002bH-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 21 Jun 2000 21:48:04 +0100 Date: Wed, 21 Jun 2000 16:47:43 -0400 (EDT) From: Nicolas Pitre To: Russell King - ARM Linux Admin cc: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Re: Compilig Kernel 2.4.0-test1-ac22-rmk1 In-Reply-To: <200006211959.UAA26501@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Wed, 21 Jun 2000, Russell King - ARM Linux Admin wrote: > I don't believe that the kernels PCMCIA support has been ported to > the assabet yet. Yes, it has. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 22 04:47:49 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 134xx7-0007sx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 22 Jun 2000 04:46:53 +0100 Received: from [139.175.250.58] (helo=mail2000.com.tw) by www.linux.org.uk with smtp (Exim 3.13 #1) id 134xx6-0007sU-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 22 Jun 2000 04:46:52 +0100 Received: By OpenMail Mailer;Thu, 22 Jun 2000 11:45:48 +0800 (CST) From: "Chien-Min Lee" Reply-To: cm_lee@mail2000.com.tw Subject: Re: Compilig Kernel 2.4.0-test1-ac22-rmk1 Message-ID: <961645547.73107.cm_lee@mail2000.com.tw> To: "nico" Cc: "linux-arm-kernel" , "linux" Date: Thu, 22 Jun 2000 11:45:47 +0800 (CST) MIME-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 8bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi: >> I don't believe that the kernels PCMCIA support has been ported to >> the assabet yet. >Yes, it has. >Nicolas BTW, Could the PCMCIA card service on kernel 2.4.0-test1-ac series be staticly linked into kernel?? I found that we can configure the PCMCIA card service to be linked into kernel when 'make config'. However, I do not know how to make my LP-E work under this static linked version... Could any one tell me?? Thanks kirk ----== Mailed via Openfind ==----- http://mail2000.com.tw/ ´£¨Ñ§K¶Oªº e-mail «H½cªA°È unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 22 10:55:30 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1353gr-0003sd-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 22 Jun 2000 10:54:29 +0100 Received: from [198.232.147.17] (helo=mail02-oak.pilot.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1353gq-0003sU-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 22 Jun 2000 10:54:28 +0100 Received: from Altera.COM (altera.com [137.57.1.1] (may be forged)) by mail02-oak.pilot.net with ESMTP id CAA20577 for ; Thu, 22 Jun 2000 02:54:27 -0700 (PDT) Received: from sj-gw01.altera.com by Altera.COM (8.8.8+Sun/SMI-4.1) id CAA07450; Thu, 22 Jun 2000 02:56:15 -0700 (PDT) Received: by sj-gw01.altera.com with Internet Mail Service (5.5.2650.21) id ; Thu, 22 Jun 2000 02:56:02 -0700 Message-ID: <82B1591A280FD411B91200D0B75B484A388674@sis-uk-msg01.altera.com> From: Clive Davies To: linux-arm-kernel@lists.arm.linux.org.uk Subject: which patches? Date: Thu, 22 Jun 2000 02:52:45 -0700 X-Mailer: Internet Mail Service (5.5.2650.21) Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I think I'm missing something here. If I start with the kernel sources in linux-2.4.0-test1, do I need to find and apply patches (something like) linux-2.4.0test-ac1 ... linux-2.4.0test-ac6 before I apply Russell's 2.4.0test1-ac6-rmk1? Or, do I start with the kernel sources and just apply 2.4.0test-ac6-rmk1 (I suspect that this is the wrong option). If its the former, can anyone tell me where to get the ac1-6 patches from? I've looked everywhere I can think of, but I can find no trace... TIA Clive unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 22 11:22:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 13547y-0004Ey-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 22 Jun 2000 11:22:30 +0100 Received: from [194.168.180.8] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 13547x-0004Eq-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 22 Jun 2000 11:22:29 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id LAA30373; Thu, 22 Jun 2000 11:22:23 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id LAA29145; Thu, 22 Jun 2000 11:20:44 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006221020.LAA29145@flint.arm.linux.org.uk> Subject: Re: which patches? To: cdavies@altera.com (Clive Davies) Date: Thu, 22 Jun 2000 11:20:44 +0100 (BST) Cc: linux-arm-kernel@lists.arm.linux.org.uk In-Reply-To: <82B1591A280FD411B91200D0B75B484A388674@sis-uk-msg01.altera.com> from "Clive Davies" at Jun 22, 2000 02:52:45 AM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Clive Davies writes: > I think I'm missing something here. If I start with the kernel sources in > linux-2.4.0-test1, do I need to find and apply patches (something like) Yes. > linux-2.4.0test-ac1 ... linux-2.4.0test-ac6 No. Just the 2.4.0test1-ac6 patch is needed. > can anyone tell me where to get the ac1-6 patches from? Your nearest kernel mirror - /pub/linux/kernel/people/alan/2.4.0test _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 22 16:17:41 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1358j3-0007zr-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 22 Jun 2000 16:17:05 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1358j2-0007zl-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 22 Jun 2000 16:17:04 +0100 Date: Thu, 22 Jun 2000 11:17:02 -0400 (EDT) From: Nicolas Pitre To: Chien-Min Lee cc: linux-arm-kernel , linux Subject: Re: Compilig Kernel 2.4.0-test1-ac22-rmk1 In-Reply-To: <961645547.73107.cm_lee@mail2000.com.tw> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Thu, 22 Jun 2000, Chien-Min Lee wrote: > Hi: > > >> I don't believe that the kernels PCMCIA support has been ported to > >> the assabet yet. > > >Yes, it has. > >Nicolas > > BTW, Could the PCMCIA card service on kernel 2.4.0-test1-ac series be staticly linked into kernel?? Kernel support may be linked into the kernel, but you'll still need the user space tools to configure your devices. > However, I do not know how to make my LP-E work under this static linked version... See http://www.cs.cmu.edu/~wearable/software/pcmcia-arm.html for more details and a link to a ramdisk image with all the required tools. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 23 16:12:02 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135V27-0003zU-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 23 Jun 2000 16:06:15 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135V26-0003zL-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 23 Jun 2000 16:06:14 +0100 Received: from crlwebmail.crl.dec.com (crlwebmail.crl.dec.com [16.11.0.2]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id LAA01362; Fri, 23 Jun 2000 11:06:07 -0400 (EDT) Received: by crlwebmail.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Fri, 23 Jun 2000 11:10:50 -0400 Message-ID: From: Jamey Hicks To: "'Tak-Shing Chan'" Cc: Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Assabet UDA1341/DMA patch Date: Fri, 23 Jun 2000 11:06:05 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk What is the status of this driver? Does anyone have it working without periodic artifacts in the sound? I sent Nico a patch that makes this work on the iPaq 3600 handheld (Bitsy). -Jamey > -----Original Message----- > From: Tak-Shing Chan [mailto:tchan.rd@idthk.com] > Sent: Tuesday, June 20, 2000 4:19 AM > To: Tak-Shing Chan > Cc: Nicolas Pitre; linux-arm-kernel@lists.arm.linux.org.uk > Subject: Re: Assabet UDA1341/DMA patch > > > Hi Nicolas, > > AUDIO_DMA_OVERSHOOT=2 and AUDIO_FRAGSIZE_DEFAULT=8190 works > fine. Another observation: odd fragsizes break things (the last > 16-bit sample became 8-bit). It looks like DMA overshoot is not > a problem at all because 8190 is the biggest possible even number. > Maybe renaming AUDIO_DMA_OVERSHOOT to AUDIO_EVEN_HACK is a better > idea. ;) > > Results so far: > * Odd fragsizes with tca=size-1 : failed > * Even fragsizes with tca=size-1: failed > * Odd fragsizes with tca=size : failed > * Even fragsizes with tca=size : passed > * Valid fragsize range : 4096-8190 > > Tak-Shing Chan > > On Tue, 20 Jun 2000, Tak-Shing Chan wrote: > > > Nicolas, > > > > I think that involves two things. First, regarding the > > meaning of DBTA--from repeated listening experiments (from 4096 > > to 8176) I'm positive that it's value and not value-1. The > > result of missing samples in digital recordings are just too > > obvious... [by the way 11.6.1.4 says TCA is in bytes.] > > > > Second, isn't the extra datum you detected caused by DMA > > overshoot (and the value is supposed to be fetched and ignored)? > > My patch explicitly takes DMA overshoot into account. > > > > It seems like it's safe to specify AUDIO_DMA_OVERSHOOT as 16 > > and AUDIO_FRAGSIZE_DEFAULT as 8176. Any confirmations? IMHO > > 8176 is big enough for most purposes. Meanwhile 64 and 4096 are > > ``failsafe'' values but not optimal. > > > > Tak-Shing Chan > > > > On Mon, 19 Jun 2000, Nicolas Pitre wrote: > > > >> However you're limitting the DMA buffer to 4096 bytes > which might be quite > >> small, especially if the system is under heavy load. > >> > >> The SA1100 manual explictly says that the DMA count > register is 13 bit > >> wide for a maximal DMA size of 8192 bytes, which is highly > adviseable for > >> audio. However 13 bits can represent a max value of > 0x1fff i.e. 8191. > >> Since most registers on the SA1100 are defined as reg = > value - 1, I > >> though it was the case for the DMA too, although it is not strictly > >> mentionned for the DMA count reg. What I've noticed > though is that even > >> if you specify 0 for the DMA count reg, you'll get at > least one datum > >> through the DMA engine which tend to confirm the -1 guess. > >> > >> Anybody can confirm this? > >> > >> > >> Nicolas > >> > > > > > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to > majordomo@lists.arm.linux.org.uk > > > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to > majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 23 17:31:23 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135WLK-00056j-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 23 Jun 2000 17:30:10 +0100 Received: from [212.250.233.4] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135WLI-00056d-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 23 Jun 2000 17:30:08 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id QAA03834 for ; Fri, 23 Jun 2000 16:45:47 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id QAA09374 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 23 Jun 2000 16:43:56 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006231543.QAA09374@flint.arm.linux.org.uk> Subject: Kernel config help text? To: linux-arm-kernel@lists.arm.linux.org.uk Date: Fri, 23 Jun 2000 16:43:55 +0100 (BST) X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Does anyone have any kernel configuration help text already written for the following options? CONFIG_SA1100_ASSABET CONFIG_ASSABET_NEPONSET CONFIG_SA1100_BITSY CONFIG_SA1100_BRUTUS CONFIG_SA1100_LART CONFIG_SA1100_THINCLIENT CONFIG_SA1100_GRAPHICSCLIENT CONFIG_SA1100_VICTOR CONFIG_PAGESIZE_16 CONFIG_CPU_ARM6 CONFIG_CPU_ARM7 CONFIG_CPU_SA110 CONFIG_DEBUG_LL CONFIG_DEBUG_DC21285_PORT CONFIG_SOUND_VIDC CONFIG_FB_SA1100 CONFIG_FONT_ACORN_8x8 CONFIG_ACORN_PARTITION CONFIG_ACORN_PARTITION_ICS CONFIG_ACORN_PARTITION_ADFS CONFIG_ACORN_PARTITION_POWERTEC CONFIG_ACORN_PARTITION_RISCIX If not, is anyone willing to write some help text for the above options? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 23 20:09:51 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135YpN-0007TJ-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 23 Jun 2000 20:09:21 +0100 Received: from [208.162.98.43] (helo=lhs.com) by www.linux.org.uk with smtp (Exim 3.13 #1) id 135YpL-0007TC-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 23 Jun 2000 20:09:20 +0100 Received: by lhs.com(Lotus SMTP MTA v4.6.6 (890.1 7-16-1999)) id 85256907.006A1C37 ; Fri, 23 Jun 2000 15:19:00 -0400 X-Lotus-FromDomain: LHS From: ebunce@lhsl.com To: Jamey Hicks cc: "'Tak-Shing Chan'" , Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Message-ID: <85256907.006A1AC1.00@lhs.com> Date: Fri, 23 Jun 2000 15:08:15 -0400 Subject: RE: Assabet UDA1341/DMA patch Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii Content-Disposition: inline Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I've got it working without audio artifacts that I can attribute to the driver. I have however been noticing some occasional lags and unresponsiveness from my Assabets with the current linux-2.4.0-test1-ac19-rmk1-np1 doing other activities, and not just with the sound. ie. I'll type to the shell with nothing beyond minimal system running and it will take a while to echo the characters and occasionally am getting a complaint from the serial driver about an overrun that I can't explain. I'd be curious to see your patch. I'd also be curious to know how much flash you have available for bootloader/kernel/boot disk in your Bitsy? Erik Jamey Hicks on 06/23/2000 11:06:05 AM To: "'Tak-Shing Chan'" cc: Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk (bcc: Erik Bunce/USER/US/LHS) Fax to: Subject: RE: Assabet UDA1341/DMA patch What is the status of this driver? Does anyone have it working without periodic artifacts in the sound? I sent Nico a patch that makes this work on the iPaq 3600 handheld (Bitsy). -Jamey > -----Original Message----- > From: Tak-Shing Chan [mailto:tchan.rd@idthk.com] > Sent: Tuesday, June 20, 2000 4:19 AM > To: Tak-Shing Chan > Cc: Nicolas Pitre; linux-arm-kernel@lists.arm.linux.org.uk > Subject: Re: Assabet UDA1341/DMA patch > > > Hi Nicolas, > > AUDIO_DMA_OVERSHOOT=2 and AUDIO_FRAGSIZE_DEFAULT=8190 works > fine. Another observation: odd fragsizes break things (the last > 16-bit sample became 8-bit). It looks like DMA overshoot is not > a problem at all because 8190 is the biggest possible even number. > Maybe renaming AUDIO_DMA_OVERSHOOT to AUDIO_EVEN_HACK is a better > idea. ;) > > Results so far: > * Odd fragsizes with tca=size-1 : failed > * Even fragsizes with tca=size-1: failed > * Odd fragsizes with tca=size : failed > * Even fragsizes with tca=size : passed > * Valid fragsize range : 4096-8190 > > Tak-Shing Chan > > On Tue, 20 Jun 2000, Tak-Shing Chan wrote: > > > Nicolas, > > > > I think that involves two things. First, regarding the > > meaning of DBTA--from repeated listening experiments (from 4096 > > to 8176) I'm positive that it's value and not value-1. The > > result of missing samples in digital recordings are just too > > obvious... [by the way 11.6.1.4 says TCA is in bytes.] > > > > Second, isn't the extra datum you detected caused by DMA > > overshoot (and the value is supposed to be fetched and ignored)? > > My patch explicitly takes DMA overshoot into account. > > > > It seems like it's safe to specify AUDIO_DMA_OVERSHOOT as 16 > > and AUDIO_FRAGSIZE_DEFAULT as 8176. Any confirmations? IMHO > > 8176 is big enough for most purposes. Meanwhile 64 and 4096 are > > ``failsafe'' values but not optimal. > > > > Tak-Shing Chan > > > > On Mon, 19 Jun 2000, Nicolas Pitre wrote: > > > >> However you're limitting the DMA buffer to 4096 bytes > which might be quite > >> small, especially if the system is under heavy load. > >> > >> The SA1100 manual explictly says that the DMA count > register is 13 bit > >> wide for a maximal DMA size of 8192 bytes, which is highly > adviseable for > >> audio. However 13 bits can represent a max value of > 0x1fff i.e. 8191. > >> Since most registers on the SA1100 are defined as reg = > value - 1, I > >> though it was the case for the DMA too, although it is not strictly > >> mentionned for the DMA count reg. What I've noticed > though is that even > >> if you specify 0 for the DMA count reg, you'll get at > least one datum > >> through the DMA engine which tend to confirm the -1 guess. > >> > >> Anybody can confirm this? > >> > >> > >> Nicolas > >> > > > > > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to > majordomo@lists.arm.linux.org.uk > > > > > > unsubscribe: body of `unsubscribe linux-arm-kernel' to > majordomo@lists.arm.linux.org.uk > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 23 20:21:56 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135Z1O-0007en-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 23 Jun 2000 20:21:46 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135Z1N-0007ec-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 23 Jun 2000 20:21:45 +0100 Received: from ims.crl.dec.com (ims.crl.dec.com [16.11.0.11]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id PAA05566; Fri, 23 Jun 2000 15:21:39 -0400 (EDT) Received: by ims.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Fri, 23 Jun 2000 15:26:37 -0400 Message-ID: From: Jamey Hicks To: "'ebunce@lhsl.com'" , Jamey Hicks Cc: "'Tak-Shing Chan'" , Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Assabet UDA1341/DMA patch Date: Fri, 23 Jun 2000 15:21:40 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Bitsy == Compaq iPAQ 3600: 32MB SDRAM, we're using 16MB of Flash. I sent you the patch from my unix account. What is needed to remove the audio artifacts (other than the ones attributable to other activity)? -Jamey unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 24 03:49:11 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135fzX-0004lx-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 24 Jun 2000 03:48:19 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135fzU-0004lW-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 24 Jun 2000 03:48:17 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id KAA12914; Sat, 24 Jun 2000 10:48:16 +0800 Date: Sat, 24 Jun 2000 10:48:16 +0800 (CST) From: Tak-Shing Chan To: Jamey Hicks cc: "'Tak-Shing Chan'" , Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jamey, Yes it's already working without any artifacts. I suppose Nico will integrate the changes in the next release. Tak-Shing Chan On Fri, 23 Jun 2000, Jamey Hicks wrote: > What is the status of this driver? Does anyone have it working without > periodic artifacts in the sound? unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 24 04:21:33 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135gVL-0005B2-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 24 Jun 2000 04:21:11 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135gVJ-0005Ad-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 24 Jun 2000 04:21:09 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id LAA13396; Sat, 24 Jun 2000 11:21:07 +0800 Date: Sat, 24 Jun 2000 11:21:07 +0800 (CST) From: Tak-Shing Chan To: Jamey Hicks cc: "'ebunce@lhsl.com'" , "'Tak-Shing Chan'" , Nicolas Pitre , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Jamey, (If you're using Nico's driver) DMA size. Tak-Shing Chan On Fri, 23 Jun 2000, Jamey Hicks wrote: > What is needed to remove the audio artifacts (other than the ones > attributable to other activity)? unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 24 04:31:04 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135gem-0005Hi-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 24 Jun 2000 04:30:56 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135gek-0005HL-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 24 Jun 2000 04:30:54 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id LAA13563; Sat, 24 Jun 2000 11:31:11 +0800 Date: Sat, 24 Jun 2000 11:31:11 +0800 (CST) From: Tak-Shing Chan To: Russell King - ARM Linux Admin cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Kernel config help text? In-Reply-To: <200006231543.QAA09374@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk I have two: > CONFIG_SA1100_ASSABET Say Y here if you are using the Intel(R) StrongARM(R) SA-1110 Microprocessor Development Board (also known as the Assabet). > CONFIG_FB_SA1100 This is a framebuffer device for the SA-1100 LCD Controller, originally written by Eric A. Thomas. See http://www.linux-fbdev.org/ for information on framebuffer devices. If you plan to use the LCD display with your SA-1100 system, say Y here. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sat Jun 24 17:35:01 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 135sr8-00051P-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sat, 24 Jun 2000 17:32:30 +0100 Received: from [158.152.220.239] (helo=tazenda.demon.co.uk ident=mail) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 135sr7-00051J-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 24 Jun 2000 17:32:29 +0100 Received: from kings-cross.london.uk.eu.org [::ffff:192.168.2.83] by tazenda.demon.co.uk with esmtp (Exim 3.11 #1 (Debian)) id 135sxD-0008TL-00; Sat, 24 Jun 2000 17:38:47 +0100 Received: from localhost ([::ffff:127.0.0.1] helo=kings-cross.london.uk.eu.org ident=phil) by kings-cross.london.uk.eu.org with esmtp (Exim 3.12 #1) id 135sxB-0006jw-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sat, 24 Jun 2000 17:38:45 +0100 X-Mailer: exmh version 2.1.1 10/15/1999 (debian) To: linux-arm-kernel@lists.arm.linux.org.uk Subject: arm7 cpu type Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Sat, 24 Jun 2000 17:38:45 +0100 From: Philip Blundell Message-Id: Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk In 2.4.0-test1-ac22-rmk2, arch/arm/config.in sets CONFIG_CPU_32v4=y for CLPS7500FE. I don't think this can be right, those chips have an ARM7 core and are v3 only. p. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 25 15:38:54 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136DX4-0002hc-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 25 Jun 2000 15:37:10 +0100 Received: from [194.168.180.100] (helo=caramon.arm.linux.org.uk ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136DX3-0002hW-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 25 Jun 2000 15:37:09 +0100 Received: from flint.arm.linux.org.uk (root@flint [192.168.0.4]) by caramon.arm.linux.org.uk (8.9.3/8.9.3) with ESMTP id PAA13191 for ; Sun, 25 Jun 2000 15:37:11 +0100 Received: (from linux@localhost) by flint.arm.linux.org.uk (8.9.3/8.9.3) id PAA26289 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 25 Jun 2000 15:34:50 +0100 From: Russell King - ARM Linux Admin Message-Id: <200006251434.PAA26289@flint.arm.linux.org.uk> Subject: Re: Kernel config help text? To: linux-arm-kernel@lists.arm.linux.org.uk Date: Sun, 25 Jun 2000 15:34:50 +0100 (BST) In-Reply-To: <200006231543.QAA09374@flint.arm.linux.org.uk> from "Russell King - ARM Linux Admin" at Jun 23, 2000 04:43:55 PM X-Mailer: ELM [version 2.5 PL1] MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk We still have the following symbols without descriptions: CONFIG_ASSABET_NEPONSET CONFIG_SA1100_BRUTUS CONFIG_SA1100_LART CONFIG_SA1100_THINCLIENT CONFIG_SA1100_GRAPHICSCLIENT CONFIG_SA1100_VICTOR CONFIG_PAGESIZE_16 CONFIG_CPU_ARM6 CONFIG_CPU_ARM7 CONFIG_CPU_SA110 CONFIG_SOUND_VIDC CONFIG_FONT_ACORN_8x8 CONFIG_ACORN_PARTITION CONFIG_ACORN_PARTITION_ICS CONFIG_ACORN_PARTITION_ADFS CONFIG_ACORN_PARTITION_POWERTEC CONFIG_ACORN_PARTITION_RISCIX Anyone willing to write some help text for the above options? _____ |_____| ------------------------------------------------- ---+---+- | | Russell King linux@arm.linux.org.uk --- --- | | | | http://www.arm.linux.org.uk/ / / | | +-+-+ --- -+- / | THE developer of ARM Linux |+| /|\ / | | | --- | +-+-+ ------------------------------------------------- /\\\ | unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Sun Jun 25 23:21:21 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136KjW-0006iE-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Sun, 25 Jun 2000 23:18:30 +0100 Received: from [24.1.10.94] (helo=maximus.xtrading.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136KjR-0006i6-00 for linux-arm-kernel@lists.arm.linux.org.uk; Sun, 25 Jun 2000 23:18:25 +0100 Received: (from fnorky@localhost) by maximus.xtrading.com (8.9.3/8.9.3) id PAA01779; Sun, 25 Jun 2000 15:07:41 -0600 Date: Sun, 25 Jun 2000 15:07:41 -0600 From: Doug Peterson To: linux-arm-kernel@lists.arm.linux.org.uk Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Kernel config help text? Message-ID: <20000625150741.A1657@6502group.org> References: <200006231543.QAA09374@flint.arm.linux.org.uk> <200006251434.PAA26289@flint.arm.linux.org.uk> Mime-Version: 1.0 Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="tThc/1wpZn/ma/RB" X-Mailer: Mutt 1.0.1i In-Reply-To: <200006251434.PAA26289@flint.arm.linux.org.uk>; from Russell King - ARM Linux Admin on Sun, Jun 25, 2000 at 03:34:50PM +0100 Organization: The StrongARM Linux Project X-URL: http://www.strongarmlinux.org X-OS: Linux 2.2.14 i686 Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk --tThc/1wpZn/ma/RB Content-Type: text/plain; charset=us-ascii On Sun, Jun 25, 2000 at 03:34:50PM +0100, Russell King - ARM Linux Admin wrote: > We still have the following symbols without descriptions: > CONFIG_SA1100_ASSABET Maybe this should be changed to CONFIG_SA1110_ASSABET as the system uses a SA-1110 not an SA-1100... Also, I'm not positive but I thing the Compaq iPAQ H3600 would also use this setting. My contacts at Intel tell me the iPAQ is basicly an Assabet. Here is my shot on the following systems. > CONFIG_ASSABET_NEPONSET Say Y here if you are using the Intel(R) StrongARM(R) SA-1110 Microprocessor Development Board (Assabet) with the SA-1111 Development Board (Nepon). > CONFIG_SA1100_BRUTUS Say Y here if you are using the Intel(R) StrongARM(R) SA-1100 Microprocessor Development Board (also known as the Brutus). > CONFIG_SA1100_LART Say Y here if you are using the Intel (R) StrongARM(R) SA-1100 based LART system. See http://www.lart.tudelft.nl/ for information on the LART. > CONFIG_SA1100_THINCLIENT Say Y here if you are using an Applied Data Systems Intel(R) StrongARM(R) SA-1100 based Thin Client SBC. See http://www.flatpanels.com/ for information on this system. > CONFIG_SA1100_GRAPHICSCLIENT Say Y here if you are using an Applied Data Systems Intel(R) StrongARM(R) SA-1100 based Graphics Client SBC. See http://www.flatpanels.com/ for information on this system. > CONFIG_SA1100_VICTOR Say Y here if you are using a Visu Aide Intel(R) StrongARM(R) SA-1100 based Victor Digital Talking Book Reader. See http://www.visuaide.com/pagevictor.en.html for information on this system. I don't have any information of the following (yet). > CONFIG_PAGESIZE_16 > CONFIG_CPU_ARM6 > CONFIG_CPU_ARM7 > CONFIG_CPU_SA110 > CONFIG_SOUND_VIDC > CONFIG_FONT_ACORN_8x8 > CONFIG_ACORN_PARTITION > CONFIG_ACORN_PARTITION_ICS > CONFIG_ACORN_PARTITION_ADFS > CONFIG_ACORN_PARTITION_POWERTEC > CONFIG_ACORN_PARTITION_RISCIX -Doug p.s. Oh yea, Hi. I'm new to the list. --tThc/1wpZn/ma/RB Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.1 (GNU/Linux) Comment: Hi there iD8DBQE5VnSb1tiWnfqonWwRAr9RAJoDPAKG3Ro9Y5S72A/gzSGHR6NIMACfYxhO 0P6uo9Fn92L7IOtfiNwAL8I= =NK79 -----END PGP SIGNATURE----- --tThc/1wpZn/ma/RB-- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 01:05:40 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136MP3-0007rs-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 01:05:29 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136MP2-0007rk-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 01:05:28 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id UAA31043; Sun, 25 Jun 2000 20:04:35 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Sun, 25 Jun 2000 20:04:35 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Jamey Hicks cc: "'Tak-Shing Chan'" , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Assabet UDA1341/DMA patch In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Fri, 23 Jun 2000, Jamey Hicks wrote: > > What is the status of this driver? Does anyone have it working without > periodic artifacts in the sound? I adjusted the DMA buffer length to 4096 since the maximum DMA length seems to be 8191 rather than 8192. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 02:25:46 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136NeR-0000G7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 02:25:27 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136NeQ-0000G1-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 02:25:26 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id VAA31082; Sun, 25 Jun 2000 21:24:35 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Sun, 25 Jun 2000 21:24:35 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Doug Peterson cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Kernel config help text? In-Reply-To: <20000625150741.A1657@6502group.org> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; CHARSET=US-ASCII Content-ID: Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Sun, 25 Jun 2000, Doug Peterson wrote: > On Sun, Jun 25, 2000 at 03:34:50PM +0100, Russell King - ARM Linux Admin wrote: > > We still have the following symbols without descriptions: > > > CONFIG_SA1100_ASSABET > Maybe this should be changed to CONFIG_SA1110_ASSABET as the system uses a > SA-1110 not an SA-1100... Also, I'm not positive but I thing the > Compaq iPAQ H3600 would also use this setting. My contacts at Intel tell > me the iPAQ is basicly an Assabet. In this case the "SA1100" should be seen a the generic architecture name which all SA11x0 CPUs have in common, like the "footbridge" or "ARC" architectures under the generic ARM architecture. For the same reason you can consider products using the 80386, 80486, Pentium and friends part of the same architecture named "i386". However the strict SA1100 CPU name appears in the bootlog and /proc/cpuinfo. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 02:46:49 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136Nyj-0000Vf-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 02:46:25 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136Nyi-0000VP-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 02:46:24 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id VAA31123; Sun, 25 Jun 2000 21:45:38 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Sun, 25 Jun 2000 21:45:38 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Russell King - ARM Linux Admin cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Kernel config help text? In-Reply-To: <200006251434.PAA26289@flint.arm.linux.org.uk> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Sun, 25 Jun 2000, Russell King - ARM Linux Admin wrote: > We still have the following symbols without descriptions: > > CONFIG_ASSABET_NEPONSET Enabling this will add suport for the extra peripherals provided by Assabet's expansion board called Neponset. The safe answer is N. Note that you can't use PCMCIA on a stand-alone Assabet with this option enabled at the moment. This should not matter in the future. WARNING: never try to use PCMCIA on the Assabet board while the Neponset expansion board is present at the risk of damaging your hardware. > CONFIG_SA1100_BRUTUS > CONFIG_SA1100_VICTOR > CONFIG_SA1100_THINCLIENT The first lines provided in Documentation/arm/SA1100/* for the above names could be used. > CONFIG_SA1100_GRAPHICSCLIENT The same text as for the ThinClient can be used. Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 10:36:23 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136VEl-0005gm-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 10:31:27 +0100 Received: from [195.134.130.75] (helo=pip5.mpl.ch) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136VEk-0005gM-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 10:31:26 +0100 Received: from elsoft.ch (pcdm.elsoft.ch [195.134.130.116]) by pip5.mpl.ch (8.9.3/8.9.3) with ESMTP id LAA23270 for ; Mon, 26 Jun 2000 11:31:05 +0200 Message-ID: <39572288.A3D01C33@elsoft.ch> Date: Mon, 26 Jun 2000 11:29:44 +0200 From: David =?iso-8859-1?Q?M=FCller?= (ELSOFT AG) Organization: ELSOFT AG X-Mailer: Mozilla 4.73 [en] (WinNT; U) X-Accept-Language: de-CH,de,en MIME-Version: 1.0 To: linux-arm-kernel@lists.arm.linux.org.uk Subject: SA and PCI cache coherency ? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi I have to make a driver for a PCI bus master device (NIC) aware of the SA PCI cache coherency problem. Is there a common guideline to do this (I've found nothing on http://www.arm.linux.org.uk/developer/)? Is there a "good" example of an already adapted driver showing me how to do it? TIA Dave unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 12:54:35 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136XSz-0007Q4-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 12:54:17 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136XSy-0007Px-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 12:54:16 +0100 Received: from crlwebmail.crl.dec.com (crlwebmail.crl.dec.com [16.11.0.2]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id HAA23368; Mon, 26 Jun 2000 07:54:18 -0400 (EDT) Received: by crlwebmail.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Mon, 26 Jun 2000 07:59:02 -0400 Message-ID: From: Jamey Hicks To: =?iso-8859-1?Q?=27David_M=FCller=27?= , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: SA and PCI cache coherency ? Date: Mon, 26 Jun 2000 07:54:20 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk David, There is a new interface in Linux for handling this problem. It is documented in Documentation/DMA-mapping.txt. This documentation does = not mention StrongArm, but use of pci_alloc_consistent() or = pci_map_single() will enable you to solve the data cache coherency problem. Also, see Documentation/pci for the new PCI driver structure that = better encapsulates probing and supports hot-plug PCI and Cardbus. We modified the ethernet driver in drivers/net/pcnet32.c to use pci_alloc_consistent so that it would work on a StrongArm = SA110/footbridge. It's in the mainstream kernel as of 2.4.0-test1. We also have = submitted a similar patch to drivers/usb/usb-ohci.c. -Jamey Hicks > -----Original Message----- > From: David M=FCller [mailto:d.mueller@elsoft.ch] > Sent: Monday, June 26, 2000 5:30 AM > To: linux-arm-kernel@lists.arm.linux.org.uk > Subject: SA and PCI cache coherency ? >=20 >=20 > Hi >=20 > I have to make a driver for a PCI bus master device (NIC) aware of > the SA PCI cache coherency problem. >=20 > Is there a common guideline to do this (I've found nothing on > http://www.arm.linux.org.uk/developer/)? >=20 > Is there a "good" example of an already adapted driver > showing me how to do it? >=20 >=20 > TIA=20 >=20 > Dave >=20 >=20 > unsubscribe: body of `unsubscribe linux-arm-kernel' to=20 > majordomo@lists.arm.linux.org.uk >=20 unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 13:25:37 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136Xwz-0007q5-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 13:25:17 +0100 Received: from [194.64.24.3] (helo=pink.ma.choin.net ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136Xwy-0007py-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 13:25:16 +0100 Received: from RR ([195.180.132.139]) by pink.ma.choin.net (8.9.3/8.9.3) with SMTP id OAA73917 for ; Mon, 26 Jun 2000 14:24:18 +0200 (CEST) (envelope-from rrascher@ids-imaging.de) Received: by RR with Microsoft Mail id <01BFDF79.FC3DD210@RR>; Mon, 26 Jun 2000 14:22:40 +0200 Message-ID: <01BFDF79.FC3DD210@RR> From: Ralf Rascher To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: PCMCIA Network Setup on Assabet Date: Mon, 26 Jun 2000 14:22:21 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, thanks to Russell and Jeff for your tips, meanwhile the kernel is running. But I have another problem, I can't setup the pcmcia network card on the assabet (I have no Neponset). I downloaded the pcmcia ramdisk from http://www.cs.cmu.edu/~wearable/software/pcmcia-arm.html but I can't mount that ramdisk, so I don't know how to set up the network. Is there another ramdisk or howto available about how to set up the pcmcia network on the assabet. Thanks Ralf unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 15:17:39 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136ZhT-0000po-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 15:17:23 +0100 Received: from [208.225.80.11] (helo=ice.ar51.net) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136ZhR-0000pi-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 15:17:21 +0100 Received: from server01.asi (akron-80-154.ar51.net [208.225.80.154]) by ice.ar51.net (8.9.3/8.8.7) with ESMTP id KAA27233; Mon, 26 Jun 2000 10:17:18 -0400 Received: by SERVER01 with Internet Mail Service (5.5.2448.0) id ; Mon, 26 Jun 2000 10:18:53 -0400 Message-ID: From: Jeff Sutherland To: "'Ralf Rascher'" , "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: RE: PCMCIA Network Setup on Assabet Date: Mon, 26 Jun 2000 10:18:44 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2448.0) Content-Type: text/plain Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk > Hello, > > thanks to Russell and Jeff for your tips, > meanwhile the kernel is running. But I have another problem, > I can't setup the pcmcia network card on the assabet > (I have no Neponset). I downloaded the pcmcia ramdisk from > http://www.cs.cmu.edu/~wearable/software/pcmcia-arm.html > but I can't mount that ramdisk, so I don't know how to set up the > network. Is there another ramdisk or howto available > about how to set up the pcmcia network on the assabet. > > Thanks > > Ralf First of all, I'm going to assume that you can boot your Assabet and have some kind of ramdisk to run. Linux/Documentation/ramdisk.txt has everything you need to know to create and manipulate ramdisk images on a Linux host. BEWARE that the ramdisk size limit by default is 4096 Megs on a i386 host. This could be the source of your problem. Nico's and John's ramdisks are larger than that. Reboot your pc with 'ramdisk=8192' at the lilo boot prompt, or add 'ramdisk=8192' to lilo.conf and rerun lilo, then you won't have to remember to do it every time you reboot. (Since that's a very infrequent occurrence with most Linux machines one is quite likely to forget custom boot keywords and parameters :D Step 1: set up PCMCIA support. The easy way is to just compile in support (instead of using modules) for PCMCIA and for the NE2000 compatible netcard network driver. Networking support also has to be enabled in the kernel. The latest patches from Nico have all this set up by default if you first do 'make assabet_config' after completing patching of the kernel. Step 2: Even if you don't use it for booting your system, you should at least study John Dorsey's ramdisk's contents. Specifically a directory, /etc/pcmcia, contains the scripts and config files necessary for starting up PCMCIA network servces. /etc/rc.d/init.d/pcmcia is the boot script that starts up PCMCIA support at boot time. Make a link in /etc/rc.d/rc3.d called S66pcmcia and point it at ../init.d/pcmcia. Two binary executables, /sbin/cardmgr and /sbin/cardctl are also necessary. Step 3: Configure the network scripts in /etc/pcmcia to suit your local environment (ip address, gateway addr, etc.). Of course you should set up all the other regular networking stuff that any typical Linux box would need, such as /etc/hosts, resolv.conf, /etc/sysconfig/network-scripts/..., the usual... Cardmgr is started by S66pcmcia and runs as a daemon. When it detects insertion of a card it id's it and locates an appropriate driver and binds it to the socket. For the Socket compact flash Ethernet card it also runs a startup script in /etc/pcmcia which brings up the network. (Start a lo interface manually or put it in a boot script somewhere, too. Makes testing handy, eg.: 'ifup lo') Sometimes cardmgr doesn't see the insertion of the card for some reason. Do 'cardctl insert' to get cardmgr to recoginze it. 'cardctl status' is useful also. HTH, //Jeff unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Mon Jun 26 16:10:43 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136aVL-0001Vg-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Mon, 26 Jun 2000 16:08:55 +0100 Received: from [205.250.170.220] (helo=xanadu.vip.ca) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136aVK-0001Va-00 for linux-arm-kernel@lists.arm.linux.org.uk; Mon, 26 Jun 2000 16:08:54 +0100 Date: Mon, 26 Jun 2000 11:02:51 -0400 (EDT) From: Nicolas Pitre To: Jeff Sutherland cc: "'Ralf Rascher'" , "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: RE: PCMCIA Network Setup on Assabet In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk On Mon, 26 Jun 2000, Jeff Sutherland wrote: > First of all, I'm going to assume that you can boot your Assabet and have > some kind of ramdisk to run. Linux/Documentation/ramdisk.txt has everything > you need to know to create and manipulate ramdisk images on a Linux host. > BEWARE that the ramdisk size limit by default is 4096 Megs on a i386 host. > This could be the source of your problem. Nico's and John's ramdisks are > larger than that. Reboot your pc with 'ramdisk=8192' at the lilo boot > prompt, or add 'ramdisk=8192' to lilo.conf and rerun lilo, then you won't > have to remember to do it every time you reboot. (Since that's a very > infrequent occurrence with most Linux machines one is quite likely to forget > custom boot keywords and parameters :D A simpler solution could use the loopback interface, like: mount ramdisk_img /mnt -o loop Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 27 08:56:38 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136qD7-0008T4-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 27 Jun 2000 08:55:09 +0100 Received: from [194.64.24.3] (helo=pink.ma.choin.net ident=root) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136qD5-0008Ss-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 27 Jun 2000 08:55:07 +0100 Received: from RR ([195.180.132.129]) by pink.ma.choin.net (8.9.3/8.9.3) with SMTP id JAA02609 for ; Tue, 27 Jun 2000 09:54:09 +0200 (CEST) (envelope-from rrascher@ids-imaging.de) Received: by RR with Microsoft Mail id <01BFE01D.6BA4D060@RR>; Tue, 27 Jun 2000 09:52:35 +0200 Message-ID: <01BFE01D.6BA4D060@RR> From: Ralf Rascher To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: PCMCIA Network Setup on Assabet Date: Tue, 27 Jun 2000 09:52:21 +0200 MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, thank you everybody for your help. I can now mount the pcmcia-ramdisk on my pc. I use a Debian 2.1 Linux with an old 2.0.36 Kernel. With that kernel I can mount Nicos Ramdisk, which is about 6.3 MBytes, but I can't mount John's pcmcia-ramdisk which is 7.7 MB. I upgraded my pc to kernel 2.2.14 and now I can mount the pcmcia-ramdisk on it. Ralf unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Tue Jun 27 19:01:34 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 136zf3-0000YV-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Tue, 27 Jun 2000 19:00:37 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 136zf2-0000YP-00 for linux-arm-kernel@lists.arm.linux.org.uk; Tue, 27 Jun 2000 19:00:36 +0100 Received: from crlwebmail.crl.dec.com (crlwebmail.crl.dec.com [16.11.0.2]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id OAA20484; Tue, 27 Jun 2000 14:00:37 -0400 (EDT) Received: by crlwebmail.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Tue, 27 Jun 2000 14:05:22 -0400 Message-ID: From: George France To: "'Doug Peterson'" , linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: Kernel config help text? Date: Tue, 27 Jun 2000 14:00:35 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello Doug; The iPAQ H3600 uses CONFIG_SA1100_BITSY. --George George France, france@crl.dec.com Cambridge Research Laboratory, Compaq Computer Corporation One Cambridgecenter MS: CRL Cambridge, MA 02142 USA -----Original Message----- From: Doug Peterson [mailto:doug@strongarmlinux.org] Sent: Sunday, June 25, 2000 5:08 PM To: linux-arm-kernel@lists.arm.linux.org.uk Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Kernel config help text? On Sun, Jun 25, 2000 at 03:34:50PM +0100, Russell King - ARM Linux Admin wrote: > We still have the following symbols without descriptions: > CONFIG_SA1100_ASSABET Maybe this should be changed to CONFIG_SA1110_ASSABET as the system uses a SA-1110 not an SA-1100... Also, I'm not positive but I thing the Compaq iPAQ H3600 would also use this setting. My contacts at Intel tell me the iPAQ is basicly an Assabet. Here is my shot on the following systems. > CONFIG_ASSABET_NEPONSET Say Y here if you are using the Intel(R) StrongARM(R) SA-1110 Microprocessor Development Board (Assabet) with the SA-1111 Development Board (Nepon). > CONFIG_SA1100_BRUTUS Say Y here if you are using the Intel(R) StrongARM(R) SA-1100 Microprocessor Development Board (also known as the Brutus). > CONFIG_SA1100_LART Say Y here if you are using the Intel (R) StrongARM(R) SA-1100 based LART system. See http://www.lart.tudelft.nl/ for information on the LART. > CONFIG_SA1100_THINCLIENT Say Y here if you are using an Applied Data Systems Intel(R) StrongARM(R) SA-1100 based Thin Client SBC. See http://www.flatpanels.com/ for information on this system. > CONFIG_SA1100_GRAPHICSCLIENT Say Y here if you are using an Applied Data Systems Intel(R) StrongARM(R) SA-1100 based Graphics Client SBC. See http://www.flatpanels.com/ for information on this system. > CONFIG_SA1100_VICTOR Say Y here if you are using a Visu Aide Intel(R) StrongARM(R) SA-1100 based Victor Digital Talking Book Reader. See http://www.visuaide.com/pagevictor.en.html for information on this system. I don't have any information of the following (yet). > CONFIG_PAGESIZE_16 > CONFIG_CPU_ARM6 > CONFIG_CPU_ARM7 > CONFIG_CPU_SA110 > CONFIG_SOUND_VIDC > CONFIG_FONT_ACORN_8x8 > CONFIG_ACORN_PARTITION > CONFIG_ACORN_PARTITION_ICS > CONFIG_ACORN_PARTITION_ADFS > CONFIG_ACORN_PARTITION_POWERTEC > CONFIG_ACORN_PARTITION_RISCIX -Doug p.s. Oh yea, Hi. I'm new to the list. unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 28 01:27:37 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1375en-0006lP-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 28 Jun 2000 01:24:45 +0100 Received: from [24.141.29.65] (helo=embeddedobjects.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1375em-0006lJ-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 28 Jun 2000 01:24:44 +0100 Received: from cgocable.net (softdnserr [::ffff:192.168.0.2]) (IDENT: hbelange) by embeddedobjects.org with esmtp; Tue, 27 Jun 2000 20:24:35 -0400 Message-ID: <3959469A.1BDDE881@cgocable.net> Date: Tue, 27 Jun 2000 20:28:10 -0400 From: Hugues Belanger X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Compaq Toolchain & 2.2.16 kernel References: <01BFE01D.6BA4D060@RR> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi all I need a little help getting started with my little project... I'm planning to port LRP (linux router project) to my LART board and I can't get even compile a kernel... I've downloaded Compaq's arm toolchain ... and extracted it to /data/lart/cross/bin ... I've added that to the path in .bash_profile... I've download kernel 2.2.16-1 and I could find any arm patch for this kernel, so I'm assuming it doesn't need any for ARM... Is that a bad assumption ? I applied all the LRP patches to the kernel tree... and complied it for i486 (that worked great)... I then changed the follwing in the top level Makefile: ARCH := arm CROSS_COMPILE = /data/lart/cross/bin/arm-linux- I get : Makefile:249: *** target file `symlinks' has both : and :: entries. Stop. when I type make menuconfig have you seen this before ? Thanks Hugues Belanger unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 28 02:33:04 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 1376hH-0007Wm-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 28 Jun 2000 02:31:23 +0100 Received: from [24.141.29.65] (helo=embeddedobjects.org) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 1376hG-0007Wf-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 28 Jun 2000 02:31:22 +0100 Received: from cgocable.net (softdnserr [::ffff:192.168.0.2]) (IDENT: hbelange) by embeddedobjects.org with esmtp; Tue, 27 Jun 2000 21:31:09 -0400 Message-ID: <39595632.B7B996A0@cgocable.net> Date: Tue, 27 Jun 2000 21:34:42 -0400 From: Hugues Belanger X-Mailer: Mozilla 4.72 [en] (X11; U; Linux 2.2.14-5.0 i586) X-Accept-Language: en MIME-Version: 1.0 To: "'linux-arm-kernel@lists.arm.linux.org.uk'" Subject: Re: Compaq Toolchain & 2.2.16 kernel References: <01BFE01D.6BA4D060@RR> <3959469A.1BDDE881@cgocable.net> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk UPDATE : OK I've found the Tool chain installation note, but I get the same error... Hugues Belanger wrote: > Hi all > > I need a little help getting started with my little project... > > I'm planning to port LRP (linux router project) to my LART board and I > can't get even compile a kernel... > > I've downloaded Compaq's arm toolchain ... and extracted it to > /data/lart/cross/bin ... > > I've added that to the path in .bash_profile... > > I've download kernel 2.2.16-1 and I could find any arm patch for this > kernel, so I'm assuming it doesn't need any for ARM... Is that a bad > assumption ? > > I applied all the LRP patches to the kernel tree... > and complied it for i486 (that worked great)... > > I then changed the follwing in the top level Makefile: > > ARCH := arm > > CROSS_COMPILE = /data/lart/cross/bin/arm-linux- > > I get : Makefile:249: *** target file `symlinks' has both : and :: > entries. Stop. > > when I type make menuconfig > > have you seen this before ? > > Thanks > > Hugues Belanger > > unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Wed Jun 28 06:54:29 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137Alk-0002S7-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Wed, 28 Jun 2000 06:52:16 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 137Ala-0002Rf-00 for linux-arm-kernel@lists.arm.linux.org.uk; Wed, 28 Jun 2000 06:52:07 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id NAA09746 for ; Wed, 28 Jun 2000 13:52:22 +0800 Date: Wed, 28 Jun 2000 13:52:22 +0800 (CST) From: Tak-Shing Chan To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Assabet touchscreen patch (cont.) Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hi Assabeters, Patch below fixes the IRQ and fops stuffs. Tak-Shing Chan --- /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c.orig Tue Jun 27 13:52:43 2000 +++ /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c Wed Jun 28 13:39:56 2000 @@ -6,7 +6,8 @@ #include #include #include -#include +#include +#include #include #ifdef DEBUG @@ -98,18 +99,11 @@ struct timer_list g_timer; struct file_operations sa1100_ts_fops = { - NULL, /* LSEEK */ - sa1100_ts_read, /* READ */ - NULL, /* WRITE */ - NULL, /* READDIR */ - sa1100_ts_poll, /* SELECT */ - sa1100_ts_ioctl, /* IOCTL */ - NULL, /* MMAP */ - sa1100_ts_open, /* OPEN */ - NULL, - sa1100_ts_release, /* CLOSE */ - NULL, - NULL + read: sa1100_ts_read, + poll: sa1100_ts_poll, + ioctl: sa1100_ts_ioctl, + open: sa1100_ts_open, + release: sa1100_ts_release }; //------------------------------------------------------------ @@ -467,16 +461,15 @@ BCR_set(BCR_CODEC_RST); /* Enable the MCP */ - Ser4MCCR0 = (1 << UGL_SA1100_MCCR0_MCE); - Ser4MCCR0 |= (1 << UGL_SA1100_MCCR0_ADM); + Ser4MCCR0 = MCCR0_MCE | MCCR0_SmpCnt; /* Enable the sa1100 interrupt on a rising edge */ /* Clear any old interrupts of sa1100 */ - GPDR |= (1 << 23); - GAFR |= (1 << 23); - GRER |= (1 << 23); - GEDR |= (1 << 23); + GPDR |= GPIO_UCB1300_IRQ; + GAFR |= GPIO_UCB1300_IRQ; + set_GPIO_IRQ_edge(GPIO_UCB1300_IRQ, GPIO_RISING_EDGE); + GEDR |= GPIO_UCB1300_IRQ; } if ( machine_is_graphicsclient()) { @@ -514,15 +507,19 @@ sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,0); sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,UGL_UCB1200_IO_TS_INT); - tmp = GEDR; - tmp |= UGL_SA1100_GRER_TS_PINS; - GEDR = tmp; +#if defined(CONFIG_SA1100_GRAPHICSCLIENT) + if (machine_is_graphicsclient()) { + tmp = GEDR; + tmp |= UGL_SA1100_GRER_TS_PINS; + GEDR = tmp; + } +#endif register_chrdev(0,"sa1100-ts",&sa1100_ts_fops); irq = -1; #if defined(CONFIG_SA1100_ASSABET) if (machine_is_assabet()) - irq = CODEC_IRQ_ASSABET; + irq = IRQ_GPIO_UCB1300_IRQ; #endif #if defined(CONFIG_SA1100_GRAPHICSCLIENT) if (machine_is_graphicsclient()) unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 29 02:33:42 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137TBF-0002Kv-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 29 Jun 2000 02:31:49 +0100 Received: from [24.200.137.248] (helo=xanadu.gn.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 137TBE-0002Kk-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 29 Jun 2000 02:31:48 +0100 Received: from localhost (nico@localhost) by xanadu.gn.com (8.9.3/8.9.3) with ESMTP id VAA10821; Wed, 28 Jun 2000 21:30:41 -0400 X-Authentication-Warning: xanadu.gn.com: nico owned process doing -bs Date: Wed, 28 Jun 2000 21:30:40 -0400 (EDT) From: Nicolas Pitre X-Sender: nico@xanadu.gn.com To: Tak-Shing Chan cc: linux-arm-kernel@lists.arm.linux.org.uk, SA-1100/Linux Subject: Re: Assabet touchscreen patch (cont.) In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk To all SA1100 driver writers out there, please read this, especially for the end... On Wed, 28 Jun 2000, Tak-Shing Chan wrote: > Hi Assabeters, > > Patch below fixes the IRQ and fops stuffs. There are few things here that are wrong regarding GPIO IRQ handling on SA1100's: > --- /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c.orig Tue Jun 27 13:52:43 2000 > +++ /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c Wed Jun 28 13:39:56 2000 > @@ -6,7 +6,8 @@ > #include > #include > #include > -#include > +#include > +#include Here it should be: #include #include > /* Enable the sa1100 interrupt on a rising edge */ > /* Clear any old interrupts of sa1100 */ > > - GPDR |= (1 << 23); > - GAFR |= (1 << 23); > - GRER |= (1 << 23); > - GEDR |= (1 << 23); > + GPDR |= GPIO_UCB1300_IRQ; > + GAFR |= GPIO_UCB1300_IRQ; > + set_GPIO_IRQ_edge(GPIO_UCB1300_IRQ, GPIO_RISING_EDGE); > + GEDR |= GPIO_UCB1300_IRQ; Here the GPIO direction should be set to input i.e.: GPDR &= ~GPIO_UCB1300_IRQ; And the GPIO Alternate Function (GAFR) must not be activated for the GPIO to act as an interrupt line. Also, since the new IRQ handling stuff, no driver code should touch to GEDR/GFER/GRER any longer. The IRQ handling code is taking care of them. > @@ -514,15 +507,19 @@ > sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,0); > sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,UGL_UCB1200_IO_TS_INT); > > - tmp = GEDR; > - tmp |= UGL_SA1100_GRER_TS_PINS; > - GEDR = tmp; > +#if defined(CONFIG_SA1100_GRAPHICSCLIENT) > + if (machine_is_graphicsclient()) { > + tmp = GEDR; > + tmp |= UGL_SA1100_GRER_TS_PINS; > + GEDR = tmp; > + } > +#endif Here neither the old nor the new version is correct. All this should disappear as mentionned above. In general, GPIO lines used as IRQ lines should be set up like this: For example, if GPIO 18 should be used as an IRQ line on a falling edge: GPDR &= ~GPIO_GPIO18; /* input direction */ set_GPIO_IRQ_edge( GPIO_GPIO18, GPIO_FALLING_EDGE ); request_irq( IRQ_GPIO18, irq_handler, ..... ); That's it! You even don't have to bother about GPIO >= 11 IRQs multiplexed on IRQ 11, etc. Everything is handled by the low-level IRQ handling code (for the curious, see include/asm/arch-sa1100/irq.h). The important thing is to *always* call set_GPIO_IRQ_edge() *before* request_irq(). Then you only need to bother with the real job in your irq_handler. Thanks! Nicolas unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 29 07:32:04 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137XoG-0005aG-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 29 Jun 2000 07:28:24 +0100 Received: from [193.145.240.2] (helo=tid.tid.es) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 137XoE-0005aA-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 29 Jun 2000 07:28:23 +0100 Received: from tid.es ([172.17.1.9]) by tid.tid.es (Netscape Messaging Server 4.15) with ESMTP id FWWKN601.V66; Thu, 29 Jun 2000 08:28:18 +0200 Message-ID: <395AEC03.458D2EFC@tid.es> Date: Thu, 29 Jun 2000 08:26:12 +0200 From: David Aparicio Campos X-Mailer: Mozilla 4.7 [en] (Win98; I) X-Accept-Language: en MIME-Version: 1.0 To: linux-arm-kernel@lists.arm.linux.org.uk Subject: 21285 memory inicialization Content-Type: multipart/mixed; boundary="------------BACABF843A2B9DDC309C214C" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk This is a multi-part message in MIME format. --------------BACABF843A2B9DDC309C214C Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hello all: I'm developing a firmware kernel loader for a SA110 + 21285 system (not an EBSA285, but a Chalice CATS). What I try to get is a very small loader & monitor with these functions: Inicialization (ARM core, RAM, and minimal coprocessor 15 + 21285 startup) Flash driver (AMD 29020 class) for on-board programming. 285 UART driver, with dumb terminal and zmodem binary upload protocol. PCI core driver, for inicialization and resource assignment. Small system monitor (read, write memory, and so on) Multipurpose load mode: test modules / "heavy" firmare modules / Linux kernel. Idea is to have a multilayer firmware, being this module the very internal one. I have based this on NetWinder BIOS and Russell King EBSA285 BIOS code. I have also seen some BLOB code, but i don't know very much about SA1100, it seems to me that there are several diferences.. *** sad story here *** I have succeed on a first version, with NetWinder based startup, but to my surprise, when I rearranged some code, I broke something. I have compared these, but I cannot find which code is damaged. Early inicialization is C - based (and compiler specific, then), very few assembler code. I tried a more specific inicialization, then studied EBSA285 BIOS (no, it seems not to work "out of the box" on this board), and changed to an 100% assembler "init.S" module, based on "crt0.S" Russell code. My idea was to merge development at some point, when I find what is going wrong with the public distribution. This version succeed in opening serial line, writing some messages, thet it died. It seems I have problems with exception management code (I am learning!!). So I carefully tried to avoid exceptions (just to have a starting point, of course I have to add it later). What I have got is a very simple monitor that: * Inicializes core & serial line. * Gives a prompt (and response :-) for memory reading and writing. When I try this, I get: 0000.0000-0400.0000: RAM space. OK to read and write (SDRAM init works). 4100.0000-4104.0000: ROM space. OK to read, but see (1). 4200.0000-4210.0000: 285 register space. OK to read and write. 7B00.0000-7C00.0000: PCI CSR space. see (2) other ranges: same than PCI CSR. (1): First thing is to read Flash ID, to select proper driver. This is done by writing a special sequence on ROM space (thank you Nicolas Pitre and Erik Mou for lighting on proper way to do this!!). My monitor hangs on first writing... (2): Accesed in order to inicialize PCI cards, my monitor hangs on first read!! ** end of sad story ** I have read SA-110 Tech Reference manual, ARM7 data sheet, and 285 Reference manual. My suspects are that It foes some unhandled "fault exception" from: Bad MMU inicialization (coprocessor #15). I tried to remove all code on this, and let reset values to go. It doesn't fix anything. Bad 285 inicialization. I'm lost on what I'm doing bad, but there are several registers to fiddle with. Solutions: - Properly handle exceptions. I don't have such black magic. - Properly initialize memory model in order to ensure access to offending zones. - Others?? Thank you all for reading such a long email!! Can anybody help?? --------------BACABF843A2B9DDC309C214C Content-Type: text/x-vcard; charset=us-ascii; name="daparic.vcf" Content-Transfer-Encoding: 7bit Content-Description: Card for David Aparicio Campos Content-Disposition: attachment; filename="daparic.vcf" begin:vcard n:Aparicio Campos;David tel;fax:+34-91-3374502 tel;work:+34-91-3374178 x-mozilla-html:TRUE org:Telefónica I+D;Gerencia 1200 adr:;;Emilio Vargas, 6;Madrid;;28043;Spain version:2.1 email;internet:daparic@tid.es title:R&D Engineer end:vcard --------------BACABF843A2B9DDC309C214C-- unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Thu Jun 29 20:22:27 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137jsW-00064M-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Thu, 29 Jun 2000 20:21:36 +0100 Received: from [192.58.206.2] (helo=crl.dec.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 137jsU-00064G-00 for linux-arm-kernel@lists.arm.linux.org.uk; Thu, 29 Jun 2000 20:21:35 +0100 Received: from crlwebmail.crl.dec.com (crlwebmail.crl.dec.com [16.11.0.2]) by crl.dec.com (8.8.8/RWD-1.2) with ESMTP id PAA07678; Thu, 29 Jun 2000 15:21:42 -0400 (EDT) Received: by crlwebmail.crl.dec.com with Internet Mail Service (5.5.2650.21) id ; Thu, 29 Jun 2000 15:26:28 -0400 Message-ID: From: George France To: "'David Aparicio Campos'" Cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: RE: 21285 memory inicialization Date: Thu, 29 Jun 2000 15:21:46 -0400 MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2650.21) Content-Type: text/plain; charset="iso-8859-1" Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello David; Try the boot loader from http://handhelds.org it has all the functions you want. Currently this boot load is used on the Compaq Personal Server which is a SA-110 and a 21285. I hope this helps, --George George France, france@crl.dec.com Cambridge Research Laboratory, Compaq Computer Corporation One Cambridgecenter MS: CRL Cambridge, MA 02142 USA -----Original Message----- From: David Aparicio Campos [mailto:daparic@tid.es] Sent: Thursday, June 29, 2000 2:26 AM To: linux-arm-kernel@lists.arm.linux.org.uk Subject: 21285 memory inicialization Hello all: I'm developing a firmware kernel loader for a SA110 + 21285 system (not an EBSA285, but a Chalice CATS). What I try to get is a very small loader & monitor with these functions: Inicialization (ARM core, RAM, and minimal coprocessor 15 + 21285 startup) Flash driver (AMD 29020 class) for on-board programming. 285 UART driver, with dumb terminal and zmodem binary upload protocol. PCI core driver, for inicialization and resource assignment. Small system monitor (read, write memory, and so on) Multipurpose load mode: test modules / "heavy" firmare modules / Linux kernel. Idea is to have a multilayer firmware, being this module the very internal one. I have based this on NetWinder BIOS and Russell King EBSA285 BIOS code. I have also seen some BLOB code, but i don't know very much about SA1100, it seems to me that there are several diferences.. *** sad story here *** I have succeed on a first version, with NetWinder based startup, but to my surprise, when I rearranged some code, I broke something. I have compared these, but I cannot find which code is damaged. Early inicialization is C - based (and compiler specific, then), very few assembler code. I tried a more specific inicialization, then studied EBSA285 BIOS (no, it seems not to work "out of the box" on this board), and changed to an 100% assembler "init.S" module, based on "crt0.S" Russell code. My idea was to merge development at some point, when I find what is going wrong with the public distribution. This version succeed in opening serial line, writing some messages, thet it died. It seems I have problems with exception management code (I am learning!!). So I carefully tried to avoid exceptions (just to have a starting point, of course I have to add it later). What I have got is a very simple monitor that: * Inicializes core & serial line. * Gives a prompt (and response :-) for memory reading and writing. When I try this, I get: 0000.0000-0400.0000: RAM space. OK to read and write (SDRAM init works). 4100.0000-4104.0000: ROM space. OK to read, but see (1). 4200.0000-4210.0000: 285 register space. OK to read and write. 7B00.0000-7C00.0000: PCI CSR space. see (2) other ranges: same than PCI CSR. (1): First thing is to read Flash ID, to select proper driver. This is done by writing a special sequence on ROM space (thank you Nicolas Pitre and Erik Mou for lighting on proper way to do this!!). My monitor hangs on first writing... (2): Accesed in order to inicialize PCI cards, my monitor hangs on first read!! ** end of sad story ** I have read SA-110 Tech Reference manual, ARM7 data sheet, and 285 Reference manual. My suspects are that It foes some unhandled "fault exception" from: Bad MMU inicialization (coprocessor #15). I tried to remove all code on this, and let reset values to go. It doesn't fix anything. Bad 285 inicialization. I'm lost on what I'm doing bad, but there are several registers to fiddle with. Solutions: - Properly handle exceptions. I don't have such black magic. - Properly initialize memory model in order to ensure access to offending zones. - Others?? Thank you all for reading such a long email!! Can anybody help?? unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 30 05:23:28 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137sHs-0004MX-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 30 Jun 2000 05:20:20 +0100 Received: from [210.184.5.125] (helo=pikachu.idthk.com) by www.linux.org.uk with esmtp (Exim 3.13 #1) id 137sHo-0004Lw-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 30 Jun 2000 05:20:17 +0100 Received: from localhost (thomas@localhost) by pikachu.idthk.com (8.9.3/8.8.7) with ESMTP id MAA18099; Fri, 30 Jun 2000 12:20:03 +0800 Date: Fri, 30 Jun 2000 12:20:03 +0800 (CST) From: Tak-Shing Chan To: Nicolas Pitre cc: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: Assabet touchscreen patch (cont.) In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Nicolas, Thanks for your correction. By the way do we really need tmp &= ~UGL_SA1100_GRER_TS_NOT_PINS? I suppose it has no effect, otherwise it leads to a second question: is it legal to use GPIO_IRQ_{fall,ris}ing_edge directly? Updated patch below. Tak-Shing Chan --- /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c.orig Tue Jun 27 13:52:43 2000 +++ /usr/src/linux-2.4.0-test2-ac1-rmk2-np1/drivers/char/sa1100_ts.c Fri Jun 30 12:06:13 2000 @@ -6,7 +6,8 @@ #include #include #include -#include +#include +#include #include #ifdef DEBUG @@ -98,18 +99,11 @@ struct timer_list g_timer; struct file_operations sa1100_ts_fops = { - NULL, /* LSEEK */ - sa1100_ts_read, /* READ */ - NULL, /* WRITE */ - NULL, /* READDIR */ - sa1100_ts_poll, /* SELECT */ - sa1100_ts_ioctl, /* IOCTL */ - NULL, /* MMAP */ - sa1100_ts_open, /* OPEN */ - NULL, - sa1100_ts_release, /* CLOSE */ - NULL, - NULL + read: sa1100_ts_read, + poll: sa1100_ts_poll, + ioctl: sa1100_ts_ioctl, + open: sa1100_ts_open, + release: sa1100_ts_release, }; //------------------------------------------------------------ @@ -467,31 +461,19 @@ BCR_set(BCR_CODEC_RST); /* Enable the MCP */ - Ser4MCCR0 = (1 << UGL_SA1100_MCCR0_MCE); - Ser4MCCR0 |= (1 << UGL_SA1100_MCCR0_ADM); + Ser4MCCR0 = MCCR0_MCE | MCCR0_SmpCnt; /* Enable the sa1100 interrupt on a rising edge */ /* Clear any old interrupts of sa1100 */ - GPDR |= (1 << 23); - GAFR |= (1 << 23); - GRER |= (1 << 23); - GEDR |= (1 << 23); + GPDR &= ~GPIO_UCB1300_IRQ; + set_GPIO_IRQ_edge(GPIO_UCB1300_IRQ, GPIO_RISING_EDGE); } if ( machine_is_graphicsclient()) { /* Enable the MCP */ - Ser4MCCR0 = (1 << UGL_SA1100_MCCR0_MCE); - - tmp = GRER; - tmp |= UGL_SA1100_GRER_TS_PINS; - tmp &= ~UGL_SA1100_GRER_TS_NOT_PINS; - GRER = tmp; - - - tmp = GEDR; - tmp |= UGL_SA1100_GRER_TS_PINS; - GEDR = tmp; + Ser4MCCR0 = MCCR0_MCE; + set_GPIO_IRQ_edge(UGL_SA1100_GRER_TS_PINS, GPIO_RISING_EDGE); } #ifdef DEBUG @@ -513,22 +495,17 @@ sa1100_ts_regread(UGL_UCB1200_IO_CLR_I); sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,0); sa1100_ts_regwrite(UGL_UCB1200_IO_CLR_I,UGL_UCB1200_IO_TS_INT); - - tmp = GEDR; - tmp |= UGL_SA1100_GRER_TS_PINS; - GEDR = tmp; - register_chrdev(0,"sa1100-ts",&sa1100_ts_fops); irq = -1; #if defined(CONFIG_SA1100_ASSABET) if (machine_is_assabet()) - irq = CODEC_IRQ_ASSABET; + irq = IRQ_GPIO_UCB1300_IRQ; #endif #if defined(CONFIG_SA1100_GRAPHICSCLIENT) if (machine_is_graphicsclient()) - irq = CODEC_IRQ_GRAPHICSCLIENT; + irq = IRQ_GPIO6; #endif - if(request_irq(irq, sa1100_ts_inthandler, 0, + if(request_irq(irq, sa1100_ts_inthandler, SA_INTERRUPT, "sa1100-ts",&g_dev_id) != 0) { printk("sa1100_ts_init: Could not get CODEC IRQ\n"); } unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk From owner-linux-arm-kernel@lists.arm.linux.org.uk Fri Jun 30 13:11:36 2000 Received: from majordomo by www.linux.org.uk with local (Exim 3.13 #1) id 137zd7-0001kV-00 for linux-arm-kernel-outgoingx@www.linux.org.uk; Fri, 30 Jun 2000 13:10:45 +0100 Received: from [210.230.34.106] (helo=so-fs.musicpod.co.jp) by www.linux.org.uk with smtp (Exim 3.13 #1) id 137zd6-0001kI-00 for linux-arm-kernel@lists.arm.linux.org.uk; Fri, 30 Jun 2000 13:10:44 +0100 Received: (qmail 23117 invoked from network); 30 Jun 2000 12:10:52 -0000 Received: from bak.side7.zetabits.com (HELO bak.zetabits.com) (imai@192.168.36.68) by so-fs.musicpod.co.jp with SMTP; 30 Jun 2000 12:10:52 -0000 Date: Fri, 30 Jun 2000 21:10:52 +0900 Message-ID: From: Kunihiko IMAI To: linux-arm-kernel@lists.arm.linux.org.uk Subject: Re: USB driver In-Reply-To: In your message of "Fri, 16 Jun 2000 17:04:04 -0400" <200006162104.RAA06343@highwind.crl.dec.com> References: <200006162104.RAA06343@highwind.crl.dec.com> User-Agent: Wanderlust/1.1.1 (Purple Rain) WEMI/1.13.7 (Shimada) CLIME/1.13.6 (=?ISO-2022-JP?B?GyRCQ2YlTj4xGyhC?=) MULE XEmacs/21.1 (patch 6) (Big Bend) (i386-vine-linux) MIME-Version: 1.0 (generated by WEMI 1.13.7 - "Shimada") Content-Type: text/plain; charset=US-ASCII Sender: owner-linux-arm-kernel@lists.arm.linux.org.uk Precedence: bulk Hello, At Fri, 16 Jun 2000 17:04:04 -0400, James Hicks wrote: > There still seems to be a problem when unplugging devices -- causing the > kernel to crash. I assume this doesn't happen on x86. I found that where this happens. But I haven't fixed it yet. The data structure is too complicated for me to trace completely. Would someone give me some advice? The crashed place is: On dl_del_list() in usb-ohci.c (with James patch + my patch), if (ed->state & ED_DEL) { /* set by sohci_free_dev */ struct ohci_device * dev = usb_to_ohci (ohci->dev[edINFO & 0x7F]); OHCI_FREE_TD (ohci, tdTailP); /* free dummy td */ ed->hwINFO = cpu_to_le32 (OHCI_ED_SKIP); ed->state = ED_NEW; /* if all eds are removed wake up sohci_free_dev */ if (!--dev->ed_cnt) wake_up (&op_wakeup); I observed it occurred `dev' was NULL. In this case, `--dev->ed_cnt' is trapped and got Oops dump. Thanks. _._. __._ _.. . .___ ._. _____ _... ._ _._ _.._. .____ .__. ... . _._ Kunihiko IMAI unsubscribe: body of `unsubscribe linux-arm-kernel' to majordomo@lists.arm.linux.org.uk