From owner-linux@cthulhu.engr.sgi.com Tue Sep 2 14:50:10 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id OAA344924; Tue, 2 Sep 1997 14:50:10 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id OAA12599 for linux-list; Tue, 2 Sep 1997 14:47:01 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id OAA12561 for ; Tue, 2 Sep 1997 14:46:57 -0700 Received: from lager.engsoc.carleton.ca (lager.engsoc.carleton.ca [134.117.69.26]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id OAA16783 for ; Tue, 2 Sep 1997 14:46:53 -0700 env-from (adevries@engsoc.carleton.ca) Received: from localhost (adevries@localhost) by lager.engsoc.carleton.ca (8.8.5/8.8.4) with SMTP id RAA06959 for ; Tue, 2 Sep 1997 17:47:13 -0400 Date: Tue, 2 Sep 1997 17:47:13 -0400 (EDT) From: Alex deVries To: linux@cthulhu.engr.sgi.com Subject: Booting off of sdb1... Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1037 Lines: 36 First off, thanks to input from Mike, I got my partition setup properly. Here's how it's done: In fx in IRIX, create an optional partition. This will create one huge partition at the beginning (sdb1), and a volume header (sdb2), which you won't use. It doesn't matter what type of partition you label sdb1, Linux doesn't recognize either. All this means is you may have to force the type recognition when doing a mount. >From within an nfs-rooted Linux, I mke2fs'd it, and copied all the files over from my nfsroot (esentially Ralf's root .cpio.gz). But, here's my problem: I've take my old kernel that boots with NFS root, and tried to boot it using: vmlinux root=/dev/sdb1 The kernel starts, and it finishes off with: Partition check: sda: sad1 sda2 sda3 sda4 sdb: sdb1 sdb2 sdb3 VFS: Mounted root (ext2 filesystem) readonly. : Can't open And then nothing. What am I missing? - Alex Alex deVries Success is realizing System Administrator attainable dreams. The EngSoc Project From owner-linux@cthulhu.engr.sgi.com Tue Sep 2 15:06:42 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA343279; Tue, 2 Sep 1997 15:06:42 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA18496 for linux-list; Tue, 2 Sep 1997 15:02:02 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA18452 for ; Tue, 2 Sep 1997 15:01:52 -0700 Received: from gatekeeper.qms.com (gatekeeper.qms.com [161.33.3.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via SMTP id PAA21543 for ; Tue, 2 Sep 1997 15:01:51 -0700 env-from (marks@sun470.sun470.rd.qms.com) Received: from sun470.rd.qms.com (sun470.qms.com) by gatekeeper.qms.com (4.1/SMI-4.1) id AA28960; Tue, 2 Sep 97 17:01:45 CDT Received: from speedy.rd.qms.com by sun470.rd.qms.com (4.1/SMI-4.1) id AA27618; Tue, 2 Sep 97 17:01:43 CDT Received: by speedy.rd.qms.com (8.8.2) id RAA26782; Tue, 2 Sep 1997 17:01:43 -0500 Date: Tue, 2 Sep 1997 17:01:43 -0500 Message-Id: <199709022201.RAA26782@speedy.rd.qms.com> From: Mark Salter To: adevries@engsoc.carleton.ca Cc: linux@cthulhu.engr.sgi.com In-Reply-To: (message from Alex deVries on Tue, 2 Sep 1997 17:47:13 -0400 (EDT)) Subject: Re: Booting off of sdb1... Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 767 Lines: 39 > The kernel starts, and it finishes off with: > Partition check: > sda: sad1 sda2 sda3 sda4 > sdb: sdb1 sdb2 sdb3 > VFS: Mounted root (ext2 filesystem) readonly. > : Can't open > And then nothing. > What am I missing? > - Alex > Alex deVries Success is realizing > System Administrator attainable dreams. > The EngSoc Project Try this patch to linux/arch/mips/sgi/prom/cmdline.c: *** cmdline.c.orig Tue Sep 2 16:59:26 1997 --- cmdline.c Tue Sep 2 15:59:04 1997 *************** *** 52,57 **** --- 52,59 ---- pic_cont: actr++; } + if (cp != &(arcs_cmdline[0])) /* get rid of trailing space */ + --cp; *cp = '\0'; #ifdef DEBUG_CMDLINE --Mark From owner-linux@cthulhu.engr.sgi.com Tue Sep 2 16:47:51 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA355101 for ; Tue, 2 Sep 1997 16:47:50 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA28519 for linux-list; Tue, 2 Sep 1997 16:46:24 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA28300 for ; Tue, 2 Sep 1997 16:45:39 -0700 Received: from lager.engsoc.carleton.ca (lager.engsoc.carleton.ca [134.117.69.26]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA23561 for ; Tue, 2 Sep 1997 16:45:37 -0700 env-from (adevries@engsoc.carleton.ca) Received: from localhost (adevries@localhost) by lager.engsoc.carleton.ca (8.8.5/8.8.4) with SMTP id TAA11131; Tue, 2 Sep 1997 19:45:59 -0400 Date: Tue, 2 Sep 1997 19:45:59 -0400 (EDT) From: Alex deVries Reply-To: Alex deVries To: Mark Salter cc: linux@cthulhu.engr.sgi.com Subject: Re: Booting off of sdb1... In-Reply-To: <199709022201.RAA26782@speedy.rd.qms.com> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1992 Lines: 48 On Tue, 2 Sep 1997, Mark Salter wrote: > > The kernel starts, and it finishes off with: > > Partition check: > > sda: sad1 sda2 sda3 sda4 > > sdb: sdb1 sdb2 sdb3 > > VFS: Mounted root (ext2 filesystem) readonly. > > : Can't open > > Try this patch to linux/arch/mips/sgi/prom/cmdline.c: Well, I would try that patch, but I am unable to compile kernels because I _always_ get this: make[1]: Entering directory `/usr/src/sgi/linux/arch/mips/kernel' mips-linux-gcc -D__KERNEL__ -I/usr/src/sgi/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -G 0 -mno-abicalls -fno-pic -D__GOGOGO__ -mcpu=r4600 -mips2 -pipe -c entry.S -o entry.o entry.S: Assembler messages: entry.S:208: Error: .previous without corresponding .section; ignored entry.S:208: Error: .previous without corresponding .section; ignored entry.S:209: Error: .previous without corresponding .section; ignored entry.S:209: Error: .previous without corresponding .section; ignored entry.S:216: Error: .previous without corresponding .section; ignored entry.S:216: Error: .previous without corresponding .section; ignored entry.S:218: Error: .previous without corresponding .section; ignored entry.S:218: Error: .previous without corresponding .section; ignored entry.S:219: Error: .previous without corresponding .section; ignored entry.S:219: Error: .previous without corresponding .section; ignored entry.S:220: Error: .previous without corresponding .section; ignored entry.S:220: Error: .previous without corresponding .section; ignored make[1]: *** [entry.o] Error 1 make[1]: Leaving directory `/usr/src/sgi/linux/arch/mips/kernel' make: *** [linuxsubdirs] Error 2 I'm using the gcc that's up on ftp.linux.sgi.com, and I'm using binutils 2.8.1, which is the latest I believe. _PLEASE_ can someone who is able to succesfully cross compile kernels post here what versions they are using, and where I can get them from. Not sorting this out brings my development efforts to a complete stop. - Alex From owner-linux@cthulhu.engr.sgi.com Tue Sep 2 17:06:51 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id RAA356209 for ; Tue, 2 Sep 1997 17:06:51 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA05086 for linux-list; Tue, 2 Sep 1997 17:05:57 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA05001 for ; Tue, 2 Sep 1997 17:05:42 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id RAA28778 for ; Tue, 2 Sep 1997 17:05:40 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id SAA19505; Tue, 2 Sep 1997 18:58:22 -0500 Date: Tue, 2 Sep 1997 18:58:22 -0500 Message-Id: <199709022358.SAA19505@athena.nuclecu.unam.mx> From: Miguel de Icaza To: adevries@engsoc.carleton.ca CC: marks@sun470.sun470.rd.qms.com, linux@cthulhu.engr.sgi.com In-reply-to: (message from Alex deVries on Tue, 2 Sep 1997 19:45:59 -0400 (EDT)) Subject: Re: Booting off of sdb1... X-Mexico: Este es un pais de orates, un pais amateur. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 938 Lines: 30 > > > Well, I would try that patch, but I am unable to compile kernels because I > _always_ get this: Probably you missed my post and Ralf's post on this issue. > I'm using the gcc that's up on ftp.linux.sgi.com, and I'm using binutils > 2.8.1, which is the latest I believe. You need to get the binutils tar file as it was made available on ftp.linux.sgi.com, not the cross-compiler kit, nor the 2.8.1 release from the FSF. > _PLEASE_ can someone who is able to succesfully cross compile kernels post > here what versions they are using, and where I can get them from. Not > sorting this out brings my development efforts to a complete stop. berenice-origin32$ mips-linux-as --version GNU assembler version 2.7 (mips-linux), using BFD version 2.7 berenice-origin32$ mycc --version 2.7.2 But then, those versions are not really the stock GNU versions, they are the ones available in ftp.linux.sgi.com Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Tue Sep 2 22:44:51 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id WAA612496 for ; Tue, 2 Sep 1997 22:44:51 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id WAA09942 for linux-list; Tue, 2 Sep 1997 22:44:19 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id WAA09934 for ; Tue, 2 Sep 1997 22:44:16 -0700 Received: from lager.engsoc.carleton.ca (lager.engsoc.carleton.ca [134.117.69.26]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id WAA07377 for ; Tue, 2 Sep 1997 22:44:14 -0700 env-from (adevries@engsoc.carleton.ca) Received: from localhost (adevries@localhost) by lager.engsoc.carleton.ca (8.8.5/8.8.4) with SMTP id BAA16207; Wed, 3 Sep 1997 01:44:34 -0400 Date: Wed, 3 Sep 1997 01:44:34 -0400 (EDT) From: Alex deVries Reply-To: Alex deVries To: Miguel de Icaza cc: linux@cthulhu.engr.sgi.com Subject: Re: Booting off of sdb1... In-Reply-To: <199709022358.SAA19505@athena.nuclecu.unam.mx> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1730 Lines: 47 On Tue, 2 Sep 1997, Miguel de Icaza wrote: > You need to get the binutils tar file as it was made available on > ftp.linux.sgi.com, not the cross-compiler kit, nor the 2.8.1 release > from the FSF. Okay, I'm getting closer. I believe the correct version is binutils-2.7.tar.gz, patched with ftp://ftp.linux.sgi.com/pub/src/binutils/binutils-2.7-4.diff.gz. I'd have used the patch for binutils-2.7.0.6, but I can't find the original anywhere. I did a ./configure --target=mips-big-linux (since mips-unknown-linux produces a functional little endian binutils, which is useless for my Indy). I end up with this error when I compile binutils: gcc -O2 -g -DTARGET_BYTES_BIG_ENDIAN -o as.new targ-cpu.o obj-format.o atof-targ.o app.o as.o atof-generic.o bignum-copy.o cond.o expr.o flonum-konst.o flonum-copy.o flonum-mult.o frags.o hash.o input-file.o input-scrub.o literal.o messages.o output-file.o read.o subsegs.o symbols.o write.o listing.o ecoff.o stabs.o sb.o macro.o e-mipself.o -L.../opcodes -lopcodes -L../bfd -lbfd ../libiberty/libiberty.a targ-cpu.o: In function `md_begin': /home/binutils-2.7/gas/targ-cpu.c:826: undefined reference to `bfd_mips_elf_swap_options_out' targ-cpu.o: In function `mips_elf_final_processing': /home/binutils-2.7/gas/targ-cpu.c:7863: undefined reference to `bfd_mips_elf32_swap_reginfo_out' /home/binutils-2.7/gas/targ-cpu.c:7879: undefined reference to `bfd_mips_elf64_swap_reginfo_out' make[1]: *** [as.new] Error 1 make[1]: Leaving directory `/home/binutils-2.7/gas' make: *** [all-gas] Error 2 NB: There's no problem with compiling with --target=mips-unknown-linux. Ideas? How have other people gotten around this? Thanks for getting me started again, Miguel. - Alex From owner-linux@cthulhu.engr.sgi.com Wed Sep 3 06:44:23 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id GAA779935 for ; Wed, 3 Sep 1997 06:44:22 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id GAA29555 for linux-list; Wed, 3 Sep 1997 06:43:42 -0700 Received: from dataserv.detroit.sgi.com (dataserv.detroit.sgi.com [169.238.128.2]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id GAA29542 for ; Wed, 3 Sep 1997 06:43:39 -0700 Received: from cygnus.detroit.sgi.com by dataserv.detroit.sgi.com via ESMTP (940816.SGI.8.6.9/930416.SGI) id JAA19245; Wed, 3 Sep 1997 09:43:34 -0400 Message-ID: <340D6986.CAD95422@cygnus.detroit.sgi.com> Date: Wed, 03 Sep 1997 09:43:34 -0400 From: Eric Kimminau Reply-To: eak@detroit.sgi.com Organization: Silicon Graphics, Inc X-Mailer: Mozilla 4.02 [en] (X11; I; IRIX 6.3 IP32) MIME-Version: 1.0 To: Mark Salter CC: adevries@engsoc.carleton.ca, linux@cthulhu.engr.sgi.com Subject: Re: Booting off of sdb1... References: <199709022201.RAA26782@speedy.rd.qms.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1730 Lines: 63 Does anyone have a built kernel with this patch applied? Thanks! Eric. Mark Salter wrote: > > > The kernel starts, and it finishes off with: > > > Partition check: > > sda: sad1 sda2 sda3 sda4 > > sdb: sdb1 sdb2 sdb3 > > VFS: Mounted root (ext2 filesystem) readonly. > > : Can't open > > > And then nothing. > > > What am I missing? > > > - Alex > > > Alex deVries Success is realizing > > System Administrator attainable dreams. > > The EngSoc Project > > Try this patch to linux/arch/mips/sgi/prom/cmdline.c: > > *** cmdline.c.orig Tue Sep 2 16:59:26 1997 > --- cmdline.c Tue Sep 2 15:59:04 1997 > *************** > *** 52,57 **** > --- 52,59 ---- > pic_cont: > actr++; > } > + if (cp != &(arcs_cmdline[0])) /* get rid of trailing space */ > + --cp; > *cp = '\0'; > > #ifdef DEBUG_CMDLINE > > --Mark -- Eric Kimminau System Engineer/RSA eak@detroit.sgi.com Silicon Graphics, Inc Voice: (248) 848-4455 39001 West 12 Mile Rd. Fax: (248) 848-5600 Farmington, MI 48331-2903 VNet Extension - 6-327-4455 "I speak my mind and no one else's." http://www.dcs.ex.ac.uk/~aba/rsa/perl-rsa-sig.html When confronted by a difficult problem, solve it by reducing it to the question, "How would the Lone Ranger handle this?" Windows 95: n. 32 bit extensions and a graphical shell for a 16 bit patch to an 8 bit operating system originally coded for a 4 bit microprocessor, written by a 2 bit company that can't stand 1 bit of competition. Author unknown. From owner-linux@cthulhu.engr.sgi.com Wed Sep 3 19:54:26 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA910118 for ; Wed, 3 Sep 1997 19:54:26 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA04333 for linux-list; Wed, 3 Sep 1997 19:53:50 -0700 Received: from oz.engr.sgi.com (oz.engr.sgi.com [150.166.61.27]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA04326; Wed, 3 Sep 1997 19:53:48 -0700 Received: (from ariel@localhost) by oz.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA24709; Wed, 3 Sep 1997 19:53:47 -0700 From: ariel@oz.engr.sgi.com (Ariel Faigon) Message-Id: <199709040253.TAA24709@oz.engr.sgi.com> Subject: Re: Linux/SGI To: ratfink@xtra.co.nz (Brendan Black) Date: Wed, 3 Sep 1997 19:53:47 -0700 (PDT) Cc: linux@cthulhu.engr.sgi.com (SGI/Linux mailing list) In-Reply-To: <340DF260.5D60941F@xtra.co.nz> from "Brendan Black" at Sep 4, 97 11:27:28 am Reply-To: ariel@cthulhu.engr.sgi.com (Ariel Faigon) Organization: Silicon Graphics Inc. X-Mailer: ELM [version 2.4 PL24 ME5a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 979 Lines: 34 : :Hi : :we don't have a spare Indy to play with, but have a :spare Challenge S (single processor R5000) : :I was wondering if you thing this machine is possible, it would be :used with a serial console (no X), for testing ;) :-- If you feel up to the challenge, sure, please go ahead. There may be a few assumptions in the current code which detects the newport graphics console that this testing may uncover. If you can fix this so it won't be Indy dependent, we win one more possible platform ! Moreover, since a challenge-S doesn't have graphics anyway Linux which is fast and compact may even outperform IRIX on some benchmarks. I would suggest that: 1) You join the mailing list: To: majordomo@engr.sgi.com Body: subscribe linux your@preferred.email.address if you have questions, ask... 2) If you have Challenge-S patches, e-mail them to the list and we can check that they didn't break anything and merge them into the source tree. -- Peace, Ariel From owner-linux@cthulhu.engr.sgi.com Wed Sep 3 20:41:46 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA912164 for ; Wed, 3 Sep 1997 20:41:46 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA12015 for linux-list; Wed, 3 Sep 1997 20:41:17 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA12010 for ; Wed, 3 Sep 1997 20:41:15 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA19593 for ; Wed, 3 Sep 1997 20:41:13 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA27779; Wed, 3 Sep 1997 22:34:20 -0500 Date: Wed, 3 Sep 1997 22:34:20 -0500 Message-Id: <199709040334.WAA27779@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@cthulhu.engr.sgi.com Subject: Need help with with irix_elfmap. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 2325 Lines: 56 I am trying to run the IRIX Xsgi server on top of Linux/SGI, and I am running into some very strange problems. If i let the program run loose, the code crashes miserably inside a function called _XSERVTransOpenCOTServer, just after the functions unlinks the file /usr/tmp/.Xshmtrans%d, and just before making the sysmp system call (this one I supposed is used as part of the usinit () libc routine). This is listing of the system calls done on IRIX for the X server around this point: mkdir("/tmp/.X11-unix", 0777) = -1 EEXIST (File exists) unlink("/tmp/.X11-unix/X0") = 0 bind(5, {sun_family=AF_UNIX, sun_path="/tmp/.X11-unix/X0"}, 19) = 0 listen(5, 5) = 0 umask(022) = 0 unlink("/usr/tmp/.Xshmtrans0") = 0 sysmp(0x420, 0x1, 0xe, 0x3d, 0) = 1 On Linux, this code is executed: mkdir("/tmp/.X11-unix", 0777) = -1 EEXIST (File exists) unlink("/tmp/.X11-unix/X0") = 0 bind(5, {sun_family=AF_UNIX, sun_path="/tmp/.X11-unix/X0"}, 19) = 0 listen(5, 5) = 0 umask(022) = 0 unlink("/usr/tmp/.Xshmtrans0") = 0 segfault Segmentation fault address = somewhere inside libc, return address suggests the usinit() libc routine was called (I guess this from an IRIX 6.4 machine where stock FSF-gdb happens to understand the symbolic information on the libc) Now, this *may* suggest that the problem is on the libc usinit() routine and that I ought to be figuring what happens inside of it, but if I run the program under gdb [1], and if I take extra care to single step around the region that calls usinit() the program won't crash at that point, it will just happen to crash at some point later at random places (ie, stepi over the code, then let it continue). So, does this ring any bell to anyone? I have the impression that the elfmap() routine in the IRIX kernel may be poking some values in the user address space and probably flushing the caches. Which is strange, since the Linux implementation uses do_mmap and that should take care of flushing any thing that should be flushed. Miguel. [1] I have a couple of patches to the gdb.tar.gz file that we got from the original DaveM-pack that I really need to commit. From owner-linux@cthulhu.engr.sgi.com Thu Sep 4 13:29:58 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id NAA1034143 for ; Thu, 4 Sep 1997 13:29:58 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id NAA17999 for linux-list; Thu, 4 Sep 1997 13:29:05 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id NAA17975 for ; Thu, 4 Sep 1997 13:29:02 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id NAA07954 for ; Thu, 4 Sep 1997 13:28:59 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id QAA08151; Thu, 4 Sep 1997 16:25:44 -0400 From: Mike Shaver Message-Id: <199709042025.QAA08151@neon.ingenia.ca> Subject: Re: Booting off of sdb1... In-Reply-To: <199709022201.RAA26782@speedy.rd.qms.com> from Mark Salter at "Sep 2, 97 05:01:43 pm" To: marks@sun470.sun470.rd.qms.com (Mark Salter) Date: Thu, 4 Sep 1997 16:25:44 -0400 (EDT) Cc: linux@cthulhu.engr.sgi.com (Linux/SGI list) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 554 Lines: 16 Thus spake Mark Salter: > Try this patch to linux/arch/mips/sgi/prom/cmdline.c: Fixed it for me. Thanks. I'll commit that to the tree and put a new kernel up on ftp.linux.sgi.com. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Chief System Architect -- will tame sendmail(8) for food #> #> "You are a very perverse individual, and I think I'd like to get to #> know you better." --- eric@reference.com From owner-linux@cthulhu.engr.sgi.com Thu Sep 4 19:54:10 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA1115337 for ; Thu, 4 Sep 1997 19:54:10 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA07587 for linux-list; Thu, 4 Sep 1997 19:52:43 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA07581 for ; Thu, 4 Sep 1997 19:52:42 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id TAA07029 for ; Thu, 4 Sep 1997 19:52:40 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA07574 for ; Thu, 4 Sep 1997 19:52:39 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id TAA24464 for ; Thu, 4 Sep 1997 19:52:37 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id VAA03103; Thu, 4 Sep 1997 21:45:31 -0500 Date: Thu, 4 Sep 1997 21:45:31 -0500 Message-Id: <199709050245.VAA03103@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: [Q: Linux/SGI] IRIX executable memory map. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 3390 Lines: 81 Hello guys, Ok, it seems our irix_elfmap routine is just fine, I just found out with a simple test case that the code is trying to access memory from the location at 0x200000 which is making my IRIX executables crash (this one is crashing inside usinit ()). Is there something magic about 0x20000 on IRIX that I should be aware of? This 0x20000 location is quite strange to me, since this is not a region of memory that happens to be part of the executable according to gdb (look at exhibit [1]). So, if it does not come from here, the obvious guess is that this memory region comes from mmaping something: either /dev/zero, or from the sys_sgi (ELFMAP) system call, but neither par nor strace show any trace of a virual mapping taking place in this region (exhibit [2] shows this). The exact location of the crash is shown on my third dump [3]. Best wishes, Miguel. Exhibit [1]: Local exec file image, from gdb. Entry point: 0x400870 0x00400134 - 0x00400147 is .interp 0x004001c8 - 0x004001e0 is .reginfo 0x004001e0 - 0x004002d8 is .dynamic 0x004002d8 - 0x004002ec is .liblist 0x0040050c - 0x0040070c is .dynsym 0x004002ec - 0x00400401 is .dynstr 0x00400404 - 0x0040050c is .hash 0x0040072c - 0x0040082c is .msym 0x0040082c - 0x00400870 is .MIPS.stubs 0x00400870 - 0x00400a00 is .text 0x00400a00 - 0x00400a20 is .init 0x10001000 - 0x10001010 is .rodata 0x10001010 - 0x10001048 is .got 0x10001048 - 0x10001054 is .bss Exhibit [2]: par output from running simple test program: 2mS usinit( 5349): execve(./usinit, 0x7fff2f64, 0x7fff2f6c) OK 5mS usinit( 5349): open(/lib/rld, O_RDONLY, 04) = 3 5mS usinit( 5349): read(3, <7f 45 4c 46 01 02 01 00 00 00 00>...,512) = 512 5mS usinit( 5349): elfmap(3, 0x7fff2d58, 2) = 0xfb60000 6mS usinit( 5349): close(3) OK 6mS usinit( 5349): getpagesize() = 4096 6mS usinit( 5349): open(/dev/zero, O_RDONLY, 01757022324) = 3 7mS usinit( 5349): mmap(0xfbc4000, 16384, PROT_WRITE|PROT_READ, MAP_PRIVATE, 3, 0) = 0xfbc4000 7mS usinit( 5349): close(3) OK 9mS usinit( 5349): syssgi(SGI_USE_FP_BCOPY, 0, 0x1dc, 0xffffffff, 0x7fff2d70, 0xfb8eee8) = 0 9mS usinit( 5349): open(/dev/zero, O_RDONLY, 01757022324) = 3 9mS usinit( 5349): mmap(0xfbc8000, 32768, PROT_WRITE|PROT_READ, MAP_PRIVATE, 3, 0) = 0xfbc8000 9mS usinit( 5349): close(3) OK 9mS usinit( 5349): syssgi(SGI_TOSSTSAVE) OK 9mS usinit( 5349): write(2, "Test\n", 5) = 5 The IRIX program crashes here when running on Linux. 10mS usinit( 5349): sysmp(MP_NPROCS) = 1 10mS usinit( 5349): open(tm, O_RDWR, 016) = 3 10mS usinit( 5349): fcntl(3, F_SETFD, 1) OK 10mS usinit( 5349): fcntl(3, F_SETLKW, type=F_WRLCK whence=SEEK_SET start=0 len=1) OK 10mS usinit( 5349): syssgi(SGI_USE_FP_BCOPY, 0x1000, 0x7fff1d5c, 0xffffffff, 0x10001008, 0x7fff2fc0) = 0 11mS usinit( 5349): read(3, ..., 4096) = 4096 [more system calls deleted] Exhibit [3]: The culprit piece of code, taken from the IRIX libc: 0xfa52a10 <_usinit+44>: lui $t6,0x20 0xfa52a14 <_usinit+48>: lw $t7,3584($t6) <- This one is the culprit This is loading something from 0x200e00 From owner-linux@cthulhu.engr.sgi.com Thu Sep 4 20:58:28 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA1115209 for ; Thu, 4 Sep 1997 20:58:28 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA16869 for linux-list; Thu, 4 Sep 1997 20:57:57 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA16865 for ; Thu, 4 Sep 1997 20:57:55 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id UAA07157 for ; Thu, 4 Sep 1997 20:57:53 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA16860 for ; Thu, 4 Sep 1997 20:57:52 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA08437 for ; Thu, 4 Sep 1997 20:57:50 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA03387; Thu, 4 Sep 1997 22:50:43 -0500 Date: Thu, 4 Sep 1997 22:50:43 -0500 Message-Id: <199709050350.WAA03387@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: IRIX emulation, getting there. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 563 Lines: 21 This is my next show stopper: There is a misterious fcntl call in the IRIX Xsgi server: it is called with the F_ALLOCSP flag, what is this flag used for in the X server setup code? Cheers, Miguel. [ Unsolved misteries -- update The magical address at 0x200000 in the IRIX executables contains a one zeroed-page, for now I have mapped an empty page at this location on every IRIX program that is ran on the system. Both the usinit() test and the Xsgi server can continue execution under this scenario now, but it is still hackish. ] From owner-linux@cthulhu.engr.sgi.com Thu Sep 4 22:01:47 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id WAA1118601 for ; Thu, 4 Sep 1997 22:01:47 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id WAA28977 for linux-list; Thu, 4 Sep 1997 22:01:14 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id WAA28961 for ; Thu, 4 Sep 1997 22:01:12 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id WAA19131 for ; Thu, 4 Sep 1997 22:01:11 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id AAA16971 for linux@engr.sgi.com; Fri, 5 Sep 1997 00:58:04 -0400 From: Mike Shaver Message-Id: <199709050458.AAA16971@neon.ingenia.ca> Subject: Kernel for local disk stuff To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Fri, 5 Sep 1997 00:58:04 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 749 Lines: 20 ftp://ftp.linux.sgi.com/pub/test/vmlinux-950905.gz works for me for mounting local disk as root. Getting a local disk set up is still a minor nightmare, because there's no simple boot image yet (though Alex is working on it), and because heavy ethernet traffic occasionally generates bus errors that lock up the box. I'm going to take a look at what causes those tomorrow, hopefully. (I'm also going to reinstall IRIX on the primary disk. mke2fs can really wreck IRIX's day. =) ) Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Linux: because every cycle counts. #> #> "I don't know what you do for a living[...]" -- perry@piermont.com #> "I change the world." -- davem@caip.rutgers.edu From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 11:52:05 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id LAA1308877 for ; Fri, 5 Sep 1997 11:52:04 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id LAA01549 for linux-list; Fri, 5 Sep 1997 11:51:11 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id LAA01531 for ; Fri, 5 Sep 1997 11:51:05 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id LAA19075 for ; Fri, 5 Sep 1997 11:51:03 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id LAA01517 for ; Fri, 5 Sep 1997 11:51:01 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id LAA21300 for ; Fri, 5 Sep 1997 11:49:28 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id NAA07307; Fri, 5 Sep 1997 13:42:08 -0500 Date: Fri, 5 Sep 1997 13:42:08 -0500 Message-Id: <199709051842.NAA07307@athena.nuclecu.unam.mx> From: Miguel de Icaza To: miguel@nuclecu.unam.mx CC: linux@fir.engr.sgi.com In-reply-to: <199709050245.VAA03103@athena.nuclecu.unam.mx> (message from Miguel de Icaza on Thu, 4 Sep 1997 21:45:31 -0500) Subject: Re: [Q: Linux/SGI] IRIX executable memory map. X-Windows: The problem for your problem. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 513 Lines: 17 IRIX unsolved misteries -- second update. I said: > Ok, it seems our irix_elfmap routine is just fine, I just found > out with a simple test case that the code is trying to access memory > from the location at 0x200000 which is making my IRIX executables > crash (this one is crashing inside usinit ()). I was debatting this with my bed sheets this morning and I believe that this page may be a special trick for IRIX sproc()ed binaries. Probably I should also turn off the cache on this page. Miguel. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 13:38:55 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id NAA1304745 for ; Fri, 5 Sep 1997 13:38:55 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id NAA11702 for linux-list; Fri, 5 Sep 1997 13:36:19 -0700 Received: from oz.engr.sgi.com (oz.engr.sgi.com [150.166.61.27]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id NAA11685 for ; Fri, 5 Sep 1997 13:36:15 -0700 Received: (from ariel@localhost) by oz.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id NAA29563 for linux@engr.sgi.com; Fri, 5 Sep 1997 13:36:13 -0700 From: ariel@oz.engr.sgi.com (Ariel Faigon) Message-Id: <199709052036.NAA29563@oz.engr.sgi.com> Subject: 2.1.53 (from Linus) To: linux@cthulhu.engr.sgi.com (SGI/Linux mailing list) Date: Fri, 5 Sep 1997 13:36:12 -0700 (PDT) Reply-To: ariel@cthulhu.engr.sgi.com (Ariel Faigon) Organization: Silicon Graphics Inc. X-Mailer: ELM [version 2.4 PL24 ME5a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 626 Lines: 21 Just thought this would be of relevance to some: Subject: Linux-2.1.53.. From: Linus Torvalds Date: Thu, 4 Sep 1997 17:10:27 -0700 Message-ID: Newsgroups: .mlist. linux-kernel linux-kernel@vger.rutgers.edu I released a 2.1.53, which fixes two major bugs in the 2.1.x series. Ingo Molnar found one (and hopefully _the_) reason for dcache corruption in d_move(), and David Miller found and fixed a TCP-related crash-inducer. In short, I hope 2.1.53 will finally be stable for people, Linus -- Peace, Ariel From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 15:33:37 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA1343284 for ; Fri, 5 Sep 1997 15:33:37 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA00467 for linux-list; Fri, 5 Sep 1997 15:31:29 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA00393 for ; Fri, 5 Sep 1997 15:31:16 -0700 Received: from gatekeeper.qms.com (gatekeeper.qms.com [161.33.3.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via SMTP id PAA03312 for ; Fri, 5 Sep 1997 15:31:09 -0700 env-from (marks@sun470.sun470.rd.qms.com) Received: from sun470.rd.qms.com (sun470.qms.com) by gatekeeper.qms.com (4.1/SMI-4.1) id AA28446; Fri, 5 Sep 97 17:31:02 CDT Received: from speedy.rd.qms.com by sun470.rd.qms.com (4.1/SMI-4.1) id AA00261; Fri, 5 Sep 97 17:31:00 CDT Received: by speedy.rd.qms.com (8.8.2) id RAA21201; Fri, 5 Sep 1997 17:30:59 -0500 Date: Fri, 5 Sep 1997 17:30:59 -0500 Message-Id: <199709052230.RAA21201@speedy.rd.qms.com> From: Mark Salter To: shaver@neon.ingenia.ca Cc: linux@cthulhu.engr.sgi.com In-Reply-To: <199709050458.AAA16971@neon.ingenia.ca> (message from Mike Shaver on Fri, 5 Sep 1997 00:58:04 -0400 (EDT)) Subject: Re: Kernel for local disk stuff Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1084 Lines: 24 Mike Shaver said: > because heavy ethernet traffic occasionally generates bus errors that > lock up the box. I'm going to take a look at what causes those > tomorrow, hopefully. These finally annoyed me to the point I started looking at it today. I was able to hack the buserr irq handler to set up a gdb frame so that gdb gets control at the instruction that was interrupted. The problem appears to be in sgiseeq.c which is no great surprise since it occurs during times of heavy network traffic. The bus error irqs always occur when interrupts are reenabled in the ret_from_sys_call after a sgiseeq irq. The hpc_ethregs tx_ctrl value is 0x1 indicating that transmit was inactive, but there was an underflow. The tx_ndptr value is 0xffffffff. The latter I think leads to the bus error. Look at kick_tx() being called from sgiseeq_tx() during the handling of the interrupt. With that value of tx_ndptr, kick_tx would end up writing to 0xbffffff0 which is not we want to do. I don't have the HPC docs, so I'm probably not going to be able to come up with a proper fix... --Mark From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 15:52:42 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA1344239 for ; Fri, 5 Sep 1997 15:52:42 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA10093 for linux-list; Fri, 5 Sep 1997 15:51:03 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA10081 for ; Fri, 5 Sep 1997 15:51:01 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id PAA20078; Fri, 5 Sep 1997 15:51:00 -0700 Date: Fri, 5 Sep 1997 15:51:00 -0700 Message-Id: <199709052251.PAA20078@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: [Q: Linux/SGI] IRIX executable memory map. In-Reply-To: <199709050245.VAA03103@athena.nuclecu.unam.mx> References: <199709050245.VAA03103@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1042 Lines: 26 Miguel de Icaza writes: > > Hello guys, > > Ok, it seems our irix_elfmap routine is just fine, I just found > out with a simple test case that the code is trying to access memory > from the location at 0x200000 which is making my IRIX executables > crash (this one is crashing inside usinit ()). > > Is there something magic about 0x20000 on IRIX that I should be > aware of? 0x20000 (not 0x200000) is automatically mapped in a process by the kernel when referenced. See on IRIX for the definitions. Most of the fields you can probably ignore, but you should set t_pid, t_rpid, and t_prid (where the latter is the value of the $prid COP0 register). Set t_pid and t_rpid to the pid of the process. (They cannot really be different in IRIX.) The library is fetching the PID from t_pid instead of calling getpid() for performance reasons. Simply automatically create the page in the page fault handler when it is first referenced. Treat as if it were an mmap() of /dev/zero for 1 page. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 15:58:29 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA1346602 for ; Fri, 5 Sep 1997 15:58:28 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA13033 for linux-list; Fri, 5 Sep 1997 15:58:10 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA13019 for ; Fri, 5 Sep 1997 15:58:08 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id PAA20120; Fri, 5 Sep 1997 15:58:07 -0700 Date: Fri, 5 Sep 1997 15:58:07 -0700 Message-Id: <199709052258.PAA20120@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: [Q: Linux/SGI] IRIX executable memory map. In-Reply-To: <199709051842.NAA07307@athena.nuclecu.unam.mx> References: <199709050245.VAA03103@athena.nuclecu.unam.mx> <199709051842.NAA07307@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 673 Lines: 17 Miguel de Icaza writes: > > IRIX unsolved misteries -- second update. > > I said: > > > Ok, it seems our irix_elfmap routine is just fine, I just found > > out with a simple test case that the code is trying to access memory > > from the location at 0x200000 which is making my IRIX executables > > crash (this one is crashing inside usinit ()). > > I was debatting this with my bed sheets this morning and I believe > that this page may be a special trick for IRIX sproc()ed binaries. > Probably I should also turn off the cache on this page. No, I think this page should be private to a given thread (sproc or process), and it can be cached. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 16:13:30 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA1348645 for ; Fri, 5 Sep 1997 16:13:30 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA18619 for linux-list; Fri, 5 Sep 1997 16:11:40 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA18590 for ; Fri, 5 Sep 1997 16:11:38 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id QAA20317; Fri, 5 Sep 1997 16:11:28 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA18437; Fri, 5 Sep 1997 16:11:28 -0700 Received: from snowcrash.cymru.net (snowcrash.cymru.net [163.164.160.3]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA14896; Fri, 5 Sep 1997 16:11:24 -0700 env-from (alan@lxorguk.ukuu.org.uk) Received: from lightning.swansea.linux.org.uk (centre.swanlink.ukuu.org.uk [137.44.10.205]) by snowcrash.cymru.net (8.8.5-q-beta3/8.7.1) with SMTP id AAA08941; Sat, 6 Sep 1997 00:10:23 +0100 Received: by lightning.swansea.linux.org.uk (Smail3.1.29.1 #2) id m0x77Rs-0005FjC; Sat, 6 Sep 97 00:05 BST Message-Id: From: alan@lxorguk.ukuu.org.uk (Alan Cox) Subject: Re: [Q: Linux/SGI] IRIX executable memory map. To: wje@fir.engr.sgi.com (William J. Earl) Date: Sat, 6 Sep 1997 00:05:56 +0100 (BST) Cc: miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com In-Reply-To: <199709052251.PAA20078@fir.engr.sgi.com> from "William J. Earl" at Sep 5, 97 03:51:00 pm Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 426 Lines: 11 > Simply automatically create the page in the page fault handler > when it is first referenced. Treat as if it were an mmap() of /dev/zero > for 1 page. Better to map it first. page fault is a far more critical path to stick conditionals into. Even better if possible is to stick it into the userspace dynamic loader for irix binaries. That however means not using the irix one or having a "preloader" before it Alan From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 16:56:35 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA59994 for ; Fri, 5 Sep 1997 16:56:34 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA02441 for linux-list; Fri, 5 Sep 1997 16:55:28 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA02424 for ; Fri, 5 Sep 1997 16:55:26 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id QAA20408; Fri, 5 Sep 1997 16:55:20 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA02349; Fri, 5 Sep 1997 16:55:15 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA26810; Fri, 5 Sep 1997 16:55:11 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id SAA09349; Fri, 5 Sep 1997 18:47:53 -0500 Date: Fri, 5 Sep 1997 18:47:53 -0500 Message-Id: <199709052347.SAA09349@athena.nuclecu.unam.mx> From: Miguel de Icaza To: wje@fir.engr.sgi.com CC: linux@fir.engr.sgi.com In-reply-to: <199709052251.PAA20078@fir.engr.sgi.com> (wje@fir.engr.sgi.com) Subject: Re: [Q: Linux/SGI] IRIX executable memory map. X-Windows: Form follows malfunction. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1212 Lines: 30 > 0x20000 (not 0x200000) is automatically mapped in a process > by the kernel when referenced. See on IRIX for > the definitions. Ok, thanks for the pointer. btw, the code seems to indicate that the address is indeed 0x200000 (it does a lui $reg,0x20). Oops. Hold on. The sys/prctl agrees with me, it is 0x200000. That structure looks very promising, I will go and fill it. > Most of the fields you can probably ignore, but you should set > t_pid, t_rpid, and t_prid (where the latter is the value of the > $prid COP0 register). Set t_pid and t_rpid to the pid of the > process. (They cannot really be different in IRIX.) Ok. There is another bit that looks promising: the t_cpu field, which is supposed to have the cpu number where the process is executing. Does IRIX update this on every context switch? > Simply automatically create the page in the page fault handler > when it is first referenced. Treat as if it were an mmap() of /dev/zero > for 1 page. Ok. The code that I used last night to work around this was to map a /dev/zero page at sys_irix_elfexec time, I think I will just poke the right values there. best wishes and thanks for this pointer! Miguel. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 16:59:44 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA1316449 for ; Fri, 5 Sep 1997 16:59:44 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA03699 for linux-list; Fri, 5 Sep 1997 16:59:26 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA03688 for ; Fri, 5 Sep 1997 16:59:24 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id QAA20430; Fri, 5 Sep 1997 16:59:22 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA03669; Fri, 5 Sep 1997 16:59:21 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA27689; Fri, 5 Sep 1997 16:57:53 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id SAA09372; Fri, 5 Sep 1997 18:49:45 -0500 Date: Fri, 5 Sep 1997 18:49:45 -0500 Message-Id: <199709052349.SAA09372@athena.nuclecu.unam.mx> From: Miguel de Icaza To: alan@lxorguk.ukuu.org.uk CC: wje@fir.engr.sgi.com, linux@fir.engr.sgi.com In-reply-to: (alan@lxorguk.ukuu.org.uk) Subject: Re: [Q: Linux/SGI] IRIX executable memory map. X-Lost: In case of doubt, make it sound convincing Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 459 Lines: 12 > Better to map it first. page fault is a far more critical path to stick > conditionals into. Even better if possible is to stick it into the userspace > dynamic loader for irix binaries. That however means not using the irix one > or having a "preloader" before it Nah. We already have an IRIX specific ELF loader, so I will just modify my existing mmap in that place. No need for extra hacks if a 10 line change to irixelf.c fixes it. Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 17:34:13 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id RAA1355826 for ; Fri, 5 Sep 1997 17:34:13 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA12884 for linux-list; Fri, 5 Sep 1997 17:33:32 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA12856 for ; Fri, 5 Sep 1997 17:33:28 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id RAA20529; Fri, 5 Sep 1997 17:33:20 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA12810; Fri, 5 Sep 1997 17:33:19 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id RAA06743; Fri, 5 Sep 1997 17:33:16 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from thoma (ralf@thoma.uni-koblenz.de [141.26.4.61]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id CAA11262; Sat, 6 Sep 1997 02:33:08 +0200 (MEST) From: Ralf Baechle Message-Id: <199709060033.CAA11262@informatik.uni-koblenz.de> Received: by thoma (SMI-8.6/KO-2.0) id CAA07341; Sat, 6 Sep 1997 02:33:06 +0200 Subject: Re: [Q: Linux/SGI] IRIX executable memory map. To: wje@fir.engr.sgi.com (William J. Earl) Date: Sat, 6 Sep 1997 02:33:05 +0200 (MET DST) Cc: miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com In-Reply-To: <199709052251.PAA20078@fir.engr.sgi.com> from "William J. Earl" at Sep 5, 97 03:51:00 pm MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 139 Lines: 8 Hi, > The library is fetching the PID from t_pid instead of calling getpid() > for performance reasons. Byte Benchmarks ;-) Ralf From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 18:32:11 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id SAA1338693 for ; Fri, 5 Sep 1997 18:32:11 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA27581 for linux-list; Fri, 5 Sep 1997 18:31:24 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA27551; Fri, 5 Sep 1997 18:31:18 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA18377; Fri, 5 Sep 1997 18:31:12 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from thoma (ralf@thoma.uni-koblenz.de [141.26.4.61]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id DAA21049; Sat, 6 Sep 1997 03:31:03 +0200 (MEST) From: Ralf Baechle Message-Id: <199709060131.DAA21049@informatik.uni-koblenz.de> Received: by thoma (SMI-8.6/KO-2.0) id DAA07368; Sat, 6 Sep 1997 03:31:01 +0200 Subject: Re: 2.1.53 (from Linus) To: ariel@cthulhu.engr.sgi.com Date: Sat, 6 Sep 1997 03:31:00 +0200 (MET DST) Cc: linux@cthulhu.engr.sgi.com In-Reply-To: <199709052036.NAA29563@oz.engr.sgi.com> from "Ariel Faigon" at Sep 5, 97 01:36:12 pm MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1480 Lines: 33 > Just thought this would be of relevance to some: > > Subject: Linux-2.1.53.. > From: Linus Torvalds > Date: Thu, 4 Sep 1997 17:10:27 -0700 > Message-ID: > Newsgroups: .mlist. linux-kernel linux-kernel@vger.rutgers.edu > > > I released a 2.1.53, which fixes two major bugs in the 2.1.x series. Ingo > Molnar found one (and hopefully _the_) reason for dcache corruption in > d_move(), and David Miller found and fixed a TCP-related crash-inducer. > > In short, I hope 2.1.53 will finally be stable for people, I've fixed a couple more showstoppers in the MIPS part of the kernel. In particular the kernel was misstreating all non-R4000/R4400 CPUs with virtual indexed caches but without a second level caches or a cachesize other than 16kbit. For I/O this might result in old data being written back to the disk. I suspect this was the source of the Inode corruption I was observing on my R5k Indy as well as the frequent crashes on my R4600 and R4300 box. The big brother, an R5230 eval board I'm hacking right now has been stable over days, same for the NCR driver. The Z8530 driver however is still as fragile as a snowflake. I've fixed a couple of other showstoppers as well. I'm currently working on cleanly integrating yet another port into the source tree. When this and some other important debugging (new init versions die) is done, I'll merge back all the stuff. Ralf From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 19:01:01 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA1373789 for ; Fri, 5 Sep 1997 19:01:01 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA04047 for linux-list; Fri, 5 Sep 1997 19:00:30 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA04019 for ; Fri, 5 Sep 1997 19:00:27 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id TAA20744; Fri, 5 Sep 1997 19:00:26 -0700 Date: Fri, 5 Sep 1997 19:00:26 -0700 Message-Id: <199709060200.TAA20744@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: [Q: Linux/SGI] IRIX executable memory map. In-Reply-To: <199709052347.SAA09349@athena.nuclecu.unam.mx> References: <199709052251.PAA20078@fir.engr.sgi.com> <199709052347.SAA09349@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1454 Lines: 35 Miguel de Icaza writes: > > > 0x20000 (not 0x200000) is automatically mapped in a process > > by the kernel when referenced. See on IRIX for > > the definitions. > > Ok, thanks for the pointer. btw, the code seems to indicate that the > address is indeed 0x200000 (it does a lui $reg,0x20). Oops. Hold > on. The sys/prctl agrees with me, it is 0x200000. Yes, you are correct. ... > > Most of the fields you can probably ignore, but you should set > > t_pid, t_rpid, and t_prid (where the latter is the value of the > > $prid COP0 register). Set t_pid and t_rpid to the pid of the > > process. (They cannot really be different in IRIX.) > > Ok. There is another bit that looks promising: the t_cpu field, which > is supposed to have the cpu number where the process is executing. > Does IRIX update this on every context switch? Yes, on a multiprocessor, but it is a constant on a uniprocessor. > > Simply automatically create the page in the page fault handler > > when it is first referenced. Treat as if it were an mmap() of /dev/zero > > for 1 page. > > Ok. The code that I used last night to work around this was to map > a /dev/zero page at sys_irix_elfexec time, I think I will just poke > the right values there. As someone else pointed out, mapping the page is sys_irix_elfexec time is probably better, since it will avoid slowing down the page fault handler. From owner-linux@cthulhu.engr.sgi.com Fri Sep 5 19:03:24 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA1377555 for ; Fri, 5 Sep 1997 19:03:24 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA04591 for linux-list; Fri, 5 Sep 1997 19:03:04 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA04576 for ; Fri, 5 Sep 1997 19:03:02 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id TAA20799; Fri, 5 Sep 1997 19:03:01 -0700 Date: Fri, 5 Sep 1997 19:03:01 -0700 Message-Id: <199709060203.TAA20799@fir.engr.sgi.com> From: "William J. Earl" To: Ralf Baechle Cc: miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com Subject: Re: [Q: Linux/SGI] IRIX executable memory map. In-Reply-To: <199709060033.CAA11262@informatik.uni-koblenz.de> References: <199709052251.PAA20078@fir.engr.sgi.com> <199709060033.CAA11262@informatik.uni-koblenz.de> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 505 Lines: 13 Ralf Baechle writes: > Hi, > > > The library is fetching the PID from t_pid instead of calling getpid() > > for performance reasons. > > Byte Benchmarks ;-) Actually not, although we have thought about it. getpid() itself is a system call, even though it could just use the PRDA. Note, however, that IRIX processes which do not use the usema stuff generally do not have the PRDA mapped, since we wanted to avoid allocating an extra page per process for processes which do not use it. From owner-linux@cthulhu.engr.sgi.com Mon Sep 8 10:55:56 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id KAA18901 for ; Mon, 8 Sep 1997 10:55:56 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id KAA06129 for linux-list; Mon, 8 Sep 1997 10:55:16 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id KAA06111 for ; Mon, 8 Sep 1997 10:55:13 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id KAA27906; Mon, 8 Sep 1997 10:55:12 -0700 Date: Mon, 8 Sep 1997 10:55:12 -0700 Message-Id: <199709081755.KAA27906@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: IRIX emulation, getting there. In-Reply-To: <199709050350.WAA03387@athena.nuclecu.unam.mx> References: <199709050350.WAA03387@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1494 Lines: 31 Miguel de Icaza writes: > > This is my next show stopper: > > There is a misterious fcntl call in the IRIX Xsgi server: > it is called with the F_ALLOCSP flag, what is this flag used for in > the X server setup code? ... This fcntl() sets the size of a file, and assures that the underlying storage is really assigned. From fcntl(2): F_FREESP Alter storage space associated with a section of the ordinary file fildes. The section is specified by a variable of data type struct flock pointed to by the third argument arg. The data type struct flock is defined in the header file [see fcntl(5)] and contains the following members: l_whence is 0, 1, or 2 to indicate that the relative offset l_start will be measured from the start of the file, the current position, or the end of the file, respectively. l_start is the offset from the position specified in l_whence. l_len is the size of the section. An l_len of 0 frees up to the end of the file; in this case, the end of file (i.e., file size) is set to the beginning of the section freed. Any data previously written into this section is no longer accessible. If the section specified is beyond the current end of file, the file is grown and filled with zeroes. The l_len field is currently ignored, and should be set to 0. F_ALLOCSP This command is identical to F_FREESP. From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 16:22:59 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA317762 for ; Wed, 10 Sep 1997 16:22:59 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA15466 for linux-list; Wed, 10 Sep 1997 16:18:19 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA15433 for ; Wed, 10 Sep 1997 16:18:15 -0700 Received: from snowcrash.cymru.net (snowcrash.cymru.net [163.164.160.3]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id PAA16059 for ; Wed, 10 Sep 1997 15:43:17 -0700 env-from (alan@lxorguk.ukuu.org.uk) Received: from lightning.swansea.linux.org.uk (centre.swanlink.ukuu.org.uk [137.44.10.205]) by snowcrash.cymru.net (8.8.5-q-beta3/8.7.1) with SMTP id XAA15858 for ; Wed, 10 Sep 1997 23:41:47 +0100 Received: by lightning.swansea.linux.org.uk (Smail3.1.29.1 #2) id m0x8vQa-0005FjC; Wed, 10 Sep 97 23:40 BST Message-Id: Date: Wed, 10 Sep 97 23:40 BST From: alan@lxorguk.ukuu.org.uk (Alan Cox) To: linux@cthulhu.engr.sgi.com Subject: Linux and SGI EFS Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 248 Lines: 7 I've been digging around the EFS stuff while I wait for my SGI to materialise. It seems however thats another task we can do nice and fast http://wwwcip.informatik.uni-erlangen.de/user/cnvogelg/proj.html has a 2.0 read only EFS module for Linux From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 19:13:20 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA344847 for ; Wed, 10 Sep 1997 19:13:20 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA11560 for linux-list; Wed, 10 Sep 1997 19:12:40 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA11556 for ; Wed, 10 Sep 1997 19:12:38 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id TAA26772 for ; Wed, 10 Sep 1997 19:12:36 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA11551 for ; Wed, 10 Sep 1997 19:12:35 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id TAA11849 for ; Wed, 10 Sep 1997 19:12:34 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id VAA25377; Wed, 10 Sep 1997 21:04:06 -0500 Date: Wed, 10 Sep 1997 21:04:06 -0500 Message-Id: <199709110204.VAA25377@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: Linux/SGI: MAP_AUTOGROW, F_ALLOCSP Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1685 Lines: 49 Hello, I am running into a new problem: the X server is doing the following: unlink ("/tmp/.Xshmtrans0"); tfd = open ("/tmp/.Xshmtrans0", O_RDWR | O_CREAT); fcntl(tfd, F_SETFD, 1); fcntl(tfd, F_SETLKW, type=F_WRLCL, whence=SEEK_SET start=0 len=1) read(6, 0x7fff1b04, 4096) = 0; fstat(6, 0x7fff18e0) = 0; lseek(6, 0, SEEK_SET) = 0 fcntl(6, F_ALLOCSP, whence=SEEK_SET, start=5224 len=0) mmap(0xe800000, 2149480, PROT_WRITE|PROT_READ, MAP_SHARED|MAP_AUTOGROW, 6, 0) = 0xe800000 OK, my problem right now is that the code is using the MAP_AUTOGROW flag for mmap. And the second problem is that I do not understand what the role for F_ALLOCSP is (thanks to all of you that sent me information on this flag), but it still does not make sense. Regarding F_ALLOCSP: I am ignoring it. Regarding mmap (..., MAP_AUTOGROW,...) I have implemented two solutions: 1. The "this does not have a chance of being accepted by the Main Penguin" approach which implemented MAP_AUTOGROW: basically: new flag, and an extra if in mm/filemap.c: filemap_nopage () routine. This flag is not even on the opengroup web pages, so it has a very little chance of being accepted just for Xsgi's sake. 2. The "this sucks, but it will get into the main kernel real fast": on irix_mmap () if the MAP_AUTOGROW flag is set, it will check if the top limit for the mmap is bigger that the current file size, if it is it does: o = sys_lseek (fd, offset + len - 1, SEEK_SET); sys_write (fd, "", 1); sys_lseek (fd, o, SEEK_SET); Any ideas, comments? I am ready to commit approach (2) ;-) Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 19:17:08 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA342172 for ; Wed, 10 Sep 1997 19:17:08 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA12236 for linux-list; Wed, 10 Sep 1997 19:16:51 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA12226 for ; Wed, 10 Sep 1997 19:16:50 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id TAA26791; Wed, 10 Sep 1997 19:16:48 -0700 Date: Wed, 10 Sep 1997 19:16:48 -0700 Message-Id: <199709110216.TAA26791@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: Linux/SGI: MAP_AUTOGROW, F_ALLOCSP In-Reply-To: <199709110204.VAA25377@athena.nuclecu.unam.mx> References: <199709110204.VAA25377@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 710 Lines: 18 Miguel de Icaza writes: ... > 2. The "this sucks, but it will get into the main > kernel real fast": on irix_mmap () if the MAP_AUTOGROW flag > is set, it will check if the top limit for the mmap is > bigger that the current file size, if it is it does: > > o = sys_lseek (fd, offset + len - 1, SEEK_SET); > sys_write (fd, "", 1); > sys_lseek (fd, o, SEEK_SET); > > Any ideas, comments? I am ready to commit approach (2) ;-) ... This seems reasonable enough. This code supports the "shm" shared memory transport of Xsgi; other process mmap() the file to communicate with the server. It is something of a hack (being not particularly secure). From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 20:12:29 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA348319 for ; Wed, 10 Sep 1997 20:12:29 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA23475 for linux-list; Wed, 10 Sep 1997 20:11:49 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA23465 for ; Wed, 10 Sep 1997 20:11:46 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id UAA26928 for ; Wed, 10 Sep 1997 20:11:44 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA23448 for ; Wed, 10 Sep 1997 20:11:39 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA27989 for ; Wed, 10 Sep 1997 20:11:37 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA25810; Wed, 10 Sep 1997 22:03:09 -0500 Date: Wed, 10 Sep 1997 22:03:09 -0500 Message-Id: <199709110303.WAA25810@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: Linux/SGI: Xsgi Shmiq/Qcntl Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 520 Lines: 17 Hello guys, Ok, 120 system calls after the last one, I need some more help. If somebody can quickly look up the IRIX X sources, I would appreciate if you could tell me how a little routine works. The source code should be in a directory called: xc/programs/Xserver/hw/sgi The code does an open on /dev/shmiq and then opens: sprintf (buf, "/dev/qcntl%d", mistery_variable) I want to know how it computes mistery_variable. I am obviously screwing something in the shmiq emulation code. Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 20:53:06 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA367918 for ; Wed, 10 Sep 1997 20:53:06 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA01925 for linux-list; Wed, 10 Sep 1997 20:51:36 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA01905 for ; Wed, 10 Sep 1997 20:51:33 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA05114 for ; Wed, 10 Sep 1997 20:51:27 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from grass (ralf@grass.uni-koblenz.de [141.26.4.65]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id FAA08065 for ; Thu, 11 Sep 1997 05:51:11 +0200 (MEST) From: Ralf Baechle Message-Id: <199709110351.FAA08065@informatik.uni-koblenz.de> Received: by grass (SMI-8.6/KO-2.0) id FAA11485; Thu, 11 Sep 1997 05:51:07 +0200 Subject: R5000 caches To: linux@cthulhu.engr.sgi.com Date: Thu, 11 Sep 1997 05:51:07 +0200 (MET DST) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 694 Lines: 14 Hi, Can anybody give me a pointer to where the R5000 caches, especially the cache instruction, are documented? My two IDT R5000 manuals don't contain the least bit of information regarding the cache instruction. I'm primarily interested in how the indexed operations select the cache set of the primary caches to operate on. On the R4600 which has 16kb per cache bit 13 selects the set. So I assume it's bit 14 on the R5000 with it's 32kb per cache? The code from David handles the R5000 like a R4000 CPU but this doesn't look very credible to me as this is a QED CPU and the other members of the R5k family like the Nevada (which run Linux now also!) have two way primary caches. Ralf From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 21:11:25 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id VAA368855 for ; Wed, 10 Sep 1997 21:11:25 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id VAA04823 for linux-list; Wed, 10 Sep 1997 21:10:36 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id VAA04815 for ; Wed, 10 Sep 1997 21:10:34 -0700 Received: from jenolan.rutgers.edu (jenolan.rutgers.edu [128.6.111.5]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id VAA08742 for ; Wed, 10 Sep 1997 21:10:33 -0700 env-from (davem@jenolan.rutgers.edu) Received: (from davem@localhost) by jenolan.rutgers.edu (8.8.5/8.8.5) id AAA14641; Thu, 11 Sep 1997 00:09:23 -0400 Date: Thu, 11 Sep 1997 00:09:23 -0400 Message-Id: <199709110409.AAA14641@jenolan.rutgers.edu> From: "David S. Miller" To: ralf@mailhost.uni-koblenz.de CC: linux@cthulhu.engr.sgi.com In-reply-to: <199709110351.FAA08065@informatik.uni-koblenz.de> (message from Ralf Baechle on Thu, 11 Sep 1997 05:51:07 +0200 (MET DST)) Subject: Re: R5000 caches Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1822 Lines: 42 From: Ralf Baechle Date: Thu, 11 Sep 1997 05:51:07 +0200 (MET DST) Can anybody give me a pointer to where the R5000 caches, especially the cache instruction, are documented? My two IDT R5000 manuals don't contain the least bit of information regarding the cache instruction. I'm primarily interested in how the indexed operations select the cache set of the primary caches to operate on. On the R4600 which has 16kb per cache bit 13 selects the set. So I assume it's bit 14 on the R5000 with it's 32kb per cache? The code from David handles the R5000 like a R4000 CPU but this doesn't look very credible to me as this is a QED CPU and the other members of the R5k family like the Nevada (which run Linux now also!) have two way primary caches. In SGI boxes, if my memory serves, the R5k's were the chips which needed the special: cli(); jump_into_64_bit_mode(); {peek,poke}_magic_physaddr(); leave_64_bit_mode(); sti(); sequence, both to enable/disable the cache and to perform flush operations. Although this could have been for the R4600. I do remember that IRIX had special code to work the R5k caches, but this might have been for the L2 cache operations only, not L1. All of this was very SGI specific and was mostly for the L2 cache operations. I think the R5k had a special "flush command" which would just pull a pin going to the cache and invalidate all the lines in one cycle (earle told me this, he may be able to elaborate). The R5k, by design at least in the SGI boxes, lacks a L2 cache, it was added externally on the SGI motherboard's, and thus all the funcy methods to access/enable/disable/flush it... Again, I could be confusing r4600 and r5k here, so who knows. Later, David "Sparc" Miller davem@caip.rutgers.edu From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 21:29:21 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id VAA368585 for ; Wed, 10 Sep 1997 21:29:21 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id VAA06982 for linux-list; Wed, 10 Sep 1997 21:28:36 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id VAA06973 for ; Wed, 10 Sep 1997 21:28:34 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id VAA11604 for ; Wed, 10 Sep 1997 21:28:28 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from grass (ralf@grass.uni-koblenz.de [141.26.4.65]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id GAA12438; Thu, 11 Sep 1997 06:28:22 +0200 (MEST) From: Ralf Baechle Message-Id: <199709110428.GAA12438@informatik.uni-koblenz.de> Received: by grass (SMI-8.6/KO-2.0) id GAA11544; Thu, 11 Sep 1997 06:28:17 +0200 Subject: Re: R5000 caches To: davem@jenolan.rutgers.edu (David S. Miller) Date: Thu, 11 Sep 1997 06:28:17 +0200 (MET DST) Cc: ralf@mailhost.uni-koblenz.de, linux@cthulhu.engr.sgi.com In-Reply-To: <199709110409.AAA14641@jenolan.rutgers.edu> from "David S. Miller" at Sep 11, 97 00:09:23 am MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 2074 Lines: 48 > In SGI boxes, if my memory serves, the R5k's were the chips which > needed the special: > > cli(); > jump_into_64_bit_mode(); > {peek,poke}_magic_physaddr(); > leave_64_bit_mode(); > sti(); > > sequence, both to enable/disable the cache and to perform flush > operations. Although this could have been for the R4600. I do Afaik it's for both CPUs. > remember that IRIX had special code to work the R5k caches, but this > might have been for the L2 cache operations only, not L1. > > All of this was very SGI specific and was mostly for the L2 cache > operations. I think the R5k had a special "flush command" which would > just pull a pin going to the cache and invalidate all the lines in one > cycle (earle told me this, he may be able to elaborate). It's the L2 enable bit for the "true", means CPU controlled L2 cache of the R5000. The fact that we're dealing with two different types of L2 caches for the R5000 complicates things slightly. By design the R5000 supports an external cache. What SGI did was more or less just ignoring that feature and using the R5000 as a R4600 replacement. That's why they're using an external cache which is controlled by the chipset. > The R5k, by design at least in the SGI boxes, lacks a L2 cache, it was > added externally on the SGI motherboard's, and thus all the funcy > methods to access/enable/disable/flush it... Again, I could be > confusing r4600 and r5k here, so who knows. Well, your memory serves right. However my problem are especially the primary caches (I've got not box with a "true" R5000 L2 cache), so that doesn't solve my problem. As far as I can tell your R4600 code seems to work for the R5000 Indy with the "fake" cache. I suspect that the problem handling the L1 caches was causing the disk corruption I observed on my Indy while on this Nevada box I'm using here SCSI works as realiable as it is supposed to be. Anyway, there is still enough fun to have. I'm planning a little lmbenchmania followed by a cycle-massacre and I'm just hacking microsecond timers for that ... Ralf From owner-linux@cthulhu.engr.sgi.com Wed Sep 10 23:55:29 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id XAA372356 for ; Wed, 10 Sep 1997 23:55:29 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id XAA20384 for linux-list; Wed, 10 Sep 1997 23:54:52 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id XAA20378 for ; Wed, 10 Sep 1997 23:54:50 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id XAA27293 for ; Wed, 10 Sep 1997 23:54:50 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id XAA20373 for ; Wed, 10 Sep 1997 23:54:48 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id XAA03510 for ; Wed, 10 Sep 1997 23:54:44 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id AAA03150; Thu, 11 Sep 1997 00:52:56 -0400 From: Mike Shaver Message-Id: <199709110452.AAA03150@neon.ingenia.ca> Subject: Re: Linux/SGI: Xsgi Shmiq/Qcntl In-Reply-To: <199709110303.WAA25810@athena.nuclecu.unam.mx> from Miguel de Icaza at "Sep 10, 97 10:03:09 pm" To: miguel@nuclecu.unam.mx (Miguel de Icaza) Date: Thu, 11 Sep 1997 00:52:56 -0400 (EDT) Cc: linux@fir.engr.sgi.com X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 766 Lines: 18 Thus spake Miguel de Icaza: > The code does an open on /dev/shmiq and then opens: sprintf (buf, > "/dev/qcntl%d", mistery_variable) I want to know how it computes > mistery_variable. I am obviously screwing something in the shmiq > emulation code. If it's like the rest of the shmiq stuff, it's probably the minor number. Remind me to test that stuff better now that I have an Indy. =) Mike -- #> Mike Shaver (shaver@ingenia.com) Information Warfare Division #> Chief Tactical and Strategic Officer "Saepe fidelis" #> #> "I like your game, but we have to change the rules." -- Anon #> From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 00:36:34 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id AAA375198 for ; Thu, 11 Sep 1997 00:36:34 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id AAA27175 for linux-list; Thu, 11 Sep 1997 00:35:49 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id AAA27168 for ; Thu, 11 Sep 1997 00:35:46 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id AAA27401; Thu, 11 Sep 1997 00:35:44 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id AAA27156; Thu, 11 Sep 1997 00:35:37 -0700 Received: from snowcrash.cymru.net (snowcrash.cymru.net [163.164.160.3]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id AAA10706; Thu, 11 Sep 1997 00:35:34 -0700 env-from (alan@lxorguk.ukuu.org.uk) Received: from lightning.swansea.linux.org.uk (centre.swanlink.ukuu.org.uk [137.44.10.205]) by snowcrash.cymru.net (8.8.5-q-beta3/8.7.1) with SMTP id IAA27653; Thu, 11 Sep 1997 08:27:46 +0100 Received: by lightning.swansea.linux.org.uk (Smail3.1.29.1 #2) id m0x93di-0005FjC; Thu, 11 Sep 97 08:26 BST Message-Id: From: alan@lxorguk.ukuu.org.uk (Alan Cox) Subject: Re: Linux/SGI: MAP_AUTOGROW, F_ALLOCSP To: wje@fir.engr.sgi.com (William J. Earl) Date: Thu, 11 Sep 1997 08:26:10 +0100 (BST) Cc: miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com In-Reply-To: <199709110216.TAA26791@fir.engr.sgi.com> from "William J. Earl" at Sep 10, 97 07:16:48 pm Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 444 Lines: 9 > This seems reasonable enough. This code supports the "shm" shared > memory transport of Xsgi; other process mmap() the file to communicate > with the server. It is something of a hack (being not particularly secure). So you use mmap() based XShm rather than the SYS5 ipc based shm Linux currently uses. Makes some sense. As to secure well last time I looked nobody had yet fixed the auth bug so X and security aren't related anyway From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 17:55:55 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id RAA521170 for ; Thu, 11 Sep 1997 17:55:55 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA11038 for linux-list; Thu, 11 Sep 1997 17:55:20 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA11023 for ; Thu, 11 Sep 1997 17:55:18 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id RAA10721; Thu, 11 Sep 1997 17:55:16 -0700 Date: Thu, 11 Sep 1997 17:55:16 -0700 Message-Id: <199709120055.RAA10721@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: Linux/SGI: Xsgi Shmiq/Qcntl In-Reply-To: <199709110303.WAA25810@athena.nuclecu.unam.mx> References: <199709110303.WAA25810@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1242 Lines: 38 Miguel de Icaza writes: > > Hello guys, > > Ok, 120 system calls after the last one, I need some more help. > If somebody can quickly look up the IRIX X sources, I would appreciate > if you could tell me how a little routine works. > > The source code should be in a directory called: > xc/programs/Xserver/hw/sgi > > The code does an open on /dev/shmiq and then opens: sprintf (buf, > "/dev/qcntl%d", mistery_variable) I want to know how it computes > mistery_variable. I am obviously screwing something in the shmiq > emulation code. fd = open( SHMIQDEVNAME, O_EXCL | O_RDWR | O_NOCTTY ) ; if ( fd < 0 ) { perror( "Failed to open shmiq device." ) ; return 0 ; } /* Now open the control character device */ if ( fstat ( fd, &sb ) < 0 ) { perror( "fstat failed -- shmiq not opened." ) ; (void) close( fd ) ; return 0 ; } (void) sprintf( cntldev, QCNTLDEVFORMAT, minor( sb.st_rdev ) ) ; The /dev/qcntl2 device is the control device for minor device 2 as returned by the clone open of /dev/shmiq. A clone device returns a different minor device for each open call; minor device numbers are reused only after they have been closed. Similary, /dev/ptc is a clone device for getting PTYs. From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 18:03:06 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA524064 for ; Thu, 11 Sep 1997 18:03:05 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA13549 for linux-list; Thu, 11 Sep 1997 18:02:44 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA13489; Thu, 11 Sep 1997 18:02:33 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id SAA10756; Thu, 11 Sep 1997 18:02:20 -0700 Date: Thu, 11 Sep 1997 18:02:20 -0700 Message-Id: <199709120102.SAA10756@fir.engr.sgi.com> From: "William J. Earl" To: Ralf Baechle Cc: davem@jenolan.rutgers.edu (David S. Miller), linux@cthulhu.engr.sgi.com Subject: Re: R5000 caches In-Reply-To: <199709110428.GAA12438@informatik.uni-koblenz.de> References: <199709110409.AAA14641@jenolan.rutgers.edu> <199709110428.GAA12438@informatik.uni-koblenz.de> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 2960 Lines: 63 Ralf Baechle writes: > > In SGI boxes, if my memory serves, the R5k's were the chips which > > needed the special: > > > > cli(); > > jump_into_64_bit_mode(); > > {peek,poke}_magic_physaddr(); > > leave_64_bit_mode(); > > sti(); > > > > sequence, both to enable/disable the cache and to perform flush > > operations. Although this could have been for the R4600. I do > > Afaik it's for both CPUs. It applies to the secondary cache on the Indy R4600SC and R5000SC and the Indigo2 R4600SC, but not to any other systems or processors. For primary caches, the set is selected by the high order bit of the cache index (virtual address bit 13 on the R4600, virtual address bit 14 on the R5000). (Note that the R10000 is different: it uses the low order address bit to select the set.) > > remember that IRIX had special code to work the R5k caches, but this > > might have been for the L2 cache operations only, not L1. > > > > All of this was very SGI specific and was mostly for the L2 cache > > operations. I think the R5k had a special "flush command" which would > > just pull a pin going to the cache and invalidate all the lines in one > > cycle (earle told me this, he may be able to elaborate). > > It's the L2 enable bit for the "true", means CPU controlled L2 cache > of the R5000. The fact that we're dealing with two different types > of L2 caches for the R5000 complicates things slightly. By design > the R5000 supports an external cache. What SGI did was more or less > just ignoring that feature and using the R5000 as a R4600 replacement. > That's why they're using an external cache which is controlled by > the chipset. Yes, that is correct for hte Indy, but the O2 R5000SC does use the real R5000 secondary cache. The L2 enable bit is set on O2 R5000SC and is not set on the Indy. > > The R5k, by design at least in the SGI boxes, lacks a L2 cache, it was > > added externally on the SGI motherboard's, and thus all the funcy > > methods to access/enable/disable/flush it... Again, I could be > > confusing r4600 and r5k here, so who knows. > > Well, your memory serves right. However my problem are especially > the primary caches (I've got not box with a "true" R5000 L2 cache), so > that doesn't solve my problem. As far as I can tell your R4600 code seems > to work for the R5000 Indy with the "fake" cache. I suspect that the > problem handling the L1 caches was causing the disk corruption I observed > on my Indy while on this Nevada box I'm using here SCSI works as realiable > as it is supposed to be. ... If you were using the index bits incorrectly, that would cause disk corruption, and, as noted above, the set selection bit differs, due to the size of the cache. Basically, if the size of the cache is X, then (X >> 1) is the set selection bit. The R4600SC secondary cache code should apply equally to the Indy R5000SC, without change. From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 18:23:18 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA531127 for ; Thu, 11 Sep 1997 18:23:18 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA19776 for linux-list; Thu, 11 Sep 1997 18:22:32 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA19764; Thu, 11 Sep 1997 18:22:30 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA02566; Thu, 11 Sep 1997 18:22:24 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from dahn (ralf@dahn.uni-koblenz.de [141.26.4.39]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id DAA12959; Fri, 12 Sep 1997 03:22:19 +0200 (MEST) From: Ralf Baechle Message-Id: <199709120122.DAA12959@informatik.uni-koblenz.de> Received: by dahn (SMI-8.6/KO-2.0) id DAA08782; Fri, 12 Sep 1997 03:22:15 +0200 Subject: Re: R5000 caches To: wje@fir.engr.sgi.com (William J. Earl) Date: Fri, 12 Sep 1997 03:22:15 +0200 (MET DST) Cc: ralf@mailhost.uni-koblenz.de, davem@jenolan.rutgers.edu, linux@cthulhu.engr.sgi.com In-Reply-To: <199709120102.SAA10756@fir.engr.sgi.com> from "William J. Earl" at Sep 11, 97 06:02:20 pm MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1190 Lines: 28 Hi, > For primary caches, the set is selected by the high order bit of > the cache index (virtual address bit 13 on the R4600, virtual > address bit 14 on the R5000). (Note that the R10000 is different: it > uses the low order address bit to select the set.) Indeed, but currently our only R10k problem is that we don't support that thing at all ... > Yes, that is correct for hte Indy, but the O2 R5000SC does > use the real R5000 secondary cache. The L2 enable bit is set on O2 > R5000SC and is not set on the Indy. > If you were using the index bits incorrectly, that would cause > disk corruption, and, as noted above, the set selection bit differs, > due to the size of the cache. Basically, if the size of the cache is > X, then (X >> 1) is the set selection bit. The R4600SC secondary > cache code should apply equally to the Indy R5000SC, without change. Ok, thanks for the information. It's what I was expecting. In that case the version which I'm just about to check in is still not R5000 safe. I'm going to fix that rsn. Quickfix for R5000 hackers: change the #define for waybit in arch/mips/mm/r4xx0.c from 0x2000 to (dcache_size >> 1). Ralf From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 18:35:30 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA535243 for ; Thu, 11 Sep 1997 18:35:30 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA23588 for linux-list; Thu, 11 Sep 1997 18:34:59 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA23573; Thu, 11 Sep 1997 18:34:56 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id SAA10912; Thu, 11 Sep 1997 18:34:54 -0700 Date: Thu, 11 Sep 1997 18:34:54 -0700 Message-Id: <199709120134.SAA10912@fir.engr.sgi.com> From: "William J. Earl" To: Ralf Baechle Cc: davem@jenolan.rutgers.edu, linux@cthulhu.engr.sgi.com Subject: Re: R5000 caches In-Reply-To: <199709120122.DAA12959@informatik.uni-koblenz.de> References: <199709120102.SAA10756@fir.engr.sgi.com> <199709120122.DAA12959@informatik.uni-koblenz.de> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1403 Lines: 27 Ralf Baechle writes: ... > > If you were using the index bits incorrectly, that would cause > > disk corruption, and, as noted above, the set selection bit differs, > > due to the size of the cache. Basically, if the size of the cache is > > X, then (X >> 1) is the set selection bit. The R4600SC secondary > > cache code should apply equally to the Indy R5000SC, without change. > > Ok, thanks for the information. It's what I was expecting. In that > case the version which I'm just about to check in is still not R5000 > safe. I'm going to fix that rsn. > > Quickfix for R5000 hackers: change the #define for waybit in > arch/mips/mm/r4xx0.c from 0x2000 to (dcache_size >> 1). By the way, in case you are not familiar with the problem, watch out for the R4600 errata in regard to index invalidate. Basically, you need to disable interrupts while using the index invalidate and index writeback invalidate instructions on R4600 Rev. 1.7, so that invalidating a given line in one set is atomic with invalidating the corresponding line in the second set. You can turn on interrupts periodically, but it is important to do both sets for a given index at the same time, without the possibility of an interrupt or exception. The errata can result in the same line winding up in both sets, leading to stale data. The bug is not present in the R4600 Rev. 2.0 or the R5000. From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 18:58:29 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA536140 for ; Thu, 11 Sep 1997 18:58:29 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA29040 for linux-list; Thu, 11 Sep 1997 18:57:54 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA29018; Thu, 11 Sep 1997 18:57:51 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA11453; Thu, 11 Sep 1997 18:57:32 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from dahn (ralf@dahn.uni-koblenz.de [141.26.4.39]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id DAA16937; Fri, 12 Sep 1997 03:57:24 +0200 (MEST) From: Ralf Baechle Message-Id: <199709120157.DAA16937@informatik.uni-koblenz.de> Received: by dahn (SMI-8.6/KO-2.0) id DAA11230; Fri, 12 Sep 1997 03:57:21 +0200 Subject: Re: R5000 caches To: wje@fir.engr.sgi.com (William J. Earl) Date: Fri, 12 Sep 1997 03:57:20 +0200 (MET DST) Cc: ralf@mailhost.uni-koblenz.de, davem@jenolan.rutgers.edu, linux@cthulhu.engr.sgi.com In-Reply-To: <199709120134.SAA10912@fir.engr.sgi.com> from "William J. Earl" at Sep 11, 97 06:34:54 pm MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 925 Lines: 17 > By the way, in case you are not familiar with the problem, watch > out for the R4600 errata in regard to index invalidate. Basically, > you need to disable interrupts while using the index invalidate > and index writeback invalidate instructions on R4600 Rev. 1.7, so that > invalidating a given line in one set is atomic with invalidating > the corresponding line in the second set. You can turn on interrupts > periodically, but it is important to do both sets for a given index > at the same time, without the possibility of an interrupt or exception. > The errata can result in the same line winding up in both sets, > leading to stale data. The bug is not present in the R4600 Rev. 2.0 > or the R5000. There is a more elegant solution for this bug. Just leave the interrupts as they are and invalidate set b before set a. Anyway, Linux currently still handles this erratum by disabling interrupts. Ralf From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 22:02:13 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id WAA06852 for ; Thu, 11 Sep 1997 22:02:13 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA19243 for linux-list; Thu, 11 Sep 1997 20:58:03 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA19216 for ; Thu, 11 Sep 1997 20:57:57 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id UAA11254 for ; Thu, 11 Sep 1997 20:57:51 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA19191 for ; Thu, 11 Sep 1997 20:57:46 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA09685 for ; Thu, 11 Sep 1997 20:57:44 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA03611; Thu, 11 Sep 1997 22:48:59 -0500 Date: Thu, 11 Sep 1997 22:48:59 -0500 Message-Id: <199709120348.WAA03611@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: Linux/SGI /dev/opengl magic ioctl X-Windows: You'd better sit down. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 658 Lines: 32 Hello guys, Ok, 2 questions 2 for tonight :-). Ok, I am getting close, just out of curiosity, what does the ioctl cmd=1 do on the /dev/opengl device? If the ioctl does not return 3, the X server complains. Right now I have this nice piece of code in the graphics driver: switch (cmd){ ... case 1: return 3; ... } which is not exactly what I would like to have, so any comments are appreciated. Second question: what does the GFX_LABEL ioctl do on the /dev/opengl device? I am missing this one in my implementation. I am around 100 system calls far from getting a nice display on Linux/SGI X. Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Thu Sep 11 21:59:40 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id VAA06678 for ; Thu, 11 Sep 1997 21:59:40 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id VAA27340 for linux-list; Thu, 11 Sep 1997 21:59:03 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id VAA27335 for ; Thu, 11 Sep 1997 21:59:01 -0700 Received: from b1mcast13.esd.sgi.com (gate-b1mcast13.esd.sgi.com [150.166.111.8]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id VAA11371 for ; Thu, 11 Sep 1997 21:59:00 -0700 Received: by b1mcast13.esd.sgi.com (940816.SGI.8.6.9/950213.SGI.AUTOCF) id VAA05184; Thu, 11 Sep 1997 21:59:00 -0700 Date: Thu, 11 Sep 1997 21:59:00 -0700 From: rtray@b1mcast13.esd.sgi.com (Robert Tray) Message-Id: <199709120459.VAA05184@b1mcast13.esd.sgi.com> To: linux@fir.engr.sgi.com, Miguel de Icaza Subject: Re: Linux/SGI /dev/opengl magic ioctl Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 938 Lines: 33 > From: Miguel de Icaza > > > Hello guys, > > Ok, 2 questions 2 for tonight :-). > > Ok, I am getting close, just out of curiosity, what does the ioctl > cmd=1 do on the /dev/opengl device? > > If the ioctl does not return 3, the X server complains. Right now > I have this nice piece of code in the graphics driver: > > switch (cmd){ > ... > > case 1: > return 3; > ... > } This is basically a version check. 1 == GFX_SYNC which is supposed to return a number which the kernel gfx driver and the Xserver agree is the current version. (The version you have is 3) > Second question: what does the GFX_LABEL ioctl do on the > /dev/opengl device? I am missing this one in my implementation. > The X server sends down the ":display.screen_number". I don't see the kernel doing anything with it. Robert Tray From owner-linux@cthulhu.engr.sgi.com Fri Sep 12 09:51:57 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id JAA30428 for ; Fri, 12 Sep 1997 09:51:57 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id JAA28003 for linux-list; Fri, 12 Sep 1997 09:50:51 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id JAA27995 for ; Fri, 12 Sep 1997 09:50:50 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id JAA23407; Fri, 12 Sep 1997 09:50:45 -0700 Date: Fri, 12 Sep 1997 09:50:45 -0700 Message-Id: <199709121650.JAA23407@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: Linux/SGI /dev/opengl magic ioctl In-Reply-To: <199709120348.WAA03611@athena.nuclecu.unam.mx> References: <199709120348.WAA03611@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1408 Lines: 45 Miguel de Icaza writes: > > Hello guys, > > Ok, 2 questions 2 for tonight :-). > > Ok, I am getting close, just out of curiosity, what does the ioctl > cmd=1 do on the /dev/opengl device? > > If the ioctl does not return 3, the X server complains. Right now > I have this nice piece of code in the graphics driver: > > switch (cmd){ > ... > > case 1: > return 3; > ... > } > > which is not exactly what I would like to have, so any comments > are appreciated. It turns out that is pretty much what the graphics driver is doing in IRIX: case GFX_SYNC : { *rvalp = GFX_SYNC_VALUE; goto out; } For GL (not OpenGL), if the process is an sproc (a process sharing and address space and file descriptors with a parent or siblings), the driver acquires a semaphore before doing an ioctl and releases it afterwards. For OpenGL, this ioctl does not really do anything at all. > Second question: what does the GFX_LABEL ioctl do on the > /dev/opengl device? I am missing this one in my implementation. It copies GFX_INFO_LABEL_SIZE (16) bytes from the argument (a char *) to the label field of the gfx_info struct for the graphics device. The label field is the choice of the current owner of the device, so it is not really significant. The /usr/gfx/gfxinfo command can print the gfx_info structure for the local device. From owner-linux@cthulhu.engr.sgi.com Fri Sep 12 23:45:09 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id XAA229702 for ; Fri, 12 Sep 1997 23:45:09 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id XAA05557 for linux-list; Fri, 12 Sep 1997 23:44:56 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id XAA05549 for ; Fri, 12 Sep 1997 23:44:54 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id XAA11718 for ; Fri, 12 Sep 1997 23:44:50 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from zaphod (ralf@zaphod.uni-koblenz.de [141.26.4.13]) by informatik.uni-koblenz.de (8.8.6/8.6.9) with SMTP id IAA29592; Sat, 13 Sep 1997 08:44:48 +0200 (MEST) From: Ralf Baechle Message-Id: <199709130644.IAA29592@informatik.uni-koblenz.de> Received: by zaphod (SMI-8.6/KO-2.0) id IAA28007; Sat, 13 Sep 1997 08:44:45 +0200 Subject: Modutils 2.1.55 Date: Sat, 13 Sep 1997 08:44:45 +0200 (MET DST) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Apparently-To: Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 399 Lines: 10 Hi all, I've upgraded the modutils to 2.1.55. Since I've seen no ill effects so far with modutils under Linux/MIPS I've sent the patches to rth for inclusion in the standard package. Patches are in the CVS archive on ftp.linux.sgi.com, as well as a source and little endian binary package. I could not connect to fnet, so for now these files will only be available on ftp.linux.sgi.com. Ralf From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 08:04:35 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id IAA528209 for ; Tue, 16 Sep 1997 08:04:35 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA28557 for linux-list; Tue, 16 Sep 1997 08:04:11 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA28548 for ; Tue, 16 Sep 1997 08:04:09 -0700 Received: from neon.ingenia.ca ([205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id IAA07140 for ; Tue, 16 Sep 1997 08:04:07 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id KAA01510 for linux@engr.sgi.com; Tue, 16 Sep 1997 10:58:42 -0400 From: Mike Shaver Message-Id: <199709161458.KAA01510@neon.ingenia.ca> Subject: EFS update To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Tue, 16 Sep 1997 10:58:41 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 856 Lines: 18 OK, I've managed to merge (read: rewrite) the EFS module code into the 2.1.55 tree from linus. I can mount my root IRIX partition, ls -l stuff, run simple binaries and the like. I don't handle symlinks correctly, but that's going to be easy enough to fix. I'm checking in my changes now, and I'll put a kernel up on ftp.linux for people to test with. This should make the installation process _much_ simpler. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Resident Linux bigot and kernel hacker. (OOPS!) #> `If you get bitten by a bug, tough luck...the one thing I won't do #> is feel sorry for you. In fact, I might ask you to do it all over #> again, just to get more information. I'm a heartless bastard.' #> -- Linus Torvalds (on development kernels) From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 08:58:38 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id IAA566252 for ; Tue, 16 Sep 1997 08:58:38 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA11878 for linux-list; Tue, 16 Sep 1997 08:58:16 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA11861 for ; Tue, 16 Sep 1997 08:58:14 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id IAA22794 for ; Tue, 16 Sep 1997 08:58:12 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id LAA02252 for linux@engr.sgi.com; Tue, 16 Sep 1997 11:52:54 -0400 From: Mike Shaver Message-Id: <199709161552.LAA02252@neon.ingenia.ca> Subject: EFS update II To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Tue, 16 Sep 1997 11:52:54 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 617 Lines: 15 OK, there's more that needs fixing than the symlink handling. I still don't handle indirect extents properly either, and I've got something a little wrong in the directory reading that gives bogus filenames on occasion. I'll let you know when I have something usable. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Commando Developer - Whatever It Takes #> #> "See, you not only have to be a good coder to create a system like #> Linux, you have to be a sneaky bastard too." - Linus Torvalds From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 14:07:12 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id OAA609990 for ; Tue, 16 Sep 1997 14:07:12 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id OAA02240 for linux-list; Tue, 16 Sep 1997 14:06:18 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id OAA02219 for ; Tue, 16 Sep 1997 14:06:12 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id OAA00062 for ; Tue, 16 Sep 1997 14:06:10 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id RAA05770 for linux@engr.sgi.com; Tue, 16 Sep 1997 17:01:05 -0400 From: Mike Shaver Message-Id: <199709162101.RAA05770@neon.ingenia.ca> Subject: EFS test kernel To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Tue, 16 Sep 1997 17:01:05 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1051 Lines: 23 ftp://ftp.linux.sgi.com/pub/test/vmlinux-970916-efs.gz has rudimentary support for EFS partition mounting, and it lets me do some basic stuff on my IRIX partition, like list small[*] directories and run binaries. I've got my IRIX root mounted at /usr/gnemul/irix (the emulation code requires that location), and have a symlink /lib/rld -> /usr/gnemul/irix/lib/rld. I sometimes get problems specifying executables by explicit path (./ls or /usr/gnemul/irix/bin/ls), but if I put the current directory in my path and chdir to there, things like ls and date and wc and uname work just fine. Please let me know if you get any problems. [*] small == not requiring indirect extents Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Ignore the man behind the curtain. #> #> "And then I realized that it never should have worked in the first #> place. Thus, it would not work again until rewritten." --- Anon. From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 18:05:01 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA660938 for ; Tue, 16 Sep 1997 18:05:01 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA09063 for linux-list; Tue, 16 Sep 1997 18:03:32 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA09050 for ; Tue, 16 Sep 1997 18:03:27 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA19431 for ; Tue, 16 Sep 1997 18:03:25 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id UAA07380 for linux@engr.sgi.com; Tue, 16 Sep 1997 20:58:22 -0400 From: Mike Shaver Message-Id: <199709170058.UAA07380@neon.ingenia.ca> Subject: EFS question To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Tue, 16 Sep 1997 20:58:22 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 982 Lines: 21 Just to make sure I understand this correctly: - when an inode has <= EFS_DIRECTEXTENTS extents, they're stored directly in the inode. - when an inode has > EFS_DIRECTEXTENTS extents, the extents in the inode refer to contiguous block regions that contain extents referring to the real data. The EFS code that Alan posted, in addition to having a curious aversion to structures, seemed to think that the first 4 bytes of dinode->di_u were the block number of a block containing extents, and thus doesn't work very well with files with numext > EFS_DIRECTEXTENTS. I just want to ensure that I make a sane fix. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> UNIX medicine man -- dark magick, cheap! #> #> When the going gets tough, the tough give cryptic error messages. #> "We believe in rough consensus and running code." From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 18:34:02 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA663800 for ; Tue, 16 Sep 1997 18:34:02 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA18839 for linux-list; Tue, 16 Sep 1997 18:33:21 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA18826; Tue, 16 Sep 1997 18:33:19 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id SAA20782; Tue, 16 Sep 1997 18:33:19 -0700 Date: Tue, 16 Sep 1997 18:33:19 -0700 Message-Id: <199709170133.SAA20782@fir.engr.sgi.com> From: "William J. Earl" To: Mike Shaver Cc: linux@cthulhu.engr.sgi.com (Linux/SGI list) Subject: Re: EFS question In-Reply-To: <199709170058.UAA07380@neon.ingenia.ca> References: <199709170058.UAA07380@neon.ingenia.ca> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 853 Lines: 18 Mike Shaver writes: > Just to make sure I understand this correctly: > - when an inode has <= EFS_DIRECTEXTENTS extents, they're stored directly > in the inode. > - when an inode has > EFS_DIRECTEXTENTS extents, the extents in the > inode refer to contiguous block regions that contain extents referring > to the real data. > > The EFS code that Alan posted, in addition to having a curious > aversion to structures, seemed to think that the first 4 bytes of > dinode->di_u were the block number of a block containing extents, and > thus doesn't work very well with files with numext > > EFS_DIRECTEXTENTS. I just want to ensure that I make a sane fix. ... The above is correct. When there are indirect extents, the offset field of the first indirect extent in the inode contains the number of indirect extents, not a file offset. From owner-linux@cthulhu.engr.sgi.com Tue Sep 16 18:47:24 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970903.SGI.8.8.7/960327.SGI.AUTOCF) via SMTP id SAA682173 for ; Tue, 16 Sep 1997 18:47:24 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA21282 for linux-list; Tue, 16 Sep 1997 18:46:56 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA21270; Tue, 16 Sep 1997 18:46:54 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA02773; Tue, 16 Sep 1997 18:46:52 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id VAA07688; Tue, 16 Sep 1997 21:41:47 -0400 From: Mike Shaver Message-Id: <199709170141.VAA07688@neon.ingenia.ca> Subject: Re: EFS question In-Reply-To: <199709170133.SAA20782@fir.engr.sgi.com> from "William J. Earl" at "Sep 16, 97 06:33:19 pm" To: wje@fir.engr.sgi.com (William J. Earl) Date: Tue, 16 Sep 1997 21:41:47 -0400 (EDT) Cc: linux@cthulhu.engr.sgi.com X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 577 Lines: 16 Thus spake William J. Earl: > When there are indirect extents, > the offset field of the first indirect extent in the inode > contains the number of indirect extents, not a file offset. _Ah!_ That makes _much_ more sense, then. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Commando Developer - Whatever It Takes #> #> "See, you not only have to be a good coder to create a system like #> Linux, you have to be a sneaky bastard too." - Linus Torvalds From owner-linux@cthulhu.engr.sgi.com Wed Sep 17 18:45:19 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id SAA66087 for ; Wed, 17 Sep 1997 18:45:19 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA23365 for linux-list; Wed, 17 Sep 1997 18:44:51 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA23361 for ; Wed, 17 Sep 1997 18:44:50 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id SAA17842 for ; Wed, 17 Sep 1997 18:44:49 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA23351 for ; Wed, 17 Sep 1997 18:44:48 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA03143 for ; Wed, 17 Sep 1997 18:44:47 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id UAA12189; Wed, 17 Sep 1997 20:34:42 -0500 Date: Wed, 17 Sep 1997 20:34:42 -0500 Message-Id: <199709180134.UAA12189@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: Linux/SGI Xsgi server: /dev/opengl ioctl NG1_SETDISPLAYMODE Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1425 Lines: 36 Hello guys, IRIX Xsgi server is now clearing the screen -and crashing since I do not have a mouse driver yet, I am quickly coding the iDev mouse driver now-. But while I am doing this, I would like to ask the list about another ioctl that the X server is issuing: the NG1_SETDISPLAYMODE, it is being called like this: ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0, 0xda0080 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 1, 0xda0100 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 2, 0xda0488 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 3, 0xda0490 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 4, 0xda0498 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 5, 0xda04a0 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 6, 0xda0500 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 7, 0xda0800 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 8, 0xda0900 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 9, 0xda0d00 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0xa, 0xda0e00 }); ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0xb, 0x4ac }); Where the structure is { int wid; unsigned int mode }. Can someone tell me what the newport driver on the IRIX kernel is doing with these calls? There is also another intersesting ioctl done later: NG1_SETGAMMARAMP0, this one takes 256 reds, 256 greens and 256 blues that I assume I have to load, do they go directly to the xmap9 lookup table? Thanks in advance, Miguel. From owner-linux@cthulhu.engr.sgi.com Wed Sep 17 19:03:57 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA75183 for ; Wed, 17 Sep 1997 19:03:57 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA27289 for linux-list; Wed, 17 Sep 1997 19:03:32 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA27280 for ; Wed, 17 Sep 1997 19:03:30 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id TAA17980; Wed, 17 Sep 1997 19:03:28 -0700 Date: Wed, 17 Sep 1997 19:03:28 -0700 Message-Id: <199709180203.TAA17980@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: Linux/SGI Xsgi server: /dev/opengl ioctl NG1_SETDISPLAYMODE In-Reply-To: <199709180134.UAA12189@athena.nuclecu.unam.mx> References: <199709180134.UAA12189@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 2753 Lines: 65 Miguel de Icaza writes: > > Hello guys, > > IRIX Xsgi server is now clearing the screen -and crashing since I > do not have a mouse driver yet, I am quickly coding the iDev mouse > driver now-. > > But while I am doing this, I would like to ask the list about > another ioctl that the X server is issuing: the NG1_SETDISPLAYMODE, it > is being called like this: > > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0, 0xda0080 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 1, 0xda0100 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 2, 0xda0488 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 3, 0xda0490 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 4, 0xda0498 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 5, 0xda04a0 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 6, 0xda0500 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 7, 0xda0800 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 8, 0xda0900 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 9, 0xda0d00 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0xa, 0xda0e00 }); > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0xb, 0x4ac }); > > Where the structure is { int wid; unsigned int mode }. > > Can someone tell me what the newport driver on the IRIX kernel is > doing with these calls? After waiting for the rex3 FIFO to drain, and the the xmap9 FIFO to not be full, it sets the xmap9 mode register for the given "wid": xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); where #define xmap9SetModeReg(rex,modereg,data24,cfreq) \ if (cfreq > 119 ) \ rex->set.dcbmode = DCB_XMAP_ALL | XM9_CRS_MODE_REG_DATA | \ DCB_DATAWIDTH_4 | W_DCB_XMAP9_PROTOCOL; \ else if (cfreq > 59) \ rex->set.dcbmode = DCB_XMAP_ALL | XM9_CRS_MODE_REG_DATA | \ DCB_DATAWIDTH_4 | WSLOW_DCB_XMAP9_PROTOCOL; \ else \ rex->set.dcbmode = DCB_XMAP_ALL | XM9_CRS_MODE_REG_DATA | \ DCB_DATAWIDTH_4 | WAYSLOW_DCB_XMAP9_PROTOCOL; \ rex->set.dcbdata0.byword = ((modereg) << 24) | (data24 & 0xffffff) > There is also another intersesting ioctl done later: > NG1_SETGAMMARAMP0, this one takes 256 reds, 256 greens and 256 blues > that I assume I have to load, do they go directly to the xmap9 lookup > table? It sets the address register for the BT445 RAMDAC to 0, and then then, for each RGB triple, does #define Bt445SetRGB( rex3, r, g, b ) \ rex3->set.dcbdata0.byword = ( (r) << 24 ) | ( (g) << 16 ) | ( (b) << 8 ) Before setting the address and before storing each RGB triple, it does #define BFIFOWAIT(rex) while ((rex)->p1.set.configmode & BFIFO); to avoid getting a FIFO overrun. From owner-linux@cthulhu.engr.sgi.com Wed Sep 17 20:33:59 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA83104 for ; Wed, 17 Sep 1997 20:33:59 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA10082 for linux-list; Wed, 17 Sep 1997 20:33:40 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA10076 for ; Wed, 17 Sep 1997 20:33:38 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id UAA18183; Wed, 17 Sep 1997 20:33:37 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA10071; Wed, 17 Sep 1997 20:33:36 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA29246; Wed, 17 Sep 1997 20:33:33 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA12615; Wed, 17 Sep 1997 22:23:30 -0500 Date: Wed, 17 Sep 1997 22:23:30 -0500 Message-Id: <199709180323.WAA12615@athena.nuclecu.unam.mx> From: Miguel de Icaza To: wje@fir.engr.sgi.com CC: linux@fir.engr.sgi.com In-reply-to: <199709180203.TAA17980@fir.engr.sgi.com> (wje@fir.engr.sgi.com) Subject: Re: Linux/SGI Xsgi server: /dev/opengl ioctl NG1_SETDISPLAYMODE X-Windows: Complex nonsolutions to simple nonproblems. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 777 Lines: 20 Another question regarding the Newport: > After waiting for the rex3 FIFO to drain, and the the xmap9 FIFO > to not be full, it sets the xmap9 mode register for the given "wid": > > xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); If the newport registers are available to the userland application, why the X server does not directly call those routines instead of relying on the kernel to perform them? Is it a convention that applications should only touch the rex3 registers and not attempt to program any of the other chips on the dcb? It would make sense since the graphics context switching would not involve the kernel peeking at the chips on the dcb what their context is and then the pain of restoring this. Miguel. From owner-linux@cthulhu.engr.sgi.com Wed Sep 17 23:03:46 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id XAA89760 for ; Wed, 17 Sep 1997 23:03:46 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id XAA00068 for linux-list; Wed, 17 Sep 1997 23:03:19 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id XAA00064 for ; Wed, 17 Sep 1997 23:03:18 -0700 Received: from b1mcast13.esd.sgi.com (gate-b1mcast13.esd.sgi.com [150.166.111.8]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id XAA18446; Wed, 17 Sep 1997 23:03:15 -0700 Received: by b1mcast13.esd.sgi.com (940816.SGI.8.6.9/950213.SGI.AUTOCF) id XAA09581; Wed, 17 Sep 1997 23:03:15 -0700 Date: Wed, 17 Sep 1997 23:03:15 -0700 From: rtray@b1mcast13.esd.sgi.com (Robert Tray) Message-Id: <199709180603.XAA09581@b1mcast13.esd.sgi.com> To: wje@fir.engr.sgi.com, Miguel de Icaza Subject: Re: Linux/SGI Xsgi server: /dev/opengl ioctl NG1_SETDISPLAYMODE Cc: linux@fir.engr.sgi.com Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 5230 Lines: 111 > From: "William J. Earl" > > Miguel de Icaza writes: > > > > But while I am doing this, I would like to ask the list about > > another ioctl that the X server is issuing: the NG1_SETDISPLAYMODE, it > > is being called like this: > > > > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0, 0xda0080 }); ... > > ioctl (/dev/opengl, NG1_SETDISPLAYMODE, { 0xb, 0x4ac }); > > > > Where the structure is { int wid; unsigned int mode }. > > > > Can someone tell me what the newport driver on the IRIX kernel is > > doing with these calls? > > After waiting for the rex3 FIFO to drain, and the the xmap9 FIFO > to not be full, it sets the xmap9 mode register for the given "wid": > > xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); > where The XMAP9 controls how the frame buffer pixels are interpreted and displayed to the monitor. That is the frame buffer may be 24 bits deep but a certain window may only be rendering 8 bits of PseudoColor (Color Index) while another window is rendering two buffers of TrueColor (RGB) split 12+12. It's also used to control which of the multiple colormaps is used to index the Pseudo- Color pixels. The X server is typically just running 8 bit PseudoColor while most GL programs run 12+12 RGB Double Buffer. The X server presets a certain number of Window IDs (wid) to be certain standard display modes ( eg. 1==Single buffer color index, 2==double buffer color index, 3==single buffer RGB, etc...). The other half of this setup involves a telling the backend hardware (VC2) which pixels use which Window ID (actually it's more frequently called Display ID (DID) in the last few years). The X server talks directly to VC2 to load a DID table which describes which XMAP9 DID/WID index every single pixel is using. (I hope that poorly worded sentence made sense.) The good news is that if you don't care about running GL programs in a separate display mode then you only have to set one DID entry in the XMAP and tell VC2 that the whole screen uses the same index into XMAP. > > There is also another intersesting ioctl done later: > > NG1_SETGAMMARAMP0, this one takes 256 reds, 256 greens and 256 blues > > that I assume I have to load, do they go directly to the xmap9 lookup > > table? > > It sets the address register for the BT445 RAMDAC to 0, and then > then, for each RGB triple, does > > #define Bt445SetRGB( rex3, r, g, b ) \ > rex3->set.dcbdata0.byword = ( (r) << 24 ) | ( (g) << 16 ) | ( (b) << 8 ) > > Before setting the address and before storing each RGB triple, it does > > #define BFIFOWAIT(rex) while ((rex)->p1.set.configmode & BFIFO); > > to avoid getting a FIFO overrun. The BrookTree 445 RAMDAC is that last chip the pixels go through on their way out to the monitor. It contains the gamma ramp to adjust the pixels for the monitors non-linearity. For Newport graphics there are two FIFOs. One FIFO paces access to the REX3 chip. All the "back end" chips are accessed via an 8 bit Display Control Bus (DCB) which is accessed throught hte REX3 dcbdata & dcbmode. You can do 64 bit writes to dcbdata but they get sent 8 bits at a time across the DCB. So you need a second FIFO (BFIFO) to pace access to the back end chips. Anyway that's the explanation for the BFIFOWAIT. > Another question regarding the Newport: > > > After waiting for the rex3 FIFO to drain, and the the xmap9 FIFO > > to not be full, it sets the xmap9 mode register for the given "wid": > > > > xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); > > If the newport registers are available to the userland application, > why the X server does not directly call those routines instead of > relying on the kernel to perform them? Is it a convention that > applications should only touch the rex3 registers and not attempt to > program any of the other chips on the dcb? The less that user code touches the DCB the better. But truthfully the biggest reason that the X server calls the kernel to set the XMAP regs is historical. In previous architectures those types of registers couldn't be set by the X server and required and ioctl. So one piece of code tends to look like the piece that it was copied from... However those registers are only preset at initialization and not touched again by the server. They are touched frequently after that by the kernel at the request of GL and to do swapbuffers. So as far as the X server is concerned performance is not a problem for those registers and it's best to avoid touching the DCB when you can. > It would make sense since the graphics context switching would not > involve the kernel peeking at the chips on the dcb what their context > is and then the pain of restoring this. Yeah, we actually strongly suspected that graphics context switches while the X server was accessing the DCB was causing the graphics to hang. We narrowed down the times that the X server accesses the DCB and that helped. Robert Tray From owner-linux@cthulhu.engr.sgi.com Thu Sep 18 09:21:36 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id JAA116173 for ; Thu, 18 Sep 1997 09:21:35 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id JAA19478 for linux-list; Thu, 18 Sep 1997 09:19:05 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id JAA19468 for ; Thu, 18 Sep 1997 09:19:04 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id JAA00605; Thu, 18 Sep 1997 09:19:03 -0700 Date: Thu, 18 Sep 1997 09:19:03 -0700 Message-Id: <199709181619.JAA00605@fir.engr.sgi.com> From: "William J. Earl" To: Miguel de Icaza Cc: linux@fir.engr.sgi.com Subject: Re: Linux/SGI Xsgi server: /dev/opengl ioctl NG1_SETDISPLAYMODE In-Reply-To: <199709180323.WAA12615@athena.nuclecu.unam.mx> References: <199709180203.TAA17980@fir.engr.sgi.com> <199709180323.WAA12615@athena.nuclecu.unam.mx> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1272 Lines: 25 Miguel de Icaza writes: > > Another question regarding the Newport: > > > After waiting for the rex3 FIFO to drain, and the the xmap9 FIFO > > to not be full, it sets the xmap9 mode register for the given "wid": > > > > xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); > > If the newport registers are available to the userland application, > why the X server does not directly call those routines instead of > relying on the kernel to perform them? Is it a convention that > applications should only touch the rex3 registers and not attempt to > program any of the other chips on the dcb? > > It would make sense since the graphics context switching would not > involve the kernel peeking at the chips on the dcb what their context > is and then the pain of restoring this. I believe that making this work reasonably would require some way of making a given user-mode primitive atomic. This can be hacked by via various schemes, but the current IRIX implementation does all sequences which must be atomic in the kernel. (For example, if I must BFIFOWAIT and then do some stores, but I context switch after the BFIFOWAIT, there is no guarantee that the FIFO is not full when I regain control.) From owner-linux@cthulhu.engr.sgi.com Thu Sep 18 11:54:39 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id LAA134306 for ; Thu, 18 Sep 1997 11:54:39 -0700 (PDT) Return-Path: Received: (from majordomo@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id LAA10462 for linux-list; Thu, 18 Sep 1997 11:54:03 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id LAA10446 for ; Thu, 18 Sep 1997 11:54:01 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id LAA09726 for ; Thu, 18 Sep 1997 11:54:00 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id OAA25739 for linux@engr.sgi.com; Thu, 18 Sep 1997 14:48:28 -0400 From: Mike Shaver Message-Id: <199709181848.OAA25739@neon.ingenia.ca> Subject: indirect EFS extent question To: linux@cthulhu.engr.sgi.com (Linux/SGI list) Date: Thu, 18 Sep 1997 14:48:28 -0400 (EDT) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1109 Lines: 27 I just read the comment in fs/efs_fs.h properly for the first time, and things are now much clearer. I think. So, is _this_ correct? /* * We're indirect, so we need to start looking in the * indirect extents. * * EFS indirect extents are really doubly indirect: * inode->di_u.di_extents (1st level extents) refers to * contiguous blocks of extents (2nd level extents). The * block of these 2nd level extents contain direct extents (or * 3rd level extents). So, for each indirect extent in * di_u.di_extents, we need to get a block of 2nd-level * extents, get a block from one of the 2nd-level extents, and * then check the extents in that final block -- which, at * long last, contains direct extents. */ Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Chief System Architect -- will tame sendmail(8) for food #> #> "You are a very perverse individual, and I think I'd like to get to #> know you better." --- eric@reference.com From owner-linux@cthulhu.engr.sgi.com Fri Sep 19 12:45:00 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id MAA325475 for ; Fri, 19 Sep 1997 12:45:00 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id MAA20716 for linux-list; Fri, 19 Sep 1997 12:44:21 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id MAA20703 for ; Fri, 19 Sep 1997 12:44:20 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id MAA15949 for ; Fri, 19 Sep 1997 12:44:17 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id MAA20682 for ; Fri, 19 Sep 1997 12:44:16 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id MAA05295 for ; Fri, 19 Sep 1997 12:39:25 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id OAA30318; Fri, 19 Sep 1997 14:28:38 -0500 Date: Fri, 19 Sep 1997 14:28:38 -0500 Message-Id: <199709191928.OAA30318@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@fir.engr.sgi.com Subject: Linux/SGI: IRIX X server -- kind-of-works-now status. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 2408 Lines: 60 Hello guys, Well, my IRIX Xsgi server is running on Linux/SGI without aborting now. It displays the nice X cursor and a black background (ie, no nice X background yet. This is caused -I believe- because I am not actually loading the xmap9 modes as kindly described by Bill. I have not yet debugged that problem (for some reason the display control bus locks out most of the kernel, I will look into fixing this on the afternoon). There are a couple of problems here and there, but at least it seems to be working, here is my list of known bugs for the emulation bits of IRIX Xsgi: 1. Keymaps are hossed, trivial to fix. 2. Remote clients can not connect (but then, I believe this is caused because of some faulty bit on query-net-devices-ioctl emulation). 3. /dev/mouse is hosed: the shmiq device is broken; trivial to fix as well. 4. psaux device is not working, I do not have any idea of how this works, so I have to sit down and read docs. 5. my xmap9 loading routine has problems talking to the display bus. This ought to be pretty easy to fix. Right now, I am using the DCB real slow protocol to talk to the xmap9, probably this is the culprit: relevant dcbmode bits: CSHOLD = 12; CSSETUP = 12; CSWIDTH = 0 On the xmap9.h include file, the macro that loads values into the xmap9 (xmapSetModeRegs) chooses the values for CSHOLD, CSSETUP and CSWIDTH based on the cfreq argument. On the sample code that Bill showed me, this value was passed like this (last parameter): xmap9SetModeReg( rex3, wid, displaymode, ng1_video_timing[bd->boardnum]->cfreq ); I am passing a 0 there which forces the cs values I showed previously. 6. The rrm code is not quite what an IRIX X server expects, so more work on this area is required. Which brings up an interesting question: I have the impression that you need to get a rendering node before you get permission to do direct rendering. My guess is that the context switch information should not be stored on the per-process task_struct but on the rendering node instead. Is this the case on IRIX? I have not really though very well about this. Thanks to all of the kind people at SGI who have helped me with explanations, documentation and helpful advise in the past for getting the X server up and running. best wishes and have a nice day, Miguel From owner-linux@cthulhu.engr.sgi.com Fri Sep 19 13:20:33 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id NAA334722 for ; Fri, 19 Sep 1997 13:20:33 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id NAA00719 for linux-list; Fri, 19 Sep 1997 13:19:30 -0700 Received: from oz.engr.sgi.com (oz.engr.sgi.com [150.166.61.27]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id NAA00705; Fri, 19 Sep 1997 13:19:28 -0700 Received: (from ariel@localhost) by oz.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) id NAA02902; Fri, 19 Sep 1997 13:19:27 -0700 (PDT) From: ariel@oz.engr.sgi.com (Ariel Faigon) Message-Id: <199709192019.NAA02902@oz.engr.sgi.com> Subject: Re: Linux/SGI: IRIX X server -- kind-of-works-now status. To: miguel@nuclecu.unam.mx (Miguel de Icaza) Date: Fri, 19 Sep 1997 13:19:27 -0700 (PDT) Cc: linux@cthulhu.engr.sgi.com (SGI/Linux mailing list) In-Reply-To: <199709191928.OAA30318@athena.nuclecu.unam.mx> from "Miguel de Icaza" at Sep 19, 97 02:28:38 pm Reply-To: ariel@cthulhu.engr.sgi.com (Ariel Faigon) Organization: Silicon Graphics Inc. X-Mailer: ELM [version 2.4 PL24 ME5a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 624 Lines: 23 Miguel, Congtratulations on getting Xsgi on Linux to this stage. To me it is pretty amazing that you were able to do all this only by looking at the HW docs, lots of syscall tracing, and so few hints from us. Keep up the great work. Please continue to update us on further progress until you have a fully functional Xsgi server working. Hope the X guys on the list would be able to help on the remaining questions. -- Peace, Ariel : : :Hello guys, : : Well, my IRIX Xsgi server is running on Linux/SGI without aborting :now. It displays the nice X cursor and a black background (ie, no :nice X background yet. : From owner-linux@cthulhu.engr.sgi.com Sun Sep 21 14:54:39 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id OAA553299 for ; Sun, 21 Sep 1997 14:54:38 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id OAA04098 for linux-list; Sun, 21 Sep 1997 14:54:17 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id OAA04093 for ; Sun, 21 Sep 1997 14:54:14 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id OAA25525 for ; Sun, 21 Sep 1997 14:54:13 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id QAA07108; Sun, 21 Sep 1997 16:43:18 -0500 Date: Sun, 21 Sep 1997 16:43:18 -0500 Message-Id: <199709212143.QAA07108@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux@cthulhu.engr.sgi.com Subject: Linux/SGI: IRIX X server -- more status. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 758 Lines: 25 Hello guys, Thanks to Robert Tray for helping me finding the proper protocol that has to be used with the xmap9 and 24 hours of rest, the X server is now displaying correctly on my machine. The reason the ioctl NG1_SETDISPLAYMODE was hanging my machine was pretty easy: on the xmap9 documentation it says that the number of entries available on the xmap9 fifo is a 3 bit number, but later on the document, it explains that this number is actually mapped like this: 000 - no entries available 001 - one entry available 011 - two entries available 010 - three entries available. Next time I will be more careful reading that. Now, I am off to test xterm and merge my code with the cvs repository at linux.sgi.com. Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Mon Sep 22 19:43:58 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA718159 for ; Mon, 22 Sep 1997 19:43:58 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA22164 for linux-list; Mon, 22 Sep 1997 19:43:36 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA22156 for ; Mon, 22 Sep 1997 19:43:35 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id TAA26643 for ; Mon, 22 Sep 1997 19:43:33 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA22148 for ; Mon, 22 Sep 1997 19:43:32 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id TAA10388 for ; Mon, 22 Sep 1997 19:43:31 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id VAA16659; Mon, 22 Sep 1997 21:32:19 -0500 Date: Mon, 22 Sep 1997 21:32:19 -0500 Message-Id: <199709230232.VAA16659@athena.nuclecu.unam.mx> From: Miguel de Icaza To: linux-mips@fnet.fr, linux@fir.engr.sgi.com Subject: Task list --preliminary list Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 794 Lines: 28 Here is a preliminary task list of things that should be done for a complete Linux/MIPS port, right now it includes by personal favorites and I am working on some of those bits, but some of the other can be done now, I have a copy of this at: http://www.nuclecu.unam.mx/~miguel/mips-task-list.txt Priority: [9] Get more userland programs compiled in RPM form [9] Merging the latest GNU libc releases. [9] Get the X libraries compiled. [5] Getting SGI mouse to work. [5] Test the STREAMS implementation of the mouse. [6] Figure why init is setting the sigprocmask to a value different from 0. This disables the debugger (SIGTRAP is disabled for all child processes). [7] Getting IRIX X server to accept connections. Please, send me additions to this list. Miguel. From owner-linux@cthulhu.engr.sgi.com Mon Sep 22 20:05:55 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA725904 for ; Mon, 22 Sep 1997 20:05:55 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA25005 for linux-list; Mon, 22 Sep 1997 20:05:34 -0700 Received: from oz.engr.sgi.com (oz.engr.sgi.com [150.166.61.27]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA24998; Mon, 22 Sep 1997 20:05:31 -0700 Received: (from ariel@localhost) by oz.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) id UAA09789; Mon, 22 Sep 1997 20:05:26 -0700 (PDT) From: ariel@oz.engr.sgi.com (Ariel Faigon) Message-Id: <199709230305.UAA09789@oz.engr.sgi.com> Subject: Re: Task list --preliminary list To: miguel@nuclecu.unam.mx (Miguel de Icaza) Date: Mon, 22 Sep 1997 20:05:26 -0700 (PDT) Cc: linux@cthulhu.engr.sgi.com (SGI/Linux mailing list) In-Reply-To: <199709230232.VAA16659@athena.nuclecu.unam.mx> from "Miguel de Icaza" at Sep 22, 97 09:32:19 pm Reply-To: ariel@cthulhu.engr.sgi.com (Ariel Faigon) Organization: Silicon Graphics Inc. X-Mailer: ELM [version 2.4 PL24 ME5a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1282 Lines: 46 My personal favorites are a drudgery to wizards but an essential necessity for most mortals: [5] Verify that the latest source tree on Llinus compiles and boots (maybe automate this with a daily build that gets tested) [4] Utility to boot Linux from IRIX (both locally and over net) [3] Stable precompiled Indy kernel + root-contents on linus [3] A working HOWTO for booting Linux on Indy All the above even without having X :-) : : :Here is a preliminary task list of things that should be done for a :complete Linux/MIPS port, right now it includes by personal favorites :and I am working on some of those bits, but some of the other can be :done now, I have a copy of this at: : : http://www.nuclecu.unam.mx/~miguel/mips-task-list.txt : :Priority: : :[9] Get more userland programs compiled in RPM form : :[9] Merging the latest GNU libc releases. : :[9] Get the X libraries compiled. : :[5] Getting SGI mouse to work. : :[5] Test the STREAMS implementation of the mouse. : :[6] Figure why init is setting the sigprocmask to a value different : from 0. This disables the debugger (SIGTRAP is disabled for : all child processes). : :[7] Getting IRIX X server to accept connections. : :Please, send me additions to this list. :Miguel. : -- Peace, Ariel From owner-linux@cthulhu.engr.sgi.com Mon Sep 22 20:47:27 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA734855 for ; Mon, 22 Sep 1997 20:47:27 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA00748 for linux-list; Mon, 22 Sep 1997 20:47:07 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA00725; Mon, 22 Sep 1997 20:47:04 -0700 Received: from athena.nuclecu.unam.mx (athena.nuclecu.unam.mx [132.248.29.9]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id UAA22027; Mon, 22 Sep 1997 20:47:02 -0700 env-from (miguel@athena.nuclecu.unam.mx) Received: (from miguel@localhost) by athena.nuclecu.unam.mx (8.8.5/8.8.5) id WAA17103; Mon, 22 Sep 1997 22:35:49 -0500 Date: Mon, 22 Sep 1997 22:35:49 -0500 Message-Id: <199709230335.WAA17103@athena.nuclecu.unam.mx> From: Miguel de Icaza To: ariel@cthulhu.engr.sgi.com CC: linux@cthulhu.engr.sgi.com In-reply-to: <199709230305.UAA09789@oz.engr.sgi.com> (ariel@oz.engr.sgi.com) Subject: Re: Task list --preliminary list X-Windows: It could be worse, but it'll take time. Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1260 Lines: 40 > [5] Verify that the latest source tree on Llinus compiles and boots > (maybe automate this with a daily build that gets tested) I checked this yesterday when I commited my code to linus. > [4] Utility to boot Linux from IRIX (both locally and over net) Well, we do not need this, since sash can load the Linux kernel from both an efs file system and from the network. I thought this had been covered in the HOWTO? Booting from local disk: 1. Put your kernel on an EFS partition say, /linux/vmlinux.gz 2. reset your Indy. 3. Enter maintenance mode (click on the button before IRIX boots). 4. at the prompt type: boot /linux/vmlinux.gz Note that I lack a manual on the ARC, so I am limited by what I figured would boot the kernel. There are some very interesting variables that allow you to boot the kernel from any file system on any partition. > [3] Stable precompiled Indy kernel + root-contents on linus I have my stable precompiled Indy kernel, I can setup a cron job to upload my latest kernel everyday :-). They are always stable. My root contents are currently pretty hosed up, as I have a mixture of IRIX, Linux and gnemul + some rpms + some broken rpms + a libc that I am not sure it is ok. Cheers, Miguel. From owner-linux@cthulhu.engr.sgi.com Mon Sep 22 20:52:52 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id UAA716007 for ; Mon, 22 Sep 1997 20:52:52 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id UAA01543 for linux-list; Mon, 22 Sep 1997 20:52:07 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id UAA01539 for ; Mon, 22 Sep 1997 20:52:06 -0700 Received: from motown.detroit.sgi.com (motown.detroit.sgi.com [169.238.128.3]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id UAA26779 for ; Mon, 22 Sep 1997 20:52:04 -0700 Received: from detroit.sgi.com by motown.detroit.sgi.com via ESMTP (950413.SGI.8.6.12/930416.SGI) id XAA25272; Mon, 22 Sep 1997 23:51:58 -0400 Message-ID: <34273C4E.9AD0AB2F@detroit.sgi.com> Date: Mon, 22 Sep 1997 23:49:34 -0400 From: Eric Kimminau Reply-To: eak@detroit.sgi.com Organization: Silicon Graphics, Inc X-Mailer: Mozilla 4.03C-SGI [en] (X11; I; IRIX 6.2 IP22) MIME-Version: 1.0 To: Miguel de Icaza CC: linux-mips@fnet.fr, linux@fir.engr.sgi.com Subject: Re: Task list --preliminary list References: <199709230232.VAA16659@athena.nuclecu.unam.mx> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1777 Lines: 48 PRI 10 Howto updated so it show how to take a system currently running IRIX and get it to a point where it is running Linux. I know it isn't truly development work, but Ive been stuck with a non-functional Indy for most of a month.My problems seem to be related to trying to NFS boot off of a remote XFS file system but I could be wrong. I haven't grabbed the latest kernel to try so that could resolve my issue. Miguel de Icaza wrote: > > Here is a preliminary task list of things that should be done for a > complete Linux/MIPS port, right now it includes by personal favorites > and I am working on some of those bits, but some of the other can be > done now, I have a copy of this at: > > http://www.nuclecu.unam.mx/~miguel/mips-task-list.txt > > Priority: > > [9] Get more userland programs compiled in RPM form > > [9] Merging the latest GNU libc releases. > > [9] Get the X libraries compiled. > > [5] Getting SGI mouse to work. > > [5] Test the STREAMS implementation of the mouse. > > [6] Figure why init is setting the sigprocmask to a value different > from 0. This disables the debugger (SIGTRAP is disabled for > all child processes). > > [7] Getting IRIX X server to accept connections. > > Please, send me additions to this list. > Miguel. -- Eric Kimminau System Engineer eak@detroit.sgi.com Silicon Graphics, Inc Vox:(810) 848-4455 39001 West 12mile Road Fax:(810)848-5600 Farmington, MI 48331-2903 "I speak my mind and no one else's." http://www.dcs.ex.ac.uk/~aba/rsa/perl-rsa-sig.html -----END PGP PUBLIC KEY BLOCK----- http://bs.mit.edu:11371/pks/lookup?op=vindex&search=Eric+A.+Kimminau&fingerprint=on From owner-linux@cthulhu.engr.sgi.com Mon Sep 22 21:09:25 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id VAA01175 for ; Mon, 22 Sep 1997 21:09:25 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id VAA03497 for linux-list; Mon, 22 Sep 1997 21:09:01 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id VAA03481 for ; Mon, 22 Sep 1997 21:08:56 -0700 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id VAA26845 for ; Mon, 22 Sep 1997 21:08:52 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id VAA03471 for ; Mon, 22 Sep 1997 21:08:51 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id VAA25930 for ; Mon, 22 Sep 1997 21:08:50 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id VAA15154; Mon, 22 Sep 1997 21:09:11 -0700 From: Ralf Baechle Message-Id: <199709230409.VAA15154@dns.cobaltmicro.com> Subject: Re: Task list --preliminary list To: linux-mips@fnet.fr Date: Mon, 22 Sep 1997 21:09:11 -0700 (PDT) Cc: miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com In-Reply-To: <34273C4E.9AD0AB2F@detroit.sgi.com> from "Eric Kimminau" at Sep 22, 97 11:49:34 pm Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 650 Lines: 14 > PRI 10 > Howto updated so it show how to take a system currently running IRIX and > get it to a point where it is running Linux. I know it isn't truly > development work, but Ive been stuck with a non-functional Indy for most > of a month.My problems seem to be related to trying to NFS boot off of a > remote XFS file system but I could be wrong. I haven't grabbed the > latest kernel to try so that could resolve my issue. This issues might have been resolved by Miguel's latest CVS commit. Your NFS problems might also have been cause by the VFS rewrite which was essentially turning the filesystems into an inferno for a limited time. Ralf From owner-linux@cthulhu.engr.sgi.com Tue Sep 23 03:49:00 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id DAA14093 for ; Tue, 23 Sep 1997 03:49:00 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id DAA24829 for linux-list; Tue, 23 Sep 1997 03:48:39 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id DAA24818 for ; Tue, 23 Sep 1997 03:48:36 -0700 Received: from dataserv.detroit.sgi.com (dataserv.detroit.sgi.com [169.238.128.2]) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) via ESMTP id DAA07935 for ; Tue, 23 Sep 1997 03:48:34 -0700 Received: from cygnus.detroit.sgi.com by dataserv.detroit.sgi.com via ESMTP (940816.SGI.8.6.9/930416.SGI) id GAA18850; Tue, 23 Sep 1997 06:48:23 -0400 Message-ID: <34279E77.E01AED70@cygnus.detroit.sgi.com> Date: Tue, 23 Sep 1997 06:48:23 -0400 From: Eric Kimminau Reply-To: eak@detroit.sgi.com Organization: Silicon Graphics, Inc X-Mailer: Mozilla 4.03C-SGI [en] (X11; I; IRIX 6.3 IP32) MIME-Version: 1.0 To: Ralf Baechle CC: linux-mips@fnet.fr, miguel@nuclecu.unam.mx, linux@fir.engr.sgi.com Subject: Re: Task list --preliminary list References: <199709230409.VAA15154@dns.cobaltmicro.com> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1319 Lines: 32 Ralf Baechle wrote: > > > PRI 10 > > Howto updated so it show how to take a system currently running IRIX and > > get it to a point where it is running Linux. I know it isn't truly > > development work, but Ive been stuck with a non-functional Indy for most > > of a month.My problems seem to be related to trying to NFS boot off of a > > remote XFS file system but I could be wrong. I haven't grabbed the > > latest kernel to try so that could resolve my issue. > > This issues might have been resolved by Miguel's latest CVS commit. > Your NFS problems might also have been cause by the VFS rewrite which > was essentially turning the filesystems into an inferno for a limited > time. > > Ralf Agreed. Im going to re-try boot later this afternoon with the new kernel. -- Eric Kimminau System Engineer/RSA eak@detroit.sgi.com Silicon Graphics, Inc Voice: (248) 848-4455 39001 West 12 Mile Rd. Fax: (248) 848-5600 Farmington, MI 48331-2903 VNet Extension - 6-327-4455 "I speak my mind and no one else's." http://www.dcs.ex.ac.uk/~aba/rsa/perl-rsa-sig.html When confronted by a difficult problem, solve it by reducing it to the question, "How would the Lone Ranger handle this?" From owner-linux@cthulhu.engr.sgi.com Tue Sep 23 11:39:32 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id LAA38381 for ; Tue, 23 Sep 1997 11:39:32 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id LAA29947 for linux-list; Tue, 23 Sep 1997 11:39:01 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id LAA29701; Tue, 23 Sep 1997 11:37:54 -0700 Received: from gatekeeper.qms.com (gatekeeper.qms.com [161.33.3.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via SMTP id LAA07105; Tue, 23 Sep 1997 11:37:52 -0700 env-from (marks@sun470.sun470.rd.qms.com) Received: from sun470.rd.qms.com (sun470.qms.com) by gatekeeper.qms.com (4.1/SMI-4.1) id AA04457; Tue, 23 Sep 97 13:37:31 CDT Received: from speedy.rd.qms.com by sun470.rd.qms.com (4.1/SMI-4.1) id AA26797; Tue, 23 Sep 97 13:37:28 CDT Received: by speedy.rd.qms.com (8.8.2) id NAA20145; Tue, 23 Sep 1997 13:37:28 -0500 Date: Tue, 23 Sep 1997 13:37:28 -0500 Message-Id: <199709231837.NAA20145@speedy.rd.qms.com> From: Mark Salter To: miguel@nuclecu.unam.mx Cc: ariel@cthulhu.engr.sgi.com, linux@cthulhu.engr.sgi.com In-Reply-To: <199709230335.WAA17103@athena.nuclecu.unam.mx> (message from Miguel de Icaza on Mon, 22 Sep 1997 22:35:49 -0500) Subject: Re: Task list --preliminary list Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 998 Lines: 25 >> [5] Verify that the latest source tree on Llinus compiles and boots >> (maybe automate this with a daily build that gets tested) > I checked this yesterday when I commited my code to linus. Yes, but I bet you didn't check it on an Indy with no scache :-) I have one of those beasts and the latest code in the cvs tree crashes early in the boot process. The code in arch/mips/mm/r4xx0.c which sets up the cache flush procs now selects r4k_flush_page_to_ram_d32i32_r4600 where earlier versions of r4xx0.c selected r4k_flush_page_to_ram_d32i32. I think the current choice is correct in that my Indy has the R4600 cache bug mentioned in IDT's errata. But the *_r4600 version also has some inline assembler under an ifdef CONFIG_SGI which appears to do something to the SGI scache. But since I have no scache, I get a bus error irq instead. It seems a bit strange that r4k_flush_page_to_ram_d32i32_r4600 has this bit of SGI specific code, but r4k_flush_page_to_ram_d32i32 does not. ??? --Mark From owner-linux@cthulhu.engr.sgi.com Tue Sep 23 15:41:08 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA14757 for ; Tue, 23 Sep 1997 15:41:07 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA26035 for linux-list; Tue, 23 Sep 1997 15:39:12 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA25993; Tue, 23 Sep 1997 15:39:03 -0700 Received: from neon.ingenia.ca (neon.ingenia.ca [205.207.220.57]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id PAA24049; Tue, 23 Sep 1997 15:38:49 -0700 env-from (shaver@neon.ingenia.ca) Received: (from shaver@localhost) by neon.ingenia.ca (8.8.5/8.7.3) id SAA02473; Tue, 23 Sep 1997 18:32:20 -0400 From: Mike Shaver Message-Id: <199709232232.SAA02473@neon.ingenia.ca> Subject: Re: Task list --preliminary list In-Reply-To: <199709230305.UAA09789@oz.engr.sgi.com> from Ariel Faigon at "Sep 22, 97 08:05:26 pm" To: ariel@cthulhu.engr.sgi.com Date: Tue, 23 Sep 1997 18:32:19 -0400 (EDT) Cc: miguel@nuclecu.unam.mx, linux@cthulhu.engr.sgi.com X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1154 Lines: 29 Thus spake Ariel Faigon: > [5] Verify that the latest source tree on Llinus compiles and boots > (maybe automate this with a daily build that gets tested) Boots on? We'll need to define a set of known-tested/supported systems, at least for now. > [4] Utility to boot Linux from IRIX (both locally and over net) Stay tuned. I'm going to be incommunicado for about a week while I move to Mountain View, but I only have one bug[*] keeping me from being able to boot off of an EFS partition, etc. That should make things a lot easier. [*] Yes, it's embarrassingly true: I still can't get indirect extents right. I'll check my code in tomorrow, in case someone smarter than me wants to take a crack at it. Mike -- #> Mike Shaver (shaver@ingenia.com) Ingenia Communications Corporation #> Resident Linux bigot and kernel hacker. (OOPS!) #> `If you get bitten by a bug, tough luck...the one thing I won't do #> is feel sorry for you. In fact, I might ask you to do it all over #> again, just to get more information. I'm a heartless bastard.' #> -- Linus Torvalds (on development kernels) From owner-linux@cthulhu.engr.sgi.com Tue Sep 23 15:41:07 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA14750 for ; Tue, 23 Sep 1997 15:41:07 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA26361 for linux-list; Tue, 23 Sep 1997 15:40:48 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA26342; Tue, 23 Sep 1997 15:40:43 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id PAA09768; Tue, 23 Sep 1997 15:40:35 -0700 Date: Tue, 23 Sep 1997 15:40:35 -0700 Message-Id: <199709232240.PAA09768@fir.engr.sgi.com> From: "William J. Earl" To: Mark Salter Cc: miguel@nuclecu.unam.mx, ariel@cthulhu.engr.sgi.com, linux@cthulhu.engr.sgi.com Subject: Re: Task list --preliminary list In-Reply-To: <199709231837.NAA20145@speedy.rd.qms.com> References: <199709230335.WAA17103@athena.nuclecu.unam.mx> <199709231837.NAA20145@speedy.rd.qms.com> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1535 Lines: 31 Mark Salter writes: > > >> [5] Verify that the latest source tree on Llinus compiles and boots > >> (maybe automate this with a daily build that gets tested) > > > I checked this yesterday when I commited my code to linus. > > Yes, but I bet you didn't check it on an Indy with no scache :-) > I have one of those beasts and the latest code in the cvs tree > crashes early in the boot process. > > The code in arch/mips/mm/r4xx0.c which sets up the cache flush > procs now selects r4k_flush_page_to_ram_d32i32_r4600 where > earlier versions of r4xx0.c selected r4k_flush_page_to_ram_d32i32. > I think the current choice is correct in that my Indy has the > R4600 cache bug mentioned in IDT's errata. But the *_r4600 version > also has some inline assembler under an ifdef CONFIG_SGI which > appears to do something to the SGI scache. But since I have > no scache, I get a bus error irq instead. > > It seems a bit strange that r4k_flush_page_to_ram_d32i32_r4600 > has this bit of SGI specific code, but r4k_flush_page_to_ram_d32i32 > does not. ??? I have not looked at the source, but the problem is probably that there needs to be two cases, one with and one without Indy R4600 scache. This is determined by reading out the EEPROM on the CPU board, and looking at the appropriate bits for "scache present" and "scache size". The processor will get a bus error trying to reference the control registers of the scache controller (via uncached memory references) if the secondary cache is not present. From owner-linux@cthulhu.engr.sgi.com Tue Sep 23 16:35:48 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA19809 for ; Tue, 23 Sep 1997 16:35:48 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA08468 for linux-list; Tue, 23 Sep 1997 16:35:17 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA08463; Tue, 23 Sep 1997 16:35:15 -0700 Received: from informatik.uni-koblenz.de (mailhost.uni-koblenz.de [141.26.4.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA10410; Tue, 23 Sep 1997 16:35:12 -0700 env-from (ralf@informatik.uni-koblenz.de) Received: from erbse (ralf@erbse.uni-koblenz.de [141.26.5.44]) by informatik.uni-koblenz.de (8.8.7/8.6.9) with SMTP id BAA03833; Wed, 24 Sep 1997 01:34:53 +0200 (MEST) From: Ralf Baechle Message-Id: <199709232334.BAA03833@informatik.uni-koblenz.de> Received: by erbse (SMI-8.6/KO-2.0) id BAA04268; Wed, 24 Sep 1997 01:34:50 +0200 Subject: Re: Task list --preliminary list To: marks@sun470.sun470.rd.qms.com (Mark Salter) Date: Wed, 24 Sep 1997 01:34:50 +0200 (MET DST) Cc: miguel@nuclecu.unam.mx, ariel@cthulhu.engr.sgi.com, linux@cthulhu.engr.sgi.com, linux-mips@fnet.fr In-Reply-To: <199709231837.NAA20145@speedy.rd.qms.com> from "Mark Salter" at Sep 23, 97 01:37:28 pm MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1510 Lines: 31 > Yes, but I bet you didn't check it on an Indy with no scache :-) > I have one of those beasts and the latest code in the cvs tree > crashes early in the boot process. > > The code in arch/mips/mm/r4xx0.c which sets up the cache flush > procs now selects r4k_flush_page_to_ram_d32i32_r4600 where > earlier versions of r4xx0.c selected r4k_flush_page_to_ram_d32i32. > I think the current choice is correct in that my Indy has the > R4600 cache bug mentioned in IDT's errata. But the *_r4600 version > also has some inline assembler under an ifdef CONFIG_SGI which > appears to do something to the SGI scache. But since I have > no scache, I get a bus error irq instead. > > It seems a bit strange that r4k_flush_page_to_ram_d32i32_r4600 > has this bit of SGI specific code, but r4k_flush_page_to_ram_d32i32 > does not. ??? The R4400 versions of the Indy have the "real" second level cache and use a SC CPU version. Opposed to that the L2 cache of the R4600 and R5000 Indies is controlled by external circuitry which needs special code. Actually since we have the same problem on a couple of other machines, SNI RM family one of them, what should be done is to break out all the #ifugly'ed SGI bits and provide some hooks for them. You L2 less machine would simply plug an empty function into these hooks. Using r4k_flush_page_to_ram_d32i32 was definately wrong, the R4600 cache is two way set associative and flush_page_to_ram needs to take care of that for all the R4600 and R5000 relatives. Ralf From owner-linux@cthulhu.engr.sgi.com Wed Sep 24 01:53:59 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id BAA76900 for ; Wed, 24 Sep 1997 01:53:59 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id BAA23945 for linux-list; Wed, 24 Sep 1997 01:53:43 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id BAA23935 for ; Wed, 24 Sep 1997 01:53:37 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id BAA27014 for ; Wed, 24 Sep 1997 01:53:35 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id AAA20896; Wed, 24 Sep 1997 00:15:11 -0700 From: Ralf Baechle Message-Id: <199709240715.AAA20896@dns.cobaltmicro.com> Subject: glibc 2.0.4 bug ... To: linux-mips@fnet.fr, linux@cthulhu.engr.sgi.com, drepper@cygnus.com Date: Wed, 24 Sep 1997 00:15:10 -0700 (PDT) Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 535 Lines: 14 Hi all, glibc 2.0.4 contains a fatal bug. It does not declare a prototype for the function llseek in unistd.h. As result GCC will (correctly) truncate the 64 bit file offset and build erroneous filesystems when building filesystems of 2GB. e2fsck will complain about read errors when trying to read blocks from the 2GB border on. Actually I wonder why I never saw a report about that on other mailinglists. Quickfix: add the following prototype for llseek(2): extern loff_t llseek (int fd, loff_t offset, int whence); Ralf From owner-linux@cthulhu.engr.sgi.com Wed Sep 24 10:17:41 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id KAA102054 for ; Wed, 24 Sep 1997 10:17:41 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id KAA13782 for linux-list; Wed, 24 Sep 1997 10:17:15 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id KAA13708 for ; Wed, 24 Sep 1997 10:16:59 -0700 Received: from cygint.cygnus.com (cygnus.com [205.180.230.20]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id KAA13866 for ; Wed, 24 Sep 1997 10:16:58 -0700 env-from (drepper@cygnus.com) Received: from localhost.cygnus.com (rtl.cygnus.com [205.180.230.21]) by cygint.cygnus.com (8.8.5/8.8.7) with ESMTP id KAA16923; Wed, 24 Sep 1997 10:16:36 -0700 (PDT) Message-Id: <199709241716.KAA16923@cygint.cygnus.com> To: ralf@cobaltmicro.com Cc: linux-mips@fnet.fr, linux@cthulhu.engr.sgi.com Subject: Re: glibc 2.0.4 bug ... In-Reply-To: Your message of "Wed, 24 Sep 1997 00:15:10 -0700 (PDT)" References: <199709240715.AAA20896@dns.cobaltmicro.com> X-Mailer: Mew version 1.06 on Emacs 19.34.1 X-fingerprint: BE 3B 21 04 BC 77 AC F0 61 92 E4 CB AC DD B9 5A Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Date: Wed, 24 Sep 1997 10:16:35 -0700 From: Ulrich Drepper Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 582 Lines: 15 From: Ralf Baechle Subject: glibc 2.0.4 bug ... Date: Wed, 24 Sep 1997 00:15:10 -0700 (PDT) > Quickfix: add the following prototype for llseek(2): > > extern loff_t llseek (int fd, loff_t offset, int whence); I've mentioned this several times: llseek is an internal function. What we need is the LFS interface which then defines lseek64. -- Uli ---------------. drepper@cygnus.com ,-. Rubensstrasse 5 Ulrich Drepper \ ,-------------------' \ 76149 Karlsruhe/Germany Cygnus Solutions `--' drepper@gnu.ai.mit.edu `------------------------ From owner-linux@cthulhu.engr.sgi.com Wed Sep 24 14:38:48 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id OAA121318 for ; Wed, 24 Sep 1997 14:38:48 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id OAA17897 for linux-list; Wed, 24 Sep 1997 14:38:22 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id OAA17889 for ; Wed, 24 Sep 1997 14:38:20 -0700 Received: from mocha.foo.org (pr119.pheasantrun.net [208.140.225.119]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id OAA06737 for ; Wed, 24 Sep 1997 14:38:19 -0700 env-from (clepple@foo.org) Received: from mug.foo.org by mocha.foo.org on Wed, 24 Sep 1997 17:37:42 -0400 (EDT) Received: from mug.foo.org (localhost [127.0.0.1]) by mug.foo.org (8.7.6/8.7.3) with ESMTP id RAA00921 for ; Wed, 24 Sep 1997 17:37:26 -0400 Message-Id: <199709242137.RAA00921@mug.foo.org> X-Mailer: exmh version 1.6.7 05/05/96 To: linux@cthulhu.engr.sgi.com Subject: Indy netboot problems Reply-to: clepple@foo.org Mime-Version: 1.0 Content-Type: text/plain Date: Wed, 24 Sep 1997 17:37:26 -0400 From: Charles Lepple Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 742 Lines: 25 Hello all, Sorry to bother you if this is a routine question, but I read the FAQ, and therein lies my problem. I'm trying to make an Indy netboot, but I'm not too familiar with the new monitor (let's just say that I knew sifex much better...Anyone remember the IRIS 3130s?). I need a way to correct the HW address, since it appears to be set to '0:0:FF:FF:0:0', and it is complaining about it. Is there any other way to change nvram settings except through the setenv/printenv commands? I hope to be contributing to this project soon, but I'm off to a fairly rocky start ;-) Thanks in advance, --Charles Lepple, http://www.ee.vt.edu/solarcar/ http://www.mitre.org/research/nanotech/ From owner-linux@cthulhu.engr.sgi.com Wed Sep 24 15:24:18 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id PAA122846 for ; Wed, 24 Sep 1997 15:24:18 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id PAA28554 for linux-list; Wed, 24 Sep 1997 15:23:55 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id PAA28548; Wed, 24 Sep 1997 15:23:53 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id PAA24127; Wed, 24 Sep 1997 15:23:52 -0700 Date: Wed, 24 Sep 1997 15:23:52 -0700 Message-Id: <199709242223.PAA24127@fir.engr.sgi.com> From: "William J. Earl" To: clepple@foo.org Cc: linux@cthulhu.engr.sgi.com Subject: Re: Indy netboot problems In-Reply-To: <199709242137.RAA00921@mug.foo.org> References: <199709242137.RAA00921@mug.foo.org> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 929 Lines: 19 Charles Lepple writes: > Hello all, > Sorry to bother you if this is a routine question, but I read the FAQ, and > therein lies my problem. > > I'm trying to make an Indy netboot, but I'm not too familiar with the new > monitor (let's just say that I knew sifex much better...Anyone remember the > IRIS 3130s?). I need a way to correct the HW address, since it appears to be > set to '0:0:FF:FF:0:0', and it is complaining about it. > > Is there any other way to change nvram settings except through the > setenv/printenv commands? ... "setenv -p xxx yyy" will set xxx permanently (in NVRAM) to yyy. Without the "-p", it just sets xxx in the environment in memory. For the hardware address ("eaddr"), however, you cannot readily change it, as it is the system serial number as well as the MAC address. I don't know any easy way for it have been corrupted; it would normally be fixed by field service. From owner-linux@cthulhu.engr.sgi.com Wed Sep 24 17:51:46 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id RAA150591 for ; Wed, 24 Sep 1997 17:51:46 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA27721 for linux-list; Wed, 24 Sep 1997 17:51:25 -0700 Received: from soyuz.wellington.sgi.com (soyuz.wellington.sgi.com [134.14.64.194]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA27714 for ; Wed, 24 Sep 1997 17:51:18 -0700 Received: from sgi.com by soyuz.wellington.sgi.com via ESMTP (950413.SGI.8.6.12/940406.SGI) id MAA19223; Thu, 25 Sep 1997 12:50:18 +1200 Message-ID: <3429B54A.ADFC1795@sgi.com> Date: Thu, 25 Sep 1997 12:50:18 +1200 From: Alistair Lambie X-Mailer: Mozilla 4.02 [en] (X11; I; IRIX 6.5-ALPHA-1274090336 IP22) MIME-Version: 1.0 To: clepple@foo.org CC: linux@cthulhu.engr.sgi.com Subject: Re: Indy netboot problems References: <199709242137.RAA00921@mug.foo.org> Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1172 Lines: 29 Charles Lepple wrote: > > Hello all, > Sorry to bother you if this is a routine question, but I read the FAQ, and > therein lies my problem. > > I'm trying to make an Indy netboot, but I'm not too familiar with the new > monitor (let's just say that I knew sifex much better...Anyone remember the > IRIS 3130s?). I need a way to correct the HW address, since it appears to be > set to '0:0:FF:FF:0:0', and it is complaining about it. > > Is there any other way to change nvram settings except through the > setenv/printenv commands? > You can not change this with setenv. If it is '0:0:FF:FF:0:0' there is a fault and you need to get the motherboard replaced. It is probably the 'Dallas Chip' that is the problem. I have twice seen where a pin gets bent on this and it returns funny MAC adresses....so if you have some hardware knowledge it may be worth checking that it is seated correctly and the pins are not bent. Cheers, Alistair -- Alistair Lambie alambie@wellington.sgi.com Silicon Graphics New Zealand SGI Voicemail: 56791 Level 5, Walsh Wrightson Tower, Ph: +64-4-802 1455 94-96 Dixon St, Wellington, NZ Fax: +64-4-802 1459 From owner-linux@cthulhu.engr.sgi.com Thu Sep 25 23:42:30 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id XAA359524 for ; Thu, 25 Sep 1997 23:42:30 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id XAA25907 for linux-list; Thu, 25 Sep 1997 23:42:07 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id XAA25900 for ; Thu, 25 Sep 1997 23:42:04 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id XAA20430 for ; Thu, 25 Sep 1997 23:42:00 -0700 env-from (ralf@mail2.cobaltmicro.com) Received: from dull.cobaltmicro.com (dull.cobaltmicro.com [209.19.61.35]) by dns.cobaltmicro.com (8.8.5/8.8.5) with ESMTP id XAA31376 for ; Thu, 25 Sep 1997 23:41:45 -0700 From: Ralf Baechle Received: (from ralf@localhost) by dull.cobaltmicro.com (8.8.5/8.8.5) id XAA09959 for linux@engr.sgi.com; Thu, 25 Sep 1997 23:39:27 -0700 Message-Id: <199709260639.XAA09959@dull.cobaltmicro.com> Subject: R3000 SGI machines To: linux@cthulhu.engr.sgi.com Date: Thu, 25 Sep 1997 23:39:27 -0700 (PDT) Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 258 Lines: 8 Hi all, I receive an increasing number of emails asking for Linux as replacement operating system for older R3000 SGI machines as well as Mips Inc. machines. Is there still documentation available for these systems that would make a port possible? Ralf From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 04:24:26 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id EAA362764 for ; Fri, 26 Sep 1997 04:24:26 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id EAA15773 for linux-list; Fri, 26 Sep 1997 04:24:10 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id EAA15755 for ; Fri, 26 Sep 1997 04:24:07 -0700 Received: from emout08.mail.aol.com (emout08.mx.aol.com [198.81.11.23]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id EAA03826 for ; Fri, 26 Sep 1997 04:24:06 -0700 env-from (LetherGlov@aol.com) From: LetherGlov@aol.com Received: (from root@localhost) by emout08.mail.aol.com (8.7.6/8.7.3/AOL-2.0.0) id HAA08289 for linux@cthulhu.engr.sgi.com; Fri, 26 Sep 1997 07:24:05 -0400 (EDT) Date: Fri, 26 Sep 1997 07:24:05 -0400 (EDT) Message-ID: <970926072404_1332144893@emout08.mail.aol.com> To: linux@cthulhu.engr.sgi.com Subject: Re: R3000 SGI machines Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 635 Lines: 13 You know I was just thinking about that this morning. I was more leaning towards the personal iris, and the crimson with the video adapter that is not supported in 6.2(the name isn't coming to me right now, GTX?). I was wondering how we got the specs from the ethernet, bus, and serial controllers from SGI. And the chip specs from mips. If I understand SGI history correctly, there was a change in the CPU after revision 3, and we are going to need to see if this is going to cause problems. another thing we need to do(I'll do it) is call ID Software and get the code for DOOM so It can be compiled for Linux. Thanks, Robbie Stone From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 08:10:53 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id IAA365097 for ; Fri, 26 Sep 1997 08:10:53 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA00293 for linux-list; Fri, 26 Sep 1997 08:10:36 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA00289 for ; Fri, 26 Sep 1997 08:10:35 -0700 Received: from rover.village.org (rover.village.org [204.144.255.49]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via SMTP id IAA16902 for ; Fri, 26 Sep 1997 08:10:21 -0700 env-from (imp@village.org) Received: from harmony [10.0.0.6] by rover.village.org with esmtp (Exim 1.71 #1) id 0xEc27-0003iK-00; Fri, 26 Sep 1997 09:10:19 -0600 Received: from harmony.village.org (localhost [127.0.0.1]) by harmony.village.org (8.8.7/8.8.3) with ESMTP id JAA22806; Fri, 26 Sep 1997 09:10:24 -0600 (MDT) Message-Id: <199709261510.JAA22806@harmony.village.org> To: Ralf Baechle Subject: Re: R3000 SGI machines Cc: linux@cthulhu.engr.sgi.com In-reply-to: Your message of "Thu, 25 Sep 1997 23:39:27 PDT." <199709260639.XAA09959@dull.cobaltmicro.com> References: <199709260639.XAA09959@dull.cobaltmicro.com> Date: Fri, 26 Sep 1997 09:10:24 -0600 From: Warner Losh Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 735 Lines: 14 In message <199709260639.XAA09959@dull.cobaltmicro.com> Ralf Baechle writes: : I receive an increasing number of emails asking for Linux as replacement : operating system for older R3000 SGI machines as well as Mips Inc. : machines. Is there still documentation available for these systems : that would make a port possible? I'll have to search my email archive (which would take some doing), but someone at SGI offered to get me the sources to RISCos, or at least the device drivers. They were donated to berkeley, who didn't integrate them into their 4.4 or 4.4lite release, was the story I was told. Might be worth looking into. Wouldn't help with some of the older SGI machines, but would help with the MIPS co boxes. Warner From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 08:50:58 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id IAA365600 for ; Fri, 26 Sep 1997 08:50:57 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA05491 for linux-list; Fri, 26 Sep 1997 08:50:37 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA05487; Fri, 26 Sep 1997 08:50:35 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id IAA10923; Fri, 26 Sep 1997 08:50:34 -0700 Date: Fri, 26 Sep 1997 08:50:34 -0700 Message-Id: <199709261550.IAA10923@fir.engr.sgi.com> From: "William J. Earl" To: Ralf Baechle Cc: linux@cthulhu.engr.sgi.com Subject: Re: R3000 SGI machines In-Reply-To: <199709260639.XAA09959@dull.cobaltmicro.com> References: <199709260639.XAA09959@dull.cobaltmicro.com> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 874 Lines: 19 Ralf Baechle writes: > Hi all, > > I receive an increasing number of emails asking for Linux as replacement > operating system for older R3000 SGI machines as well as Mips Inc. > machines. Is there still documentation available for these systems > that would make a port possible? There is some, which I could track down, on the same basis as the Indy documentation. I don't know how complete the files are, however. There is a family resemblance in much of Indigo system, especially the I/O. The Personal IRIS and Power Series machines, which were VME-based, are fairly different (except for graphics), but are relatively similar to the MIPS VME systems, such as the M1000 and M2000. The Power Series are MP systems, and have some special purpose interprocessor synchronization hardware. The documentation for the VME systems will be harder to locate. From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 08:56:23 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id IAA363922 for ; Fri, 26 Sep 1997 08:56:23 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA06197 for linux-list; Fri, 26 Sep 1997 08:56:12 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA06182; Fri, 26 Sep 1997 08:56:10 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id IAA10936; Fri, 26 Sep 1997 08:56:09 -0700 Date: Fri, 26 Sep 1997 08:56:09 -0700 Message-Id: <199709261556.IAA10936@fir.engr.sgi.com> From: "William J. Earl" To: LetherGlov@aol.com Cc: linux@cthulhu.engr.sgi.com Subject: Re: R3000 SGI machines In-Reply-To: <970926072404_1332144893@emout08.mail.aol.com> References: <970926072404_1332144893@emout08.mail.aol.com> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1191 Lines: 21 LetherGlov@aol.com writes: > You know I was just thinking about that this morning. I was more leaning > towards the personal iris, and the crimson with the video adapter that is not > supported in 6.2(the name isn't coming to me right now, GTX?). I was > wondering how we got the specs from the ethernet, bus, and serial controllers > from SGI. And the chip specs from mips. If I understand SGI history > correctly, there was a change in the CPU after revision 3, and we are going > to need to see if this is going to cause problems. another thing we need to > do(I'll do it) is call ID Software and get the code for DOOM so It can be > compiled for Linux. The MIPS processor specifications are public, and available as printed books. Also, many of them are online at http://www.sgi.com/MIPS/. With the approval of the appropriate division general manager, I supplied the people doing the Indy port of linux with copies of the Indy hardware documentation. Since the Personal IRIX and Crimson use mostly VME controllers. those drivers should carry over from other VME ports. I don't know if I can arrange to provide documentation for the graphics controllers. From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 08:58:32 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id IAA364624 for ; Fri, 26 Sep 1997 08:58:31 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id IAA06492 for linux-list; Fri, 26 Sep 1997 08:58:21 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id IAA06485; Fri, 26 Sep 1997 08:58:19 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id IAA10949; Fri, 26 Sep 1997 08:58:19 -0700 Date: Fri, 26 Sep 1997 08:58:19 -0700 Message-Id: <199709261558.IAA10949@fir.engr.sgi.com> From: "William J. Earl" To: Warner Losh Cc: Ralf Baechle , linux@cthulhu.engr.sgi.com Subject: Re: R3000 SGI machines In-Reply-To: <199709261510.JAA22806@harmony.village.org> References: <199709260639.XAA09959@dull.cobaltmicro.com> <199709261510.JAA22806@harmony.village.org> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1022 Lines: 19 Warner Losh writes: > In message <199709260639.XAA09959@dull.cobaltmicro.com> Ralf Baechle writes: > : I receive an increasing number of emails asking for Linux as replacement > : operating system for older R3000 SGI machines as well as Mips Inc. > : machines. Is there still documentation available for these systems > : that would make a port possible? > > I'll have to search my email archive (which would take some doing), > but someone at SGI offered to get me the sources to RISCos, or at > least the device drivers. They were donated to berkeley, who didn't > integrate them into their 4.4 or 4.4lite release, was the story I was > told. Might be worth looking into. Wouldn't help with some of the > older SGI machines, but would help with the MIPS co boxes. The RISCos driver source was supplied with every system, and we did donate the source to Berkeley. Most of the VME controllers were common to MIPS and SGI VME systems, so most of the drivers, other than graphics, would be covered. From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 11:59:10 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id LAA367854 for ; Fri, 26 Sep 1997 11:59:10 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id LAA13897 for linux-list; Fri, 26 Sep 1997 11:58:52 -0700 Received: from dragon.engr.sgi.com (dragon.engr.sgi.com [150.166.63.33]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id LAA13889 for ; Fri, 26 Sep 1997 11:58:51 -0700 Received: (from sr@localhost) by dragon.engr.sgi.com (970321.SGI.8.8.5/970502.SGI.AUTOCF) id LAA67409 for linux@engr.sgi.com; Fri, 26 Sep 1997 11:58:50 -0700 (PDT) From: sr@dragon.engr.sgi.com (Steve Rikli) Message-Id: <199709261858.LAA67409@dragon.engr.sgi.com> Subject: Re: R3000 SGI machines To: linux@cthulhu.engr.sgi.com Date: Fri, 26 Sep 1997 11:58:50 -0700 (PDT) In-Reply-To: <199709261556.IAA10936@fir.engr.sgi.com> from "William J. Earl" at Sep 26, 97 08:56:09 am X-Mailer: ELM [version 2.4 PL24 ME5a] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1910 Lines: 42 William J. Earl wrote: > LetherGlov@aol.com writes: >> >> You know I was just thinking about that this morning. I was more leaning >> towards the personal iris, and the crimson with the video adapter that is not >> supported in 6.2(the name isn't coming to me right now, GTX?). I was >> wondering how we got the specs from the ethernet, bus, and serial controllers >> from SGI. And the chip specs from mips. If I understand SGI history >> correctly, there was a change in the CPU after revision 3, and we are going >> to need to see if this is going to cause problems. another thing we need to >> do(I'll do it) is call ID Software and get the code for DOOM so It can be >> compiled for Linux. > > The MIPS processor specifications are public, and available as printed > books. Also, many of them are online at http://www.sgi.com/MIPS/. > > With the approval of the appropriate division general manager, I > supplied the people doing the Indy port of linux with copies of the > Indy hardware documentation. > > Since the Personal IRIX and Crimson use mostly VME controllers. those > drivers should carry over from other VME ports. I don't know if I can > arrange to provide documentation for the graphics controllers. This may not be exactly the kind of thing some of y'all are looking for, but since we're talking about doc & info sources for old SGI hardware I felt I should point this one out: http://www.geocities.com/SiliconValley/Pines/2258/4dfaq.html Several handfulls of general notes about the 4D and/or other R3K platforms. Hope someone finds it useful. cheers, sr. -- || Steve Rikli ||| Any sufficiently advanced technology is || || Systems Administrator ||| indistinguishable from a rigged demo. || || NSD, EIS Infrastructure ||| || || Silicon Graphics, Inc. ||| - Andy Finkel || From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 12:11:53 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id MAA364789 for ; Fri, 26 Sep 1997 12:11:53 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id MAA17162 for linux-list; Fri, 26 Sep 1997 12:11:33 -0700 Received: from oz.engr.sgi.com (oz.engr.sgi.com [150.166.61.27]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id MAA17090; Fri, 26 Sep 1997 12:11:30 -0700 Received: from sgi.com (localhost [127.0.0.1]) by oz.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via ESMTP id MAA23057; Fri, 26 Sep 1997 12:11:30 -0700 (PDT) Message-ID: <342C08E1.DF562E36@sgi.com> Date: Fri, 26 Sep 1997 12:11:29 -0700 From: Ariel Faigon Organization: Silicon Graphics, Inc. X-Mailer: Mozilla 4.03C-SGI [en] (X11; I; IRIX 6.5-ALPHA-1274090336 IP32) MIME-Version: 1.0 To: acorda@geocities.com CC: linux@cthulhu.engr.sgi.com Subject: Muchos thankyous (4D FAQ)! Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 316 Lines: 12 Dear AJC, Your 4D-IRIS / old SGI hardware documentation is a treasure. Thank you for puting it online. I linked to it from the www.linux.sgi.com web site with the hope that one day someone would port SGI/Linux to these old machines. Thanks to Steve Rikli for pointing this site to teh linux list. -- Peace, Ariel From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 17:36:55 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id RAA548840 for ; Fri, 26 Sep 1997 17:36:54 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA05453 for linux-list; Fri, 26 Sep 1997 17:36:37 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA05449; Fri, 26 Sep 1997 17:36:35 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id RAA04775; Fri, 26 Sep 1997 17:36:25 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id RAA02682; Fri, 26 Sep 1997 17:36:25 -0700 From: Ralf Baechle Message-Id: <199709270036.RAA02682@dns.cobaltmicro.com> Subject: Re: R3000 SGI machines To: wje@fir.engr.sgi.com (William J. Earl) Date: Fri, 26 Sep 1997 17:36:25 -0700 (PDT) Cc: imp@village.org, ralf@cobaltmicro.com, linux@cthulhu.engr.sgi.com, ab@ivm.net In-Reply-To: <199709261558.IAA10949@fir.engr.sgi.com> from "William J. Earl" at Sep 26, 97 08:58:19 am Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1277 Lines: 26 > Warner Losh writes: > > In message <199709260639.XAA09959@dull.cobaltmicro.com> Ralf Baechle writes: > > : I receive an increasing number of emails asking for Linux as replacement > > : operating system for older R3000 SGI machines as well as Mips Inc. > > : machines. Is there still documentation available for these systems > > : that would make a port possible? > > > > I'll have to search my email archive (which would take some doing), > > but someone at SGI offered to get me the sources to RISCos, or at > > least the device drivers. They were donated to berkeley, who didn't > > integrate them into their 4.4 or 4.4lite release, was the story I was > > told. Might be worth looking into. Wouldn't help with some of the > > older SGI machines, but would help with the MIPS co boxes. > > The RISCos driver source was supplied with every system, and we > did donate the source to Berkeley. Most of the VME controllers were > common to MIPS and SGI VME systems, so most of the drivers, other than > graphics, would be covered. I think I can still get access to RISC/os 5.x system, maybe also old RISC/os 4.51 tapes. What files should I try to look for? Andy, I think you still have Resi in use, don't you? Any driver sources included? Ralf From owner-linux@cthulhu.engr.sgi.com Fri Sep 26 19:24:51 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id TAA552061 for ; Fri, 26 Sep 1997 19:24:51 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id TAA17129 for linux-list; Fri, 26 Sep 1997 19:24:34 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id TAA17124 for ; Fri, 26 Sep 1997 19:24:32 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id TAA24091 for ; Fri, 26 Sep 1997 19:24:30 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id TAA03105; Fri, 26 Sep 1997 19:24:36 -0700 From: Ralf Baechle Message-Id: <199709270224.TAA03105@dns.cobaltmicro.com> Subject: Re: MIPS R2k/3k machines.... To: linux-mips@fnet.fr, linux@cthulhu.engr.sgi.com Date: Fri, 26 Sep 1997 19:24:35 -0700 (PDT) In-Reply-To: from "Lucien Murray-Pitts" at Sep 27, 97 02:55:42 am Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 894 Lines: 22 Hi, > I too, like the several other people, want to help port Linux to > the older 2k and 3k systems BUT like all the others I am hindered by a > total LACK of documentation. I did manage to find one nice chap who gave > me some docco. but unfortuantely it was for the R4Ks, however... I assume you only refer to the system documentation. The CPU documentation is available for free from {sgigate,www}.sgi.com and a couple of Mips silicon manufacturers like {ftp,www}.idt.com and others more. > Does ANYONE have ANY technical INFORMATION which might be of use > to all of us port of linux. > > No matter how small, silly or whatever I'd like to hear about it. I asked some people at SGI and they'll try to dig out what still exists. Somebody came up with a quite nice link for owners of historic SGI machines. Check out the site, there is a link to it on www.linux.sgi.com. Ralf From owner-linux@cthulhu.engr.sgi.com Sun Sep 28 17:15:38 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id RAA630323 for ; Sun, 28 Sep 1997 17:15:38 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id RAA27419 for linux-list; Sun, 28 Sep 1997 17:15:13 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id RAA27415 for ; Sun, 28 Sep 1997 17:15:10 -0700 Received: from server3.syd.mail.ozemail.net (server3.syd.mail.ozemail.net [203.108.7.41]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id RAA23820 for ; Sun, 28 Sep 1997 17:15:06 -0700 env-from (Regan@bde.com.au) Received: from oznet02.ozemail.com.au (oznet02.ozemail.com.au [203.2.192.124]) by server3.syd.mail.ozemail.net (8.8.4/8.6.12) with ESMTP id KAA03720; Mon, 29 Sep 1997 10:14:57 +1000 (EST) Received: from bde.com.au (gamers.ozemail.com.au [203.7.189.162]) by oznet02.ozemail.com.au (8.8.4/8.6.12) with SMTP id KAA21074; Mon, 29 Sep 1997 10:14:53 +1000 (EST) Received: by bde.com.au with NT SMTP Gateway ver 31 id <342EF445@bde.com.au>; Mon, 29 Sep 97 11:20:21 S From: Regan Russell To: linux-mips , linux , linux-mips-request Subject: RE: MIPS R2k/3k machines.... Date: Mon, 29 Sep 97 11:11:00 S Message-ID: <342EF445@bde.com.au> X-Mailer: Microsoft Mail V3.0 Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 1523 Lines: 46 Look for "The mips programmers refenence manual". (Faquar and Bunce ?) ( R2000 / R3000 ) "The MIPS RISC architecture". (Kane ?) ( mentions R4000 but contrates on the R2000 / R3000 ) Both books were availible recently. I have both books at home, (I do complete crap in windoze for a living), also look at xspim and spim the simulator for the R2000, R3000. Think it was on sunsite... Hope this helps.. PS. Anyone got a spare motherboard for a DS3100 ? ---------- From: linux-mips-request Sent: Friday, September 26, 1997 7:24 PM To: linux-mips; linux Subject: Re: MIPS R2k/3k machines.... Hi, > I too, like the several other people, want to help port Linux to > the older 2k and 3k systems BUT like all the others I am hindered by a > total LACK of documentation. I did manage to find one nice chap who gave > me some docco. but unfortuantely it was for the R4Ks, however... I assume you only refer to the system documentation. The CPU documentation is available for free from {sgigate,www}.sgi.com and a couple of Mips silicon manufacturers like {ftp,www}.idt.com and others more. > Does ANYONE have ANY technical INFORMATION which might be of use > to all of us port of linux. > > No matter how small, silly or whatever I'd like to hear about it. I asked some people at SGI and they'll try to dig out what still exists. Somebody came up with a quite nice link for owners of historic SGI machines. Check out the site, there is a link to it on www.linux.sgi.com. Ralf From owner-linux@cthulhu.engr.sgi.com Tue Sep 30 16:38:42 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA885124 for ; Tue, 30 Sep 1997 16:38:42 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA08871 for linux-list; Tue, 30 Sep 1997 16:37:47 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA08859 for ; Tue, 30 Sep 1997 16:37:42 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id QAA00538 for ; Tue, 30 Sep 1997 16:36:23 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id QAA22417 for linux@engr.sgi.com; Tue, 30 Sep 1997 16:36:24 -0700 From: Ralf Baechle Message-Id: <199709302336.QAA22417@dns.cobaltmicro.com> Subject: IRIX ELF docs To: linux@cthulhu.engr.sgi.com Date: Tue, 30 Sep 1997 16:36:24 -0700 (PDT) Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 455 Lines: 11 Hi all, our current linker is producing IRIX flavored ELF binaries, not MIPS ABI. We still seem to have some bugs in the dynamic linker and these are now pretty close to the top on my to do list. However I've got not documentation about the IRIX binary format, so I'm pretty much relying on reverse engineering for fixing them. Does anybody have a pointer to documentation or documentation about IRIX ELF flavoured o32 bit object file format? Ralf From owner-linux@cthulhu.engr.sgi.com Tue Sep 30 16:42:52 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id QAA887770 for ; Tue, 30 Sep 1997 16:42:52 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id QAA09624 for linux-list; Tue, 30 Sep 1997 16:42:40 -0700 Received: from fir.engr.sgi.com (fir.engr.sgi.com [150.166.49.183]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id QAA09620; Tue, 30 Sep 1997 16:42:38 -0700 Received: (from wje@localhost) by fir.engr.sgi.com (950413.SGI.8.6.12/950213.SGI.AUTOCF) id QAA14355; Tue, 30 Sep 1997 16:42:37 -0700 Date: Tue, 30 Sep 1997 16:42:37 -0700 Message-Id: <199709302342.QAA14355@fir.engr.sgi.com> From: "William J. Earl" To: Ralf Baechle Cc: linux@cthulhu.engr.sgi.com Subject: Re: IRIX ELF docs In-Reply-To: <199709302336.QAA22417@dns.cobaltmicro.com> References: <199709302336.QAA22417@dns.cobaltmicro.com> Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 759 Lines: 15 Ralf Baechle writes: > Hi all, > > our current linker is producing IRIX flavored ELF binaries, not MIPS > ABI. We still seem to have some bugs in the dynamic linker and these > are now pretty close to the top on my to do list. However I've got > not documentation about the IRIX binary format, so I'm pretty much > relying on reverse engineering for fixing them. Does anybody have > a pointer to documentation or documentation about IRIX ELF flavoured > o32 bit object file format? IRIX ELF O32 (dynamic) object files are MIPS ABI object files. There are optional extra sections, to support features such as "quickstart" (which allows RLD to skip some of the fixups at startup time), but the required parts are as defined by the MIPS ABI. From owner-linux@cthulhu.engr.sgi.com Tue Sep 30 18:07:11 1997 Received: from cthulhu.engr.sgi.com (cthulhu.engr.sgi.com [192.26.80.2]) by neteng.engr.sgi.com (970321.SGI.8.8.5/960327.SGI.AUTOCF) via SMTP id SAA911247 for ; Tue, 30 Sep 1997 18:07:11 -0700 (PDT) Return-Path: Received: (from majordomo-owner@localhost) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) id SAA23852 for linux-list; Tue, 30 Sep 1997 18:05:09 -0700 Received: from sgi.sgi.com (sgi.engr.sgi.com [192.26.80.37]) by cthulhu.engr.sgi.com (950413.SGI.8.6.12/960327.SGI.AUTOCF) via ESMTP id SAA23785 for ; Tue, 30 Sep 1997 18:05:00 -0700 Received: from dns.cobaltmicro.com ([209.19.61.1]) by sgi.sgi.com (950413.SGI.8.6.12/970507) via ESMTP id SAA22581 for ; Tue, 30 Sep 1997 18:03:29 -0700 env-from (ralf@dns.cobaltmicro.com) Received: (from ralf@localhost) by dns.cobaltmicro.com (8.8.5/8.8.5) id SAA22736; Tue, 30 Sep 1997 18:02:51 -0700 From: Ralf Baechle Message-Id: <199710010102.SAA22736@dns.cobaltmicro.com> Subject: For stability freaks ... To: linux-mips@fnet.fr, linux@cthulhu.engr.sgi.com Date: Tue, 30 Sep 1997 18:02:51 -3100 (PDT) Content-Type: text Sender: owner-linux@cthulhu.engr.sgi.com Precedence: bulk Content-Length: 578 Lines: 17 Ok, I said I'd touch 2.0.x for MIPS again when hell freezes. It was damn cold weather the last days and hell has froozen ... [root@(none) /]# uname -a Linux (none) 2.0.30 #389 Tue Sep 30 17:47:39 PDT 1997 mips unknown [root@(none) /]# cat /proc/cpuinfo cpu : MIPS cpu model : Nevada V1.0 system type : Cray YMP [just kidding ...] BogoMIPS : 131.89 byteorder : little endian unaligned accesses : 0 wait instruction : yes microsecond timers : yes extra interrupt vector : yes Ralf