Re: About that Alpha kernel bug mentioned in beta3 errata
tags 237884 pending quit On Mon, Apr 05, 2004 at 06:41:07PM +0200, Bjoern Brill wrote: Result: CD drive is properly detected, symlink in /dev/cdroms looks alright. (Should it matter: this is on a EV5 machine.) Thank you very much. The following patch from glibc should fix the problem. -- Debian GNU/Linux 3.0 is out! ( http://www.debian.org/ ) Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED] Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt Index: arch/alpha/lib/stxncpy.S === RCS file: /home/gondolin/herbert/src/CVS/debian/kernel-source-2.4/arch/alpha/lib/stxncpy.S,v retrieving revision 1.4 diff -u -r1.4 stxncpy.S --- a/arch/alpha/lib/stxncpy.S 3 Sep 2003 10:26:49 - 1.4 +++ b/arch/alpha/lib/stxncpy.S 9 Apr 2004 10:23:30 - @@ -313,9 +313,9 @@ lda t2, -1 # e0: for creating masks later beq t12, $u_head# .. e1 : - extql t2, a1, t2 # e0: + mskqh t2, t5, t2 # e0: begin src validity mask cmpbge zero, t1, t8# .. e1 : is there a zero? - andnot t2, t6, t12 # e0: dest mask for a single word copy + extql t2, a1, t2 # e0: or t8, t10, t5 # .. e1 : test for end-of-count too cmpbge zero, t2, t3# e0: cmoveq a2, t5, t8 # .. e1 : @@ -330,14 +330,14 @@ ldq_u t0, 0(a0) # e0: negqt8, t6 # .. e1 : build bitmask of bytes = zero mskqh t1, t4, t1 # e0: - and t6, t8, t2 # .. e1 : - subqt2, 1, t6 # e0: - or t6, t2, t8 # e1: + and t6, t8, t12 # .. e1 : + subqt12, 1, t6 # e0: + or t6, t12, t8 # e1: - zapnot t12, t8, t12# e0: prepare source word; mirror changes + zapnot t2, t8, t2 # e0: prepare source word; mirror changes zapnot t1, t8, t1 # .. e1 : to source validity mask - andnot t0, t12, t0 # e0: zero place for source to reside + andnot t0, t2, t0 # e0: zero place for source to reside or t0, t1, t0 # e1: and put it there stq_u t0, 0(a0) # e0: ret (t9)# .. e1 :
Re: About that Alpha kernel bug mentioned in beta3 errata
On Sun, 4 Apr 2004, Herbert Xu wrote: Bjoern Brill [EMAIL PROTECTED] wrote: The symlink gets created in drivers/cdrom/cdrom.c:register_cdrom(): if (pos = 0) { char vname[16]; sprintf (vname, cdrom%d, cdi-number); strncpy (rname + pos, ../, 3); devfs_mk_symlink (devfs_handle, vname, DEVFS_FL_DEFAULT, rname + pos, slave, NULL); devfs_auto_unregister (cdi-de, slave); } That piece of code is shared by all kinds of cdroms, and all architectures... so how can it be broken on alpha and not on x86, and with SCSI but not IDE? I think the strncpy (rname + pos, ../, 3); call fails, and strncpy is arch specific. This would mean that, in the end, alpha stxncpy.S is broken again. I don't know enough alpha assembly Good catch. Can you either build a new kernel with stxncpy.S disabled, or try the image from http://gondor.apana.org.au/~herbert/debian/kernel/binary/2.4.25/alpha-generic/ and see if the problem goes away? If it does, then I'll turn stxncpy off until it's fixed. Bingo. I've put your test kernel and the initrd.gz from the same installer CD as before (sarge-alpha-businesscard.iso, Mar 31 daily build) somewhere in my /boot partition and pointed aboot to them, together with the installer's default boot arguments (ramdisk_size=16384 root=/dev/ram devfs=mount,dall), while having the installer CD in the drive. This should ensure nothing has changed except replacing the 2.4.25-1-generic kernel with your test kernel. Result: CD drive is properly detected, symlink in /dev/cdroms looks alright. (Should it matter: this is on a EV5 machine.) Regards, Bjoern -- Bjorn Brill [EMAIL PROTECTED] Frankfurt am Main, Germany -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: About that Alpha kernel bug mentioned in beta3 errata
Bjoern Brill [EMAIL PROTECTED] wrote: The symlink gets created in drivers/cdrom/cdrom.c:register_cdrom(): if (pos = 0) { char vname[16]; sprintf (vname, cdrom%d, cdi-number); strncpy (rname + pos, ../, 3); devfs_mk_symlink (devfs_handle, vname, DEVFS_FL_DEFAULT, rname + pos, slave, NULL); devfs_auto_unregister (cdi-de, slave); } That piece of code is shared by all kinds of cdroms, and all architectures... so how can it be broken on alpha and not on x86, and with SCSI but not IDE? I think the strncpy (rname + pos, ../, 3); call fails, and strncpy is arch specific. This would mean that, in the end, alpha stxncpy.S is broken again. I don't know enough alpha assembly Good catch. Can you either build a new kernel with stxncpy.S disabled, or try the image from http://gondor.apana.org.au/~herbert/debian/kernel/binary/2.4.25/alpha-generic/ and see if the problem goes away? If it does, then I'll turn stxncpy off until it's fixed. Cheers, -- Debian GNU/Linux 3.0 is out! ( http://www.debian.org/ ) Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED] Home Page: http://gondor.apana.org.au/~herbert/ PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: About that Alpha kernel bug mentioned in beta3 errata
On Wed, 31 Mar 2004, Steve Langasek wrote: Hi Bjoern, On Tue, Mar 30, 2004 at 12:44:48AM +0200, Bjoern Brill wrote: if that bug (alpha cannot install from SCSI cdrom) is the one mentioned in http://lists.debian.org/debian-boot/2004/debian-boot-200401/msg00374.html, then I've run into it as well with a daily build shortly before d-i beta3. Is this fixed now? If not here is some additional info: [...] Yes, this is the same bug. The beta3 on alpha released with a 2.4.24 kernel. Could I ask you to test out a more recent daily snapshot, to confirm for me whether this bug still exists in the 2.4.25-based images? Since I have no SCSI cdrom's at hand, I can't exactly test for this, though I'm happy to work on debugging the code if that's where the problem lies. I'm just trying the latest daily build (dated Mar 31 23:03) of sarge-alpha-businesscard.iso: ~ # uname -a Linux kruemel 2.4.25-1-generic #1 Tue Feb 24 08:10:58 EST 2004 alpha unknown The bug is still there. I can work around it as previously described. (Not that this would buy me much, installation did and still does fail later on due to broken packages and other stuff. This time we fail to install libc6.1_2.3.2.ds1-11 because it tries to overwrite /etc/default/devpts which is also in package initscripts. Last time the netinst iso had no network driver modules. Before that aboot was uninstallable because aboot-base was missing. Isn't there a way I can _declare_ a step has completed successfully and clean up the breakage myself? Hrrm well.) If it turns out to be a gcc optimization problem, I'm afraid I'm out of my depth where debugging kernels is concerned. So am I. But I think compiler problems are usually cared for not that different from bugs: since one can't do much about the actual problem unless the compiler gets fixed (which can take quite some time), one works around it by slightly restating or reordering the code and hoping a different optimization case kicks in and the error goes away. Regards, Bjoern -- Bjorn Brill [EMAIL PROTECTED] Frankfurt am Main, Germany -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: About that Alpha kernel bug mentioned in beta3 errata
Hi Bjoern, On Tue, Mar 30, 2004 at 12:44:48AM +0200, Bjoern Brill wrote: if that bug (alpha cannot install from SCSI cdrom) is the one mentioned in http://lists.debian.org/debian-boot/2004/debian-boot-200401/msg00374.html, then I've run into it as well with a daily build shortly before d-i beta3. Is this fixed now? If not here is some additional info: On my system (an AS 2100A Lynx btw, cool these are supported now): # cd /dev/cdroms # ls -l lr-xr-xr-x 1 root root 34 Jan 1 1970 cdrom0 - yyyscsi/host0/bus0/target6/lun0/cd [actually, the 'y's are y with dieresis, 0xff in ISO-8859-1] Yes, this is the same bug. The beta3 on alpha released with a 2.4.24 kernel. Could I ask you to test out a more recent daily snapshot, to confirm for me whether this bug still exists in the 2.4.25-based images? Since I have no SCSI cdrom's at hand, I can't exactly test for this, though I'm happy to work on debugging the code if that's where the problem lies. If it turns out to be a gcc optimization problem, I'm afraid I'm out of my depth where debugging kernels is concerned. Thanks, -- Steve Langasek postmodern programmer signature.asc Description: Digital signature