Differences between revisions 43 and 44
Revision 43 as of 2006-06-25 07:15:17
Size: 4672
Editor: JurijSmakov
Comment:
Revision 44 as of 2006-07-03 00:09:48
Size: 4840
Editor: JurijSmakov
Comment:
Deletions are marked like this. Additions are marked like this.
Line 9: Line 9:
 * Investigate the feasibility of backporting stuff from 2.6.17 to 2.6.16 for sparc64. This is becoming more pressing, since a) it is possible that 2.6.17 will not come out in time for freeze and stuff; and b) 2.6.17-rc seem to be doing much better w/respect SMP stability (see below).  * Investigate the feasibility of backporting stuff from 2.6.17 to 2.6.16 for sparc64. This is becoming more pressing, since a) it is possible that 2.6.17 will not come out in time for freeze and stuff; and b) 2.6.17-rc seem to be doing much better w/respect SMP stability (see below). We now have 2.6.17 in unstable, but it's not yet clear whether it will make it to testing.
Line 15: Line 15:
 * Update kernel-handbook, decide what to do about the missing chapters, and finally package it. That will probably take a while.  * Update kernel-handbook, decide what to do about the missing chapters, and finally package it. That will probably take a while. It's currently postponed since the kernel packaging still haven't stabilized.

Hi, I am Debian Developer, involved in the following areas:

Current "To Do" list:

  • Investigate the feasibility of backporting stuff from 2.6.17 to 2.6.16 for sparc64. This is becoming more pressing, since a) it is possible that 2.6.17 will not come out in time for freeze and stuff; and b) 2.6.17-rc seem to be doing much better w/respect SMP stability (see below). We now have 2.6.17 in unstable, but it's not yet clear whether it will make it to testing.
  • Add drivers for T2000/T1000 to installer (mptfusion stuff) and for Blade 2500. Actually, the stuff for Blade2500 is already there (alim15x3 for IDE, sym53c8xx and scsi_transport_spi for SCSI) are already included, so something else is clearly wrong. Asked for more info. The T1000/T2000 are not supported by our kernels currently anyway, so it's not too urgent.
  • Try to build a custom kernel which would produce more debugging info for the infamous sparc64 SMP crash ([http://bugs.debian.org/361214 bug 361214]). Should try building 2.6.17 from git, rumour is that a lot of sparc SMP changes went into it, so there is hope. Built one, asked Blars to test. Still crashes for Blars, but buildds are reportedly stable with 2.6.17-rc1 (woohoo!).

  • Investigate the serial console corruption on Ultra5 with 2.6, reported by Joey Hess. All symbols come out as a bunch of X'es. Hm, not broken for me. Latest 2.6.16 works fine modulo the usual occasional garbage on the console (I thought it was fixed long ago :-/). As Joey reports that it works with lower speed, I'm putting it on a back burner for now.
    • Re this, the machine was running at 115200 baud, and I switched it down to 9600, and it's now ok. I don't know why the 2.4 kernel worked at the high speed and 2.6 didn't, but there we are. --JoeyHess

  • Update kernel-handbook, decide what to do about the missing chapters, and finally package it. That will probably take a while. It's currently postponed since the kernel packaging still haven't stabilized.

Done

  • DMA errors on sparc32, upstream bug [http://bugzilla.kernel.org/show_bug.cgi?id=6344 6344]. Sent a message to d-s asking to test a presumably fixed kernel. [http://lists.debian.org/debian-sparc/2006/04/msg00045.html Thread]. Looks like my initial theory did not pan out. Tried booting 2.6.17-rc3, but did not get too far with it. Martin Habets is looking at it. Fix by Bob Breuer is included in 2.6.16-15 and will be in 2.6.17-2.

  • Update and upload linux-kernel-di-sparc-2.6 to the latest kernel. Done by Frans Pop (thanks!)
  • Check why SILO built with newer compiler hangs on sparc32. Would be really great if we could make it work, because SILO is currently the only package in the archive which build-depends on gcc-2.95 :-P Found the [http://lists.debian.org/debian-sparc/2006/06/msg00010.html problem], fix included in the SILO git repo, should appear in upstream 1.4.13 soon. 1.4.13 not appeared yet, so I've uploaded a fixed version 1.4.12-1.

  • Investigate recent firefox problems on sparc ([http://bugs.debian.org/362170 bug 362170]). Got a good backtrace, sent to a bug. Dies due to unaligned access, [https://bugzilla.mozilla.org/show_bug.cgi?id=161826 upstream bug] for this problem exists. Produced a seemingly working patch, posted to the upstream bug for review. Patch has been incorporated into Debian's firefox 1.5.dfsg+1.5.0.4-1, will probably get included upstream sooner or later.

  • Fix the base-installer patch in [http://bugs.debian.org/358532 bug 358532]. Submitted a [http://lists.debian.org/debian-boot/2006/04/msg00520.html part of the solution] to d-b for review. Fixed by ?FransPop (thanks!)

  • Gnustep problems on sparc, bug [http://bugs.debian.org/362192 bug 362192]. Identified the source of the problem, sent followup to the bug.

  • Patches from Fabio for [http://people.ubuntu.com/~fabbione/sparc-fix-raid.dpatch RAID support] and [http://people.ubuntu.com/~fabbione/sparc-no-chs-in-udeb.dpatch libparted]. Former is submitted as a wishlist bug against parted, the latter is more problematic. It's more a quick and dirty solution than a real fix. The d-i's parted server should really be taught how to handle libparted's exceptions. ?FransPop is aware of patches, he should be able to take better care of them :-).

  • Finish the draft of the out-of-tree kernel module policy, produce a sample future-policy-compliant package. Didn't complete, too much sparc stuff is currently going on. Hopefully someone will pick up the ball.