Andrew Morton announced Linux 2.6.11-mm3, saying:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.11/2.6.11-mm3/
- A new version of the "acpi poweroff fix". People who were having trouble with ACPI poweroff, please test and report.
A very large update to the CFQ I/O scheduler. Treat with caution, run benchmarks. Remember that the I/O scheduler can be selected on a per-disk basis with
echo as > /sys/block/sda/queue/scheduler
echo deadline > /sys/block/sda/queue/scheduler
echo cfq > /sys/block/sda/queue/scheduler- video-for-linux update
Helge Hafting reported:
2.6.11-mm1 and earlier: mouse appear as /dev/input/mouse0
2.6.11-mm3: mouse appear as /dev/input/mouse1No big problem, one change to xorg.conf and I got the mouse back. I guess it wasn't supposed to change like that though?
This is a mouse connected to the ps2 port, also appearing as /dev/psaux
Dmitry Torokhov recommended, I'd recommend
using /dev/input/mice unless you want to _exclude_ some of your input
devices. It will get data from all you mice at once and is always
available.
He explained, Vojtech
activated scroll handling in keyboard code by default so now your keyboard is
mapped to the mouse0 and the mouse moved to mouse1.
Andrew replied,
We cannot ship a kernel with this change, surely?
Our users would come hunting for us with pitchforks.
But Vojtech
Pavlik replied:
Mouse device numbers are defined to be unstable because of hotplug.
Most users use /dev/input/mice, where this won't have impact.
The officially correct solution is to use udev to get stable device names.
The change is easily reverted - just change the 'atkbd.scroll' default value.
Marcelo Tosatti announced Linux 2.4.30-rc1, saying, Here goes the first release candidate for v2.4.30. It contains
a small number of fixes, including a fix for recently discovered ppp DoS
(CAN-2005-0384).
Andrew Morton announced Linux 2.6.12-rc1-mm1, saying:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.12-rc1/2.6.12-rc1-mm1/
- We might have a fix here for the recent AGP/DRM problems. If you were having problems with that, please test and report.
- An update to the hfs and hfsplus filesystems.
- Lots more pcmcia changes.
- Linus is away this week. Not a lot more should be going into 2.6.12 now and I have a list of ~140 bugs, many of which are post-2.6.10 regressions. We should fix these.
Russell King's ears pricked up at the mention of a bug list; he asked,
Is this your own personal bug list, or is it
accessible anywhere?
Andrew replied, It's
just an email folder at present.
... USB,
ALSA, Input, ACPI and suspend are the usual culprits.
Jan Engelhardt reported, I have observed that
squashfs, when mounted, does not return any "." or ".." pseudo-directories
upon readdir. Could this be added? Would there be any objections?
Jesper Juhl replied, I can't say if there will
be any objections or not, but if that's something that people want, then
I'd like to take a stab at implementing it - could be fun and I'd love to
learn a little more about that area of the kernel, so I'll have a go at it
if noone screams.
He dove into the code, and Pietro Zuco agreed that
this functionality would be useful.
The discussion spiraled around from here; at one point Andreas Schwab
remarked, "." and ".." are handled in the VFS.
No filesystem code ever sees them during lookup.
Elsewhere, Phil
Lougher mentioned:
Cramfs also doesn't store '.' and '..', which is where I got the idea from in the first place when originally implementing Squashfs.
Filesystems don't need to store '.' or ''..' in the filesystem, as they're never looked up by the VFS - as mentioned elsewhere in this thread, the VFS handles '.' and '..' internally.
Not storing the redundant '.' and '..' entries within the filesystem achieves a small but nonetheless useful space saving.
He added, The lack of '.' and '..' entries
hasn't caused any problems despite cramfs/squashfs being used for a large
number of years. I'm inclined to believe any application that _relies_
on seeing '.' and '..' returned by readdir is broken. This situation is
easily fixed within the application rather than forcing the filesystem to
unnecessarily fake '.' and '..' entries which are never used.
There was very little support for Phil's position however. H. Peter Anvin
said rhetorically at one point, Are you seriously
suggesting changing our behaviour of all the conventional filesystems to a
non-Unix behaviour, to match cramfs and squashfs?
Andrew Morton announced Linux 2.6.12-rc1-mm2, saying:
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.12-rc1/2.6.12-rc1-mm2/
- Added David Miller's networking tree to the -mm lineup as bk-net.patch.
Added Herbert Xu's crypto development tree to the -mm lineup as bk-cryptodev.patch.
-mm kernels now aggregate Linus's tree and 34 subsystem trees. Usually they are pulled 3-4 hours before the release of the -mm kernel.
Usually it is possible to determine the latest cset from each tree by looking at the first couple of lines of the relevant patch in the broken-out/ directory. Although sometimes it isn't there if I had to massage the diff.
- There may be an x86_64 problem here, although it works for me. If it fails early in boot, try reverting x86_64-separate-amd-cmp-detection-from-hyper-threading.patch
- There's some work here on the recent USB PM resume bugs. If you had problems there, please test and be sure to cc linux-usb-devel@lists.sourceforge.net in any reports.
- Some fixes for the recent DRM problems.
- Big DVB update
- md updates
- nfs4 server updates
- Lots more fixes
- Lots more bugs.
Regarding the aggregation of the various subsystem trees and Linus's tree into the -mm series, Lee Revell asked:
Do you notify the subsystem maintainers ahead of time so that critical fixes can be pushed to BK?
I am thinking of the recent ALSA example, where the emu10k1 driver was b0rked in 2.6.12-mm1, but the fix had been in ALSA CVS for a week.
Andrew replied, Occasionally I'll go out and ping
people, but almost always the subsystem guys know what the development cycle
is, and they appropriately decide which code should go in, and when.
He added, regarding ALSA, We've been discussing
how to get ALSA CVS into ALSA bk more promptly.
Marcelo Tosatti announced Linux 2.4.30-rc2, saying:
Here goes the second release candidate for v2.4.30.
It contains a bunch of security updates (ext2 mkdir leak, af_bluetooth range checking, isofs corrupt media, load_elf_library DoS), an ia64 update, another round of networking fixes, amongst others.
If nothing terrible shows up, this will become v2.4.30.
Chris Wright announced Linux 2.6.11.6, saying, With some pending security fixes it's time to for a -stable update.
So, here's 2.6.11.6, in the normal kernel.org places. This includes
some security fixes, esp. one which closes a local root exploit in
bluetooth.
Marcelo Tosatti announced Linux 2.4.30-rc3, saying, A nasty typo happened while merging v2.6 load_elf_library() DoS fix,
which could leap to oopses.
Marcelo Tosatti announced Linux 2.4.30-rc4, saying:
Here goes -rc4 to fix a couple of regressions have been confirmed:
- ext3 IO EH changes need more work
- Netfilter bogus mc_list deletion
Hopefully this will become final in a day or two.