On Fri, 26 Mar 2010 23:07:27 +0900
Tejun Heo <tj(a)kernel.org> wrote:
> 2.6.34-rcX has a random memory corruption bug which is showing up as
> various boot failures. Yinghai has a patch.
>
> http://thread.gmane.org/gmane.linux.kernel/963616/focus=964914
Is this patch in 2.6.34-rc2.git3 already? Because it still crashes
early during boot for me.
Thanks,
seife
--
Stefan Seyfried
"Any ideas, John?"
"Well, surrounding them's out."
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
On 31.3.2010 16:43, Jan Engelhardt wrote:
>
> On Wednesday 2010-03-31 16:33, Michal Marek wrote:
>>
>> Hi Jan,
>>
>> I merged the branch (commit 54e2592), but I reverted the /usr/src/linux
>> -> /usr/src/linux-src move.
>
> Hm I don't have 54e2592 in my tree at all. You seemed to have merged
> something very outdated. I will forward you the reply I sent last to the
> list again for you to look at.
54e2592 is my commit merging it, the topmost commit of your branch was
113d492 (Produce kernel-devel%variant), sorry for not being clear. It
will appear on gitorious soon.
>> I agree that it would make it harder for
>> users to shoot themselves in the foot, but the costs are
>> 1) bugreports / complaints from users who can't find their kernel source
>> 2) duplication both on the media and in the installed system
>> 3) some packages rebuild files from /usr/src/linux/, OK, these can be
>> easily fixed.
>>
>> 1) is what worries me most, but I guess coolo would not be all that
>> happy about an additional 7MB package on the DVD.
>
> You could drop kernel-source-vanilla... or switch to xzdio.9 payload :-)
> [my .src.rpm is almost 10MB smaller than those in kotd]
kernel-source-vanilla is not on the DVD I think, but both kernel-source
and kernel-devel are important enough to put there. Thanks for the hint
about xzdio.9, what is the impact of this on build and install time?
Michal
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
Greetings.
The following is a submission to reduce the number of files that will
be installed when one wants to compile kernel modules, offering
saving both in time and space. Disks are touted as cheap, but let's
face it, spending ~340 MB for kernel sources means you have 340 less
for your personal files. The time savings should be obvious too: less
files to install, and also less files to index — locatedb, quota,
beagle/etc., you name it. This is especially beneficial for
developers and OBS or OBS-like instances where packages are not only
built in random order(*), but where the chroots are usually torn down
everytime.
(*) Calling build locally and manually, without --clean, and with the
same buildroot everytime would allow for a time optimization when
all KMPs are built after another, as big packages like kernel-* would
not have to be reinstalled/deleted everytime.
With this patch(set) — there is a bit of cleanup upfront —, the
"kernel-devel" package is introduced which contains all (and only)
the necessary files to support external module building. The entire
source for personal recompilation orgies is moved out to
"kernel-source", which now behaves like "kernel-source-vanilla"
already did.
The standalone sources' (kernel-source) path has been changed so as
to not overlap with those used for module building, because people
always trash their KMP build environment for the shipped kernel when
they ran cloneconfig.
Another desired side-effect of the so-codified separation into
the kernel-devel package is that dangerous calls that would cause
above-mentioned trashing are now completely inhibited.
This modification has been tested with various kinds of free and
non-free packages and kernel versions already. The latter is
especially important because they are the most likely ones to
break because of their awkward-to-horrid Makefiles. These have been
made available for a long time already at jftp.medozas.de; the list
includes at least:
fglrx-kmp-default-8.681~2.6.31.11~jen93-jen0.x86_64.rpm
ndiswrapper-kmp-default-1.55~2.6.31.11~jen93-3.4.jen2.x86_64.rpm
nvidiagfx-kmp-default-190.53~2.6.31.11~jen93-jen8.x86_64.rpm
nvidiagfx-legacy173-kmp-default-173.14.22~2.6.31.11~jen93-jen8.x86_64.rpm
nvidiagfx-legacy71-kmp-default-71.86.11~2.6.31.11~jen93-jen8.x86_64.rpm
nvidiagfx-legacy96-kmp-default-96.43.14~2.6.31.11~jen93-jen8.x86_64.rpm
ttyrpld-kmp-default-2.60~2.6.31.11~jen93-jen4.x86_64.rpm
virtualbox-modules-kmp-default-3.1.2_56127~2.6.31.12~jen93-jen0.x86_64.rpm
vmware-modules-kmp-default-6.5.3~185404~2.6.31.12~jen93-jen3.x86_64.rpm
xtables-addons-kmp-default-1.22~2.6.31.12~jen93-jen0.x86_64.rpm
It was also tested with 2.6.32 and — since somebody decided out of
the blue to rebase last week :-) — just now with .33-rc6 too.
The resulting kernel-devel package is 10x smaller in installed size
and has roughly 3x fewer files.
The following changes since commit 69aad1a30e31d4f33ab56a1f195c7c4229c9dd66:
Nikanth Karthikesan (1):
- patches.fixes/dm-stripe-zero-stripes: dm-stripe: return -EINVAL if stripe count is zero (bnc#576312).
are available in the git repository at:
git://dev.medozas.de/suse-kernel less-source
Jan Engelhardt (3):
misc: use proper tar options
kernel-source: use more macros
kernel-source: split devel files and full-source
doc/README.SUSE | 4 +-
kernel-source.changes | 7 +++++
rpm/compute-PATCHVERSION.sh | 2 +-
rpm/group-source-files | 33 +++++++++++++++++++++++++
rpm/kernel-binary.spec.in | 12 ++++----
rpm/kernel-source.spec.in | 56 +++++++++++++++++++++++++++++++++++++++----
rpm/kernel-syms.spec.in | 4 +-
scripts/tar-up.sh | 2 +-
8 files changed, 103 insertions(+), 17 deletions(-)
create mode 100755 rpm/group-source-files
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
The iwlagn driver in kernel 2.6.33 used in 11.3 M4 is affected by a bug that
results in network disruptions caused by a timeout in a firmware watchdog. This
is reported in the kernel bugzilla
(https://bugzilla.kernel.org/show_bug.cgi?id=15374) where patches are posted.
The changes do not eliminate the firmware errors, but the fixup is a lot less
disruptive to the network connection.
This bug is reported in https://bugzilla.novell.com/show_bug.cgi?id=592399.
The above patches are already in kernel 2.6.34.
Larry
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hi all -
2.6.34-rc2 was tagged over the weekend so I completed the integration
with the master branch. KOTD and Kernel:HEAD kernels will be building
shortly and once I've gotten a successful round of builds I'll submit to
Factory.
Since it's an -rc2 release there will probably be some hiccups, and we'd
love to get feedback on them. If you run into an issue it would be
awesome if you could also confirm whether the issue is in the -vanilla
kernel so we know whether to file upstream bugs or not.
Thanks and happy testing!
- -Jeff
- --
Jeff Mahoney
SUSE Labs
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)
Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org/
iEYEARECAAYFAkunn8IACgkQLPWxlyuTD7J1pgCcCbp/SaGGFJygNfd+f5C+sEI1
UkQAn3heWkXC7KNUG+ib1Lfl/gRSZ3z+
=hf4s
-----END PGP SIGNATURE-----
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
Hi,
I'm desperately trying to get a pass through pci device to a 11.2 kvm
instance going on a openSUSE 11.1/i586 host, which is a VT-d capable
machine: Xeon X3460, Asus P7F-E motherboard, Intel 3420 PCH chipset, 12 GB
Ram, with the relevant options enabled in BIOS of course...
According to:
http://www.linux-kvm.org/page/How_to_assign_devices_with_VT-d_in_KVM
I need to enable the IOMMU, but no matter, what I provide as kernel
parameter to iommu (or do not provide iommu at all), iommu keeps out of
sight:
~# dmesg | grep -e DMAR -e IOMMU
ACPI: DMAR BF78E0C0, 0090 (r1 AMI OEMDMAR 1 MSFT 97)
Tentatively running a 11.2 x86_64 installation has it:
~# grep -e DMAR -e IOMMU /boot-11.2-x86_64.msg
<4>[ 0.000000] ACPI: DMAR 00000000bf78e0c0 00090 (v01 AMI OEMDMAR
00000001 MSFT 00000097)
<6>[ 0.046487] DMAR:Host address width 36
<6>[ 0.046491] DMAR:DRHD base: 0x000000fed90000 flags: 0x1
<6>[ 0.046499] IOMMU fed90000: ver 1:0 cap c90780106f0462 ecap f020e3
<6>[ 0.046502] DMAR:RMRR base: 0x000000000e4000 end: 0x000000000e7fff
<6>[ 0.046506] DMAR:RMRR base: 0x000000bf7ec000 end: 0x000000bf7fffff
Probably, the 11.1/x86_64 kernel has it, too.
Do I have a chance to get this going with 11.1/i586 (e.g. with only a
modified kernel)? Or is it possible to run an x86_64 kernel with a i586
userspace? Unfortunately, for some reason I have to avoid upgrading this
system to 11.2 or x86_64 by all means.
TIA,
Pete
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
As part of my testing of the 11.3 Mx releases, I attempt to connect from my
laptop to my AP using all my devices. In 11.3 M4, it worked with drivers rtl8187
and b43; however, my Dell DW1450, which uses p54usb, could not connect. It could
scan, but it was clearly only doing a passive scan as only channel 1 was seen.
My conclusion is that it cannot transmit.
In further testing, that same kernel RPM also failed on 11.2; whereas my
home-built 2.6.34-rc2 kernel works. To further complicate matters, after I
downloaded the openSUSE version of 2.6.33-6.3 source, and built it on 11.2, it
worked. I then suspected a difference between gcc 4.4 and 4.5; however, I built
that version on 11.3 M4 using gcc 4.5.0 and it worked.
AFAICT, the only bad version is the one in the Factory repo. Both x64_64 and
i386 architectures have been tested. I have no idea what could have happened,
but I will certainly be watching to see if the issue is cleared.
Larry
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
Status update:
I installed recent kernel-vanilla yesterday first time and it has runned without any problems for over 24 hours now.
Only openSUSE patched kernel flavors have problems ...
Johannes
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org
Hi,
I installed Kernel 2.6.34-rc2 on three systems (all 11.2 with Kernel:HEAD repo). Two of them do not have any problems - they seem even a bit faster compared to 2.6.33.
But on one system (MSI K7T Turbo [MS-6330] with a AMD Duron 900) the kernel hangs at an early stage (all flavors). It is so early that I cannot provide any logs and only the reset button works then.
Can you please tell me what I have to do for providing useful information?
Btw. I am not really familiar with kexec and kdump ...
Thanks.
Johannes
--
To unsubscribe, e-mail: opensuse-kernel+unsubscribe(a)opensuse.org
For additional commands, e-mail: opensuse-kernel+help(a)opensuse.org