With 2.6.25-rc6-git-something I have experienced stuck keys. When typing
a sentence a character repeats aboooooooout 10 times, or as 'about'
above ;). I tested with a new kernel today, 2.6.25-rc7-git-something and
the problem is still present.With 2.6.25-rc3 the keyboard is fine, and I've not tried with any kernel
in between.As a side note, timestamps gets mixed up in dmesg output in a 1-2 second
span. Maybe related?This is on a HP dv2140eu laptop, x86_64 SMP. Please cc me if you want
configuration details or testing to be done.regards, Richard
--
I should add that while watching a movie I noticed that video stutters
every now and then, on an irregular basis. It lasts for approx a second
each time.
--
Update
I've been trying to reproduce this bug all morning, and could not
reproduce it without firefox running. I then saw with dmesg that the
flash plugin had crashed:[ 2170.872049] npviewer.bin[9053]: segfault at 0 ip f6dc9348 sp ffd3e788
error 4 in libflashplayer.so[f6d2c000+66b000]This would explain why it can take several days between me experiencing
the bug, but it doesn't explain the mixed up timestamps in kern.logEven if flash/npviewer is at fault, IMHO it shouldn't cause microlockups
that I experienceregards, Richard
--
Hi Richard,
what does your .config look like? Do you have CONFIG_GCGROUP_SCHED
enabled? (please note that this is different from CONFIG_GROUP_SCHED). If
so, please try to reproduce without it.--
Jiri Kosina
SUSE Labs--
CGROUP_SCHED depends on GROUP_SCHED. The stutters are not due to the
group scheduling extensions. (or the scheduler AFAICT).Thanks,
--
regards,
Dhaval
--
There is no such thing in my config at all, at least not with
CONFIG_GROUP_SCHED disabled. If you meant CONFIG_CGROUP_SCHED it was not
enabled at any time. Config attachedregards, Richard
Can you check if you have CONFIG_GROUP_SCHED enabled? If so, can you
disable and test?Thanks,
--
regards,
Dhaval
--
Yes it's enabled, will try without groupsched.
regards, Richard
--
Same behavior without group_sched
regards, Richard
--
Is it easy to reproduce? Would it be possible for you to bisect?
Thanks,
--
regards,
Dhaval
--
I have a pretty good idea how to trigger it now, but it still takes some
time of typing to actually see it. It's easy to see if it fails, but not
the other way around.I will give it a try to at least narrow it down, but it will take some
time..regards, Richard
--
This doesn't sound good at all.
Ingo, Thomas, are there any obvious candidates in the x86 and/or scheduler area
that might cause it to happen?Rafael
--
Does the weirdness persist if you boot with "io_delay=0x80" as a kernel
parameter? (it should if -rc3 really was fine, but...)Rene.
--
I'm testing that param at the moment, and while writing this response it
happened again, meaning it didn't change anything..I can't say for sure rc3 wasn't affected, I just never noticed the
behavior during the 2-3 weeks I used it.regards, Richard
--
Haven't been following the list lately, but I believe I read something about
a scheduler / X.org interaction problem. Afraid that's all from me though...Rene.
--
| Greg Kroah-Hartman | [PATCH 004/196] Chinese: add translation of SubmittingPatches |
| Tarkan Erimer | Re: Dual-Licensing Linux Kernel with GPL V2 and GPL V3 |
| Matt Mackall | Re: [PATCH] x86: fix unconditional arch/x86/kernel/pcspeaker.c compiling |
| James Bottomley | Re: Integration of SCST in the mainstream Linux kernel |
| Gerrit Renker | [PATCH 27/37] dccp: Integration of dynamic feature activation - part 2 (server side) |
| David Miller | [GIT]: Networking |
| Jarek Poplawski | [PATCH] pkt_sched: Destroy gen estimators under rtnl_lock(). |
| Natalie Protasevich | [BUG] New Kernel Bugs |
git: | |
