You are not logged in.
Romashka wrote:Yes, it can be triggered in different ways, there's even test program for it.
link?
Sadly, I don't remember where I read that, I forgot to post link at that time here.
I've found the same bugreport at linux-nfs.org: http://bugzilla.linux-nfs.org/show_bug.cgi?id=130 (patch),
and a message thread with this patch included: http://linux-nfs.org/pipermail/nfsv4/20 … 05460.html - I don't see here if Linus pulled all those changes, it seems it's included in 2.6.20-rc1 only.
Also, see here: http://lkml.org/lkml/2005/12/17/71
and "[libata] use kmap_atomic(KM_IRQ0) in SCSI simulator" in 2.6.20-rc2 changelog.
Googling on this issue has shown many kmap_atomic()-related issues that result in "kernel BUG in .../highmem.c".
I don't know if our bug is exactly the same as http://lkml.org/lkml/2006/12/5/167
Last edited by Romashka (2007-01-27 12:52:54)
to live is to die
Offline
http://lkml.org/lkml/2006/12/26/91 - xfs/udev issue? I don't think so.
http://linux-nfs.org/pipermail/nfsv4/20 … 05289.html - same thing.
This is definetely not boot-only issue.
http://permalink.gmane.org/gmane.linux.kernel/103417 - is same fix needed for 2.6.19?
Hmm... the same issue was in 2.6.9: http://sourceware.org/ml/systemtap/2006 … 00485.html
http://lkml.org/lkml/2006/12/12/155 - what did he mean by saying:
With mem=896M highmem-enabled kernels work OK.
?
He set mem=896M in kernel boot option or what?
maybe this is mistake and he is talking about 880MB limit memory limit on x86?
Offline
Romashka wrote:http://lkml.org/lkml/2006/12/12/155 - what did he mean by saying:
With mem=896M highmem-enabled kernels work OK.
?
He set mem=896M in kernel boot option or what?maybe this is mistake and he is talking about 880MB limit memory limit on x86?
What is that 880M limit? I never heard about it.
to live is to die
Offline
What is that 880M limit? I never heard about it.
http://kerneltrap.org/node/2450
This describes it a bit more- it has to do with kernel high memory.
Offline