Messages in this thread | | | Date | Fri, 20 Jul 2012 22:52:40 +0200 | Subject | Re: oops in kernel ( 3.4.x -> 3.5rc ) | From | nicolas prochazka <> |
| |
Hello the problem is occured with : - linux kernel 3.4.5 i do not test with 3.4.0 / 1 / 2 / 3 / 4 , but i can if you want - linux kernel 3.5rc6 rc7 / do not test with other rc.
the problem is not occured with : linux kernel 3.3.4 / 3.3.8
These servers are used for : - starting a lot of virtual machine with qemu-kvm ( ~ 40 ) ( lot of select i think) - do a lot of network tests with openvswitch
I can test a kernel 3.4.x before and after a commit id (?) to find a regression.
Regards, Nicolas.
2012/7/20 Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>: > On Fri, Jul 20, 2012 at 09:21:53AM -0400, Dave Jones wrote: >> On Fri, Jul 20, 2012 at 11:56:06AM +0200, nicolas prochazka wrote: >> >> > [ 2384.900061] BUG: unable to handle kernel paging request at 000000010000002f >> >> That '1' looks like a random bit flip. Try running memtest86. >> > > Looks more a 32-bit value of 1 followed by a 32-bit value of 0x2f. Most > likely a pointer to some other piece of a struct. However, taking a look > at fs/files.c code, nothing seems suspicious. > > Nicolas, it wasn't clear to me if you had problems with 3.4 too. There > has been some changes in fs/files.c on 3.4-rc1 in the piece of code > where you hit the problem. > > What does your system exercise? Any chance you are using a lot of > select, which has also been changed in those same patches to fs/files.c? > > Regards. > Cascardo. > > >> > [ 2384.910010] Pid: 23838, comm: queue.sh Tainted: G D W >> >> This wasn't the first problem either. >> >> > [ 2397.885344] BUG: unable to handle kernel paging request at 000000010000003b >> >> Looks like the same flipped bit. >> >> Dave >> >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in >> the body of a message to majordomo@vger.kernel.org >> More majordomo info at http://vger.kernel.org/majordomo-info.html >> Please read the FAQ at http://www.tux.org/lkml/ >> >
| |