Wrong DIF guard tag on ext2 write
in [Kernel]
|
Prev: KVM, Fix QEMU-KVM is killed by guest SRAO MCE (resend)
Next: drivers/watchdog: Eliminate a NULL pointer dereference
From: Christof Schmitt on 31 May 2010 07:30 When running tests with an ext2 filesystem on a device using DIF/DIX integrity data, i sometimes see invalid guard tags on write requests. To track down the problem, i patched the function sd_prep_fn in sd.c to verify the IP checksums in the guard tags against the actual data. Sometimes there is a mismatch and the write request fails when the HBA checks the guard tag. Since the guard tags are created in Linux, it seems that the data attached to the write request changes between the generation in bio_integrity_generate and the call to sd_prep_fn. Using ext3 or ext4 instead of ext2 does not show the problem. There is a bugzilla open at Redhat with the same symptom, but there is no data or activity: https://bugzilla.redhat.com/show_bug.cgi?id=574266 What would be the best way to track down this problem? -- Christof Schmitt -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Christof Schmitt on 31 May 2010 07:40 On Mon, May 31, 2010 at 01:28:17PM +0200, Christof Schmitt wrote: > When running tests with an ext2 filesystem on a device using DIF/DIX > integrity data, i sometimes see invalid guard tags on write requests. > To track down the problem, i patched the function sd_prep_fn in sd.c > to verify the IP checksums in the guard tags against the actual data. > Sometimes there is a mismatch and the write request fails when the HBA > checks the guard tag. > > Since the guard tags are created in Linux, it seems that the data > attached to the write request changes between the generation in > bio_integrity_generate and the call to sd_prep_fn. > > Using ext3 or ext4 instead of ext2 does not show the problem. > > There is a bugzilla open at Redhat with the same symptom, but there is > no data or activity: > https://bugzilla.redhat.com/show_bug.cgi?id=574266 > > What would be the best way to track down this problem? One more thing: The test is running with a 2.6.34 kernel, the problem in the bugzilla is reported for 2.6.33. Christof Schmitt -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Martin K. Petersen on 31 May 2010 10:30 >>>>> "Christof" == Christof Schmitt <christof.schmitt(a)de.ibm.com> writes: Christof> Since the guard tags are created in Linux, it seems that the Christof> data attached to the write request changes between the Christof> generation in bio_integrity_generate and the call to Christof> sd_prep_fn. Yep, known bug. Page writeback locking is messed up for buffer_head users. The extNfs folks volunteered to look into this a while back but I don't think they have found the time yet. Christof> Using ext3 or ext4 instead of ext2 does not show the problem. Last I looked there were still code paths in ext3 and ext4 that permitted pages to be changed during flight. I guess you've just been lucky. -- Martin K. Petersen Oracle Linux Engineering -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Christof Schmitt on 31 May 2010 10:50 On Mon, May 31, 2010 at 10:20:44AM -0400, Martin K. Petersen wrote: > >>>>> "Christof" == Christof Schmitt <christof.schmitt(a)de.ibm.com> writes: > > Christof> Since the guard tags are created in Linux, it seems that the > Christof> data attached to the write request changes between the > Christof> generation in bio_integrity_generate and the call to > Christof> sd_prep_fn. > > Yep, known bug. Page writeback locking is messed up for buffer_head > users. The extNfs folks volunteered to look into this a while back but > I don't think they have found the time yet. Thanks for the info. This means that this bug appears with all filesystems? > > > Christof> Using ext3 or ext4 instead of ext2 does not show the problem. > > Last I looked there were still code paths in ext3 and ext4 that > permitted pages to be changed during flight. I guess you've just been > lucky. ext3 looks good so far. I see the problem also with ext4, so i spoke too early on that one. I will start a longer testrun with ext3 to see if and when the problem appears with ext3 in my setup. -- Christof Schmitt -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
From: Nick Piggin on 31 May 2010 10:50
On Mon, May 31, 2010 at 10:20:44AM -0400, Martin K. Petersen wrote: > >>>>> "Christof" == Christof Schmitt <christof.schmitt(a)de.ibm.com> writes: > > Christof> Since the guard tags are created in Linux, it seems that the > Christof> data attached to the write request changes between the > Christof> generation in bio_integrity_generate and the call to > Christof> sd_prep_fn. > > Yep, known bug. Page writeback locking is messed up for buffer_head > users. The extNfs folks volunteered to look into this a while back but > I don't think they have found the time yet. What do you mean by messed up? Allowing modifications to the page while it is under writeback? This is deliberate of course and not limited to buffer_head users either. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo(a)vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ |