Linus Torvalds wrote:
> It would be interesting to see what triggered it, since it apparently
> worked before. So yes, a bisection would be great.
Tejun, all the problematic patches are yours - so adding you.
Neil, since the problem only occurs whilst an xfs filesystem is mounted on a
raid6 array, I've cc'ed you too...
Got as far as I could...
I've run 9 or 10 kernels/bisects and got to a point with 8 of Tejun's changesets
where it wouldn't compile:
drivers/ata/sata_via.c:120: error: `ata_scsi_device_suspend' undeclared here
(not in a function)
drivers/ata/sata_via.c:120: error: initializer element is not constant
drivers/ata/sata_via.c:120: error: (near initialization for `svia_sht.suspend')
drivers/ata/sata_via.c:121: error: `ata_scsi_device_resume' undeclared here (not
in a function)
drivers/ata/sata_via.c:121: error: initializer element is not constant
drivers/ata/sata_via.c:121: error: (near initialization for `svia_sht.resume')
make: *** [drivers/ata/sata_via.o] Error 1
git bisect visualise gave:
git reset --hard 8575b814097af648dad284bd3087875a11b13d18
git reset --hard e92351bb53c0849fabfa80be53cbf3b0aa166e54
git reset --hard 3a32a8e96694a243ec7e7feb6d76dfc4b1fe90c1
git reset --hard 9666f4009c22f6520ac3fb8a19c9e32ab973e828
to step through - non compiled
git reset --hard 1d30c33d8d07868199560b24f10ed6280e78a89c
compiled and hung on resume.
given the first patch identified is
9666f4009c22f6520ac3fb8a19c9e32ab973e828: "libata: reimplement suspend/resume
support using sdev->manage_start_stop"
That seems a good candidate...
Incidentally, when I compile 1d30c33d8d07868199560b24f10ed6280e78a89c (far side
of the implicated changesets) if I umount my xfs over raid6 filesystem (no lvm
as I said in the OP) the resume succeeds.
PS I hope I've interpreted bisect correctly - first use and all that...