Messages in this thread | | | From | Jeff Moyer <> | Subject | Re: [PATCH V5 00/30] loop: Issue O_DIRECT aio using bio_vec | Date | Tue, 22 Jan 2013 11:14:27 -0500 |
| |
Dave Kleikamp <dave.kleikamp@oracle.com> writes:
> On 01/18/2013 03:48 PM, Jeff Moyer wrote: >> Dave Kleikamp <dave.kleikamp@oracle.com> writes: >> >>> On 01/18/2013 11:56 AM, Jeff Moyer wrote: >>>> Jeff Moyer <jmoyer@redhat.com> writes: >>>> >>>>> ./check -g aio >>>>> >>>>> and here is the summary: >>>>> >>>>> Ran: 112 113 198 207 208 209 210 211 212 239 240 >>>>> Failures: 112 198 207 239 240 >>>>> Failed 5 of 11 tests >>>>> >>>>> To be fair, I have not yet run this on a kernel without your changes. >>>> >>>> I re-ran the same test against a vanilla 3.8.0-rc4 kernel and it passed: >>>> >>>> Ran: 112 113 198 207 208 209 210 211 212 239 240 >>>> Passed all 11 tests >>>> >>>> I'll start digging into the failures. >>> >>> Thanks, I'll take a look at it too. >> >> Note that the failures occur on regular block devices, too; this isn't >> restricted to loop. The fix I mentioned in reply to patch 17 actually >> fixes all of the failures reported in the aio group. I'm going to do >> some more thorough testing over the weekend. > > Thanks much. I'll definitely add this to my testing.
After fixing up that thinko, a full run of xfstests on an xfs file system backed by a loop device (which itself is a file on an xfs file system) shows no regressions.
Cheers, Jeff
| |