Diff to fix pjdfstest tests
yaneurabeya at gmail.com
Wed Apr 9 22:57:10 UTC 2014
(CCing the maintainer, pjd@)
On Wed, Apr 9, 2014 at 3:51 PM, Peel, Casey <casey.peel at isilon.com> wrote:
> In attempting to get pjdfstest working I found that some of the tests have easily-corrected failures:
> One of which is a plan error: ftruncate/00.t runs 26 tests but only specifies 21 in the plan.
> Tests that create a small memory disk and fill it to ENOSPC don't validate that the memory disk mounts and will happily attempt to fill up your real partition. The core of the failing is that the memory disk isn't large enough for the subsequent newfs to succeed and thus the mount fails.
> vbox1009# mdconfig -a -n -t malloc -s 256k
> vbox1009# newfs /dev/md0
> /dev/md0: 0.2MB (512 sectors) block size 32768, fragment size 4096
> using 1 cylinder groups of 0.25MB, 8 blks, 128 inodes.
> super-block backups (for fsck -b #) at:
> first cylinder group ran out of space
> vbox1009# mount /dev/md0 /mnt
> mount: /dev/md0: Invalid argument
> I've increased the size of the memdisk and added a hard exit to the test if the mount fails.
> The open/16.t tests are all failing. The tests expect a ELOOP exit code but I'm getting EMLINK. I haven't validated if this is a test error or an error on the not-exactly-FreeBSD-10 code I'm running, so I haven't included those tests in this diff and just made them TODO in my code until I figure it out.
> root # prove -r --nocount /svn/branch/src/tools/regression/pjdfstest/tests
> All tests successful.
> Files=206, Tests=12100, 839 wallclock secs ( 3.04 usr 2.71 sys + 40.46 cusr 374.61 csys = 420.82 CPU)
> Result: PASS
> root #
Have you looked at what I did on
https://github.com/yaneurabeya/pjdfstest (in particular I was working
on the onefs branch back when I was with EMC)?
The EMLINK issue was a known problem with the tests:
. There are some other commits you probably want to grab and apply to
the local tree after validating them:
https://github.com/yaneurabeya/pjdfstest/commits/master . It would be
nice if these were in the upstream github repo/in FreeBSD, but I
didn't get around to setting up a pull request. I'll see about doing
More information about the freebsd-testing