On 1/23/20 11:14 AM, Anand Jain wrote:
This patch is causing regression in the test case generic/027
and generic/275 with MKFS_OPTIONS="-n64K" on x86_64.
Both of these tests, test FS behavior at ENOSPC.
In generic/027, it fails to delete a file with ENOSPC
+rm: cannot remove '/mnt/scratch/testdir/6/file_1598': No space left on
device
In generic/275, it failed to create at least 128k size file after
deleting 256k sized file. Failure may be fair taking into the cow part,
but any idea why it could be successful before this patch?
+du: cannot access '/mnt/scratch/tmp1': No such file or directory
+stat: cannot stat '/mnt/scratch/tmp1': No such file or directory
These fail on misc-next as well.
I spot checked this and it appears because before we just weren't getting as
full before, and now we are.
These failures are the next thing on my list, I'll likely have them fixed early
next week. Thanks,
Josef
