On Thu, Dec 12, 2019 at 01:54:43PM +0800, Qu Wenruo wrote:
On 2019/12/12 下午1:37, Naohiro Aota wrote:
On Thu, Dec 12, 2019 at 01:30:34PM +0800, Qu Wenruo wrote:
[BUG]
With latest master, btrfs/09[58] fails like:
btrfs/095 2s ... - output mismatch (see
xfstests-dev/results//btrfs/095.out.bad)
--- tests/btrfs/095.out 2019-12-12 13:23:24.266697540 +0800
+++ xfstests-dev/results//btrfs/095.out.bad 2019-12-12
13:23:29.340030879 +0800
@@ -4,32 +4,32 @@
File contents before power failure:
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
*
-207 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
+771 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
*
-226 bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb bb
...
(Run 'diff -u xfstests-dev/tests/btrfs/095.out
xfstests-dev/results//btrfs/095.out.bad' to see the entire diff)
btrfs/098 2s ... - output mismatch (see
xfstests-dev/results//btrfs/098.out.bad)
--- tests/btrfs/098.out 2019-12-12 13:23:24.266697540 +0800
+++ xfstests-dev/results//btrfs/098.out.bad 2019-12-12
13:23:31.306697545 +0800
@@ -3,20 +3,20 @@
File contents before power failure:
0 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
*
-144 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
+537 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
*
-151 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
...
(Run 'diff -u xfstests-dev/tests/btrfs/098.out
xfstests-dev/results//btrfs/098.out.bad' to see the entire diff)
Ran: btrfs/095 btrfs/098
Failures: btrfs/095 btrfs/098
Failed 2 of 2 tests
[CAUSE]
It looks like commit 37520a314bd4 ("fstests: Don't use gawk's strtonum")
is making _filter_od doing stupid filtering.
I sent a fix to the list. That commit is parsing od's offsets as decimal
which actually is octal.
https://lore.kernel.org/fstests/20191212031152.1906287-1-naohiro.aota@xxxxxxx/T/#u
Oh, that's much better.
Although that _filter_od still seems can't handle hex.
I agree with you that ocatal makes no sense these days. I'm considering to
extend _filter_od to take an argument to specify "oct|dec|hex" so that
callers can use "od -A x" and see outputs in hex (or decimal).
Thanks,
