If the path to the file being bmap'd has a [0-9] in it,
the output filter matches it and we get golden output
failure. Be more specific on the match.
Merge of master-melb:xfs-cmds:31170a by kenmcd.
Be more specific when trying to match extent output lines.
/tmp might be small, might not support files larger than 2GB,
etc, so trying to host loopback images of 100GB filesystems
will break in some situations. We should use $TEST_DIR for the
images.
Note: remounting of loopback images has a major bug (in mount)
that "leaks" loopback device references. Hence all the hackery
with losetup to work around this.
Merge of master-melb:xfs-cmds:31169a by kenmcd.
Don't use /tmp for hosting loopback images
made to test 008,the file sizes got much larger. It appears
that randholes actually reads the entire file, so this has
slowed the test down by a factor of ten (all file sizes
were increased by 10x). This means the test is now taking
about 18 minutes to run on a UML session, and all the time
is spent reading the files.
Instead, scale the file size based on the page size. We know
how many holes we are trying to produce and the I/O size
being used to produce them, so the size of the files can be
finely tuned. Assuming a decent random distribution, if the
number of blocks in the file is 4x the page size and the
I/O size is page sized, this means that every I/O should
generate a new hole and we'll only get a small amount of
adjacent extents. This has passed over 10 times on ia64
w/ 64k page and another 15 times on UML with 4k page.
UML runtime is down from ~1000s to 5s, ia64 runtime is down from
~30s to 7s.
Merge of master-melb:xfs-cmds:31168a by kenmcd.
Greatly reduce runtime by reducing filesizes down to sane minimum.
have a better idea of predicting when it will get to the end
of the log. This way we can handle a change in log traffic in
the future. A test to keep an eye on log traffic is more of
a performance test than should be done elsewhere.
Merge of master-melb:xfs-cmds:31053a by kenmcd.
Sample the log traffic to work out its data rate so that we
have a better idea of predicting when it will get to the end
of the log. This way we can handle a change in log traffic in
the future.
Files are too small for 64k sized I/Os to reliably produce 50
extents +/-10%. Increase the size of the files to increase
reliability.
Merge of master-melb:xfs-cmds:30996a by kenmcd.
Fix intermittent failure on 64k pagesize.
Flags has extra bits set in it xfs_bmap output when
the filesyste is using stripe alignment. Fix the test
to only look at the unwritten flag. Capture the xfs_bmap
output as well so failures can be debugged easily.
Merge of master-melb:xfs-cmds:30994a by kenmcd.
fix filter for stripe aligned filesystems.
We should use XFS_IO_PROG variable rather than calling xfs_io
directly and relying on teh path to pick up the correct binary
for the test.
Merge of master-melb:xfs-cmds:30993a by kenmcd.
Use XFS_IO_PROG, not xfs_io.
Merge of master-melb:xfs-cmds:30992a by kenmcd.
export the language that all tests will run under to prevent
language settings from affecting test output and giving
spurious failures when comparing against golden output.
instead of di_forkoff.
Merge of master-melb:xfs-cmds:30863a by kenmcd.
Test out bug in xfs_attr_shortform_bytesfit for its use of m_attroffset
instead of di_forkoff.
Kill the output of uninitialised values when punching a hole.
This prevents the test from passing on just about everything.
Merge of master-melb:xfs-cmds:30762a by kenmcd.
remove roffp and rlenp output for punch operations.
Test 091 assumes a direct I/O alignment of 512 bytes,
a hold over from 2.4 kernels. On 2.6. kernels, direct
I/O needs to be aligned to the sector size the filesystem
was mkfs'd with.
Merge of master-melb:xfs-cmds:30760a by kenmcd.
Teach 091 about 2.6 kernels and grab the sector size from the
xfs_info output.
More recent versions of ln (version >= 6.0) have a different error
output. update the filter to handle this.
Merge of master-melb:xfs-cmds:30705a by kenmcd.
update filter to handle new ln error output.