xfstests 068: Add mmap load
PATCH 1/1 xfstests 068: Add mmap load
Test 068 is the go-to test for freeze deadlock coverage; unfortunately it only uses fsstress, which doesn't do any mmap IO.
Using the existing fstest binary gets us a cheap mmap exerciser as well.
---
I think it's fair to modify this test vs. making a new one - it's never passed yet, and will only start passing in kernel 3.6, so it shouldn't cause new test regressions to add mmap ops.
And I can verify that this finds mmap bugs; I had a backport which messed up ext4's freeze vs. mmap handling, and this exposes it. :)
068 +15 -0
@@ -90,6 +90,21 @@ touch $tmp.running rmdir $STRESS_DIR } & +# start fstest -m loop in a background block; this gets us mmap coverage +{ + FSTEST_DIR="$SCRATCH_MNT/fstest_test_dir" + mkdir "$FSTEST_DIR" + + procs=2 + while [ -f "$tmp.running" ] + do + src/fstest -m -p $FSTEST_DIR -n $procs -f 4 > /dev/null 2>&1 + done + + rm -rf $FSTEST_DIR/* + rmdir $FSTEST_DIR +} & + i=0 let ITERATIONS=$ITERATIONS-1
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 31 lines · Eric Sandeen
Test 068 is the go-to test for freeze deadlock coverage; unfortunately it only uses fsstress, which doesn't do any mmap IO.
Using the existing fstest binary gets us a cheap mmap exerciser as well.
Signed-off-by Eric Sandeen <[email protected]>---
I think it's fair to modify this test vs. making a new one - it's never passed yet, and will only start passing in kernel 3.6, so it shouldn't cause new test regressions to add mmap ops.
And I can verify that this finds mmap bugs; I had a backport which messed up ext4's freeze vs. mmap handling, and this exposes it. :)
068 +4
-0@@ -90,6 +90,21 @@ touch $tmp.running rmdir $STRESS_DIR } & +# start fstest -m loop in a background block; this gets us mmap coverage +{ + FSTEST_DIR="$SCRATCH_MNT/fstest_test_dir" + mkdir "$FSTEST_DIR"
068 touch $tmp.running · line 106
Quoted 10 lines
068 +10
-0+ + procs=2 + while [ -f "$tmp.running" ] + do + src/fstest -m -p $FSTEST_DIR -n $procs -f 4 > /dev/null 2>&1 + done + + rm -rf $FSTEST_DIR/* + rmdir $FSTEST_DIR +} &
Looks good.
Signature · 7 lines
Dave Chinner [email protected] _______________________________________________ xfs mailing list [email protected] http://oss.sgi.com/mailman/listinfo/xfs
Quoted 8 lines · Eric Sandeen
Test 068 is the go-to test for freeze deadlock coverage; unfortunately it only uses fsstress, which doesn't do any mmap IO.
Using the existing fstest binary gets us a cheap mmap exerciser as well.
Signed-off-by Eric Sandeen <[email protected]>
Looks good,
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs
Quoted 10 lines · Eric Sandeen
Test 068 is the go-to test for freeze deadlock coverage; unfortunately it only uses fsstress, which doesn't do any mmap IO.
Using the existing fstest binary gets us a cheap mmap exerciser as well.
Signed-off-by Eric Sandeen<[email protected]>---
Quoted 8 lines
I think it's fair to modify this test vs. making a new one - it's never passed yet, and will only start passing in kernel 3.6, so it shouldn't cause new test regressions to add mmap ops.
And I can verify that this finds mmap bugs; I had a backport which messed up ext4's freeze vs. mmap handling, and this exposes it. :)
Committed to git://oss.sgi.com/xfs/cmds/xfs.git, master branch.
--Mark.
_______________________________________________
xfs mailing list
[email protected]
http://oss.sgi.com/mailman/listinfo/xfs