2013-04-02 14:27:33

by Mel Gorman

[permalink] [raw]
Subject: Excessive stall times on ext4 in 3.9-rc2

I'm testing a page-reclaim-related series on my laptop that is partially
aimed at fixing long stalls when doing metadata-intensive operations on
low memory such as a git checkout. I've been running 3.9-rc2 with the
series applied but found that the interactive performance was awful even
when there was plenty of free memory.

I activated a monitor from mmtests that logs when a process is stuck for
a long time in D state and found that there are a lot of stalls in ext4.
The report first states that processes have been stalled for a total of
6498 seconds on IO which seems like a lot. Here is a breakdown of the
recorded events.

Time stalled in this event: 566745 ms
Event count: 181
git sleep_on_buffer 1236 ms
git sleep_on_buffer 1161 ms
imapd sleep_on_buffer 3111 ms
cp sleep_on_buffer 10745 ms
cp sleep_on_buffer 5036 ms
cp sleep_on_buffer 4370 ms
cp sleep_on_buffer 1682 ms
cp sleep_on_buffer 8207 ms
cp sleep_on_buffer 5312 ms
cp sleep_on_buffer 1563 ms
patch sleep_on_buffer 1172 ms
patch sleep_on_buffer 4585 ms
patch sleep_on_buffer 3541 ms
patch sleep_on_buffer 4155 ms
patch sleep_on_buffer 3120 ms
cc1 sleep_on_buffer 1107 ms
cc1 sleep_on_buffer 1291 ms
cc1 sleep_on_buffer 1125 ms
cc1 sleep_on_buffer 1257 ms
imapd sleep_on_buffer 1424 ms
patch sleep_on_buffer 1126 ms
mutt sleep_on_buffer 4804 ms
patch sleep_on_buffer 3489 ms
patch sleep_on_buffer 4242 ms
cp sleep_on_buffer 1942 ms
cp sleep_on_buffer 2670 ms
cp sleep_on_buffer 1071 ms
cp sleep_on_buffer 1676 ms
cp sleep_on_buffer 1058 ms
cp sleep_on_buffer 1382 ms
cp sleep_on_buffer 2196 ms
cp sleep_on_buffer 1017 ms
cp sleep_on_buffer 1096 ms
cp sleep_on_buffer 1203 ms
cp sleep_on_buffer 1307 ms
cp sleep_on_buffer 1676 ms
cp sleep_on_buffer 1024 ms
cp sleep_on_buffer 1270 ms
cp sleep_on_buffer 1200 ms
cp sleep_on_buffer 1674 ms
cp sleep_on_buffer 1202 ms
cp sleep_on_buffer 2260 ms
cp sleep_on_buffer 1685 ms
cp sleep_on_buffer 1921 ms
cp sleep_on_buffer 1434 ms
cp sleep_on_buffer 1346 ms
cp sleep_on_buffer 2132 ms
cp sleep_on_buffer 1304 ms
cp sleep_on_buffer 1328 ms
cp sleep_on_buffer 1419 ms
cp sleep_on_buffer 1882 ms
cp sleep_on_buffer 1172 ms
cp sleep_on_buffer 1299 ms
cp sleep_on_buffer 1806 ms
cp sleep_on_buffer 1297 ms
cp sleep_on_buffer 1484 ms
cp sleep_on_buffer 1313 ms
cp sleep_on_buffer 1342 ms
cp sleep_on_buffer 1320 ms
cp sleep_on_buffer 1147 ms
cp sleep_on_buffer 1346 ms
cp sleep_on_buffer 2391 ms
cp sleep_on_buffer 1128 ms
cp sleep_on_buffer 1386 ms
cp sleep_on_buffer 1505 ms
cp sleep_on_buffer 1664 ms
cp sleep_on_buffer 1290 ms
cp sleep_on_buffer 1532 ms
cp sleep_on_buffer 1719 ms
cp sleep_on_buffer 1149 ms
cp sleep_on_buffer 1364 ms
cp sleep_on_buffer 1397 ms
cp sleep_on_buffer 1213 ms
cp sleep_on_buffer 1171 ms
cp sleep_on_buffer 1352 ms
cp sleep_on_buffer 3000 ms
cp sleep_on_buffer 4866 ms
cp sleep_on_buffer 5863 ms
cp sleep_on_buffer 3951 ms
cp sleep_on_buffer 3469 ms
cp sleep_on_buffer 2172 ms
cp sleep_on_buffer 21366 ms
cp sleep_on_buffer 28856 ms
cp sleep_on_buffer 1212 ms
cp sleep_on_buffer 2326 ms
cp sleep_on_buffer 1357 ms
cp sleep_on_buffer 1482 ms
cp sleep_on_buffer 1372 ms
cp sleep_on_buffer 1475 ms
cp sleep_on_buffer 1540 ms
cp sleep_on_buffer 2993 ms
cp sleep_on_buffer 1269 ms
cp sleep_on_buffer 1478 ms
cp sleep_on_buffer 1137 ms
cp sleep_on_buffer 1114 ms
cp sleep_on_buffer 1137 ms
cp sleep_on_buffer 1616 ms
cp sleep_on_buffer 1291 ms
cp sleep_on_buffer 1336 ms
cp sleep_on_buffer 2440 ms
cp sleep_on_buffer 1058 ms
cp sleep_on_buffer 1825 ms
cp sleep_on_buffer 1320 ms
cp sleep_on_buffer 2556 ms
cp sleep_on_buffer 2463 ms
cp sleep_on_buffer 2563 ms
cp sleep_on_buffer 1218 ms
cp sleep_on_buffer 2862 ms
cp sleep_on_buffer 1484 ms
cp sleep_on_buffer 1039 ms
cp sleep_on_buffer 5180 ms
cp sleep_on_buffer 2584 ms
cp sleep_on_buffer 1357 ms
cp sleep_on_buffer 4492 ms
cp sleep_on_buffer 1111 ms
cp sleep_on_buffer 3992 ms
cp sleep_on_buffer 4205 ms
cp sleep_on_buffer 4980 ms
cp sleep_on_buffer 6303 ms
imapd sleep_on_buffer 8473 ms
cp sleep_on_buffer 7128 ms
cp sleep_on_buffer 4740 ms
cp sleep_on_buffer 10236 ms
cp sleep_on_buffer 1210 ms
cp sleep_on_buffer 2670 ms
cp sleep_on_buffer 11461 ms
cp sleep_on_buffer 5946 ms
cp sleep_on_buffer 7144 ms
cp sleep_on_buffer 2205 ms
cp sleep_on_buffer 25904 ms
cp sleep_on_buffer 1766 ms
cp sleep_on_buffer 9823 ms
cp sleep_on_buffer 1849 ms
cp sleep_on_buffer 1380 ms
cp sleep_on_buffer 2524 ms
cp sleep_on_buffer 2389 ms
cp sleep_on_buffer 1996 ms
cp sleep_on_buffer 10396 ms
cp sleep_on_buffer 2020 ms
cp sleep_on_buffer 1132 ms
cc1 sleep_on_buffer 1182 ms
cp sleep_on_buffer 1195 ms
cp sleep_on_buffer 1179 ms
cp sleep_on_buffer 7301 ms
cp sleep_on_buffer 8328 ms
cp sleep_on_buffer 6922 ms
cp sleep_on_buffer 10555 ms
Cache I/O sleep_on_buffer 11963 ms
cp sleep_on_buffer 2368 ms
cp sleep_on_buffer 6905 ms
cp sleep_on_buffer 1686 ms
cp sleep_on_buffer 1219 ms
cp sleep_on_buffer 1793 ms
cp sleep_on_buffer 1899 ms
cp sleep_on_buffer 6412 ms
cp sleep_on_buffer 2799 ms
cp sleep_on_buffer 1316 ms
cp sleep_on_buffer 1211 ms
git sleep_on_buffer 1328 ms
imapd sleep_on_buffer 4242 ms
imapd sleep_on_buffer 2754 ms
imapd sleep_on_buffer 4496 ms
imapd sleep_on_buffer 4603 ms
imapd sleep_on_buffer 7929 ms
imapd sleep_on_buffer 8851 ms
imapd sleep_on_buffer 2016 ms
imapd sleep_on_buffer 1019 ms
imapd sleep_on_buffer 1138 ms
git sleep_on_buffer 1510 ms
git sleep_on_buffer 1366 ms
git sleep_on_buffer 3445 ms
git sleep_on_buffer 2704 ms
git sleep_on_buffer 2057 ms
git sleep_on_buffer 1202 ms
git sleep_on_buffer 1293 ms
cat sleep_on_buffer 1505 ms
imapd sleep_on_buffer 1263 ms
imapd sleep_on_buffer 1347 ms
imapd sleep_on_buffer 2910 ms
git sleep_on_buffer 1210 ms
git sleep_on_buffer 1199 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff811105e3>] do_generic_file_read.constprop.35+0x363/0x440
[<ffffffff81111359>] generic_file_aio_read+0xd9/0x220
[<ffffffff81172b53>] do_sync_read+0xa3/0xe0
[<ffffffff8117327b>] vfs_read+0xab/0x170
[<ffffffff8117338d>] sys_read+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Some of those stalls are awful -- 28 seconds to update atime seems
excessive. This is with relatime in use

[email protected]:~ > mount | grep sd
/dev/sda8 on / type ext4 (rw,relatime,nobarrier,data=ordered)
/dev/sda6 on /home type ext4 (rw,relatime,nobarrier,data=ordered)
/dev/sda5 on /usr/src type ext4 (rw,relatime,nobarrier,data=ordered)

/tmp is mounted as tmpfs so I doubt it's a small write problem.

Time stalled in this event: 466201 ms
Event count: 45
git sleep_on_buffer 1011 ms
git sleep_on_buffer 29540 ms
git sleep_on_buffer 1485 ms
git sleep_on_buffer 1244 ms
git sleep_on_buffer 17896 ms
git sleep_on_buffer 1882 ms
git sleep_on_buffer 18249 ms
mv sleep_on_buffer 2107 ms
mv sleep_on_buffer 12655 ms
mv sleep_on_buffer 4290 ms
mv sleep_on_buffer 2640 ms
patch sleep_on_buffer 2433 ms
patch sleep_on_buffer 2305 ms
patch sleep_on_buffer 3672 ms
git sleep_on_buffer 16663 ms
git sleep_on_buffer 16516 ms
git sleep_on_buffer 16168 ms
git sleep_on_buffer 1382 ms
git sleep_on_buffer 1695 ms
git sleep_on_buffer 1301 ms
git sleep_on_buffer 22039 ms
git sleep_on_buffer 19077 ms
git sleep_on_buffer 1208 ms
git sleep_on_buffer 20237 ms
git sleep_on_buffer 1284 ms
git sleep_on_buffer 19518 ms
git sleep_on_buffer 1959 ms
git sleep_on_buffer 27574 ms
git sleep_on_buffer 9708 ms
git sleep_on_buffer 1968 ms
git sleep_on_buffer 23600 ms
git sleep_on_buffer 12578 ms
git sleep_on_buffer 19573 ms
git sleep_on_buffer 2257 ms
git sleep_on_buffer 19068 ms
git sleep_on_buffer 2833 ms
git sleep_on_buffer 3182 ms
git sleep_on_buffer 22496 ms
git sleep_on_buffer 14030 ms
git sleep_on_buffer 1722 ms
git sleep_on_buffer 25652 ms
git sleep_on_buffer 15730 ms
git sleep_on_buffer 19096 ms
git sleep_on_buffer 1529 ms
git sleep_on_buffer 3149 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177ba9>] vfs_lstat+0x19/0x20
[<ffffffff81177d15>] sys_newlstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

These are directory lookups which might be a bit more reasonable to
stall on but stalls of 30 seconds seems way out of order. Unfortuantely
I do not have a comparison with older kernels but even when interactive
performance was bad on older kernels, it did not feel *this* bad.

The rest of the mail is just the remaining stalls recorded. They are a
lot of them and they are all really high. Is this a known issue? It's
not necessarily an ext4 issue and could be an IO scheduler or some other
writeback change too. I've been offline for a while so could have missed
similar bug reports and/or fixes.

Time stalled in this event: 437040 ms
Event count: 106
git wait_on_page_bit 1517 ms
git wait_on_page_bit 2694 ms
git wait_on_page_bit 2829 ms
git wait_on_page_bit 2796 ms
git wait_on_page_bit 2625 ms
git wait_on_page_bit 14350 ms
git wait_on_page_bit 4529 ms
xchat wait_on_page_bit 1928 ms
akregator wait_on_page_bit 1116 ms
akregator wait_on_page_bit 3556 ms
cat wait_on_page_bit 5311 ms
sequence-patch. wait_on_page_bit 2555 ms
pool wait_on_page_bit 1485 ms
git wait_on_page_bit 6778 ms
git wait_on_page_bit 3464 ms
git wait_on_page_bit 2189 ms
pool wait_on_page_bit 3657 ms
compare-kernels wait_on_page_bit 5729 ms
compare-kernels wait_on_page_bit 4446 ms
git wait_on_page_bit 2011 ms
xchat wait_on_page_bit 6250 ms
git wait_on_page_bit 2761 ms
git wait_on_page_bit 1157 ms
xchat wait_on_page_bit 2670 ms
pool wait_on_page_bit 5964 ms
xchat wait_on_page_bit 1805 ms
play wait_on_page_bit 1800 ms
xchat wait_on_page_bit 12008 ms
cat wait_on_page_bit 3642 ms
sequence-patch. wait_on_page_bit 2309 ms
sequence-patch. wait_on_page_bit 5430 ms
cat wait_on_page_bit 2614 ms
sequence-patch. wait_on_page_bit 2220 ms
git wait_on_page_bit 3505 ms
git wait_on_page_bit 4181 ms
mozStorage #2 wait_on_page_bit 1012 ms
xchat wait_on_page_bit 1966 ms
pool wait_on_page_bit 14217 ms
pool wait_on_page_bit 3728 ms
xchat wait_on_page_bit 1896 ms
play wait_on_page_bit 8731 ms
mutt wait_on_page_bit 14378 ms
play wait_on_page_bit 1208 ms
Cache I/O wait_on_page_bit 1174 ms
xchat wait_on_page_bit 1141 ms
mozStorage #2 wait_on_page_bit 1161 ms
mozStorage #2 wait_on_page_bit 6727 ms
Cache I/O wait_on_page_bit 7559 ms
mozStorage #2 wait_on_page_bit 4630 ms
Cache I/O wait_on_page_bit 4642 ms
mozStorage #2 wait_on_page_bit 1764 ms
mozStorage #2 wait_on_page_bit 2357 ms
Cache I/O wait_on_page_bit 3694 ms
xchat wait_on_page_bit 8484 ms
mozStorage #2 wait_on_page_bit 3958 ms
mozStorage #2 wait_on_page_bit 2067 ms
Cache I/O wait_on_page_bit 2728 ms
xchat wait_on_page_bit 4115 ms
Cache I/O wait_on_page_bit 7738 ms
xchat wait_on_page_bit 7279 ms
Cache I/O wait_on_page_bit 4366 ms
mozStorage #2 wait_on_page_bit 2040 ms
mozStorage #2 wait_on_page_bit 1102 ms
mozStorage #2 wait_on_page_bit 4628 ms
Cache I/O wait_on_page_bit 5127 ms
akregator wait_on_page_bit 2897 ms
Cache I/O wait_on_page_bit 1429 ms
mozStorage #3 wait_on_page_bit 1465 ms
git wait_on_page_bit 2830 ms
git wait_on_page_bit 2508 ms
mutt wait_on_page_bit 4955 ms
pool wait_on_page_bit 4495 ms
mutt wait_on_page_bit 7429 ms
akregator wait_on_page_bit 3744 ms
mutt wait_on_page_bit 11632 ms
pool wait_on_page_bit 11632 ms
sshd wait_on_page_bit 16035 ms
mutt wait_on_page_bit 16254 ms
mutt wait_on_page_bit 3253 ms
mutt wait_on_page_bit 3254 ms
git wait_on_page_bit 2644 ms
git wait_on_page_bit 2434 ms
git wait_on_page_bit 8364 ms
git wait_on_page_bit 1618 ms
git wait_on_page_bit 5990 ms
git wait_on_page_bit 2663 ms
git wait_on_page_bit 1102 ms
git wait_on_page_bit 1160 ms
git wait_on_page_bit 1161 ms
git wait_on_page_bit 1608 ms
git wait_on_page_bit 2100 ms
git wait_on_page_bit 2215 ms
git wait_on_page_bit 1231 ms
git wait_on_page_bit 2274 ms
git wait_on_page_bit 6081 ms
git wait_on_page_bit 6877 ms
git wait_on_page_bit 2035 ms
git wait_on_page_bit 2568 ms
git wait_on_page_bit 4475 ms
pool wait_on_page_bit 1253 ms
mv sleep_on_buffer 1036 ms
git wait_on_page_bit 1876 ms
git wait_on_page_bit 2332 ms
git wait_on_page_bit 2840 ms
git wait_on_page_bit 1850 ms
git wait_on_page_bit 3943 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff8110f3aa>] generic_perform_write+0xca/0x210
[<ffffffff8110f548>] generic_file_buffered_write+0x58/0x90
[<ffffffff81110f96>] __generic_file_aio_write+0x1b6/0x3b0
[<ffffffff8111120a>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ea3a3>] ext4_file_write+0x83/0xd0
[<ffffffff81172a73>] do_sync_write+0xa3/0xe0
[<ffffffff811730fe>] vfs_write+0xae/0x180
[<ffffffff8117341d>] sys_write+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 417840 ms
Event count: 56
xchat sleep_on_buffer 8571 ms
xchat sleep_on_buffer 1772 ms
xchat sleep_on_buffer 4063 ms
xchat sleep_on_buffer 16290 ms
xchat sleep_on_buffer 3201 ms
compare-kernels sleep_on_buffer 1698 ms
xchat sleep_on_buffer 14631 ms
xchat sleep_on_buffer 12970 ms
xchat sleep_on_buffer 4182 ms
xchat sleep_on_buffer 5449 ms
Cache I/O sleep_on_buffer 4079 ms
xchat sleep_on_buffer 8246 ms
xchat sleep_on_buffer 6530 ms
xchat sleep_on_buffer 2041 ms
xchat sleep_on_buffer 15815 ms
pool sleep_on_buffer 4115 ms
tee sleep_on_buffer 2057 ms
xchat sleep_on_buffer 4814 ms
tee sleep_on_buffer 66037 ms
Cache I/O sleep_on_buffer 6601 ms
xchat sleep_on_buffer 10208 ms
tee sleep_on_buffer 6064 ms
Cache I/O sleep_on_buffer 2008 ms
xchat sleep_on_buffer 5257 ms
git sleep_on_buffer 2032 ms
xchat sleep_on_buffer 2313 ms
tee sleep_on_buffer 5287 ms
Cache I/O sleep_on_buffer 1650 ms
akregator sleep_on_buffer 1154 ms
tee sleep_on_buffer 10362 ms
xchat sleep_on_buffer 6208 ms
xchat sleep_on_buffer 4405 ms
Cache I/O sleep_on_buffer 8580 ms
mozStorage #2 sleep_on_buffer 6573 ms
tee sleep_on_buffer 10180 ms
Cache I/O sleep_on_buffer 7691 ms
mozStorage #3 sleep_on_buffer 5502 ms
xchat sleep_on_buffer 2339 ms
Cache I/O sleep_on_buffer 3819 ms
sshd sleep_on_buffer 7252 ms
tee sleep_on_buffer 11422 ms
Cache I/O sleep_on_buffer 1661 ms
bash sleep_on_buffer 10905 ms
git sleep_on_buffer 1277 ms
git sleep_on_buffer 18599 ms
git sleep_on_buffer 1189 ms
git sleep_on_buffer 22945 ms
pool sleep_on_buffer 17753 ms
git sleep_on_buffer 1367 ms
git sleep_on_buffer 2223 ms
git sleep_on_buffer 1280 ms
git sleep_on_buffer 2061 ms
git sleep_on_buffer 1034 ms
pool sleep_on_buffer 18189 ms
git sleep_on_buffer 1344 ms
xchat sleep_on_buffer 2545 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118b868>] file_update_time+0x98/0x100
[<ffffffff81110f5c>] __generic_file_aio_write+0x17c/0x3b0
[<ffffffff8111120a>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ea3a3>] ext4_file_write+0x83/0xd0
[<ffffffff81172a73>] do_sync_write+0xa3/0xe0
[<ffffffff811730fe>] vfs_write+0xae/0x180
[<ffffffff8117341d>] sys_write+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 283964 ms
Event count: 27
git sleep_on_buffer 19088 ms
git sleep_on_buffer 1177 ms
git sleep_on_buffer 30745 ms
git sleep_on_buffer 4782 ms
git sleep_on_buffer 11435 ms
git sleep_on_buffer 2816 ms
git sleep_on_buffer 5088 ms
git-merge sleep_on_buffer 18801 ms
git sleep_on_buffer 1415 ms
git sleep_on_buffer 16005 ms
git sleep_on_buffer 2178 ms
git sleep_on_buffer 14354 ms
git sleep_on_buffer 12612 ms
git sleep_on_buffer 2785 ms
git sleep_on_buffer 15498 ms
git sleep_on_buffer 15331 ms
git sleep_on_buffer 1151 ms
git sleep_on_buffer 1320 ms
git sleep_on_buffer 8787 ms
git sleep_on_buffer 2199 ms
git sleep_on_buffer 1006 ms
git sleep_on_buffer 23644 ms
git sleep_on_buffer 2407 ms
git sleep_on_buffer 1169 ms
git sleep_on_buffer 25022 ms
git sleep_on_buffer 18651 ms
git sleep_on_buffer 24498 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811fb6cf>] ext4_orphan_add+0x10f/0x1f0
[<ffffffff811fc6cb>] ext4_unlink+0x32b/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 266300 ms
Event count: 69
git sleep_on_buffer 2773 ms
akregator sleep_on_buffer 1957 ms
git sleep_on_buffer 1417 ms
imapd sleep_on_buffer 9532 ms
imapd sleep_on_buffer 57801 ms
pool sleep_on_buffer 7761 ms
imapd sleep_on_buffer 1444 ms
patch sleep_on_buffer 3872 ms
imapd sleep_on_buffer 6422 ms
imapd sleep_on_buffer 1748 ms
pool sleep_on_buffer 10552 ms
imapd sleep_on_buffer 10114 ms
imapd sleep_on_buffer 7575 ms
mutt sleep_on_buffer 3901 ms
bzip2 sleep_on_buffer 1104 ms
imapd sleep_on_buffer 4983 ms
imapd sleep_on_buffer 1746 ms
mutt sleep_on_buffer 1881 ms
imapd sleep_on_buffer 1067 ms
imapd sleep_on_buffer 1863 ms
imapd sleep_on_buffer 1508 ms
imapd sleep_on_buffer 1508 ms
offlineimap sleep_on_buffer 1385 ms
imapd sleep_on_buffer 1653 ms
imapd sleep_on_buffer 1179 ms
imapd sleep_on_buffer 3473 ms
imapd sleep_on_buffer 10130 ms
vim sleep_on_buffer 1690 ms
imapd sleep_on_buffer 3102 ms
dconf-service sleep_on_buffer 5097 ms
imapd sleep_on_buffer 2888 ms
cp sleep_on_buffer 1036 ms
imapd sleep_on_buffer 22501 ms
rsync sleep_on_buffer 5026 ms
imapd sleep_on_buffer 2897 ms
rsync sleep_on_buffer 1200 ms
akregator sleep_on_buffer 4780 ms
Cache I/O sleep_on_buffer 1433 ms
imapd sleep_on_buffer 2588 ms
akregator sleep_on_buffer 1576 ms
vi sleep_on_buffer 2086 ms
firefox sleep_on_buffer 4718 ms
imapd sleep_on_buffer 1158 ms
git sleep_on_buffer 2073 ms
git sleep_on_buffer 1017 ms
git sleep_on_buffer 1616 ms
git sleep_on_buffer 1043 ms
imapd sleep_on_buffer 1746 ms
imapd sleep_on_buffer 1007 ms
git sleep_on_buffer 1146 ms
git sleep_on_buffer 1916 ms
git sleep_on_buffer 1059 ms
git sleep_on_buffer 1801 ms
git sleep_on_buffer 1208 ms
git sleep_on_buffer 1486 ms
git sleep_on_buffer 1806 ms
git sleep_on_buffer 1295 ms
git sleep_on_buffer 1461 ms
git sleep_on_buffer 1371 ms
git sleep_on_buffer 2010 ms
git sleep_on_buffer 1622 ms
git sleep_on_buffer 1453 ms
git sleep_on_buffer 1392 ms
git sleep_on_buffer 1329 ms
git sleep_on_buffer 1773 ms
git sleep_on_buffer 1750 ms
git sleep_on_buffer 2354 ms
imapd sleep_on_buffer 3201 ms
imapd sleep_on_buffer 2240 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811ebe24>] __ext4_new_inode+0x294/0x10c0
[<ffffffff811fac5b>] ext4_create+0xbb/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 240174 ms
Event count: 34
systemd-journal sleep_on_buffer 1321 ms
systemd-journal sleep_on_buffer 4851 ms
systemd-journal sleep_on_buffer 3341 ms
systemd-journal sleep_on_buffer 17219 ms
systemd-journal sleep_on_buffer 3190 ms
systemd-journal sleep_on_buffer 13420 ms
systemd-journal sleep_on_buffer 23421 ms
systemd-journal sleep_on_buffer 4987 ms
systemd-journal sleep_on_buffer 16358 ms
systemd-journal sleep_on_buffer 2734 ms
mozStorage #2 sleep_on_buffer 1454 ms
systemd-journal sleep_on_buffer 4524 ms
mozStorage #2 sleep_on_buffer 1211 ms
systemd-journal sleep_on_buffer 1711 ms
systemd-journal sleep_on_buffer 2158 ms
mkdir wait_on_page_bit_killable 1084 ms
systemd-journal sleep_on_buffer 5673 ms
mozStorage #2 sleep_on_buffer 1800 ms
systemd-journal sleep_on_buffer 5586 ms
mozStorage #2 sleep_on_buffer 3199 ms
nm-dhcp-client. wait_on_page_bit_killable 1060 ms
mozStorage #2 sleep_on_buffer 6669 ms
systemd-journal sleep_on_buffer 3603 ms
systemd-journal sleep_on_buffer 7666 ms
systemd-journal sleep_on_buffer 13961 ms
systemd-journal sleep_on_buffer 9063 ms
systemd-journal sleep_on_buffer 4120 ms
systemd-journal sleep_on_buffer 3328 ms
systemd-journal sleep_on_buffer 12093 ms
systemd-journal sleep_on_buffer 5464 ms
systemd-journal sleep_on_buffer 12649 ms
systemd-journal sleep_on_buffer 23460 ms
systemd-journal sleep_on_buffer 13123 ms
systemd-journal sleep_on_buffer 4673 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118b868>] file_update_time+0x98/0x100
[<ffffffff811f539c>] ext4_page_mkwrite+0x5c/0x470
[<ffffffff8113740e>] do_wp_page+0x5ce/0x7d0
[<ffffffff81139598>] handle_pte_fault+0x1c8/0x200
[<ffffffff8113a731>] handle_mm_fault+0x271/0x390
[<ffffffff81597959>] __do_page_fault+0x169/0x520
[<ffffffff81597d19>] do_page_fault+0x9/0x10
[<ffffffff81594488>] page_fault+0x28/0x30
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 212304 ms
Event count: 41
pool sleep_on_buffer 1216 ms
pool sleep_on_buffer 36361 ms
cp sleep_on_buffer 5034 ms
git sleep_on_buffer 2344 ms
gnuplot sleep_on_buffer 1733 ms
gnuplot sleep_on_buffer 2303 ms
gnuplot sleep_on_buffer 1982 ms
gnuplot sleep_on_buffer 2491 ms
gnuplot sleep_on_buffer 1520 ms
gnuplot sleep_on_buffer 1209 ms
gnuplot sleep_on_buffer 1188 ms
gnuplot sleep_on_buffer 1654 ms
gnuplot sleep_on_buffer 1403 ms
gnuplot sleep_on_buffer 1386 ms
gnuplot sleep_on_buffer 1899 ms
gnuplot sleep_on_buffer 2673 ms
gnuplot sleep_on_buffer 2158 ms
gnuplot sleep_on_buffer 1780 ms
gnuplot sleep_on_buffer 1624 ms
gnuplot sleep_on_buffer 1704 ms
gnuplot sleep_on_buffer 2207 ms
gnuplot sleep_on_buffer 2557 ms
gnuplot sleep_on_buffer 1692 ms
gnuplot sleep_on_buffer 1686 ms
gnuplot sleep_on_buffer 1258 ms
offlineimap sleep_on_buffer 1217 ms
pool sleep_on_buffer 13434 ms
offlineimap sleep_on_buffer 30091 ms
offlineimap sleep_on_buffer 9048 ms
offlineimap sleep_on_buffer 13754 ms
offlineimap sleep_on_buffer 36560 ms
offlineimap sleep_on_buffer 1465 ms
cp sleep_on_buffer 1525 ms
cp sleep_on_buffer 2193 ms
DOM Worker sleep_on_buffer 5563 ms
DOM Worker sleep_on_buffer 3597 ms
cp sleep_on_buffer 1261 ms
git sleep_on_buffer 1427 ms
git sleep_on_buffer 1097 ms
git sleep_on_buffer 1232 ms
offlineimap sleep_on_buffer 5778 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff81180bd8>] lookup_open+0xc8/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 211510 ms
Event count: 20
flush-8:0 sleep_on_buffer 29387 ms
flush-8:0 sleep_on_buffer 2159 ms
flush-8:0 sleep_on_buffer 8593 ms
flush-8:0 sleep_on_buffer 3143 ms
flush-8:0 sleep_on_buffer 4641 ms
flush-8:0 sleep_on_buffer 17279 ms
flush-8:0 sleep_on_buffer 2210 ms
flush-8:0 sleep_on_buffer 15948 ms
flush-8:0 sleep_on_buffer 4686 ms
flush-8:0 sleep_on_buffer 7027 ms
flush-8:0 sleep_on_buffer 17871 ms
flush-8:0 sleep_on_buffer 3262 ms
flush-8:0 sleep_on_buffer 7311 ms
flush-8:0 sleep_on_buffer 11255 ms
flush-8:0 sleep_on_buffer 5693 ms
flush-8:0 sleep_on_buffer 8628 ms
flush-8:0 sleep_on_buffer 10917 ms
flush-8:0 sleep_on_buffer 17497 ms
flush-8:0 sleep_on_buffer 15750 ms
flush-8:0 sleep_on_buffer 18253 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d506>] ext4_split_extent_at+0xb6/0x390
[<ffffffff8121e038>] ext4_split_extent.isra.47+0x108/0x130
[<ffffffff8121e3ae>] ext4_ext_convert_to_initialized+0x15e/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 201192 ms
Event count: 23
imapd sleep_on_buffer 3770 ms
imapd sleep_on_buffer 37050 ms
make sleep_on_buffer 5342 ms
compare-mmtests sleep_on_buffer 1774 ms
scp sleep_on_buffer 2478 ms
scp sleep_on_buffer 2368 ms
imapd sleep_on_buffer 3163 ms
pool sleep_on_buffer 2033 ms
imapd sleep_on_buffer 1311 ms
imapd sleep_on_buffer 11011 ms
imapd sleep_on_buffer 1345 ms
imapd sleep_on_buffer 20545 ms
imapd sleep_on_buffer 19511 ms
imapd sleep_on_buffer 20863 ms
imapd sleep_on_buffer 32313 ms
imapd sleep_on_buffer 6984 ms
imapd sleep_on_buffer 8152 ms
imapd sleep_on_buffer 3038 ms
imapd sleep_on_buffer 8032 ms
imapd sleep_on_buffer 3649 ms
imapd sleep_on_buffer 2195 ms
imapd sleep_on_buffer 1848 ms
mv sleep_on_buffer 2417 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177bc6>] vfs_stat+0x16/0x20
[<ffffffff81177ce5>] sys_newstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 169878 ms
Event count: 56
git wait_on_page_bit 8573 ms
git wait_on_page_bit 2986 ms
git wait_on_page_bit 1811 ms
git wait_on_page_bit 2623 ms
git wait_on_page_bit 1419 ms
git wait_on_page_bit 1244 ms
git wait_on_page_bit 1134 ms
git wait_on_page_bit 5825 ms
git wait_on_page_bit 3567 ms
git wait_on_page_bit 1119 ms
git wait_on_page_bit 1375 ms
git wait_on_page_bit 3726 ms
git wait_on_page_bit 2670 ms
git wait_on_page_bit 4141 ms
git wait_on_page_bit 3858 ms
git wait_on_page_bit 6684 ms
git wait_on_page_bit 5355 ms
gen-report.sh wait_on_page_bit 4747 ms
git wait_on_page_bit 6752 ms
git wait_on_page_bit 1229 ms
git wait_on_page_bit 4409 ms
git wait_on_page_bit 3101 ms
git wait_on_page_bit 1817 ms
git wait_on_page_bit 1687 ms
git wait_on_page_bit 3683 ms
git wait_on_page_bit 2031 ms
git wait_on_page_bit 2138 ms
git wait_on_page_bit 1513 ms
git wait_on_page_bit 1804 ms
git wait_on_page_bit 2559 ms
git wait_on_page_bit 7958 ms
git wait_on_page_bit 6265 ms
git wait_on_page_bit 1261 ms
git wait_on_page_bit 4018 ms
git wait_on_page_bit 1450 ms
git wait_on_page_bit 1821 ms
git wait_on_page_bit 3186 ms
git wait_on_page_bit 1513 ms
git wait_on_page_bit 3215 ms
git wait_on_page_bit 1262 ms
git wait_on_page_bit 8188 ms
git sleep_on_buffer 1019 ms
git wait_on_page_bit 5233 ms
git wait_on_page_bit 1842 ms
git wait_on_page_bit 1378 ms
git wait_on_page_bit 1386 ms
git wait_on_page_bit 2016 ms
git wait_on_page_bit 1901 ms
git wait_on_page_bit 2750 ms
git sleep_on_buffer 1152 ms
git wait_on_page_bit 1169 ms
git wait_on_page_bit 1371 ms
git wait_on_page_bit 1916 ms
git wait_on_page_bit 1630 ms
git wait_on_page_bit 8286 ms
git wait_on_page_bit 1112 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff8111d620>] truncate_inode_pages+0x10/0x20
[<ffffffff8111d677>] truncate_pagecache+0x47/0x70
[<ffffffff811f2f4d>] ext4_setattr+0x17d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 167244 ms
Event count: 118
folder-markup.s sleep_on_buffer 2055 ms
folder-markup.s sleep_on_buffer 3917 ms
mv sleep_on_buffer 1025 ms
folder-markup.s sleep_on_buffer 1670 ms
folder-markup.s sleep_on_buffer 1144 ms
folder-markup.s sleep_on_buffer 1063 ms
folder-markup.s sleep_on_buffer 1385 ms
folder-markup.s sleep_on_buffer 1753 ms
folder-markup.s sleep_on_buffer 1351 ms
folder-markup.s sleep_on_buffer 1143 ms
folder-markup.s sleep_on_buffer 1581 ms
folder-markup.s sleep_on_buffer 1747 ms
folder-markup.s sleep_on_buffer 1241 ms
folder-markup.s sleep_on_buffer 1419 ms
folder-markup.s sleep_on_buffer 1429 ms
folder-markup.s sleep_on_buffer 1112 ms
git sleep_on_buffer 1190 ms
git sleep_on_buffer 1190 ms
git sleep_on_buffer 1050 ms
git sleep_on_buffer 1463 ms
git sleep_on_buffer 1376 ms
folder-markup.s sleep_on_buffer 1481 ms
folder-markup.s sleep_on_buffer 1424 ms
folder-markup.s sleep_on_buffer 1633 ms
folder-markup.s sleep_on_buffer 1012 ms
folder-markup.s sleep_on_buffer 1706 ms
folder-markup.s sleep_on_buffer 1246 ms
folder-markup.s sleep_on_buffer 1275 ms
git sleep_on_buffer 1484 ms
git sleep_on_buffer 1216 ms
git sleep_on_buffer 1065 ms
git sleep_on_buffer 1455 ms
folder-markup.s sleep_on_buffer 1063 ms
folder-markup.s sleep_on_buffer 3059 ms
folder-markup.s sleep_on_buffer 1140 ms
folder-markup.s sleep_on_buffer 1353 ms
mv sleep_on_buffer 1050 ms
folder-markup.s sleep_on_buffer 1209 ms
git sleep_on_buffer 1341 ms
scp sleep_on_buffer 4975 ms
folder-markup.s sleep_on_buffer 1743 ms
folder-markup.s sleep_on_buffer 1280 ms
folder-markup.s sleep_on_buffer 2140 ms
folder-markup.s sleep_on_buffer 1138 ms
folder-markup.s sleep_on_buffer 1140 ms
folder-markup.s sleep_on_buffer 1162 ms
folder-markup.s sleep_on_buffer 1023 ms
git sleep_on_buffer 2174 ms
git sleep_on_buffer 1306 ms
git sleep_on_buffer 1224 ms
git sleep_on_buffer 1359 ms
git sleep_on_buffer 1551 ms
git sleep_on_buffer 1068 ms
git sleep_on_buffer 1367 ms
git sleep_on_buffer 1292 ms
git sleep_on_buffer 1369 ms
git sleep_on_buffer 1554 ms
git sleep_on_buffer 1273 ms
git sleep_on_buffer 1365 ms
mv sleep_on_buffer 1107 ms
folder-markup.s sleep_on_buffer 1519 ms
folder-markup.s sleep_on_buffer 1253 ms
folder-markup.s sleep_on_buffer 1195 ms
mv sleep_on_buffer 1091 ms
git sleep_on_buffer 1147 ms
git sleep_on_buffer 1271 ms
git sleep_on_buffer 1056 ms
git sleep_on_buffer 1134 ms
git sleep_on_buffer 1252 ms
git sleep_on_buffer 1352 ms
git sleep_on_buffer 1449 ms
folder-markup.s sleep_on_buffer 1732 ms
folder-markup.s sleep_on_buffer 1332 ms
folder-markup.s sleep_on_buffer 1450 ms
git sleep_on_buffer 1102 ms
git sleep_on_buffer 1771 ms
git sleep_on_buffer 1225 ms
git sleep_on_buffer 1089 ms
git sleep_on_buffer 1083 ms
folder-markup.s sleep_on_buffer 1071 ms
folder-markup.s sleep_on_buffer 1186 ms
folder-markup.s sleep_on_buffer 1170 ms
git sleep_on_buffer 1249 ms
git sleep_on_buffer 1255 ms
folder-markup.s sleep_on_buffer 1563 ms
folder-markup.s sleep_on_buffer 1258 ms
git sleep_on_buffer 2066 ms
git sleep_on_buffer 1493 ms
git sleep_on_buffer 1515 ms
git sleep_on_buffer 1380 ms
git sleep_on_buffer 1238 ms
git sleep_on_buffer 1393 ms
git sleep_on_buffer 1040 ms
git sleep_on_buffer 1986 ms
git sleep_on_buffer 1293 ms
git sleep_on_buffer 1209 ms
git sleep_on_buffer 1098 ms
git sleep_on_buffer 1091 ms
git sleep_on_buffer 1701 ms
git sleep_on_buffer 2237 ms
git sleep_on_buffer 1810 ms
folder-markup.s sleep_on_buffer 1166 ms
folder-markup.s sleep_on_buffer 2064 ms
folder-markup.s sleep_on_buffer 1285 ms
folder-markup.s sleep_on_buffer 1129 ms
folder-markup.s sleep_on_buffer 1080 ms
git sleep_on_buffer 1277 ms
git sleep_on_buffer 1280 ms
folder-markup.s sleep_on_buffer 1298 ms
folder-markup.s sleep_on_buffer 1355 ms
folder-markup.s sleep_on_buffer 1043 ms
folder-markup.s sleep_on_buffer 1204 ms
git sleep_on_buffer 1068 ms
git sleep_on_buffer 1654 ms
git sleep_on_buffer 1380 ms
git sleep_on_buffer 1289 ms
git sleep_on_buffer 1442 ms
git sleep_on_buffer 1299 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9dd2>] ext4_dx_add_entry+0xc2/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fa9b6>] ext4_add_nondir+0x26/0x80
[<ffffffff811fac9f>] ext4_create+0xff/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 135113 ms
Event count: 116
flush-8:16 get_request 1274 ms
flush-8:16 get_request 1098 ms
flush-8:16 get_request 1079 ms
flush-8:16 get_request 1234 ms
flush-8:16 get_request 1229 ms
flush-8:16 get_request 1056 ms
flush-8:16 get_request 1096 ms
flush-8:16 get_request 1092 ms
flush-8:16 get_request 1099 ms
flush-8:16 get_request 1057 ms
flush-8:16 get_request 1103 ms
flush-8:16 get_request 1207 ms
flush-8:16 get_request 1087 ms
flush-8:16 get_request 1060 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1196 ms
flush-8:16 get_request 1453 ms
flush-8:16 get_request 1084 ms
flush-8:16 get_request 1051 ms
flush-8:16 get_request 1084 ms
flush-8:16 get_request 1132 ms
flush-8:16 get_request 1164 ms
flush-8:16 get_request 1063 ms
flush-8:16 get_request 1221 ms
flush-8:16 get_request 1074 ms
flush-8:16 get_request 1099 ms
flush-8:16 get_request 1077 ms
flush-8:16 get_request 1243 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1078 ms
flush-8:16 get_request 1101 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1056 ms
flush-8:16 get_request 1333 ms
flush-8:16 get_request 1103 ms
flush-8:16 get_request 1216 ms
flush-8:16 get_request 1108 ms
flush-8:16 get_request 1109 ms
flush-8:16 get_request 1113 ms
flush-8:16 get_request 1349 ms
flush-8:16 get_request 1086 ms
flush-8:16 get_request 1070 ms
flush-8:16 get_request 1064 ms
flush-8:16 get_request 1091 ms
flush-8:16 get_request 1064 ms
flush-8:16 get_request 1222 ms
flush-8:16 get_request 1103 ms
flush-8:16 get_request 1434 ms
flush-8:16 get_request 1124 ms
flush-8:16 get_request 1359 ms
flush-8:16 get_request 1060 ms
flush-8:16 get_request 1057 ms
flush-8:16 get_request 1066 ms
flush-8:16 get_request 1357 ms
flush-8:16 get_request 1089 ms
flush-8:16 get_request 1071 ms
flush-8:16 get_request 1196 ms
flush-8:16 get_request 1091 ms
flush-8:16 get_request 1203 ms
flush-8:16 get_request 1100 ms
flush-8:16 get_request 1208 ms
flush-8:16 get_request 1113 ms
flush-8:16 get_request 1260 ms
flush-8:16 get_request 1480 ms
flush-8:16 get_request 1054 ms
flush-8:16 get_request 1211 ms
flush-8:16 get_request 1101 ms
flush-8:16 get_request 1098 ms
flush-8:16 get_request 1190 ms
flush-8:16 get_request 1046 ms
flush-8:16 get_request 1066 ms
flush-8:16 get_request 1204 ms
flush-8:16 get_request 1076 ms
flush-8:16 get_request 1094 ms
flush-8:16 get_request 1094 ms
flush-8:16 get_request 1081 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1193 ms
flush-8:16 get_request 1066 ms
flush-8:16 get_request 1069 ms
flush-8:16 get_request 1081 ms
flush-8:16 get_request 1107 ms
flush-8:16 get_request 1375 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1068 ms
flush-8:16 get_request 1077 ms
flush-8:16 get_request 1108 ms
flush-8:16 get_request 1080 ms
flush-8:16 get_request 1098 ms
flush-8:16 get_request 1063 ms
flush-8:16 get_request 1074 ms
flush-8:16 get_request 1072 ms
flush-8:16 get_request 1038 ms
flush-8:16 get_request 1058 ms
flush-8:16 get_request 1202 ms
flush-8:16 get_request 1359 ms
flush-8:16 get_request 1190 ms
flush-8:16 get_request 1497 ms
flush-8:16 get_request 2173 ms
flush-8:16 get_request 1199 ms
flush-8:16 get_request 1358 ms
flush-8:16 get_request 1384 ms
flush-8:16 get_request 1355 ms
flush-8:16 get_request 1327 ms
flush-8:16 get_request 1312 ms
flush-8:16 get_request 1318 ms
flush-8:16 get_request 1093 ms
flush-8:16 get_request 1265 ms
flush-8:16 get_request 1155 ms
flush-8:16 get_request 1107 ms
flush-8:16 get_request 1263 ms
flush-8:16 get_request 1104 ms
flush-8:16 get_request 1122 ms
flush-8:16 get_request 1578 ms
flush-8:16 get_request 1089 ms
flush-8:16 get_request 1075 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff812a4b1f>] generic_make_request.part.59+0x6f/0xa0
[<ffffffff812a5050>] generic_make_request+0x60/0x70
[<ffffffff812a50c7>] submit_bio+0x67/0x130
[<ffffffff811f6014>] ext4_io_submit+0x24/0x60
[<ffffffff811f2265>] ext4_writepage+0x135/0x220
[<ffffffff81119292>] __writepage+0x12/0x40
[<ffffffff81119a96>] write_cache_pages+0x206/0x460
[<ffffffff81119d35>] generic_writepages+0x45/0x70
[<ffffffff8111ac15>] do_writepages+0x25/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 115939 ms
Event count: 23
bash sleep_on_buffer 3076 ms
du sleep_on_buffer 2364 ms
du sleep_on_buffer 1515 ms
git sleep_on_buffer 1706 ms
rm sleep_on_buffer 10595 ms
find sleep_on_buffer 2048 ms
rm sleep_on_buffer 9146 ms
rm sleep_on_buffer 8220 ms
rm sleep_on_buffer 6080 ms
cp sleep_on_buffer 6302 ms
ls sleep_on_buffer 1225 ms
cp sleep_on_buffer 6279 ms
cp sleep_on_buffer 1164 ms
cp sleep_on_buffer 3365 ms
cp sleep_on_buffer 2191 ms
cp sleep_on_buffer 1367 ms
du sleep_on_buffer 4155 ms
cp sleep_on_buffer 3906 ms
cp sleep_on_buffer 4758 ms
rsync sleep_on_buffer 6575 ms
git sleep_on_buffer 1688 ms
git sleep_on_buffer 26470 ms
git sleep_on_buffer 1744 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff811849b2>] vfs_readdir+0xc2/0xe0
[<ffffffff81184ae9>] sys_getdents+0x89/0x110
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 101122 ms
Event count: 18
flush-8:0 sleep_on_buffer 21732 ms
flush-8:0 sleep_on_buffer 2211 ms
flush-8:0 sleep_on_buffer 1480 ms
flush-8:0 sleep_on_buffer 16292 ms
flush-8:0 sleep_on_buffer 2975 ms
flush-8:0 sleep_on_buffer 7025 ms
flush-8:0 sleep_on_buffer 5535 ms
flush-8:0 sleep_on_buffer 1885 ms
flush-8:0 sleep_on_buffer 1329 ms
flush-8:0 sleep_on_buffer 1374 ms
flush-8:0 sleep_on_buffer 1490 ms
flush-8:0 sleep_on_buffer 16341 ms
flush-8:0 sleep_on_buffer 14939 ms
flush-8:0 sleep_on_buffer 1202 ms
flush-8:0 sleep_on_buffer 1262 ms
flush-8:0 sleep_on_buffer 1121 ms
flush-8:0 sleep_on_buffer 1571 ms
flush-8:0 sleep_on_buffer 1358 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d506>] ext4_split_extent_at+0xb6/0x390
[<ffffffff8121e038>] ext4_split_extent.isra.47+0x108/0x130
[<ffffffff8121e3ae>] ext4_ext_convert_to_initialized+0x15e/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 98613 ms
Event count: 8
git sleep_on_buffer 14529 ms
git sleep_on_buffer 4477 ms
git sleep_on_buffer 10045 ms
git sleep_on_buffer 11068 ms
git sleep_on_buffer 18777 ms
git sleep_on_buffer 9434 ms
git sleep_on_buffer 12262 ms
git sleep_on_buffer 18021 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f4e95>] ext4_evict_inode+0x1e5/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 94944 ms
Event count: 11
git sleep_on_buffer 16110 ms
git sleep_on_buffer 6508 ms
git sleep_on_buffer 23186 ms
git sleep_on_buffer 25228 ms
git-merge sleep_on_buffer 1672 ms
konqueror sleep_on_buffer 1411 ms
git sleep_on_buffer 1803 ms
git sleep_on_buffer 15397 ms
git sleep_on_buffer 1276 ms
git sleep_on_buffer 1012 ms
git sleep_on_buffer 1341 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811fc3e1>] ext4_unlink+0x41/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 93658 ms
Event count: 26
flush-8:0 sleep_on_buffer 1294 ms
flush-8:0 sleep_on_buffer 2856 ms
flush-8:0 sleep_on_buffer 3764 ms
flush-8:0 sleep_on_buffer 5086 ms
flush-8:0 sleep_on_buffer 1203 ms
flush-8:0 sleep_on_buffer 1289 ms
flush-8:0 sleep_on_buffer 1264 ms
flush-8:0 sleep_on_buffer 1252 ms
flush-8:0 sleep_on_buffer 2997 ms
flush-8:0 sleep_on_buffer 2765 ms
flush-8:0 sleep_on_buffer 4235 ms
flush-8:0 sleep_on_buffer 5205 ms
flush-8:0 sleep_on_buffer 6971 ms
flush-8:0 sleep_on_buffer 4155 ms
ps wait_on_page_bit_killable 1054 ms
flush-8:0 sleep_on_buffer 3719 ms
flush-8:0 sleep_on_buffer 10283 ms
flush-8:0 sleep_on_buffer 3068 ms
flush-8:0 sleep_on_buffer 2000 ms
flush-8:0 sleep_on_buffer 2264 ms
flush-8:0 sleep_on_buffer 3623 ms
flush-8:0 sleep_on_buffer 12954 ms
flush-8:0 sleep_on_buffer 6579 ms
flush-8:0 sleep_on_buffer 1245 ms
flush-8:0 sleep_on_buffer 1293 ms
flush-8:0 sleep_on_buffer 1240 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 92353 ms
Event count: 11
flush-8:0 sleep_on_buffer 2192 ms
flush-8:0 sleep_on_buffer 2088 ms
flush-8:0 sleep_on_buffer 1460 ms
flush-8:0 sleep_on_buffer 1241 ms
flush-8:0 sleep_on_buffer 1986 ms
flush-8:0 sleep_on_buffer 1331 ms
flush-8:0 sleep_on_buffer 2192 ms
flush-8:0 sleep_on_buffer 3327 ms
flush-8:0 sleep_on_buffer 73408 ms
flush-8:0 sleep_on_buffer 1229 ms
flush-253:0 sleep_on_buffer 1899 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 91515 ms
Event count: 7
flush-8:0 sleep_on_buffer 7128 ms
flush-8:0 sleep_on_buffer 18731 ms
flush-8:0 sleep_on_buffer 12643 ms
flush-8:0 sleep_on_buffer 28149 ms
flush-8:0 sleep_on_buffer 5728 ms
flush-8:0 sleep_on_buffer 18040 ms
git wait_on_page_bit 1096 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121e658>] ext4_ext_convert_to_initialized+0x408/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4601>] write_cache_pages_da+0x421/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 86251 ms
Event count: 76
imapd wait_on_page_bit_killable 1088 ms
imapd wait_on_page_bit_killable 1092 ms
git wait_on_page_bit_killable 1616 ms
git wait_on_page_bit_killable 1114 ms
play wait_on_page_bit_killable 1019 ms
play wait_on_page_bit_killable 1012 ms
play wait_on_page_bit_killable 1223 ms
play wait_on_page_bit_killable 1223 ms
play wait_on_page_bit_killable 1034 ms
play wait_on_page_bit_killable 1034 ms
play wait_on_page_bit_killable 1096 ms
play wait_on_page_bit_killable 1096 ms
play wait_on_page_bit_killable 1093 ms
play wait_on_page_bit_killable 1093 ms
vim wait_on_page_bit_killable 1084 ms
dbus-daemon-lau wait_on_page_bit_killable 1076 ms
play wait_on_page_bit_killable 1097 ms
play wait_on_page_bit_killable 1097 ms
git wait_on_page_bit_killable 1005 ms
systemd-journal wait_on_page_bit_killable 1252 ms
systemd-journal wait_on_page_bit_killable 1158 ms
git wait_on_page_bit_killable 1237 ms
git wait_on_page_bit_killable 1043 ms
git wait_on_page_bit_killable 1068 ms
git wait_on_page_bit_killable 1070 ms
git wait_on_page_bit_killable 1070 ms
git wait_on_page_bit_killable 1097 ms
git wait_on_page_bit_killable 1055 ms
git wait_on_page_bit_killable 1252 ms
git wait_on_page_bit_killable 1187 ms
git wait_on_page_bit_killable 1069 ms
git wait_on_page_bit_killable 1194 ms
git wait_on_page_bit_killable 1035 ms
git wait_on_page_bit_killable 1046 ms
git wait_on_page_bit_killable 1024 ms
git wait_on_page_bit_killable 1124 ms
git wait_on_page_bit_killable 1293 ms
git wait_on_page_bit_killable 1184 ms
git wait_on_page_bit_killable 1269 ms
git wait_on_page_bit_killable 1268 ms
git wait_on_page_bit_killable 1088 ms
git wait_on_page_bit_killable 1093 ms
git wait_on_page_bit_killable 1013 ms
git wait_on_page_bit_killable 1034 ms
git wait_on_page_bit_killable 1018 ms
git wait_on_page_bit_killable 1185 ms
git wait_on_page_bit_killable 1258 ms
git wait_on_page_bit_killable 1006 ms
git wait_on_page_bit_killable 1061 ms
git wait_on_page_bit_killable 1108 ms
git wait_on_page_bit_killable 1006 ms
git wait_on_page_bit_killable 1012 ms
git wait_on_page_bit_killable 1210 ms
git wait_on_page_bit_killable 1239 ms
git wait_on_page_bit_killable 1146 ms
git wait_on_page_bit_killable 1106 ms
git wait_on_page_bit_killable 1063 ms
git wait_on_page_bit_killable 1070 ms
git wait_on_page_bit_killable 1041 ms
git wait_on_page_bit_killable 1052 ms
git wait_on_page_bit_killable 1237 ms
git wait_on_page_bit_killable 1117 ms
git wait_on_page_bit_killable 1086 ms
git wait_on_page_bit_killable 1051 ms
git wait_on_page_bit_killable 1029 ms
runlevel wait_on_page_bit_killable 1019 ms
evolution wait_on_page_bit_killable 1384 ms
evolution wait_on_page_bit_killable 1144 ms
firefox wait_on_page_bit_killable 1537 ms
git wait_on_page_bit_killable 1017 ms
evolution wait_on_page_bit_killable 1015 ms
evolution wait_on_page_bit_killable 1523 ms
ps wait_on_page_bit_killable 1394 ms
kio_http wait_on_page_bit_killable 1010 ms
plugin-containe wait_on_page_bit_killable 1522 ms
qmmp wait_on_page_bit_killable 1170 ms
[<ffffffff811115c8>] wait_on_page_bit_killable+0x78/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff81111a78>] filemap_fault+0x3d8/0x410
[<ffffffff8113599a>] __do_fault+0x6a/0x530
[<ffffffff811394be>] handle_pte_fault+0xee/0x200
[<ffffffff8113a731>] handle_mm_fault+0x271/0x390
[<ffffffff81597959>] __do_page_fault+0x169/0x520
[<ffffffff81597d19>] do_page_fault+0x9/0x10
[<ffffffff81594488>] page_fault+0x28/0x30
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 78888 ms
Event count: 10
git sleep_on_buffer 1019 ms
git sleep_on_buffer 2031 ms
git sleep_on_buffer 2109 ms
git sleep_on_buffer 5858 ms
git sleep_on_buffer 15181 ms
git sleep_on_buffer 22771 ms
git sleep_on_buffer 2331 ms
git sleep_on_buffer 1341 ms
git sleep_on_buffer 24648 ms
git sleep_on_buffer 1599 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb052>] ext4_delete_entry+0x62/0x120
[<ffffffff811fc495>] ext4_unlink+0xf5/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 77568 ms
Event count: 12
git sleep_on_buffer 2592 ms
git sleep_on_buffer 1312 ms
git sleep_on_buffer 1974 ms
git sleep_on_buffer 2508 ms
git sleep_on_buffer 1245 ms
git sleep_on_buffer 20990 ms
git sleep_on_buffer 14782 ms
git sleep_on_buffer 2026 ms
git sleep_on_buffer 1880 ms
git sleep_on_buffer 2174 ms
git sleep_on_buffer 24451 ms
git sleep_on_buffer 1634 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811fc633>] ext4_unlink+0x293/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 73950 ms
Event count: 21
pool wait_on_page_bit 5626 ms
git sleep_on_buffer 1077 ms
pool wait_on_page_bit 1040 ms
offlineimap wait_on_page_bit 1083 ms
pool wait_on_page_bit 1044 ms
pool wait_on_page_bit 7268 ms
pool wait_on_page_bit 9900 ms
pool wait_on_page_bit 3530 ms
offlineimap wait_on_page_bit 18212 ms
git wait_on_page_bit 1101 ms
git wait_on_page_bit 1402 ms
git sleep_on_buffer 1037 ms
pool wait_on_page_bit 1107 ms
git sleep_on_buffer 1106 ms
pool wait_on_page_bit 11643 ms
pool wait_on_page_bit 1272 ms
evolution wait_on_page_bit 1471 ms
pool wait_on_page_bit 1458 ms
pool wait_on_page_bit 1331 ms
git sleep_on_buffer 1082 ms
offlineimap wait_on_page_bit 1160 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff81110c50>] filemap_write_and_wait_range+0x60/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 70700 ms
Event count: 27
flush-8:0 sleep_on_buffer 1735 ms
flush-8:0 sleep_on_buffer 1720 ms
flush-8:0 sleep_on_buffer 3099 ms
flush-8:0 sleep_on_buffer 1321 ms
flush-8:0 sleep_on_buffer 3276 ms
flush-8:0 sleep_on_buffer 4215 ms
flush-8:0 sleep_on_buffer 1412 ms
flush-8:0 sleep_on_buffer 1049 ms
flush-8:0 sleep_on_buffer 2320 ms
flush-8:0 sleep_on_buffer 8076 ms
flush-8:0 sleep_on_buffer 2210 ms
flush-8:0 sleep_on_buffer 1204 ms
flush-8:0 sleep_on_buffer 1262 ms
flush-8:0 sleep_on_buffer 1995 ms
flush-8:0 sleep_on_buffer 1675 ms
flush-8:0 sleep_on_buffer 4219 ms
flush-8:0 sleep_on_buffer 4027 ms
flush-8:0 sleep_on_buffer 3452 ms
flush-8:0 sleep_on_buffer 6020 ms
flush-8:0 sleep_on_buffer 1318 ms
flush-8:0 sleep_on_buffer 1065 ms
flush-8:0 sleep_on_buffer 1148 ms
flush-8:0 sleep_on_buffer 1230 ms
flush-8:0 sleep_on_buffer 4479 ms
flush-8:0 sleep_on_buffer 1580 ms
flush-8:0 sleep_on_buffer 4551 ms
git sleep_on_buffer 1042 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121a97b>] __ext4_ext_dirty.isra.40+0x7b/0x80
[<ffffffff8121d34a>] ext4_ext_insert_extent+0x31a/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 59807 ms
Event count: 37
mv sleep_on_buffer 1439 ms
mv sleep_on_buffer 1490 ms
mv sleep_on_buffer 1876 ms
mv sleep_on_buffer 1240 ms
mv sleep_on_buffer 1897 ms
mv sleep_on_buffer 2089 ms
mv sleep_on_buffer 1375 ms
mv sleep_on_buffer 1386 ms
mv sleep_on_buffer 1442 ms
mv sleep_on_buffer 1682 ms
mv sleep_on_buffer 1188 ms
offlineimap sleep_on_buffer 2247 ms
mv sleep_on_buffer 1262 ms
mv sleep_on_buffer 8930 ms
mv sleep_on_buffer 1392 ms
mv sleep_on_buffer 1536 ms
mv sleep_on_buffer 1064 ms
mv sleep_on_buffer 1303 ms
mv sleep_on_buffer 1487 ms
mv sleep_on_buffer 1331 ms
mv sleep_on_buffer 1757 ms
mv sleep_on_buffer 1069 ms
mv sleep_on_buffer 1183 ms
mv sleep_on_buffer 1548 ms
mv sleep_on_buffer 1090 ms
mv sleep_on_buffer 1770 ms
mv sleep_on_buffer 1002 ms
mv sleep_on_buffer 1199 ms
mv sleep_on_buffer 1066 ms
mv sleep_on_buffer 1275 ms
mv sleep_on_buffer 1198 ms
mv sleep_on_buffer 1653 ms
mv sleep_on_buffer 1197 ms
mv sleep_on_buffer 1275 ms
mv sleep_on_buffer 1317 ms
mv sleep_on_buffer 1025 ms
mv sleep_on_buffer 1527 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fba26>] ext4_rename+0x276/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 59307 ms
Event count: 15
git sleep_on_buffer 3293 ms
git sleep_on_buffer 1350 ms
git sleep_on_buffer 2132 ms
git sleep_on_buffer 1018 ms
git sleep_on_buffer 16069 ms
git sleep_on_buffer 5478 ms
offlineimap sleep_on_buffer 1138 ms
imapd sleep_on_buffer 1927 ms
imapd sleep_on_buffer 6417 ms
offlineimap sleep_on_buffer 6241 ms
offlineimap sleep_on_buffer 1549 ms
rsync sleep_on_buffer 3776 ms
rsync sleep_on_buffer 2516 ms
git sleep_on_buffer 1025 ms
git sleep_on_buffer 5378 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121c4ad>] ext4_ext_tree_init+0x2d/0x40
[<ffffffff811ecc06>] __ext4_new_inode+0x1076/0x10c0
[<ffffffff811fac5b>] ext4_create+0xbb/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 58651 ms
Event count: 4
git sleep_on_buffer 13070 ms
git sleep_on_buffer 18222 ms
git sleep_on_buffer 13508 ms
git sleep_on_buffer 13851 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fc898>] ext4_orphan_del+0x1a8/0x1e0
[<ffffffff811f4fbb>] ext4_evict_inode+0x30b/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 56275 ms
Event count: 14
git sleep_on_buffer 1116 ms
git sleep_on_buffer 1347 ms
git sleep_on_buffer 1258 ms
git sleep_on_buffer 3471 ms
git sleep_on_buffer 3348 ms
git sleep_on_buffer 1185 ms
git sleep_on_buffer 1423 ms
git sleep_on_buffer 2662 ms
git sleep_on_buffer 8693 ms
git sleep_on_buffer 8223 ms
git sleep_on_buffer 4792 ms
git sleep_on_buffer 2553 ms
git sleep_on_buffer 2550 ms
git sleep_on_buffer 13654 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f9c6e>] add_dirent_to_buf+0x12e/0x1d0
[<ffffffff811f9e38>] ext4_dx_add_entry+0x128/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fa9b6>] ext4_add_nondir+0x26/0x80
[<ffffffff811fac9f>] ext4_create+0xff/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 55128 ms
Event count: 12
dconf-service sleep_on_buffer 1918 ms
pool sleep_on_buffer 10558 ms
pool sleep_on_buffer 1957 ms
pool sleep_on_buffer 1903 ms
pool sleep_on_buffer 1187 ms
offlineimap sleep_on_buffer 2077 ms
URL Classifier sleep_on_buffer 3924 ms
offlineimap sleep_on_buffer 2573 ms
StreamT~ns #343 sleep_on_buffer 11686 ms
DOM Worker sleep_on_buffer 2215 ms
pool sleep_on_buffer 4513 ms
offlineimap sleep_on_buffer 10617 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 53464 ms
Event count: 4
play sleep_on_buffer 6853 ms
play sleep_on_buffer 15340 ms
play sleep_on_buffer 24793 ms
play sleep_on_buffer 6478 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811f313c>] ext4_setattr+0x36c/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811712bd>] chown_common+0xbd/0xd0
[<ffffffff81172417>] sys_fchown+0xb7/0xd0
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 51867 ms
Event count: 3
flush-8:0 sleep_on_buffer 42842 ms
flush-8:0 sleep_on_buffer 2026 ms
flush-8:0 sleep_on_buffer 6999 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 49716 ms
Event count: 8
pool sleep_on_buffer 4642 ms
offlineimap sleep_on_buffer 4279 ms
evolution sleep_on_buffer 5182 ms
rsync sleep_on_buffer 5599 ms
git sleep_on_buffer 8338 ms
StreamT~ns #343 sleep_on_buffer 2216 ms
git sleep_on_buffer 2844 ms
git sleep_on_buffer 16616 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812291ba>] ext4_free_blocks+0x36a/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f513e>] ext4_evict_inode+0x48e/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 42396 ms
Event count: 5
git sleep_on_buffer 1115 ms
git sleep_on_buffer 15407 ms
git sleep_on_buffer 9114 ms
git sleep_on_buffer 1076 ms
git sleep_on_buffer 15684 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f4e95>] ext4_evict_inode+0x1e5/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 41836 ms
Event count: 29
git sleep_on_buffer 1326 ms
git sleep_on_buffer 1017 ms
git sleep_on_buffer 1077 ms
git sleep_on_buffer 2618 ms
git sleep_on_buffer 1058 ms
git sleep_on_buffer 1321 ms
git sleep_on_buffer 1199 ms
git sleep_on_buffer 1067 ms
git sleep_on_buffer 1227 ms
git sleep_on_buffer 1101 ms
git sleep_on_buffer 1105 ms
git sleep_on_buffer 1048 ms
git sleep_on_buffer 1254 ms
git sleep_on_buffer 1866 ms
git sleep_on_buffer 1768 ms
git sleep_on_buffer 1613 ms
git sleep_on_buffer 1690 ms
git sleep_on_buffer 1189 ms
git sleep_on_buffer 1063 ms
git sleep_on_buffer 1022 ms
git sleep_on_buffer 2039 ms
git sleep_on_buffer 1898 ms
git sleep_on_buffer 1422 ms
git sleep_on_buffer 1678 ms
git sleep_on_buffer 1285 ms
git sleep_on_buffer 2058 ms
git sleep_on_buffer 1336 ms
git sleep_on_buffer 1364 ms
git sleep_on_buffer 2127 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9dd2>] ext4_dx_add_entry+0xc2/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fae2c>] ext4_link+0xfc/0x1b0
[<ffffffff81181e33>] vfs_link+0x113/0x1c0
[<ffffffff811828a4>] sys_linkat+0x174/0x1c0
[<ffffffff81182909>] sys_link+0x19/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 41493 ms
Event count: 2
flush-8:0 sleep_on_buffer 28180 ms
flush-8:0 sleep_on_buffer 13313 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d506>] ext4_split_extent_at+0xb6/0x390
[<ffffffff8121e038>] ext4_split_extent.isra.47+0x108/0x130
[<ffffffff8121e3ae>] ext4_ext_convert_to_initialized+0x15e/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4601>] write_cache_pages_da+0x421/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 40644 ms
Event count: 30
flush-8:16 get_request 1797 ms
flush-8:16 get_request 1334 ms
flush-8:16 get_request 1288 ms
flush-8:16 get_request 1741 ms
flush-8:16 get_request 2518 ms
flush-8:16 get_request 1752 ms
flush-8:16 get_request 1069 ms
flush-8:16 get_request 1487 ms
flush-8:16 get_request 1000 ms
flush-8:16 get_request 1270 ms
flush-8:16 get_request 1223 ms
flush-8:16 get_request 1384 ms
flush-8:16 get_request 1082 ms
flush-8:16 get_request 1195 ms
flush-8:16 get_request 1163 ms
flush-8:16 get_request 1605 ms
flush-8:16 get_request 1110 ms
flush-8:16 get_request 1249 ms
flush-8:16 get_request 2064 ms
flush-8:16 get_request 1073 ms
flush-8:16 get_request 1238 ms
flush-8:16 get_request 1215 ms
flush-8:16 get_request 1075 ms
flush-8:16 get_request 1532 ms
flush-8:16 get_request 1586 ms
flush-8:16 get_request 1165 ms
flush-8:16 get_request 1129 ms
flush-8:16 get_request 1098 ms
flush-8:16 get_request 1099 ms
flush-8:16 get_request 1103 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff812a4b1f>] generic_make_request.part.59+0x6f/0xa0
[<ffffffff812a5050>] generic_make_request+0x60/0x70
[<ffffffff812a50c7>] submit_bio+0x67/0x130
[<ffffffff811f6014>] ext4_io_submit+0x24/0x60
[<ffffffff811f2265>] ext4_writepage+0x135/0x220
[<ffffffff81119292>] __writepage+0x12/0x40
[<ffffffff81119a96>] write_cache_pages+0x206/0x460
[<ffffffff81119d35>] generic_writepages+0x45/0x70
[<ffffffff8111ac15>] do_writepages+0x25/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 39571 ms
Event count: 8
kio_http sleep_on_buffer 23133 ms
vi sleep_on_buffer 4288 ms
git sleep_on_buffer 1410 ms
mutt sleep_on_buffer 2302 ms
mutt sleep_on_buffer 2299 ms
Cache I/O sleep_on_buffer 1283 ms
gpg sleep_on_buffer 3265 ms
git sleep_on_buffer 1591 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb67b>] ext4_orphan_add+0xbb/0x1f0
[<ffffffff811fc6cb>] ext4_unlink+0x32b/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 38769 ms
Event count: 6
rsync sleep_on_buffer 3513 ms
rsync sleep_on_buffer 3570 ms
git sleep_on_buffer 26211 ms
git sleep_on_buffer 1657 ms
git sleep_on_buffer 2184 ms
git sleep_on_buffer 1634 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811fbd0c>] ext4_rename+0x55c/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 34812 ms
Event count: 4
acroread wait_on_page_bit 11968 ms
acroread wait_on_page_bit 7121 ms
acroread wait_on_page_bit 3126 ms
acroread wait_on_page_bit 12597 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff8111d620>] truncate_inode_pages+0x10/0x20
[<ffffffff8111d677>] truncate_pagecache+0x47/0x70
[<ffffffff811f2f4d>] ext4_setattr+0x17d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff811c2996>] compat_sys_open+0x16/0x20
[<ffffffff8159d81c>] sysenter_dispatch+0x7/0x21
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 34740 ms
Event count: 4
systemd-journal sleep_on_buffer 1126 ms
systemd-journal sleep_on_buffer 29206 ms
systemd-journal sleep_on_buffer 1787 ms
systemd-journal sleep_on_buffer 2621 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff8121fd1f>] ext4_fallocate+0x1cf/0x420
[<ffffffff81171b32>] do_fallocate+0x112/0x190
[<ffffffff81171c02>] sys_fallocate+0x52/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 33158 ms
Event count: 32
mv sleep_on_buffer 1043 ms
git wait_on_page_bit 1150 ms
cc1 sleep_on_buffer 1062 ms
git wait_on_page_bit 1055 ms
flush-8:16 get_request 1091 ms
mktexlsr sleep_on_buffer 1152 ms
imapd sleep_on_buffer 1004 ms
flush-8:16 get_request 1087 ms
flush-8:16 get_request 1104 ms
sleep wait_on_page_bit_killable 1142 ms
git wait_on_page_bit_killable 1108 ms
git wait_on_page_bit_killable 1007 ms
git wait_on_page_bit_killable 1074 ms
git wait_on_page_bit_killable 1050 ms
nm-dhcp-client. wait_on_page_bit_killable 1069 ms
uname wait_on_page_bit_killable 1086 ms
sed wait_on_page_bit_killable 1101 ms
git wait_on_page_bit_killable 1057 ms
grep wait_on_page_bit_killable 1045 ms
imapd sleep_on_buffer 1032 ms
git sleep_on_buffer 1015 ms
folder-markup.s sleep_on_buffer 1048 ms
git wait_on_page_bit 1086 ms
git sleep_on_buffer 1041 ms
git sleep_on_buffer 1048 ms
git wait_on_page_bit 1063 ms
git sleep_on_buffer 1083 ms
series2git sleep_on_buffer 1073 ms
git wait_on_page_bit 1093 ms
git wait_on_page_bit 1071 ms
git wait_on_page_bit 1018 ms

Time stalled in this event: 32109 ms
Event count: 23
flush-8:16 get_request 1475 ms
flush-8:16 get_request 1431 ms
flush-8:16 get_request 1027 ms
flush-8:16 get_request 2019 ms
flush-8:16 get_request 1021 ms
flush-8:16 get_request 1013 ms
flush-8:16 get_request 1093 ms
flush-8:16 get_request 1178 ms
flush-8:16 get_request 1051 ms
flush-8:16 get_request 1296 ms
flush-8:16 get_request 1525 ms
flush-8:16 get_request 1083 ms
flush-8:16 get_request 1654 ms
flush-8:16 get_request 1583 ms
flush-8:16 get_request 1405 ms
flush-8:16 get_request 2004 ms
flush-8:16 get_request 2203 ms
flush-8:16 get_request 1980 ms
flush-8:16 get_request 1211 ms
flush-8:16 get_request 1116 ms
flush-8:16 get_request 1071 ms
flush-8:16 get_request 1255 ms
flush-8:16 get_request 1415 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff812a4b1f>] generic_make_request.part.59+0x6f/0xa0
[<ffffffff812a5050>] generic_make_request+0x60/0x70
[<ffffffff812a50c7>] submit_bio+0x67/0x130
[<ffffffff811a30fb>] submit_bh+0xfb/0x130
[<ffffffff811a6058>] __block_write_full_page+0x1c8/0x340
[<ffffffff811a62a3>] block_write_full_page_endio+0xd3/0x110
[<ffffffff811a62f0>] block_write_full_page+0x10/0x20
[<ffffffff811aa0c3>] blkdev_writepage+0x13/0x20
[<ffffffff81119292>] __writepage+0x12/0x40
[<ffffffff81119a96>] write_cache_pages+0x206/0x460
[<ffffffff81119d35>] generic_writepages+0x45/0x70
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 31440 ms
Event count: 6
pool sleep_on_buffer 13120 ms
scp sleep_on_buffer 5297 ms
scp sleep_on_buffer 3769 ms
scp sleep_on_buffer 2870 ms
cp sleep_on_buffer 5153 ms
git sleep_on_buffer 1231 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811ebe24>] __ext4_new_inode+0x294/0x10c0
[<ffffffff811fb456>] ext4_mkdir+0x146/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 30241 ms
Event count: 4
git sleep_on_buffer 10480 ms
evince sleep_on_buffer 1309 ms
git sleep_on_buffer 17269 ms
git sleep_on_buffer 1183 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811eb7cf>] ext4_free_inode+0x22f/0x5f0
[<ffffffff811f4fe1>] ext4_evict_inode+0x331/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 28375 ms
Event count: 4
flush-8:0 sleep_on_buffer 7042 ms
flush-8:0 sleep_on_buffer 1900 ms
flush-8:0 sleep_on_buffer 1746 ms
flush-8:0 sleep_on_buffer 17687 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121e658>] ext4_ext_convert_to_initialized+0x408/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4601>] write_cache_pages_da+0x421/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 26600 ms
Event count: 4
systemd-journal sleep_on_buffer 2463 ms
systemd-journal sleep_on_buffer 2988 ms
systemd-journal sleep_on_buffer 19520 ms
systemd-journal sleep_on_buffer 1629 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb67b>] ext4_orphan_add+0xbb/0x1f0
[<ffffffff8121f9e1>] ext4_ext_truncate+0x71/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff81171979>] do_sys_ftruncate.constprop.14+0x109/0x170
[<ffffffff81171a09>] sys_ftruncate+0x9/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 25557 ms
Event count: 2
flush-253:0 sleep_on_buffer 2782 ms
flush-253:0 sleep_on_buffer 22775 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119ddb2>] wb_do_writeback+0xb2/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 25493 ms
Event count: 5
git sleep_on_buffer 15264 ms
git sleep_on_buffer 2091 ms
git sleep_on_buffer 2507 ms
git sleep_on_buffer 1218 ms
git sleep_on_buffer 4413 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811eb7cf>] ext4_free_inode+0x22f/0x5f0
[<ffffffff811f4fe1>] ext4_evict_inode+0x331/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 25420 ms
Event count: 8
Cache I/O sleep_on_buffer 8766 ms
pool sleep_on_buffer 1851 ms
rsync sleep_on_buffer 2738 ms
imapd sleep_on_buffer 1697 ms
evolution sleep_on_buffer 2829 ms
pool sleep_on_buffer 2854 ms
firefox sleep_on_buffer 2326 ms
imapd sleep_on_buffer 2359 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811ec0e8>] __ext4_new_inode+0x558/0x10c0
[<ffffffff811fac5b>] ext4_create+0xbb/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 24833 ms
Event count: 9
kswapd0 wait_on_page_bit 2147 ms
kswapd0 wait_on_page_bit 1483 ms
kswapd0 wait_on_page_bit 1393 ms
kswapd0 wait_on_page_bit 1844 ms
kswapd0 wait_on_page_bit 1920 ms
kswapd0 wait_on_page_bit 3606 ms
kswapd0 wait_on_page_bit 7155 ms
kswapd0 wait_on_page_bit 1189 ms
kswapd0 wait_on_page_bit 4096 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811228cf>] shrink_inactive_list+0x15f/0x4a0
[<ffffffff811230cc>] shrink_lruvec+0x13c/0x260
[<ffffffff81123256>] shrink_zone+0x66/0x180
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff8112451b>] balance_pgdat+0x33b/0x4b0
[<ffffffff811247a6>] kswapd+0x116/0x230
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 23799 ms
Event count: 19
jbd2/sdb1-8 wait_on_page_bit 1077 ms
jbd2/sdb1-8 wait_on_page_bit 1126 ms
jbd2/sdb1-8 wait_on_page_bit 1197 ms
jbd2/sdb1-8 wait_on_page_bit 1101 ms
jbd2/sdb1-8 wait_on_page_bit 1160 ms
jbd2/sdb1-8 wait_on_page_bit 1594 ms
jbd2/sdb1-8 wait_on_page_bit 1364 ms
jbd2/sdb1-8 wait_on_page_bit 1094 ms
jbd2/sdb1-8 wait_on_page_bit 1141 ms
jbd2/sdb1-8 wait_on_page_bit 1309 ms
jbd2/sdb1-8 wait_on_page_bit 1325 ms
jbd2/sdb1-8 wait_on_page_bit 1415 ms
jbd2/sdb1-8 wait_on_page_bit 1331 ms
jbd2/sdb1-8 wait_on_page_bit 1372 ms
jbd2/sdb1-8 wait_on_page_bit 1187 ms
jbd2/sdb1-8 wait_on_page_bit 1472 ms
jbd2/sdb1-8 wait_on_page_bit 1192 ms
jbd2/sdb1-8 wait_on_page_bit 1080 ms
jbd2/sdb1-8 wait_on_page_bit 1262 ms
[<ffffffff8110f0e0>] wait_on_page_bit+0x70/0x80
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff8110f2a3>] filemap_fdatawait+0x23/0x30
[<ffffffff8123a78c>] journal_finish_inode_data_buffers+0x6c/0x170
[<ffffffff8123b376>] jbd2_journal_commit_transaction+0x706/0x13c0
[<ffffffff81240513>] kjournald2+0xb3/0x240
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 22392 ms
Event count: 2
rsync sleep_on_buffer 3595 ms
git sleep_on_buffer 18797 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811fb960>] ext4_rename+0x1b0/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 21612 ms
Event count: 3
flush-8:0 sleep_on_buffer 13971 ms
flush-8:0 sleep_on_buffer 3795 ms
flush-8:0 sleep_on_buffer 3846 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811efadd>] ext4_da_update_reserve_space+0x1cd/0x280
[<ffffffff8121f88a>] ext4_ext_map_blocks+0x91a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 21313 ms
Event count: 6
git sleep_on_buffer 1261 ms
git sleep_on_buffer 2135 ms
systemd-journal sleep_on_buffer 13451 ms
git sleep_on_buffer 1203 ms
git sleep_on_buffer 1180 ms
git sleep_on_buffer 2083 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff811e99fd>] ext4_file_mmap+0x3d/0x50
[<ffffffff81140175>] mmap_region+0x325/0x590
[<ffffffff811406f8>] do_mmap_pgoff+0x318/0x440
[<ffffffff8112ba05>] vm_mmap_pgoff+0xa5/0xd0
[<ffffffff8113ee84>] sys_mmap_pgoff+0xa4/0x180
[<ffffffff81006b8d>] sys_mmap+0x1d/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 19298 ms
Event count: 3
flush-8:0 sleep_on_buffer 14371 ms
flush-8:0 sleep_on_buffer 1545 ms
flush-8:0 sleep_on_buffer 3382 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d24c>] ext4_ext_insert_extent+0x21c/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 19044 ms
Event count: 2
akregator sleep_on_buffer 12495 ms
imapd sleep_on_buffer 6549 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb67b>] ext4_orphan_add+0xbb/0x1f0
[<ffffffff8121f9e1>] ext4_ext_truncate+0x71/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 18957 ms
Event count: 5
flush-8:0 sleep_on_buffer 2120 ms
flush-8:0 sleep_on_buffer 1668 ms
flush-8:0 sleep_on_buffer 2679 ms
flush-8:0 sleep_on_buffer 4561 ms
flush-8:0 sleep_on_buffer 7929 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121a97b>] __ext4_ext_dirty.isra.40+0x7b/0x80
[<ffffffff8121d34a>] ext4_ext_insert_extent+0x31a/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 18341 ms
Event count: 6
imapd sleep_on_buffer 5018 ms
imapd sleep_on_buffer 1541 ms
acroread sleep_on_buffer 5963 ms
git sleep_on_buffer 3274 ms
git sleep_on_buffer 1387 ms
git sleep_on_buffer 1158 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811ea201>] ext4_release_file+0x61/0xd0
[<ffffffff811742a0>] __fput+0xb0/0x240
[<ffffffff81174439>] ____fput+0x9/0x10
[<ffffffff81065dc7>] task_work_run+0x97/0xd0
[<ffffffff81002cbc>] do_notify_resume+0x9c/0xb0
[<ffffffff8159c46a>] int_signal+0x12/0x17
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 18310 ms
Event count: 17
cp sleep_on_buffer 1061 ms
cp sleep_on_buffer 1032 ms
cp sleep_on_buffer 1072 ms
cp sleep_on_buffer 1039 ms
cp sleep_on_buffer 1035 ms
cp sleep_on_buffer 1167 ms
cp sleep_on_buffer 1029 ms
cp sleep_on_buffer 1108 ms
cp sleep_on_buffer 1009 ms
cp sleep_on_buffer 1113 ms
cp sleep_on_buffer 1113 ms
cp sleep_on_buffer 1029 ms
free wait_on_page_bit_killable 1067 ms
imapd sleep_on_buffer 1103 ms
cat sleep_on_buffer 1180 ms
imapd sleep_on_buffer 1005 ms
git sleep_on_buffer 1148 ms
[<ffffffff8110ef12>] __lock_page_killable+0x62/0x70
[<ffffffff81110507>] do_generic_file_read.constprop.35+0x287/0x440
[<ffffffff81111359>] generic_file_aio_read+0xd9/0x220
[<ffffffff81172b53>] do_sync_read+0xa3/0xe0
[<ffffffff8117327b>] vfs_read+0xab/0x170
[<ffffffff8117338d>] sys_read+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 18275 ms
Event count: 2
systemd-journal sleep_on_buffer 1594 ms
systemd-journal sleep_on_buffer 16681 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff81228bbd>] ext4_mb_new_blocks+0x1fd/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff8121fd1f>] ext4_fallocate+0x1cf/0x420
[<ffffffff81171b32>] do_fallocate+0x112/0x190
[<ffffffff81171c02>] sys_fallocate+0x52/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 17970 ms
Event count: 2
pool sleep_on_buffer 12739 ms
pool sleep_on_buffer 5231 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9bc4>] add_dirent_to_buf+0x84/0x1d0
[<ffffffff811fa7e4>] ext4_add_entry+0x124/0x2d0
[<ffffffff811fa9b6>] ext4_add_nondir+0x26/0x80
[<ffffffff811fac9f>] ext4_create+0xff/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 17925 ms
Event count: 1
git sleep_on_buffer 17925 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f003b>] ext4_getblk+0x5b/0x1f0
[<ffffffff811f01e1>] ext4_bread+0x11/0x80
[<ffffffff811f758d>] ext4_append+0x5d/0x120
[<ffffffff811fb243>] ext4_init_new_dir+0x83/0x150
[<ffffffff811fb48d>] ext4_mkdir+0x17d/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 17421 ms
Event count: 1
git sleep_on_buffer 17421 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f9c6e>] add_dirent_to_buf+0x12e/0x1d0
[<ffffffff811fa7e4>] ext4_add_entry+0x124/0x2d0
[<ffffffff811fb4bd>] ext4_mkdir+0x1ad/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 17385 ms
Event count: 7
git sleep_on_buffer 1409 ms
git sleep_on_buffer 1128 ms
git sleep_on_buffer 6323 ms
rsync sleep_on_buffer 4503 ms
git sleep_on_buffer 1204 ms
mv sleep_on_buffer 1190 ms
git sleep_on_buffer 1628 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811fb960>] ext4_rename+0x1b0/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 13983 ms
Event count: 3
patch sleep_on_buffer 1511 ms
cp sleep_on_buffer 2096 ms
git sleep_on_buffer 10376 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811ec0e8>] __ext4_new_inode+0x558/0x10c0
[<ffffffff811fb456>] ext4_mkdir+0x146/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 13603 ms
Event count: 4
git sleep_on_buffer 2160 ms
gen-report.sh sleep_on_buffer 4730 ms
evolution sleep_on_buffer 4697 ms
git sleep_on_buffer 2016 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811fb6cf>] ext4_orphan_add+0x10f/0x1f0
[<ffffffff811f31a4>] ext4_setattr+0x3d4/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 13264 ms
Event count: 8
ls sleep_on_buffer 1116 ms
ls sleep_on_buffer 1756 ms
ls sleep_on_buffer 1901 ms
ls sleep_on_buffer 2033 ms
ls sleep_on_buffer 1373 ms
ls sleep_on_buffer 3046 ms
offlineimap sleep_on_buffer 1011 ms
imapd sleep_on_buffer 1028 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177bc6>] vfs_stat+0x16/0x20
[<ffffffff81177ce5>] sys_newstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 12710 ms
Event count: 6
git sleep_on_buffer 1364 ms
git sleep_on_buffer 1612 ms
git sleep_on_buffer 4321 ms
git sleep_on_buffer 2185 ms
git sleep_on_buffer 2126 ms
git sleep_on_buffer 1102 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff8122509a>] ext4_mb_load_buddy+0x26a/0x350
[<ffffffff8122676b>] ext4_mb_find_by_goal+0x9b/0x2d0
[<ffffffff81227109>] ext4_mb_regular_allocator+0x59/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 12397 ms
Event count: 7
jbd2/dm-0-8 sleep_on_buffer 1516 ms
jbd2/dm-0-8 sleep_on_buffer 1153 ms
jbd2/dm-0-8 sleep_on_buffer 1307 ms
jbd2/dm-0-8 sleep_on_buffer 1518 ms
jbd2/dm-0-8 sleep_on_buffer 1513 ms
jbd2/dm-0-8 sleep_on_buffer 1516 ms
jbd2/dm-0-8 sleep_on_buffer 3874 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff8123b488>] jbd2_journal_commit_transaction+0x818/0x13c0
[<ffffffff81240513>] kjournald2+0xb3/0x240
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 12361 ms
Event count: 4
git sleep_on_buffer 1076 ms
scp sleep_on_buffer 1517 ms
rsync sleep_on_buffer 5018 ms
rsync sleep_on_buffer 4750 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f9c6e>] add_dirent_to_buf+0x12e/0x1d0
[<ffffffff811fa7e4>] ext4_add_entry+0x124/0x2d0
[<ffffffff811fa9b6>] ext4_add_nondir+0x26/0x80
[<ffffffff811fac9f>] ext4_create+0xff/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 12175 ms
Event count: 3
patch sleep_on_buffer 1546 ms
patch sleep_on_buffer 7218 ms
patch sleep_on_buffer 3411 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9dd2>] ext4_dx_add_entry+0xc2/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fb4bd>] ext4_mkdir+0x1ad/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 11862 ms
Event count: 4
bash sleep_on_buffer 5441 ms
offlineimap sleep_on_buffer 2780 ms
pool sleep_on_buffer 1529 ms
pool sleep_on_buffer 2112 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb67b>] ext4_orphan_add+0xbb/0x1f0
[<ffffffff811f31a4>] ext4_setattr+0x3d4/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 11695 ms
Event count: 1
git sleep_on_buffer 11695 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811fb960>] ext4_rename+0x1b0/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 11452 ms
Event count: 8
compare-mmtests sleep_on_buffer 1407 ms
compare-mmtests sleep_on_buffer 1439 ms
find sleep_on_buffer 2063 ms
git sleep_on_buffer 1128 ms
cp sleep_on_buffer 1041 ms
rsync sleep_on_buffer 1533 ms
rsync sleep_on_buffer 1070 ms
FileLoader sleep_on_buffer 1771 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f0227>] ext4_bread+0x57/0x80
[<ffffffff811f7b21>] __ext4_read_dirblock+0x41/0x1d0
[<ffffffff811f849b>] htree_dirblock_to_tree+0x3b/0x1a0
[<ffffffff811f8d7f>] ext4_htree_fill_tree+0x7f/0x220
[<ffffffff811e8d67>] ext4_dx_readdir+0x1a7/0x440
[<ffffffff811e9572>] ext4_readdir+0x422/0x4e0
[<ffffffff811849a0>] vfs_readdir+0xb0/0xe0
[<ffffffff81184ae9>] sys_getdents+0x89/0x110
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 9483 ms
Event count: 3
offlineimap sleep_on_buffer 1768 ms
dconf-service sleep_on_buffer 6600 ms
git sleep_on_buffer 1115 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811fb8b5>] ext4_rename+0x105/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 8201 ms
Event count: 1
systemd-journal sleep_on_buffer 8201 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff8121fd1f>] ext4_fallocate+0x1cf/0x420
[<ffffffff81171b32>] do_fallocate+0x112/0x190
[<ffffffff81171c02>] sys_fallocate+0x52/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 7699 ms
Event count: 2
git sleep_on_buffer 3475 ms
git sleep_on_buffer 4224 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fc898>] ext4_orphan_del+0x1a8/0x1e0
[<ffffffff811f4fbb>] ext4_evict_inode+0x30b/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 7564 ms
Event count: 2
tar sleep_on_buffer 1286 ms
rm sleep_on_buffer 6278 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811fc3e1>] ext4_unlink+0x41/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff811825dd>] sys_unlinkat+0x1d/0x40
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 6596 ms
Event count: 1
acroread sleep_on_buffer 6596 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff811105e3>] do_generic_file_read.constprop.35+0x363/0x440
[<ffffffff81111359>] generic_file_aio_read+0xd9/0x220
[<ffffffff81172b53>] do_sync_read+0xa3/0xe0
[<ffffffff8117327b>] vfs_read+0xab/0x170
[<ffffffff8117338d>] sys_read+0x4d/0x90
[<ffffffff8159d81c>] sysenter_dispatch+0x7/0x21
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 6589 ms
Event count: 1
tar sleep_on_buffer 6589 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9dd2>] ext4_dx_add_entry+0xc2/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fb4bd>] ext4_mkdir+0x1ad/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 6272 ms
Event count: 6
pool wait_on_page_bit 1005 ms
pool wait_on_page_bit 1015 ms
StreamT~ns #908 sleep_on_buffer 1086 ms
Cache I/O wait_on_page_bit 1091 ms
StreamT~ns #138 wait_on_page_bit 1046 ms
offlineimap sleep_on_buffer 1029 ms
[<ffffffff810a04ed>] futex_wait+0x17d/0x270
[<ffffffff810a21ac>] do_futex+0x7c/0x1b0
[<ffffffff810a241d>] sys_futex+0x13d/0x190
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 6237 ms
Event count: 1
offlineimap sleep_on_buffer 6237 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812291ba>] ext4_free_blocks+0x36a/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 6192 ms
Event count: 4
ls sleep_on_buffer 1679 ms
ls sleep_on_buffer 1746 ms
ls sleep_on_buffer 1076 ms
ls sleep_on_buffer 1691 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff811ef20d>] __ext4_get_inode_loc+0x3dd/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177bc6>] vfs_stat+0x16/0x20
[<ffffffff81177ce5>] sys_newstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 5989 ms
Event count: 3
flush-8:0 sleep_on_buffer 1184 ms
flush-8:0 sleep_on_buffer 1548 ms
flush-8:0 sleep_on_buffer 3257 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81227556>] ext4_mb_mark_diskspace_used+0x76/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 5770 ms
Event count: 1
git sleep_on_buffer 5770 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121bf74>] ext4_ext_rm_leaf+0x1e4/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f513e>] ext4_evict_inode+0x48e/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 4477 ms
Event count: 2
offlineimap sleep_on_buffer 2154 ms
DOM Worker sleep_on_buffer 2323 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff81180bd8>] lookup_open+0xc8/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 4428 ms
Event count: 3
compare-mmtests sleep_on_buffer 1725 ms
compare-mmtests sleep_on_buffer 1634 ms
cp sleep_on_buffer 1069 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177ba9>] vfs_lstat+0x19/0x20
[<ffffffff81177d15>] sys_newlstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 4168 ms
Event count: 3
git sleep_on_buffer 1866 ms
git sleep_on_buffer 1070 ms
git sleep_on_buffer 1232 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff8122509a>] ext4_mb_load_buddy+0x26a/0x350
[<ffffffff81227247>] ext4_mb_regular_allocator+0x197/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3940 ms
Event count: 2
evolution sleep_on_buffer 1978 ms
git sleep_on_buffer 1962 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812291ba>] ext4_free_blocks+0x36a/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3802 ms
Event count: 2
git sleep_on_buffer 1933 ms
git sleep_on_buffer 1869 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81171cd7>] sys_faccessat+0x97/0x220
[<ffffffff81171e73>] sys_access+0x13/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3792 ms
Event count: 3
cc1 sleep_on_buffer 1161 ms
compare-mmtests sleep_on_buffer 1088 ms
cc1 sleep_on_buffer 1543 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117e76a>] link_path_walk+0x7ca/0x8e0
[<ffffffff81181596>] path_openat+0x96/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3783 ms
Event count: 2
compare-mmtests sleep_on_buffer 2237 ms
compare-mmtests sleep_on_buffer 1546 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117e76a>] link_path_walk+0x7ca/0x8e0
[<ffffffff81181596>] path_openat+0x96/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117285f>] sys_openat+0xf/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3692 ms
Event count: 2
git sleep_on_buffer 1667 ms
git sleep_on_buffer 2025 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff8122509a>] ext4_mb_load_buddy+0x26a/0x350
[<ffffffff8122676b>] ext4_mb_find_by_goal+0x9b/0x2d0
[<ffffffff81227109>] ext4_mb_regular_allocator+0x59/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3533 ms
Event count: 1
pool sleep_on_buffer 3533 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f9dd2>] ext4_dx_add_entry+0xc2/0x590
[<ffffffff811fa925>] ext4_add_entry+0x265/0x2d0
[<ffffffff811fbf16>] ext4_rename+0x766/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 3329 ms
Event count: 3
folder-markup.s sleep_on_buffer 1147 ms
imapd sleep_on_buffer 1053 ms
gnuplot sleep_on_buffer 1129 ms
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2861 ms
Event count: 2
chmod sleep_on_buffer 1227 ms
chmod sleep_on_buffer 1634 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811f313c>] ext4_setattr+0x36c/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff8117137b>] chmod_common+0xab/0xb0
[<ffffffff811721a1>] sys_fchmodat+0x41/0xa0
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2822 ms
Event count: 1
gnome-terminal sleep_on_buffer 2822 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811eb856>] ext4_free_inode+0x2b6/0x5f0
[<ffffffff811f4fe1>] ext4_evict_inode+0x331/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81174368>] __fput+0x178/0x240
[<ffffffff81174439>] ____fput+0x9/0x10
[<ffffffff81065dc7>] task_work_run+0x97/0xd0
[<ffffffff81002cbc>] do_notify_resume+0x9c/0xb0
[<ffffffff8159c46a>] int_signal+0x12/0x17
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2769 ms
Event count: 1
imapd sleep_on_buffer 2769 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811f313c>] ext4_setattr+0x36c/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff8117137b>] chmod_common+0xab/0xb0
[<ffffffff811721a1>] sys_fchmodat+0x41/0xa0
[<ffffffff81172214>] sys_chmod+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2727 ms
Event count: 1
mv sleep_on_buffer 2727 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f9c6e>] add_dirent_to_buf+0x12e/0x1d0
[<ffffffff811fa7e4>] ext4_add_entry+0x124/0x2d0
[<ffffffff811fbf16>] ext4_rename+0x766/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2675 ms
Event count: 1
flush-8:0 sleep_on_buffer 2675 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2658 ms
Event count: 1
patch sleep_on_buffer 2658 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121c4ad>] ext4_ext_tree_init+0x2d/0x40
[<ffffffff811ecc06>] __ext4_new_inode+0x1076/0x10c0
[<ffffffff811fb456>] ext4_mkdir+0x146/0x2b0
[<ffffffff81181b42>] vfs_mkdir+0xa2/0x120
[<ffffffff81182533>] sys_mkdirat+0xa3/0xf0
[<ffffffff81182594>] sys_mkdir+0x14/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2603 ms
Event count: 2
flush-8:0 sleep_on_buffer 1162 ms
flush-8:0 sleep_on_buffer 1441 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d24c>] ext4_ext_insert_extent+0x21c/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2580 ms
Event count: 2
rm sleep_on_buffer 1265 ms
rm sleep_on_buffer 1315 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff811e8265>] ext4_read_block_bitmap+0x35/0x60
[<ffffffff8122908c>] ext4_free_blocks+0x23c/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f513e>] ext4_evict_inode+0x48e/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff811825dd>] sys_unlinkat+0x1d/0x40
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2542 ms
Event count: 2
flush-8:16 get_request 1316 ms
flush-8:16 get_request 1226 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff812a4b1f>] generic_make_request.part.59+0x6f/0xa0
[<ffffffff812a5050>] generic_make_request+0x60/0x70
[<ffffffff812a50c7>] submit_bio+0x67/0x130
[<ffffffff811a30fb>] submit_bh+0xfb/0x130
[<ffffffff811a6058>] __block_write_full_page+0x1c8/0x340
[<ffffffff811a62a3>] block_write_full_page_endio+0xd3/0x110
[<ffffffff811a62f0>] block_write_full_page+0x10/0x20
[<ffffffff811aa0c3>] blkdev_writepage+0x13/0x20
[<ffffffff81119292>] __writepage+0x12/0x40
[<ffffffff81119a96>] write_cache_pages+0x206/0x460
[<ffffffff81119d35>] generic_writepages+0x45/0x70
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2504 ms
Event count: 1
acroread sleep_on_buffer 2504 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff811105e3>] do_generic_file_read.constprop.35+0x363/0x440
[<ffffffff81111359>] generic_file_aio_read+0xd9/0x220
[<ffffffff81172b53>] do_sync_read+0xa3/0xe0
[<ffffffff8117327b>] vfs_read+0xab/0x170
[<ffffffff8117338d>] sys_read+0x4d/0x90
[<ffffffff8159dc79>] ia32_sysret+0x0/0x5
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2477 ms
Event count: 2
git sleep_on_buffer 1200 ms
firefox sleep_on_buffer 1277 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812291ba>] ext4_free_blocks+0x36a/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f513e>] ext4_evict_inode+0x48e/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2168 ms
Event count: 2
xchat sleep_on_buffer 1096 ms
xchat sleep_on_buffer 1072 ms
[<ffffffff81185476>] do_poll.isra.7+0x1c6/0x290
[<ffffffff81186331>] do_sys_poll+0x191/0x200
[<ffffffff81186466>] sys_poll+0x66/0x100
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2156 ms
Event count: 2
git sleep_on_buffer 1076 ms
git sleep_on_buffer 1080 ms
[<ffffffff811383b2>] unmap_single_vma+0x82/0x100
[<ffffffff81138c2c>] unmap_vmas+0x4c/0xa0
[<ffffffff811408f0>] exit_mmap+0x90/0x170
[<ffffffff81043ee5>] mmput.part.27+0x45/0x110
[<ffffffff81043fcd>] mmput+0x1d/0x30
[<ffffffff8104be22>] exit_mm+0x132/0x180
[<ffffffff8104bfc5>] do_exit+0x155/0x460
[<ffffffff8104c34f>] do_group_exit+0x3f/0xa0
[<ffffffff8104c3c2>] sys_exit_group+0x12/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2141 ms
Event count: 2
imapd sleep_on_buffer 1057 ms
ntpd wait_on_page_bit_killable 1084 ms
[<ffffffff81185a99>] do_select+0x4c9/0x5d0
[<ffffffff81185d58>] core_sys_select+0x1b8/0x2f0
[<ffffffff811860d6>] sys_select+0xb6/0x100
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2130 ms
Event count: 2
git sleep_on_buffer 1110 ms
git sleep_on_buffer 1020 ms
[<ffffffff811f4ccb>] ext4_evict_inode+0x1b/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81182c4a>] sys_renameat+0x33a/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2092 ms
Event count: 1
flush-8:0 sleep_on_buffer 2092 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121a97b>] __ext4_ext_dirty.isra.40+0x7b/0x80
[<ffffffff8121d69b>] ext4_split_extent_at+0x24b/0x390
[<ffffffff8121e038>] ext4_split_extent.isra.47+0x108/0x130
[<ffffffff8121e3ae>] ext4_ext_convert_to_initialized+0x15e/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2079 ms
Event count: 2
offlineimap sleep_on_buffer 1030 ms
pool wait_on_page_bit 1049 ms
[<ffffffff811ea6e5>] ext4_sync_file+0x205/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2066 ms
Event count: 2
folder-markup.s sleep_on_buffer 1024 ms
tee sleep_on_buffer 1042 ms
[<ffffffff8117b90e>] pipe_read+0x20e/0x340
[<ffffffff81172b53>] do_sync_read+0xa3/0xe0
[<ffffffff8117327b>] vfs_read+0xab/0x170
[<ffffffff8117338d>] sys_read+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 2047 ms
Event count: 1
Cache I/O sleep_on_buffer 2047 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812291e1>] ext4_free_blocks+0x391/0xc10
[<ffffffff8121bd16>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121bf95>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121dcbc>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8121fb0b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff81171979>] do_sys_ftruncate.constprop.14+0x109/0x170
[<ffffffff81171a09>] sys_ftruncate+0x9/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1977 ms
Event count: 1
patch sleep_on_buffer 1977 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117e76a>] link_path_walk+0x7ca/0x8e0
[<ffffffff8117ede3>] path_lookupat+0x53/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177ba9>] vfs_lstat+0x19/0x20
[<ffffffff81177d15>] sys_newlstat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1839 ms
Event count: 1
compare-mmtests sleep_on_buffer 1839 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117e76a>] link_path_walk+0x7ca/0x8e0
[<ffffffff81181596>] path_openat+0x96/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1819 ms
Event count: 1
cp sleep_on_buffer 1819 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff811a4bc7>] write_dirty_buffer+0x67/0x70
[<ffffffff8123d035>] __flush_batch+0x45/0xa0
[<ffffffff8123dad6>] jbd2_log_do_checkpoint+0x1d6/0x220
[<ffffffff8123dba1>] __jbd2_log_wait_for_space+0x81/0x190
[<ffffffff812382d0>] start_this_handle+0x2e0/0x3e0
[<ffffffff81238590>] jbd2__journal_start.part.8+0x90/0x190
[<ffffffff812386d5>] jbd2__journal_start+0x45/0x50
[<ffffffff812205d1>] __ext4_journal_start_sb+0x81/0x170
[<ffffffff811ebf61>] __ext4_new_inode+0x3d1/0x10c0
[<ffffffff811fac5b>] ext4_create+0xbb/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1664 ms
Event count: 1
flush-8:0 sleep_on_buffer 1664 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d125>] ext4_ext_insert_extent+0xf5/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1635 ms
Event count: 1
flush-8:0 sleep_on_buffer 1635 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1591 ms
Event count: 1
imapd sleep_on_buffer 1591 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811f96f9>] ext4_lookup.part.31+0x29/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8117ca84>] lookup_hash+0x14/0x20
[<ffffffff8117fae3>] do_unlinkat+0xf3/0x260
[<ffffffff81182611>] sys_unlink+0x11/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1529 ms
Event count: 1
ls sleep_on_buffer 1529 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f0227>] ext4_bread+0x57/0x80
[<ffffffff811f7b21>] __ext4_read_dirblock+0x41/0x1d0
[<ffffffff811f7f3d>] dx_probe+0x3d/0x410
[<ffffffff811f8dce>] ext4_htree_fill_tree+0xce/0x220
[<ffffffff811e8d67>] ext4_dx_readdir+0x1a7/0x440
[<ffffffff811e9572>] ext4_readdir+0x422/0x4e0
[<ffffffff811849a0>] vfs_readdir+0xb0/0xe0
[<ffffffff81184ae9>] sys_getdents+0x89/0x110
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1523 ms
Event count: 1
gnuplot sleep_on_buffer 1523 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121fad7>] ext4_ext_truncate+0x167/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1519 ms
Event count: 1
find sleep_on_buffer 1519 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f0227>] ext4_bread+0x57/0x80
[<ffffffff811f7b21>] __ext4_read_dirblock+0x41/0x1d0
[<ffffffff811f849b>] htree_dirblock_to_tree+0x3b/0x1a0
[<ffffffff811f8e42>] ext4_htree_fill_tree+0x142/0x220
[<ffffffff811e8d67>] ext4_dx_readdir+0x1a7/0x440
[<ffffffff811e9572>] ext4_readdir+0x422/0x4e0
[<ffffffff811849a0>] vfs_readdir+0xb0/0xe0
[<ffffffff81184ae9>] sys_getdents+0x89/0x110
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1509 ms
Event count: 1
git sleep_on_buffer 1509 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff811e8265>] ext4_read_block_bitmap+0x35/0x60
[<ffffffff81227533>] ext4_mb_mark_diskspace_used+0x53/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1470 ms
Event count: 1
rm sleep_on_buffer 1470 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eb4d0>] ext4_read_inode_bitmap+0x400/0x4d0
[<ffffffff811eb7ab>] ext4_free_inode+0x20b/0x5f0
[<ffffffff811f4fe1>] ext4_evict_inode+0x331/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff8117fbe1>] do_unlinkat+0x1f1/0x260
[<ffffffff811825dd>] sys_unlinkat+0x1d/0x40
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1462 ms
Event count: 1
imapd sleep_on_buffer 1462 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811fbb37>] ext4_rename+0x387/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1457 ms
Event count: 1
git sleep_on_buffer 1457 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff811e8265>] ext4_read_block_bitmap+0x35/0x60
[<ffffffff81227533>] ext4_mb_mark_diskspace_used+0x53/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1395 ms
Event count: 1
flush-8:0 sleep_on_buffer 1395 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1387 ms
Event count: 1
git sleep_on_buffer 1387 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1378 ms
Event count: 1
gnuplot sleep_on_buffer 1378 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811ea201>] ext4_release_file+0x61/0xd0
[<ffffffff811742a0>] __fput+0xb0/0x240
[<ffffffff81174439>] ____fput+0x9/0x10
[<ffffffff81065de4>] task_work_run+0xb4/0xd0
[<ffffffff8104bffa>] do_exit+0x18a/0x460
[<ffffffff8104c34f>] do_group_exit+0x3f/0xa0
[<ffffffff8104c3c2>] sys_exit_group+0x12/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1337 ms
Event count: 1
git sleep_on_buffer 1337 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff81224c2e>] ext4_mb_init_group+0x9e/0x100
[<ffffffff81224d97>] ext4_mb_good_group+0x107/0x1a0
[<ffffffff81227233>] ext4_mb_regular_allocator+0x183/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4185>] mpage_add_bh_to_extent+0x45/0xa0
[<ffffffff811f4505>] write_cache_pages_da+0x325/0x4b0
[<ffffffff811f49e5>] ext4_da_writepages+0x355/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1309 ms
Event count: 1
flush-8:0 sleep_on_buffer 1309 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff8121a97b>] __ext4_ext_dirty.isra.40+0x7b/0x80
[<ffffffff8121d69b>] ext4_split_extent_at+0x24b/0x390
[<ffffffff8121e038>] ext4_split_extent.isra.47+0x108/0x130
[<ffffffff8121e3ae>] ext4_ext_convert_to_initialized+0x15e/0x590
[<ffffffff8121ee7b>] ext4_ext_handle_uninitialized_extents+0x2fb/0x3c0
[<ffffffff8121f547>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1284 ms
Event count: 1
cp sleep_on_buffer 1284 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b789>] update_time+0x79/0xc0
[<ffffffff8118ba31>] touch_atime+0x161/0x170
[<ffffffff81177e71>] sys_readlinkat+0xe1/0x120
[<ffffffff81177ec6>] sys_readlink+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1277 ms
Event count: 1
git sleep_on_buffer 1277 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff81224c2e>] ext4_mb_init_group+0x9e/0x100
[<ffffffff81224d97>] ext4_mb_good_group+0x107/0x1a0
[<ffffffff81227233>] ext4_mb_regular_allocator+0x183/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff81110c3a>] filemap_write_and_wait_range+0x4a/0x70
[<ffffffff811ea54a>] ext4_sync_file+0x6a/0x2d0
[<ffffffff811a1758>] do_fsync+0x58/0x80
[<ffffffff811a1abb>] sys_fsync+0xb/0x10
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1235 ms
Event count: 1
cp sleep_on_buffer 1235 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812275bf>] ext4_mb_mark_diskspace_used+0xdf/0x4d0
[<ffffffff81228c6f>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8122d7d0>] ext4_alloc_blocks+0x140/0x2b0
[<ffffffff8122d995>] ext4_alloc_branch+0x55/0x2c0
[<ffffffff8122ecb9>] ext4_ind_map_blocks+0x299/0x500
[<ffffffff811efd43>] ext4_map_blocks+0x1b3/0x450
[<ffffffff811f23e7>] _ext4_get_block+0x87/0x170
[<ffffffff811f2501>] ext4_get_block+0x11/0x20
[<ffffffff811a65bf>] __block_write_begin+0x1af/0x4d0
[<ffffffff811f1969>] ext4_write_begin+0x159/0x410
[<ffffffff8110f3aa>] generic_perform_write+0xca/0x210
[<ffffffff8110f548>] generic_file_buffered_write+0x58/0x90
[<ffffffff81110f96>] __generic_file_aio_write+0x1b6/0x3b0
[<ffffffff8111120a>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ea3a3>] ext4_file_write+0x83/0xd0
[<ffffffff81172a73>] do_sync_write+0xa3/0xe0
[<ffffffff811730fe>] vfs_write+0xae/0x180
[<ffffffff8117341d>] sys_write+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1182 ms
Event count: 1
imapd sleep_on_buffer 1182 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb052>] ext4_delete_entry+0x62/0x120
[<ffffffff811fbfea>] ext4_rename+0x83a/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1181 ms
Event count: 1
systemd-journal sleep_on_buffer 1181 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff8121a8f2>] ext4_ext_get_access.isra.39+0x22/0x30
[<ffffffff8121d125>] ext4_ext_insert_extent+0xf5/0x420
[<ffffffff8121f60a>] ext4_ext_map_blocks+0x69a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff8121fd1f>] ext4_fallocate+0x1cf/0x420
[<ffffffff81171b32>] do_fallocate+0x112/0x190
[<ffffffff81171c02>] sys_fallocate+0x52/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1160 ms
Event count: 1
rm sleep_on_buffer 1160 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811f9505>] ext4_find_entry+0x325/0x4f0
[<ffffffff811fc169>] ext4_rmdir+0x39/0x270
[<ffffffff8117dbf8>] vfs_rmdir.part.32+0xa8/0xf0
[<ffffffff8117fc8a>] vfs_rmdir+0x3a/0x50
[<ffffffff8117fe63>] do_rmdir+0x1c3/0x1e0
[<ffffffff811825ed>] sys_unlinkat+0x2d/0x40
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1108 ms
Event count: 1
mutt sleep_on_buffer 1108 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811eb7cf>] ext4_free_inode+0x22f/0x5f0
[<ffffffff811f4fe1>] ext4_evict_inode+0x331/0x4c0
[<ffffffff8118bcbf>] evict+0xaf/0x1b0
[<ffffffff8118c543>] iput_final+0xd3/0x160
[<ffffffff8118c609>] iput+0x39/0x50
[<ffffffff81187248>] dentry_iput+0x98/0xe0
[<ffffffff81188ac8>] dput+0x128/0x230
[<ffffffff81174368>] __fput+0x178/0x240
[<ffffffff81174439>] ____fput+0x9/0x10
[<ffffffff81065dc7>] task_work_run+0x97/0xd0
[<ffffffff81002cbc>] do_notify_resume+0x9c/0xb0
[<ffffffff8159c46a>] int_signal+0x12/0x17
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1106 ms
Event count: 1
flush-8:0 sleep_on_buffer 1106 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f51b1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811efadd>] ext4_da_update_reserve_space+0x1cd/0x280
[<ffffffff8121f88a>] ext4_ext_map_blocks+0x91a/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119de90>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff810690eb>] kthread+0xbb/0xc0
[<ffffffff8159c0fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1081 ms
Event count: 1
imapd sleep_on_buffer 1081 ms
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811fb67b>] ext4_orphan_add+0xbb/0x1f0
[<ffffffff8121f9e1>] ext4_ext_truncate+0x71/0x1e0
[<ffffffff811ef535>] ext4_truncate.part.59+0xd5/0xf0
[<ffffffff811f0614>] ext4_truncate+0x34/0x90
[<ffffffff811f2f5d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d132>] notify_change+0x1f2/0x3c0
[<ffffffff811715d9>] do_truncate+0x59/0xa0
[<ffffffff8117d186>] handle_truncate+0x66/0xa0
[<ffffffff81181306>] do_last+0x626/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1079 ms
Event count: 1
git sleep_on_buffer 1079 ms
[<ffffffff812a5050>] generic_make_request+0x60/0x70
[<ffffffff812a50c7>] submit_bio+0x67/0x130
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1074 ms
Event count: 1
cp sleep_on_buffer 1074 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a35ae>] __lock_buffer+0x2e/0x30
[<ffffffff81239def>] do_get_write_access+0x43f/0x4b0
[<ffffffff81239fab>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220839>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f2b88>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f2bf9>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811ec749>] __ext4_new_inode+0xbb9/0x10c0
[<ffffffff811fac5b>] ext4_create+0xbb/0x190
[<ffffffff81180aa5>] vfs_create+0xb5/0x120
[<ffffffff81180c4e>] lookup_open+0x13e/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff811815b3>] path_openat+0xb3/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff8117284c>] sys_open+0x1c/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1072 ms
Event count: 1
du sleep_on_buffer 1072 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff8117ca63>] __lookup_hash+0x33/0x40
[<ffffffff8158464f>] lookup_slow+0x40/0xa4
[<ffffffff8117efb2>] path_lookupat+0x222/0x780
[<ffffffff8117f53f>] filename_lookup+0x2f/0xc0
[<ffffffff81182074>] user_path_at_empty+0x54/0xa0
[<ffffffff811820cc>] user_path_at+0xc/0x10
[<ffffffff81177b39>] vfs_fstatat+0x49/0xa0
[<ffffffff81177d45>] sys_newfstatat+0x15/0x30
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1034 ms
Event count: 1
git sleep_on_buffer 1034 ms
[<ffffffff8110ef82>] __lock_page+0x62/0x70
[<ffffffff8110fe71>] find_lock_page+0x51/0x80
[<ffffffff8110ff04>] grab_cache_page_write_begin+0x64/0xd0
[<ffffffff811f1ca4>] ext4_da_write_begin+0x84/0x2e0
[<ffffffff8110f3aa>] generic_perform_write+0xca/0x210
[<ffffffff8110f548>] generic_file_buffered_write+0x58/0x90
[<ffffffff81110f96>] __generic_file_aio_write+0x1b6/0x3b0
[<ffffffff8111120a>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ea3a3>] ext4_file_write+0x83/0xd0
[<ffffffff81172a73>] do_sync_write+0xa3/0xe0
[<ffffffff811730fe>] vfs_write+0xae/0x180
[<ffffffff8117341d>] sys_write+0x4d/0x90
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1031 ms
Event count: 1
git sleep_on_buffer 1031 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811e7818>] ext4_wait_block_bitmap+0xb8/0xc0
[<ffffffff8122462e>] ext4_mb_init_cache+0x1ce/0x730
[<ffffffff8122509a>] ext4_mb_load_buddy+0x26a/0x350
[<ffffffff81227247>] ext4_mb_regular_allocator+0x197/0x430
[<ffffffff81228db6>] ext4_mb_new_blocks+0x3f6/0x490
[<ffffffff8121f471>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811efe65>] ext4_map_blocks+0x2d5/0x450
[<ffffffff811f3f0a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4a10>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac0b>] do_writepages+0x1b/0x30
[<ffffffff81110be9>] __filemap_fdatawrite_range+0x49/0x50
[<ffffffff811114b7>] filemap_flush+0x17/0x20
[<ffffffff811f0354>] ext4_alloc_da_blocks+0x44/0xa0
[<ffffffff811fb960>] ext4_rename+0x1b0/0x980
[<ffffffff8117d4ed>] vfs_rename_other+0xcd/0x120
[<ffffffff81180126>] vfs_rename+0xb6/0x240
[<ffffffff81182c96>] sys_renameat+0x386/0x3d0
[<ffffffff81182cf6>] sys_rename+0x16/0x20
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1029 ms
Event count: 1
git wait_on_page_bit_killable 1029 ms
[<ffffffff815966d9>] kretprobe_trampoline+0x25/0x4c
[<ffffffff81111728>] filemap_fault+0x88/0x410
[<ffffffff81135d69>] __do_fault+0x439/0x530
[<ffffffff811394be>] handle_pte_fault+0xee/0x200
[<ffffffff8113a731>] handle_mm_fault+0x271/0x390
[<ffffffff81597a20>] __do_page_fault+0x230/0x520
[<ffffffff81594ec5>] do_device_not_available+0x15/0x20
[<ffffffff8159d50e>] device_not_available+0x1e/0x30
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1017 ms
Event count: 1
npviewer.bin sleep_on_buffer 1017 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff811eefee>] __ext4_get_inode_loc+0x1be/0x3f0
[<ffffffff811f0d2e>] ext4_iget+0x7e/0x940
[<ffffffff811f9796>] ext4_lookup.part.31+0xc6/0x140
[<ffffffff811f9835>] ext4_lookup+0x25/0x30
[<ffffffff8117c628>] lookup_real+0x18/0x50
[<ffffffff81180bd8>] lookup_open+0xc8/0x1d0
[<ffffffff81180fe7>] do_last+0x307/0x820
[<ffffffff8118182a>] path_openat+0x32a/0x4a0
[<ffffffff8118210d>] do_filp_open+0x3d/0xa0
[<ffffffff81172749>] do_sys_open+0xf9/0x1e0
[<ffffffff811c2996>] compat_sys_open+0x16/0x20
[<ffffffff8159dc79>] ia32_sysret+0x0/0x5
[<ffffffffffffffff>] 0xffffffffffffffff

Time stalled in this event: 1016 ms
Event count: 1
rm sleep_on_buffer 1016 ms
[<ffffffff815966b4>] kretprobe_trampoline+0x0/0x4c
[<ffffffff811a3566>] __wait_on_buffer+0x26/0x30
[<ffffffff8123d8f0>] __wait_cp_io+0xd0/0xe0
[<ffffffff8123da23>] jbd2_log_do_checkpoint+0x123/0x220
[<ffffffff8123dba1>] __jbd2_log_wait_for_space+0x81/0x190
[<ffffffff812382d0>] start_this_handle+0x2e0/0x3e0
[<ffffffff81238590>] jbd2__journal_start.part.8+0x90/0x190
[<ffffffff812386d5>] jbd2__journal_start+0x45/0x50
[<ffffffff812205d1>] __ext4_journal_start_sb+0x81/0x170
[<ffffffff811fc44c>] ext4_unlink+0xac/0x350
[<ffffffff8117daef>] vfs_unlink.part.31+0x7f/0xe0
[<ffffffff8117f9d7>] vfs_unlink+0x37/0x50
[<ffffffff8117fbff>] do_unlinkat+0x20f/0x260
[<ffffffff811825dd>] sys_unlinkat+0x1d/0x40
[<ffffffff8159c1ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff


--
Mel Gorman
SUSE Labs


2013-04-02 14:59:26

by Jiri Slaby

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On 04/02/2013 04:27 PM, Mel Gorman wrote:
> I'm testing a page-reclaim-related series on my laptop that is partially
> aimed at fixing long stalls when doing metadata-intensive operations on
> low memory such as a git checkout. I've been running 3.9-rc2 with the
> series applied but found that the interactive performance was awful even
> when there was plenty of free memory.
>
> I activated a monitor from mmtests that logs when a process is stuck for
> a long time in D state and found that there are a lot of stalls in ext4.
> The report first states that processes have been stalled for a total of
> 6498 seconds on IO which seems like a lot. Here is a breakdown of the
> recorded events.

Just a note that I am indeed using ext4 on the affected machine for all
filesystems I have except for an efi partition...

--
js
suse labs

2013-04-02 15:03:46

by Zheng Liu

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

Hi Mel,

Thanks for reporting it.

On 04/02/2013 10:27 PM, Mel Gorman wrote:
> I'm testing a page-reclaim-related series on my laptop that is partially
> aimed at fixing long stalls when doing metadata-intensive operations on
> low memory such as a git checkout. I've been running 3.9-rc2 with the
> series applied but found that the interactive performance was awful even
> when there was plenty of free memory.
>
> I activated a monitor from mmtests that logs when a process is stuck for
> a long time in D state and found that there are a lot of stalls in ext4.
> The report first states that processes have been stalled for a total of
> 6498 seconds on IO which seems like a lot. Here is a breakdown of the
> recorded events.

In this merge window, we add a status tree as a extent cache. Meanwhile
a es_cache shrinker is registered to try to reclaim from this cache when
we are under a high memory pressure. So I suspect that the root cause
is this shrinker. Could you please tell me how to reproduce this
problem? If I understand correctly, I can run mmtest to reproduce this
problem, right?

Thanks,
- Zheng

2013-04-02 15:07:40

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 03:27:17PM +0100, Mel Gorman wrote:
> I'm testing a page-reclaim-related series on my laptop that is partially
> aimed at fixing long stalls when doing metadata-intensive operations on
> low memory such as a git checkout. I've been running 3.9-rc2 with the
> series applied but found that the interactive performance was awful even
> when there was plenty of free memory.

Can you try 3.9-rc4 or later and see if the problem still persists?
There were a number of ext4 issues especially around low memory
performance which weren't resolved until -rc4.

Thanks,

- Ted

2013-04-02 15:15:24

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 11:06:51AM -0400, Theodore Ts'o wrote:
>
> Can you try 3.9-rc4 or later and see if the problem still persists?
> There were a number of ext4 issues especially around low memory
> performance which weren't resolved until -rc4.

Actually, sorry, I took a closer look and I'm not as sure going to
-rc4 is going to help (although we did have some ext4 patches to fix a
number of bugs that flowed in as late as -rc4).

Can you send us the patch that you used to get record these long stall
times? And I assume you're using a laptop drive? 5400RPM or 7200RPM?

- Ted

2013-04-02 15:16:04

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 11:03:36PM +0800, Zheng Liu wrote:
> Hi Mel,
>
> Thanks for reporting it.
>
> On 04/02/2013 10:27 PM, Mel Gorman wrote:
> > I'm testing a page-reclaim-related series on my laptop that is partially
> > aimed at fixing long stalls when doing metadata-intensive operations on
> > low memory such as a git checkout. I've been running 3.9-rc2 with the
> > series applied but found that the interactive performance was awful even
> > when there was plenty of free memory.
> >
> > I activated a monitor from mmtests that logs when a process is stuck for
> > a long time in D state and found that there are a lot of stalls in ext4.
> > The report first states that processes have been stalled for a total of
> > 6498 seconds on IO which seems like a lot. Here is a breakdown of the
> > recorded events.
>
> In this merge window, we add a status tree as a extent cache. Meanwhile
> a es_cache shrinker is registered to try to reclaim from this cache when
> we are under a high memory pressure.

Ok.

> So I suspect that the root cause
> is this shrinker. Could you please tell me how to reproduce this
> problem? If I understand correctly, I can run mmtest to reproduce this
> problem, right?
>

This is normal desktop usage with some development thrown in, nothing
spectacular but nothing obviously reproducible either unfortuantely. I
just noticed that some git operations were taking abnormally long, mutt
was very slow opening mail, applications like mozilla were very slow to
launch etc. and dug a little further. I haven't checked if regression
tests under mmtests captured something similar yet.

--
Mel Gorman
SUSE Labs

2013-04-02 18:20:48

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

So I tried to reproduce the problem, and so I installed systemtap
(bleeding edge, since otherwise it won't work with development
kernel), and then rebuilt a kernel with all of the necessary CONFIG
options enabled:

CONFIG_DEBUG_INFO, CONFIG_KPROBES, CONFIG_RELAY, CONFIG_DEBUG_FS,
CONFIG_MODULES, CONFIG_MODULE_UNLOAD

I then pulled down mmtests, and tried running watch-dstate.pl, which
is what I sasume you were using, and I got a reminder of why I've
tried very hard to use systemtap:

semantic error: while resolving probe point: identifier 'kprobe' at /tmp/stapdjN4_l:18:7
source: probe kprobe.function("get_request_wait")
^

semantic error: no match
semantic error: while resolving probe point: identifier 'kprobe' at :74:8
source: }probe kprobe.function("get_request_wait").return
^

Pass 2: analysis failed. [man error::pass2]
Unexpected exit of STAP script at ./watch-dstate.pl line 296.

I have no clue what to do next. Can you give me a hint?

Thanks,

- Ted

2013-04-02 23:17:05

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

I've tried doing some quick timing, and if it is a performance
regression, it's not a recent one --- or I haven't been able to
reproduce what Mel is seeing. I tried the following commands while
booted into 3.2, 3.8, and 3.9-rc3 kernels:

time git clone ...
rm .git/index ; time git reset

I did this a number of git repo's; including one that was freshly
cloned, and one that had around 3 dozen patches applied via git am (so
there were a bunch of loose objects). And I tried doing this on an
SSD and a 5400rpm HDD, and I did it with all of the in-memory cache
flushed via "git 3 > /proc/sys/vm/drop_caches". The worst case was
doing a "time git reset" after deleting the .git/index file after
applying all of Kent Overstreet's recent AIO patches that had been
sent out for review. It took around 55 seconds, on 3.2, 3.8 and
3.9-rc3. That is pretty horrible, but for me that's the reason why I
use SSD's.

Mel, how bad is various git commands that you are trying? Have you
tried using time to get estimates of how long a git clone or other git
operation is taking?

- Ted

2013-04-03 10:19:32

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 11:14:36AM -0400, Theodore Ts'o wrote:
> On Tue, Apr 02, 2013 at 11:06:51AM -0400, Theodore Ts'o wrote:
> >
> > Can you try 3.9-rc4 or later and see if the problem still persists?
> > There were a number of ext4 issues especially around low memory
> > performance which weren't resolved until -rc4.
>
> Actually, sorry, I took a closer look and I'm not as sure going to
> -rc4 is going to help (although we did have some ext4 patches to fix a
> number of bugs that flowed in as late as -rc4).
>

I'm running with -rc5 now. I have not noticed much interactivity problems
as such but the stall detection script reported that mutt stalled for
20 seconds opening an inbox and imapd blocked for 59 seconds doing path
lookups, imaps blocked again for 12 seconds doing an atime update, an RSS
reader blocked for 3.5 seconds writing a file. etc.

There has been no reclaim activity in the system yet and 2G is still free
so it's very unlikely to be a page or slab reclaim problem.

> Can you send us the patch that you used to get record these long stall
> times?

No patch but it depends on systemtap which you are already aware is a wreck
to work with and frequently breaks between kernel versions for a variety of
reasons. Minimally, it is necessary to revert commit ba6fdda4 (profiling:
Remove unused timer hook) to get systemtap working. I've reported this
problem to the patch author and the systemtap mailing list.

Other workarounds are necessary so I updated mmtests in git and at
http://www.csn.ul.ie/~mel/projects/mmtests/mmtests-0.10-mmtests-0.01.tar.gz
. Download and untar it

1. stap can be "fixed" by running bin/stap-fix.sh . It will try and run
a one-liner stap script and if that fails it'll try very crude workarounds.
Your milage may vary considerably

2. If you want to run the monitor script yourself, it's
sudo monitors/watch-dstate.pl | tee /tmp/foo.log

but be aware the log may be truncated due to buffeering. Optionally you
can avoid the buffered write problem by running mmtests as

sudo ./run-mmtests.sh --config configs/config-monitor-interactive stall-debug

and the log will be in work/log/dstate-stall-debug-monitor.gz

3. Summarise the report with

cat /tmp/foo.log | subreport/stap-dstate-frequency

I'll be digging through other mmtests results shortly to see if I already
have a better reproduction case that is eligible for bisection but those
results are based on different machines so no guarantees of success.

> And I assume you're using a laptop drive? 5400RPM or 7200RPM?
>

Yes, laptop drive, 7200RPM. CFQ scheduler. Drive queue depth is 32.

/dev/sda:

ATA device, with non-removable media
Model Number: ST9320423AS
Serial Number: 5VH5M0LY
Firmware Revision: 0003LVM1
Transport: Serial
Standards:
Used: unknown (minor revision code 0x0029)
Supported: 8 7 6 5
Likely used: 8
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 268435455
LBA48 user addressable sectors: 625142448
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
device size with M = 1024*1024: 305245 MBytes
device size with M = 1000*1000: 320072 MBytes (320 GB)
cache/buffer size = 16384 KBytes
Nominal Media Rotation Rate: 7200
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: 128
Recommended acoustic management value: 254, current value: 0
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* DOWNLOAD_MICROCODE
* Advanced Power Management feature set
SET_MAX security extension
* 48-bit Address feature set
* Device Configuration Overlay feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* 64-bit World wide name
* IDLE_IMMEDIATE with UNLOAD
* Write-Read-Verify feature set
* WRITE_UNCORRECTABLE_EXT command
* {READ,WRITE}_DMA_EXT_GPL commands
* Segmented DOWNLOAD_MICROCODE
* Gen1 signaling speed (1.5Gb/s)
* Gen2 signaling speed (3.0Gb/s)
* Native Command Queueing (NCQ)
* Phy event counters
Device-initiated interface power management
* Software settings preservation
* SMART Command Transport (SCT) feature set
* SCT Read/Write Long (AC1), obsolete
* SCT Error Recovery Control (AC3)
* SCT Features Control (AC4)
* SCT Data Tables (AC5)
unknown 206[12] (vendor specific)
Security:
Master password revision code = 65534
supported
not enabled
not locked
frozen
not expired: security count
supported: enhanced erase
70min for SECURITY ERASE UNIT. 70min for ENHANCED SECURITY ERASE UNIT.
Logical Unit WWN Device Identifier: 5000c5002f2d395d
NAA : 5
IEEE OUI : 000c50
Unique ID : 02f2d395d
Checksum: correct

--
Mel Gorman
SUSE Labs

2013-04-03 12:06:27

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Wed, Apr 03, 2013 at 11:19:25AM +0100, Mel Gorman wrote:
>
> I'm running with -rc5 now. I have not noticed much interactivity problems
> as such but the stall detection script reported that mutt stalled for
> 20 seconds opening an inbox and imapd blocked for 59 seconds doing path
> lookups, imaps blocked again for 12 seconds doing an atime update, an RSS
> reader blocked for 3.5 seconds writing a file. etc.

If imaps blocked for 12 seconds during an atime update, combined with
everything else, at a guess it got caught by something holding up a
journal commit. Could you try enabling the jbd2_run_stats tracepoint
and grabbing the trace log? This will give you statistics on how long
(in milliseconds) each of the various phases of a jbd2 commit is
taking, i.e.:

jbd2/sdb1-8-327 [002] .... 39681.874661: jbd2_run_stats: dev 8,17 tid 7163786 wait 0 request_delay 0 running 3530 locked 0 flushing 0 logging 0 handle_count 75 blocks 8 blocks_logged 9
jbd2/sdb1-8-327 [003] .... 39682.514153: jbd2_run_stats: dev 8,17 tid 7163787 wait 0 request_delay 0 running 640 locked 0 flushing 0 logging 0 handle_count 39 blocks 12 blocks_logged 13
jbd2/sdb1-8-327 [000] .... 39687.665609: jbd2_run_stats: dev 8,17 tid 7163788 wait 0 request_delay 0 running 5150 locked 0 flushing 0 logging 0 handle_count 60 blocks 13 blocks_logged 14
jbd2/sdb1-8-327 [000] .... 39693.200453: jbd2_run_stats: dev 8,17 tid 7163789 wait 0 request_delay 0 running 4840 locked 0 flushing 0 logging 0 handle_count 53 blocks 10 blocks_logged 11
jbd2/sdb1-8-327 [001] .... 39695.061657: jbd2_run_stats: dev 8,17 tid 7163790 wait 0 request_delay 0 running 1860 locked 0 flushing 0 logging 0 handle_count 124 blocks 19 blocks_logged 20

In the above sample each journal commit is running for no more than 5
seconds or so (since that's the default jbd2 commit timeout; if a
transaction is running for less than 5 seconds, then either we ran out
of room in the journal, and the blocks_logged number will be high, or
a commit was forced by something such as an fsync call).

If an atime update is getting blocked by 12 seconds, then it would be
interesting to see if a journal commit is running for significantly
longer than 5 seconds, or if one of the other commit phases is taking
significant amounts of time. (On the example above they are all
taking no time, since I ran this on a relatively uncontended system;
only a single git operation taking place.)

Something else that might be worth trying is grabbing a lock_stat
report and see if something is sitting on an ext4 or jbd2 mutex for a
long time.

Finally, as I mentioned I tried some rather simplistic tests and I
didn't notice any difference between a 3.2 kernel and a 3.8/3.9-rc5
kernel. Assuming you can get a version of systemtap that
simultaneously works on 3.2 and 3.9-rc5 :-P, and chance you could do a
quick experiment and see if you're seeing a difference on your setup?

Thanks!!

- Ted

2013-04-03 15:15:44

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Wed, Apr 03, 2013 at 08:05:30AM -0400, Theodore Ts'o wrote:
> On Wed, Apr 03, 2013 at 11:19:25AM +0100, Mel Gorman wrote:
> >
> > I'm running with -rc5 now. I have not noticed much interactivity problems
> > as such but the stall detection script reported that mutt stalled for
> > 20 seconds opening an inbox and imapd blocked for 59 seconds doing path
> > lookups, imaps blocked again for 12 seconds doing an atime update, an RSS
> > reader blocked for 3.5 seconds writing a file. etc.
>
> If imaps blocked for 12 seconds during an atime update, combined with
> everything else, at a guess it got caught by something holding up a
> journal commit.

It's a possibility.

I apologise but I forgot that mail is stored on a crypted partition on
this machine. It's formatted ext4 but dmcrypt could be making this problem
worse if it's stalling ext4 waiting to encrypt/decrypt data due to either
a scheduler or workqueue change.

> Could you try enabling the jbd2_run_stats tracepoint
> and grabbing the trace log? This will give you statistics on how long
> (in milliseconds) each of the various phases of a jbd2 commit is
> taking, i.e.:
>
> jbd2/sdb1-8-327 [002] .... 39681.874661: jbd2_run_stats: dev 8,17 tid 7163786 wait 0 request_delay 0 running 3530 locked 0 flushing 0 logging 0 handle_count 75 blocks 8 blocks_logged 9
> jbd2/sdb1-8-327 [003] .... 39682.514153: jbd2_run_stats: dev 8,17 tid 7163787 wait 0 request_delay 0 running 640 locked 0 flushing 0 logging 0 handle_count 39 blocks 12 blocks_logged 13
> jbd2/sdb1-8-327 [000] .... 39687.665609: jbd2_run_stats: dev 8,17 tid 7163788 wait 0 request_delay 0 running 5150 locked 0 flushing 0 logging 0 handle_count 60 blocks 13 blocks_logged 14
> jbd2/sdb1-8-327 [000] .... 39693.200453: jbd2_run_stats: dev 8,17 tid 7163789 wait 0 request_delay 0 running 4840 locked 0 flushing 0 logging 0 handle_count 53 blocks 10 blocks_logged 11
> jbd2/sdb1-8-327 [001] .... 39695.061657: jbd2_run_stats: dev 8,17 tid 7163790 wait 0 request_delay 0 running 1860 locked 0 flushing 0 logging 0 handle_count 124 blocks 19 blocks_logged 20
>

Attached as well as the dstate summary that was recorded at the same
time. It's not quite as compelling but I'll keep the monitor running and
see if something falls out. I didn't find anything useful in the existing
mmtests tests that could be used to bisect this but not many of them are
focused on IO.

> In the above sample each journal commit is running for no more than 5
> seconds or so (since that's the default jbd2 commit timeout; if a
> transaction is running for less than 5 seconds, then either we ran out
> of room in the journal, and the blocks_logged number will be high, or
> a commit was forced by something such as an fsync call).
>

I didn't see anything majorly compelling in the jbd tracepoints but I'm
not 100% sure I'm looking for the right thing either. I also recorded
/proc/latency_stat and there were some bad sync latencies from the file
as you can see here

3 4481 1586 jbd2_log_wait_commit ext4_sync_file vfs_fsync sys_msync system_call_fastpath
3 11325 4373 sleep_on_page wait_on_page_bit kretprobe_trampoline filemap_write_and_wait_range ext4_sync_file vfs_fsync sys_msync system_call_fastpath
85 1130707 14904 jbd2_journal_stop jbd2_journal_force_commit ext4_force_commit ext4_sync_file do_fsync sys_fsync system_call_fastpath
1 2161073 2161073 start_this_handle jbd2__journal_start.part.8 jbd2__journal_start __ext4_journal_start_sb ext4_da_writepages do_writepages __filemap_fdatawrite_range filemap_write_and_wait_range ext4_sync_file do_fsync sys_fsync system_call_fastpath
118 7798435 596184 jbd2_log_wait_commit jbd2_journal_stop jbd2_journal_force_commit ext4_force_commit ext4_sync_file do_fsync sys_fsync system_call_fastpath
599 15496449 3405822 sleep_on_page wait_on_page_bit kretprobe_trampoline filemap_write_and_wait_range ext4_sync_file do_fsync sys_fsync system_call_fastpath
405 28572881 2619592 jbd2_log_wait_commit ext4_sync_file do_fsync sys_fsync system_call_fastpath


> If an atime update is getting blocked by 12 seconds, then it would be
> interesting to see if a journal commit is running for significantly
> longer than 5 seconds, or if one of the other commit phases is taking
> significant amounts of time. (On the example above they are all
> taking no time, since I ran this on a relatively uncontended system;
> only a single git operation taking place.)
>
> Something else that might be worth trying is grabbing a lock_stat
> report and see if something is sitting on an ext4 or jbd2 mutex for a
> long time.
>

Ok, if nothing useful falls out in this session I'll enable lock
debugging. latency_stat on its own would not be enough to conclude that
a problem was related to lock contention.

> Finally, as I mentioned I tried some rather simplistic tests and I
> didn't notice any difference between a 3.2 kernel and a 3.8/3.9-rc5
> kernel. Assuming you can get a version of systemtap that
> simultaneously works on 3.2 and 3.9-rc5 :-P, and chance you could do a
> quick experiment and see if you're seeing a difference on your setup?
>

stap-fix.sh should be able to kick systemtap sufficiently hard for
either 3.2 or 3.9-rc5 to keep it working. I'll keep digging when I can.

--
Mel Gorman
SUSE Labs


Attachments:
(No filename) (5.18 kB)
dstate-summary.txt (19.91 kB)
ftrace-debug-stalls-monitor.gz (13.36 kB)
Download all attachments

2013-04-03 15:22:55

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 07:16:13PM -0400, Theodore Ts'o wrote:
> I've tried doing some quick timing, and if it is a performance
> regression, it's not a recent one --- or I haven't been able to
> reproduce what Mel is seeing. I tried the following commands while
> booted into 3.2, 3.8, and 3.9-rc3 kernels:
>
> time git clone ...
> rm .git/index ; time git reset
>

FWIW, I had run a number if git checkout based tests over time and none
of them revealed anything useful. Granted it was on other machines but I
don't think it's git on its own. It's a combination that leads to this
problem. Maybe it's really an IO scheduler problem and I need to figure
out what combination triggers it.

> <SNIP>
>
> Mel, how bad is various git commands that you are trying? Have you
> tried using time to get estimates of how long a git clone or other git
> operation is taking?
>

Unfortunately, the milage varies considerably and it's not always
possible to time the operation. It may be that one occasion that opening
a mail takes an abnormal length time with git operations occasionally
making it far worse.

--
Mel Gorman
SUSE Labs

2013-04-05 22:18:19

by Jiri Slaby

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On 04/03/2013 12:19 PM, Mel Gorman wrote:
> On Tue, Apr 02, 2013 at 11:14:36AM -0400, Theodore Ts'o wrote:
>> On Tue, Apr 02, 2013 at 11:06:51AM -0400, Theodore Ts'o wrote:
>>>
>>> Can you try 3.9-rc4 or later and see if the problem still persists?
>>> There were a number of ext4 issues especially around low memory
>>> performance which weren't resolved until -rc4.
>>
>> Actually, sorry, I took a closer look and I'm not as sure going to
>> -rc4 is going to help (although we did have some ext4 patches to fix a
>> number of bugs that flowed in as late as -rc4).
>>
>
> I'm running with -rc5 now. I have not noticed much interactivity problems
> as such but the stall detection script reported that mutt stalled for
> 20 seconds opening an inbox and imapd blocked for 59 seconds doing path
> lookups, imaps blocked again for 12 seconds doing an atime update, an RSS
> reader blocked for 3.5 seconds writing a file. etc.
>
> There has been no reclaim activity in the system yet and 2G is still free
> so it's very unlikely to be a page or slab reclaim problem.

Ok, so now I'm runnning 3.9.0-rc5-next-20130404, it's not that bad, but
it still sucks. Updating a kernel in a VM still results in "Your system
is too SLOW to play this!" by mplayer and frame dropping.

3.5G out of 6G memory used, the rest is I/O cache.

I have 7200RPM disks in my desktop.

--
js
suse labs

2013-04-05 23:17:25

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Sat, Apr 06, 2013 at 12:18:11AM +0200, Jiri Slaby wrote:
> Ok, so now I'm runnning 3.9.0-rc5-next-20130404, it's not that bad, but
> it still sucks. Updating a kernel in a VM still results in "Your system
> is too SLOW to play this!" by mplayer and frame dropping.

What was the first kernel where you didn't have the problem? Were you
using the 3.8 kernel earlier, and did you see the interactivity
problems there?

What else was running in on your desktop at the same time? How was
the file system mounted, and can you send me the output of dumpe2fs -h
/dev/XXX? Oh, and what options were you using to when you kicked off
the VM?

The other thing that would be useful was to enable the jbd2_run_stats
tracepoint and to send the output of the trace log when you notice the
interactivity problems.

Thanks,

- Ted

2013-04-06 07:29:55

by Jiri Slaby

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On 04/06/2013 01:16 AM, Theodore Ts'o wrote:
> On Sat, Apr 06, 2013 at 12:18:11AM +0200, Jiri Slaby wrote:
>> Ok, so now I'm runnning 3.9.0-rc5-next-20130404, it's not that bad, but
>> it still sucks. Updating a kernel in a VM still results in "Your system
>> is too SLOW to play this!" by mplayer and frame dropping.
>
> What was the first kernel where you didn't have the problem? Were you
> using the 3.8 kernel earlier, and did you see the interactivity
> problems there?

I'm not sure, as I am using -next like for ever. But sure, there was a
kernel which didn't ahve this problem.

> What else was running in on your desktop at the same time?

Nothing, just VM (kernel update from console) and mplayer2 on the host.
This is more-or-less reproducible with these two.

> How was
> the file system mounted,

Both are actually a single device /dev/sda5:
/dev/sda5 on /win type ext4 (rw,noatime,data=ordered)

Should I try writeback?

> and can you send me the output of dumpe2fs -h
> /dev/XXX?

dumpe2fs 1.42.7 (21-Jan-2013)
Filesystem volume name: <none>
Last mounted on: /win
Filesystem UUID: cd4bf4d2-bc32-4777-a437-ee24c4ee5f1b
Filesystem magic number: 0xEF53
Filesystem revision #: 1 (dynamic)
Filesystem features: has_journal ext_attr resize_inode dir_index
filetype needs_recovery extent flex_bg sparse_super large_file huge_file
uninit_bg dir_nlink extra_isize
Filesystem flags: signed_directory_hash
Default mount options: user_xattr acl
Filesystem state: clean
Errors behavior: Continue
Filesystem OS type: Linux
Inode count: 30507008
Block count: 122012416
Reserved block count: 0
Free blocks: 72021328
Free inodes: 30474619
First block: 0
Block size: 4096
Fragment size: 4096
Reserved GDT blocks: 994
Blocks per group: 32768
Fragments per group: 32768
Inodes per group: 8192
Inode blocks per group: 512
RAID stride: 32747
Flex block group size: 16
Filesystem created: Fri Sep 7 20:44:21 2012
Last mount time: Thu Apr 4 12:22:01 2013
Last write time: Thu Apr 4 12:22:01 2013
Mount count: 256
Maximum mount count: -1
Last checked: Sat Sep 8 21:13:28 2012
Check interval: 0 (<none>)
Lifetime writes: 1011 GB
Reserved blocks uid: 0 (user root)
Reserved blocks gid: 0 (group root)
First inode: 11
Inode size: 256
Required extra isize: 28
Desired extra isize: 28
Journal inode: 8
Default directory hash: half_md4
Directory Hash Seed: b6ad3f8b-72ce-49d6-92cb-abccd7dbe98e
Journal backup: inode blocks
Journal features: journal_incompat_revoke
Journal size: 128M
Journal length: 32768
Journal sequence: 0x00054dc7
Journal start: 8193

> Oh, and what options were you using to when you kicked off
> the VM?

qemu-kvm -k en-us -smp 2 -m 1200 -soundhw hda -usb -usbdevice tablet
-net user -net nic,model=e1000 -serial pty -balloon virtio -hda x.img

> The other thing that would be useful was to enable the jbd2_run_stats
> tracepoint and to send the output of the trace log when you notice the
> interactivity problems.

Ok, I will try.

thanks,
--
js
suse labs

2013-04-06 07:37:49

by Jiri Slaby

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On 04/06/2013 09:29 AM, Jiri Slaby wrote:
> On 04/06/2013 01:16 AM, Theodore Ts'o wrote:
>> On Sat, Apr 06, 2013 at 12:18:11AM +0200, Jiri Slaby wrote:
>>> Ok, so now I'm runnning 3.9.0-rc5-next-20130404, it's not that bad, but
>>> it still sucks. Updating a kernel in a VM still results in "Your system
>>> is too SLOW to play this!" by mplayer and frame dropping.
>>
>> What was the first kernel where you didn't have the problem? Were you
>> using the 3.8 kernel earlier, and did you see the interactivity
>> problems there?
>
> I'm not sure, as I am using -next like for ever. But sure, there was a
> kernel which didn't ahve this problem.
>
>> What else was running in on your desktop at the same time?
>
> Nothing, just VM (kernel update from console) and mplayer2 on the host.
> This is more-or-less reproducible with these two.

Ok,
dd if=/dev/zero of=xxx
is enough instead of "kernel update".

Writeback mount doesn't help.

>> How was
>> the file system mounted,
>
> Both are actually a single device /dev/sda5:
> /dev/sda5 on /win type ext4 (rw,noatime,data=ordered)
>
> Should I try writeback?
>
>> and can you send me the output of dumpe2fs -h
>> /dev/XXX?
>
> dumpe2fs 1.42.7 (21-Jan-2013)
> Filesystem volume name: <none>
> Last mounted on: /win
> Filesystem UUID: cd4bf4d2-bc32-4777-a437-ee24c4ee5f1b
> Filesystem magic number: 0xEF53
> Filesystem revision #: 1 (dynamic)
> Filesystem features: has_journal ext_attr resize_inode dir_index
> filetype needs_recovery extent flex_bg sparse_super large_file huge_file
> uninit_bg dir_nlink extra_isize
> Filesystem flags: signed_directory_hash
> Default mount options: user_xattr acl
> Filesystem state: clean
> Errors behavior: Continue
> Filesystem OS type: Linux
> Inode count: 30507008
> Block count: 122012416
> Reserved block count: 0
> Free blocks: 72021328
> Free inodes: 30474619
> First block: 0
> Block size: 4096
> Fragment size: 4096
> Reserved GDT blocks: 994
> Blocks per group: 32768
> Fragments per group: 32768
> Inodes per group: 8192
> Inode blocks per group: 512
> RAID stride: 32747
> Flex block group size: 16
> Filesystem created: Fri Sep 7 20:44:21 2012
> Last mount time: Thu Apr 4 12:22:01 2013
> Last write time: Thu Apr 4 12:22:01 2013
> Mount count: 256
> Maximum mount count: -1
> Last checked: Sat Sep 8 21:13:28 2012
> Check interval: 0 (<none>)
> Lifetime writes: 1011 GB
> Reserved blocks uid: 0 (user root)
> Reserved blocks gid: 0 (group root)
> First inode: 11
> Inode size: 256
> Required extra isize: 28
> Desired extra isize: 28
> Journal inode: 8
> Default directory hash: half_md4
> Directory Hash Seed: b6ad3f8b-72ce-49d6-92cb-abccd7dbe98e
> Journal backup: inode blocks
> Journal features: journal_incompat_revoke
> Journal size: 128M
> Journal length: 32768
> Journal sequence: 0x00054dc7
> Journal start: 8193
>
>> Oh, and what options were you using to when you kicked off
>> the VM?
>
> qemu-kvm -k en-us -smp 2 -m 1200 -soundhw hda -usb -usbdevice tablet
> -net user -net nic,model=e1000 -serial pty -balloon virtio -hda x.img
>
>> The other thing that would be useful was to enable the jbd2_run_stats
>> tracepoint and to send the output of the trace log when you notice the
>> interactivity problems.
>
> Ok, I will try.
>
> thanks,
>


--
js
suse labs

2013-04-06 08:20:10

by Jiri Slaby

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On 04/06/2013 09:37 AM, Jiri Slaby wrote:
> On 04/06/2013 09:29 AM, Jiri Slaby wrote:
>> On 04/06/2013 01:16 AM, Theodore Ts'o wrote:
>>> On Sat, Apr 06, 2013 at 12:18:11AM +0200, Jiri Slaby wrote:
>>>> Ok, so now I'm runnning 3.9.0-rc5-next-20130404, it's not that bad, but
>>>> it still sucks. Updating a kernel in a VM still results in "Your system
>>>> is too SLOW to play this!" by mplayer and frame dropping.
>>>
>>> What was the first kernel where you didn't have the problem? Were you
>>> using the 3.8 kernel earlier, and did you see the interactivity
>>> problems there?
>>
>> I'm not sure, as I am using -next like for ever. But sure, there was a
>> kernel which didn't ahve this problem.
>>
>>> What else was running in on your desktop at the same time?
>>
>> Nothing, just VM (kernel update from console) and mplayer2 on the host.
>> This is more-or-less reproducible with these two.
>
> Ok,
> dd if=/dev/zero of=xxx
> is enough instead of "kernel update".
>
> Writeback mount doesn't help.
>
>>> How was
>>> the file system mounted,
>>
>> Both are actually a single device /dev/sda5:
>> /dev/sda5 on /win type ext4 (rw,noatime,data=ordered)
>>
>> Should I try writeback?
>>
>>> and can you send me the output of dumpe2fs -h
>>> /dev/XXX?
>>
>> dumpe2fs 1.42.7 (21-Jan-2013)
>> Filesystem volume name: <none>
>> Last mounted on: /win
>> Filesystem UUID: cd4bf4d2-bc32-4777-a437-ee24c4ee5f1b
>> Filesystem magic number: 0xEF53
>> Filesystem revision #: 1 (dynamic)
>> Filesystem features: has_journal ext_attr resize_inode dir_index
>> filetype needs_recovery extent flex_bg sparse_super large_file huge_file
>> uninit_bg dir_nlink extra_isize
>> Filesystem flags: signed_directory_hash
>> Default mount options: user_xattr acl
>> Filesystem state: clean
>> Errors behavior: Continue
>> Filesystem OS type: Linux
>> Inode count: 30507008
>> Block count: 122012416
>> Reserved block count: 0
>> Free blocks: 72021328
>> Free inodes: 30474619
>> First block: 0
>> Block size: 4096
>> Fragment size: 4096
>> Reserved GDT blocks: 994
>> Blocks per group: 32768
>> Fragments per group: 32768
>> Inodes per group: 8192
>> Inode blocks per group: 512
>> RAID stride: 32747
>> Flex block group size: 16
>> Filesystem created: Fri Sep 7 20:44:21 2012
>> Last mount time: Thu Apr 4 12:22:01 2013
>> Last write time: Thu Apr 4 12:22:01 2013
>> Mount count: 256
>> Maximum mount count: -1
>> Last checked: Sat Sep 8 21:13:28 2012
>> Check interval: 0 (<none>)
>> Lifetime writes: 1011 GB
>> Reserved blocks uid: 0 (user root)
>> Reserved blocks gid: 0 (group root)
>> First inode: 11
>> Inode size: 256
>> Required extra isize: 28
>> Desired extra isize: 28
>> Journal inode: 8
>> Default directory hash: half_md4
>> Directory Hash Seed: b6ad3f8b-72ce-49d6-92cb-abccd7dbe98e
>> Journal backup: inode blocks
>> Journal features: journal_incompat_revoke
>> Journal size: 128M
>> Journal length: 32768
>> Journal sequence: 0x00054dc7
>> Journal start: 8193
>>
>>> Oh, and what options were you using to when you kicked off
>>> the VM?
>>
>> qemu-kvm -k en-us -smp 2 -m 1200 -soundhw hda -usb -usbdevice tablet
>> -net user -net nic,model=e1000 -serial pty -balloon virtio -hda x.img
>>
>>> The other thing that would be useful was to enable the jbd2_run_stats
>>> tracepoint and to send the output of the trace log when you notice the
>>> interactivity problems.
>>
>> Ok, I will try.

Inline here, as well as attached:
# tracer: nop
#
# entries-in-buffer/entries-written: 46/46 #P:2
#
# _-----=> irqs-off
# / _----=> need-resched
# | / _---=> hardirq/softirq
# || / _--=> preempt-depth
# ||| / delay
# TASK-PID CPU# |||| TIMESTAMP FUNCTION
# | | | |||| | |
jbd2/sda5-8-10969 [000] .... 387.054319: jbd2_run_stats: dev
259,655360 tid 348892 wait 0 request_delay 0 running 5728 locked 0
flushing 0 logging 28 handle_count 10 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 392.594132: jbd2_run_stats: dev
259,655360 tid 348893 wait 0 request_delay 0 running 5300 locked 0
flushing 0 logging 64 handle_count 75944 blocks 1 blocks_logged 2
jbd2/md2-8-959 [000] .... 396.249990: jbd2_run_stats: dev 9,2
tid 382990 wait 0 request_delay 0 running 5500 locked 0 flushing 0
logging 220 handle_count 3 blocks 1 blocks_logged 2
jbd2/md1-8-1826 [000] .... 397.205670: jbd2_run_stats: dev 9,1
tid 1081270 wait 0 request_delay 0 running 5760 locked 0 flushing 0
logging 200 handle_count 2 blocks 0 blocks_logged 0
jbd2/sda5-8-10969 [000] .... 397.563660: jbd2_run_stats: dev
259,655360 tid 348894 wait 0 request_delay 0 running 5000 locked 0
flushing 0 logging 32 handle_count 89397 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 403.679552: jbd2_run_stats: dev
259,655360 tid 348895 wait 0 request_delay 0 running 5000 locked 1040
flushing 0 logging 112 handle_count 148224 blocks 1 blocks_logged 2
jbd2/md1-8-1826 [000] .... 407.981693: jbd2_run_stats: dev 9,1
tid 1081271 wait 0 request_delay 0 running 5064 locked 0 flushing 0
logging 152 handle_count 198 blocks 20 blocks_logged 21
jbd2/md2-8-959 [000] .... 408.111339: jbd2_run_stats: dev 9,2
tid 382991 wait 0 request_delay 0 running 5156 locked 2268 flushing 0
logging 124 handle_count 5 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 408.823650: jbd2_run_stats: dev
259,655360 tid 348896 wait 0 request_delay 0 running 5156 locked 0
flushing 0 logging 100 handle_count 63257 blocks 1 blocks_logged 2
jbd2/md1-8-1826 [000] .... 411.385104: jbd2_run_stats: dev 9,1
tid 1081272 wait 0 request_delay 0 running 3236 locked 0 flushing 0
logging 116 handle_count 42 blocks 7 blocks_logged 8
jbd2/md1-8-1826 [000] .... 412.590289: jbd2_run_stats: dev 9,1
tid 1081273 wait 0 request_delay 0 running 124 locked 0 flushing 0
logging 740 handle_count 7 blocks 5 blocks_logged 6
jbd2/md2-8-959 [000] .... 413.087300: jbd2_run_stats: dev 9,2
tid 382992 wait 0 request_delay 0 running 5012 locked 0 flushing 0
logging 92 handle_count 12 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 414.047500: jbd2_run_stats: dev
259,655360 tid 348897 wait 0 request_delay 0 running 5004 locked 32
flushing 0 logging 292 handle_count 104485 blocks 4 blocks_logged 5
jbd2/md2-8-959 [000] .... 418.301823: jbd2_run_stats: dev 9,2
tid 382993 wait 0 request_delay 0 running 5024 locked 0 flushing 0
logging 284 handle_count 4 blocks 0 blocks_logged 0
jbd2/md1-8-1826 [001] .... 418.384624: jbd2_run_stats: dev 9,1
tid 1081274 wait 0 request_delay 0 running 5416 locked 0 flushing 0
logging 384 handle_count 393 blocks 14 blocks_logged 15
jbd2/sda5-8-10969 [000] .... 418.599524: jbd2_run_stats: dev
259,655360 tid 348898 wait 0 request_delay 0 running 4736 locked 0
flushing 0 logging 112 handle_count 43360 blocks 17 blocks_logged 18
jbd2/md1-8-1826 [001] .... 418.711491: jbd2_run_stats: dev 9,1
tid 1081275 wait 0 request_delay 0 running 40 locked 0 flushing 0
logging 48 handle_count 4 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 422.444437: jbd2_run_stats: dev
259,655360 tid 348899 wait 0 request_delay 0 running 3684 locked 0
flushing 0 logging 144 handle_count 62564 blocks 22 blocks_logged 23
jbd2/sda5-8-10969 [000] .... 427.903435: jbd2_run_stats: dev
259,655360 tid 348900 wait 0 request_delay 0 running 5332 locked 0
flushing 0 logging 128 handle_count 118362 blocks 19 blocks_logged 20
jbd2/sda5-8-10969 [000] .... 431.981049: jbd2_run_stats: dev
259,655360 tid 348901 wait 0 request_delay 0 running 3976 locked 0
flushing 0 logging 100 handle_count 88833 blocks 13 blocks_logged 14
jbd2/md1-8-1826 [001] .... 437.291566: jbd2_run_stats: dev 9,1
tid 1081276 wait 0 request_delay 0 running 244 locked 0 flushing 0
logging 380 handle_count 5 blocks 6 blocks_logged 7
jbd2/sda5-8-10969 [000] .... 437.342205: jbd2_run_stats: dev
259,655360 tid 348902 wait 0 request_delay 0 running 5016 locked 0
flushing 0 logging 344 handle_count 134290 blocks 13 blocks_logged 14
jbd2/sda5-8-10969 [000] .... 441.879748: jbd2_run_stats: dev
259,655360 tid 348903 wait 0 request_delay 0 running 3624 locked 0
flushing 0 logging 76 handle_count 81013 blocks 13 blocks_logged 14
jbd2/sda5-8-10969 [000] .... 447.059645: jbd2_run_stats: dev
259,655360 tid 348904 wait 0 request_delay 0 running 5048 locked 0
flushing 0 logging 128 handle_count 127735 blocks 13 blocks_logged 14
jbd2/sda5-8-10969 [001] .... 447.667205: jbd2_run_stats: dev
259,655360 tid 348905 wait 0 request_delay 0 running 580 locked 0
flushing 0 logging 156 handle_count 131 blocks 4 blocks_logged 5
jbd2/sda5-8-10969 [001] .... 453.156101: jbd2_run_stats: dev
259,655360 tid 348906 wait 0 request_delay 0 running 5308 locked 0
flushing 0 logging 184 handle_count 109134 blocks 16 blocks_logged 17
jbd2/sda5-8-10969 [001] .... 456.546335: jbd2_run_stats: dev
259,655360 tid 348907 wait 0 request_delay 0 running 3248 locked 0
flushing 0 logging 228 handle_count 66315 blocks 10 blocks_logged 11
jbd2/md2-8-959 [001] .... 458.812838: jbd2_run_stats: dev 9,2
tid 382994 wait 0 request_delay 0 running 5052 locked 92 flushing 0
logging 232 handle_count 8 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 462.113411: jbd2_run_stats: dev
259,655360 tid 348908 wait 0 request_delay 0 running 5292 locked 4
flushing 0 logging 268 handle_count 139470 blocks 14 blocks_logged 15
jbd2/md2-8-959 [001] .... 463.012109: jbd2_run_stats: dev 9,2
tid 382995 wait 0 request_delay 0 running 4380 locked 0 flushing 0
logging 52 handle_count 3 blocks 0 blocks_logged 0
jbd2/sda5-8-10969 [000] .... 463.012121: jbd2_run_stats: dev
259,655360 tid 348909 wait 0 request_delay 0 running 1116 locked 0
flushing 0 logging 52 handle_count 5 blocks 4 blocks_logged 5
jbd2/sda5-8-10969 [001] .... 468.229949: jbd2_run_stats: dev
259,655360 tid 348910 wait 0 request_delay 0 running 5012 locked 0
flushing 0 logging 204 handle_count 134170 blocks 18 blocks_logged 19
jbd2/md2-8-959 [000] .... 473.230180: jbd2_run_stats: dev 9,2
tid 382996 wait 0 request_delay 0 running 5116 locked 0 flushing 0
logging 268 handle_count 3 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 473.422616: jbd2_run_stats: dev
259,655360 tid 348911 wait 0 request_delay 0 running 5292 locked 0
flushing 0 logging 108 handle_count 84844 blocks 15 blocks_logged 16
jbd2/md1-8-1826 [000] .... 477.503164: jbd2_run_stats: dev 9,1
tid 1081277 wait 0 request_delay 0 running 5580 locked 0 flushing 0
logging 852 handle_count 124 blocks 4 blocks_logged 5
jbd2/sda5-8-10969 [000] .... 479.048020: jbd2_run_stats: dev
259,655360 tid 348912 wait 0 request_delay 0 running 5000 locked 212
flushing 0 logging 416 handle_count 139926 blocks 17 blocks_logged 18
jbd2/md1-8-1826 [000] .... 482.570545: jbd2_run_stats: dev 9,1
tid 1081278 wait 0 request_delay 0 running 5316 locked 0 flushing 0
logging 604 handle_count 11 blocks 0 blocks_logged 0
jbd2/sda5-8-10969 [001] .... 484.456879: jbd2_run_stats: dev
259,655360 tid 348913 wait 0 request_delay 0 running 5284 locked 0
flushing 0 logging 544 handle_count 40620 blocks 11 blocks_logged 12
jbd2/sda5-8-10969 [001] .... 486.014655: jbd2_run_stats: dev
259,655360 tid 348914 wait 0 request_delay 0 running 1540 locked 108
flushing 0 logging 456 handle_count 55965 blocks 4 blocks_logged 5
jbd2/sda5-8-10969 [001] .... 491.082420: jbd2_run_stats: dev
259,655360 tid 348915 wait 0 request_delay 0 running 5160 locked 0
flushing 0 logging 368 handle_count 33509 blocks 12 blocks_logged 13
jbd2/md1-8-1826 [000] .... 494.688094: jbd2_run_stats: dev 9,1
tid 1081279 wait 0 request_delay 0 running 5828 locked 0 flushing 0
logging 716 handle_count 2 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 497.548126: jbd2_run_stats: dev
259,655360 tid 348916 wait 0 request_delay 0 running 5020 locked 36
flushing 0 logging 1780 handle_count 1481 blocks 13 blocks_logged 14
jbd2/md2-8-959 [000] .... 500.647267: jbd2_run_stats: dev 9,2
tid 382997 wait 0 request_delay 0 running 5272 locked 244 flushing 0
logging 432 handle_count 5 blocks 1 blocks_logged 2
jbd2/sda5-8-10969 [000] .... 501.134535: jbd2_run_stats: dev
259,655360 tid 348917 wait 0 request_delay 0 running 5040 locked 0
flushing 0 logging 328 handle_count 755 blocks 4 blocks_logged 5
jbd2/md1-8-1826 [001] .... 502.020846: jbd2_run_stats: dev 9,1
tid 1081280 wait 0 request_delay 0 running 5896 locked 0 flushing 0
logging 52 handle_count 20 blocks 5 blocks_logged 6
jbd2/md2-8-959 [000] .... 505.989307: jbd2_run_stats: dev 9,2
tid 382998 wait 0 request_delay 0 running 5756 locked 0 flushing 0
logging 20 handle_count 8 blocks 1 blocks_logged 2

thanks,
--
js
suse labs


Attachments:
trace (9.34 kB)

2013-04-06 13:16:03

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Sat, Apr 06, 2013 at 09:29:48AM +0200, Jiri Slaby wrote:
>
> I'm not sure, as I am using -next like for ever. But sure, there was a
> kernel which didn't ahve this problem.

Any chance you could try rolling back to 3.2 or 3.5 to see if you can
get a starting point? Even a high-level bisection search would be
helpful to give us a hint.

>Ok,
> dd if=/dev/zero of=xxx
>is enough instead of "kernel update".

Was the dd running in the VM or in the host OS? Basically, is running
the VM required?

> Nothing, just VM (kernel update from console) and mplayer2 on the host.
> This is more-or-less reproducible with these two.

No browser or anything else running that might be introducing a stream
of fsync()'s?

> jbd2/sda5-8-10969 [000] .... 403.679552: jbd2_run_stats: dev
>259,655360 tid 348895 wait 0 request_delay 0 running 5000 locked 1040
>flushing 0 logging 112 handle_count 148224 blocks 1 blocks_logged 2

> jbd2/md2-8-959 [000] .... 408.111339: jbd2_run_stats: dev 9,2
>tid 382991 wait 0 request_delay 0 running 5156 locked 2268 flushing 0
>logging 124 handle_count 5 blocks 1 blocks_logged 2

OK, so this is interesting. The commit is stalling for 1 second in
the the transaction commit on sda5, and then very shortly thereafter
for 2.2 seconds on md2, while we are trying to lock down the
transaction. What that means is that we are waiting for all of the
transaction handles opened against that particular commit to complete
before we can let the transaction commit proceed.

Is md2 sharing the same disk spindle as sda5? And to which disk were
you doing the "dd if=/dev/zero of=/dev/XXX" command?

If I had to guess what's going on, the disk is accepting a huge amount
of writes to its track buffer, and then occasionally it is going out
to lunch trying to write all of this data to the disk platter. This
is not (always) happening when we do the commit (with its attended
cache flush command), but in a few cases, we are doing a read command
which is getting stalled. There are a few cases where we start a
transaction handle, and then discover that we need to read in a disk
block, and if that read stalls for a long period of time, it will hold
the transaction handle open, and this will in turn stall the commit.

If you were to grab a blocktrace, I suspect that is what you will
find; that it's actually a read command which is stalling at some
point, correlated with when we are trying to start transaction commit.

- Ted

2013-04-07 22:00:41

by Frank Ch. Eigler

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2


Hi -


tytso wrote:

> So I tried to reproduce the problem, and so I installed systemtap
> (bleeding edge, since otherwise it won't work with development
> kernel), and then rebuilt a kernel with all of the necessary CONFIG
> options enabled:
>
> CONFIG_DEBUG_INFO, CONFIG_KPROBES, CONFIG_RELAY, CONFIG_DEBUG_FS,
> CONFIG_MODULES, CONFIG_MODULE_UNLOAD
> [...]

That sounds about right.


> I then pulled down mmtests, and tried running watch-dstate.pl, which
> is what I assume you were using [...]

I just took a look at the mmtests, particularly the stap-fix.sh stuff.
The heroics therein are really not called for. git kernel developers
should use git systemtap, as has always been the case. All
compatibility hacks in stap-fix.sh have already been merged, in many
cases for months.


> [...]
> semantic error: while resolving probe point: identifier 'kprobe' at /tmp/stapdjN4_l:18:7
> source: probe kprobe.function("get_request_wait")
> ^
> Pass 2: analysis failed. [man error::pass2]
> Unexpected exit of STAP script at ./watch-dstate.pl line 296.
> I have no clue what to do next. Can you give me a hint?

You should see the error::pass2 man page, which refers to
error::reporting, which refers to involving stap folks and running
stap-report to gather needed info.

But in this case, that's unnecessary: the problem is most likely that
the get_request_wait function does not actually exist any longer, since

commit a06e05e6afab70b4b23c0a7975aaeae24b195cd6
Author: Tejun Heo <[email protected]>
Date: Mon Jun 4 20:40:55 2012 -0700

block: refactor get_request[_wait]()


Systemtap could endavour to list roughly-matching functions that do
exist, if you think that's be helpful.


- FChE

2013-04-08 08:36:51

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Sun, Apr 07, 2013 at 05:59:06PM -0400, Frank Ch. Eigler wrote:
>
> Hi -
>
>
> tytso wrote:
>
> > So I tried to reproduce the problem, and so I installed systemtap
> > (bleeding edge, since otherwise it won't work with development
> > kernel), and then rebuilt a kernel with all of the necessary CONFIG
> > options enabled:
> >
> > CONFIG_DEBUG_INFO, CONFIG_KPROBES, CONFIG_RELAY, CONFIG_DEBUG_FS,
> > CONFIG_MODULES, CONFIG_MODULE_UNLOAD
> > [...]
>
> That sounds about right.
>
>
> > I then pulled down mmtests, and tried running watch-dstate.pl, which
> > is what I assume you were using [...]
>
> I just took a look at the mmtests, particularly the stap-fix.sh stuff.
> The heroics therein are really not called for. git kernel developers
> should use git systemtap, as has always been the case. All
> compatibility hacks in stap-fix.sh have already been merged, in many
> cases for months.
>

At one point in the past this used to be the case but then systemtap had to
be compiled as part of automated tests across different kernel versions. It
could have been worked around in various ways or even installed manually
when machines were deployed but stap-fix.sh generally took less time to
keep working.

>
> > [...]
> > semantic error: while resolving probe point: identifier 'kprobe' at /tmp/stapdjN4_l:18:7
> > source: probe kprobe.function("get_request_wait")
> > ^
> > Pass 2: analysis failed. [man error::pass2]
> > Unexpected exit of STAP script at ./watch-dstate.pl line 296.
> > I have no clue what to do next. Can you give me a hint?
>
> You should see the error::pass2 man page, which refers to
> error::reporting, which refers to involving stap folks and running
> stap-report to gather needed info.
>
> But in this case, that's unnecessary: the problem is most likely that
> the get_request_wait function does not actually exist any longer, since
>
> commit a06e05e6afab70b4b23c0a7975aaeae24b195cd6
> Author: Tejun Heo <[email protected]>
> Date: Mon Jun 4 20:40:55 2012 -0700
>
> block: refactor get_request[_wait]()
>

Yes, this was indeed the problem. The next version of watch-dstate.pl
treated get_request_wait() as a function that may or may not exist. It
uses /proc/kallsyms to figure it out.

Thanks.

--
Mel Gorman
SUSE Labs

2013-04-08 10:54:26

by Frank Ch. Eigler

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

Hi, Mel -

> > [...] git kernel developers
> > should use git systemtap, as has always been the case. [...]
>
> At one point in the past this used to be the case but then systemtap had to
> be compiled as part of automated tests across different kernel versions. It
> could have been worked around in various ways or even installed manually
> when machines were deployed but stap-fix.sh generally took less time to
> keep working.

OK, if that works for you. Keep in mind though that newer versions of
systemtap retain backward-compatibility for ancient versions of the
kernel, so git systemtap should work on those older versions just
fine.


> [...]
> Yes, this was indeed the problem. The next version of watch-dstate.pl
> treated get_request_wait() as a function that may or may not exist. It
> uses /proc/kallsyms to figure it out.

... or you can use the "?" punctuation in the script to have
systemtap adapt:

probe kprobe.function("get_request_wait") ? { ... }


- FChE

2013-04-08 11:02:03

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Sun, Apr 07, 2013 at 05:59:06PM -0400, Frank Ch. Eigler wrote:
> > semantic error: while resolving probe point: identifier 'kprobe' at /tmp/stapdjN4_l:18:7
> > source: probe kprobe.function("get_request_wait")
> > ^
> > Pass 2: analysis failed. [man error::pass2]
> > Unexpected exit of STAP script at ./watch-dstate.pl line 296.
> > I have no clue what to do next. Can you give me a hint?

Is there any reason why the error message couldn't be simplified, to
something as "kernel symbol not found"? I wasn't sure if the problem
was that there was some incompatibility between a recent change with
kprobe and systemtap, or parse failure in the systemtap script, etc.

> Systemtap could endavour to list roughly-matching functions that do
> exist, if you think that's be helpful.

If the goal is ease of use, I suspect the more important thing that
systemtap could do is to make its error messages more easily
understandable, instead of pointing the user to read a man page where
the user then has to figure out which one of a number of failure
scenarios were caused by a particularly opaque error message. (The
man page doesn't even say that "semantic error while resolving probe
point" means that a kernel function doesn't exist -- especially
complaining about the kprobe identifier points the user in the wrong
direction.)

- Ted

2013-04-10 11:07:12

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 02, 2013 at 11:06:51AM -0400, Theodore Ts'o wrote:
> On Tue, Apr 02, 2013 at 03:27:17PM +0100, Mel Gorman wrote:
> > I'm testing a page-reclaim-related series on my laptop that is partially
> > aimed at fixing long stalls when doing metadata-intensive operations on
> > low memory such as a git checkout. I've been running 3.9-rc2 with the
> > series applied but found that the interactive performance was awful even
> > when there was plenty of free memory.
>
> Can you try 3.9-rc4 or later and see if the problem still persists?
> There were a number of ext4 issues especially around low memory
> performance which weren't resolved until -rc4.
>

I experimented with this for a while. -rc6 "feels" much better where
-rc2 felt it would stall for prolonged periods of time but it could be
my imagination too. It does appear that queue depth and await times are
slowly increasing for various reasons.

It's always been the case for me that metadata intensive and write activities
in the background (opening maildir + cache cold git checkout) would stall
the machine for periods of time. This time around, I timed how long it
takes gnome-terminal to open, run find on a directory and exit again while
a cache cold git checkout and a maildir folder were running

v3.0.66
count time
471 5
23 10
11 15
14 20
4 25
8 30
3 35

v3.7
636 5
20 10
13 15
11 20
7 25
1 30
3 35
1 40
1 45

v3.8
count time
394 5
10 10
12 15
8 20
9 25
6 30
2 35
3 40

v3.9-rc6
count time
481 5
14 10
9 15
12 20
8 25
4 30
2 35
3 40
1 45
1 50
1 140

This shows that kernel 3.7 was able to open the terminal in 5 seconds or
less 636 times during the test. Very broadly speaking, v3.0.66 is snappier
and generally able to open the terminal and do some work faster. v3.9-rc6 is
sometimes much slower such as when it took 140 seconds to open the terminal
but not consistently slow enough to allow it to be reliably bisected.

Further, whatever my perceptions are telling me, the fact is that git
checkouts are not obviously worse. However, queue depth and IO wait
times are higher but it's gradual and would not obviously make a very bad
impression. See here;

v3.0.66 checkout:278 depth:387.36 await: 878.97 launch:29.39 max_launch:34.20
v3.7 checkout:268 depth:439.96 await: 971.39 launch:29.46 max_launch:40.42
v3.8 checkout:275 depth:598.12 await:1280.62 launch:31.95 max_launch:38.50
v3.9-rc6 checkout:266 depth:540.74 await:1182.10 launch:45.39 max_launch:138.14

Cache cold git checkout times are roughly comparable but average queue depth
has been increasing and average wait times in v3.8 and v3.9-rc6 are higher
in comparison to v3.0.66. The average time it takes to launch a terminal
and do something with it is also increasing. Unfortunately, these results
are not always perfectly reproducible and it cannot be reliably bisected.

That said, the worst IO wait times (in milliseconds) are getting higher

await r_await w_await
v3.0.66 5811.24 39.19 28309.72
v3.7 7508.79 46.36 36318.96
v3.8 7083.35 47.55 35305.46
v3.9-rc2 9211.14 35.25 34560.08
v3.9-rc6 7499.53 95.21 122780.43

Worst-case small read times have almost doubled. A worst case write
delay was 122 seconds in v3.9-rc6!

The average wait times are also not painting a pretty picture

await r_await w_await
v3.0.66 878.97 7.79 6975.51
v3.7 971.39 7.84 7745.57
v3.8 1280.63 7.75 10306.62
v3.9-rc2 1280.37 7.55 7687.20
v3.9-rc6 1182.11 8.11 13869.67

That is indicating that average wait times have almost doubled since
v3.7. Even though -rc2 felt bad, it's not obviously reflected in the await
figures which is partially what makes bisecting this difficult. At least
you can get an impression of the wait times from this smoothened graph
showing await times from iostat

http://www.csn.ul.ie/~mel/postings/interactivity-20130410/await-times-smooth.png

Again, while one can see the wait times are worse, it's not generally
worse enough to pinpoint it to a single change.

Other observations

On my laptop, pm-utils was setting dirty_background_ratio to 5% and
dirty_ratio to 10% away from the expected defaults of 10% and 20%. Any
of the changes related to dirty balancing could have affected how often
processes get dirty rate-limited.

During major activity there is likely to be "good" behaviour
with stalls roughly every 30 seconds roughly corresponding to
dirty_expire_centiseconds. As you'd expect, the flusher thread is stuck
when this happens.

237 ? 00:00:00 flush-8:0
[<ffffffff811a35b9>] sleep_on_buffer+0x9/0x10
[<ffffffff811a35ee>] __lock_buffer+0x2e/0x30
[<ffffffff8123a21f>] do_get_write_access+0x43f/0x4b0
[<ffffffff8123a3db>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220b89>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff812278a4>] ext4_mb_mark_diskspace_used+0x74/0x4d0
[<ffffffff81228fbf>] ext4_mb_new_blocks+0x2af/0x490
[<ffffffff8121f7c1>] ext4_ext_map_blocks+0x501/0xa00
[<ffffffff811f0065>] ext4_map_blocks+0x2d5/0x470
[<ffffffff811f412a>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f4c30>] ext4_da_writepages+0x380/0x620
[<ffffffff8111ac3b>] do_writepages+0x1b/0x30
[<ffffffff811998f0>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119bf9a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c1e6>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c48b>] wb_writeback+0x27b/0x330
[<ffffffff8119c5d7>] wb_check_old_data_flush+0x97/0xa0
[<ffffffff8119de49>] wb_do_writeback+0x149/0x1d0
[<ffffffff8119df53>] bdi_writeback_thread+0x83/0x280
[<ffffffff8106901b>] kthread+0xbb/0xc0
[<ffffffff8159d47c>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

For other stalls it looks like journal collisions like this;

USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
mel 9593 4.9 0.2 583212 20576 pts/2 Dl+ 11:49 0:00 gnome-terminal --disable-
[<ffffffff81238693>] start_this_handle+0x2c3/0x3e0
[<ffffffff81238970>] jbd2__journal_start.part.8+0x90/0x190
[<ffffffff81238ab5>] jbd2__journal_start+0x45/0x50
[<ffffffff81220921>] __ext4_journal_start_sb+0x81/0x170
[<ffffffff811f53cb>] ext4_dirty_inode+0x2b/0x60
[<ffffffff8119a84e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff811f335c>] ext4_setattr+0x36c/0x640
[<ffffffff8118cf72>] notify_change+0x1f2/0x3c0
[<ffffffff81170f7d>] chown_common+0xbd/0xd0
[<ffffffff811720d7>] sys_fchown+0xb7/0xd0
[<ffffffff8159d52d>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 758 0.0 0.0 0 0 ? D 11:16 0:00
[jbd2/sda6-8]
[<ffffffff8123b28a>] jbd2_journal_commit_transaction+0x1ea/0x13c0
[<ffffffff81240943>] kjournald2+0xb3/0x240
[<ffffffff8106901b>] kthread+0xbb/0xc0
[<ffffffff8159d47c>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

So for myself I can increase the dirty limits, the writeback expire times
and maybe up the journal commit interval from the default of 5 seconds and
see what that "feels" like over the next few days but it still leaves the
fact that worst-case IO wait times in default configurations appear to be
getting worse over time.

--
Mel Gorman
SUSE Labs

2013-04-10 13:13:35

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Wed, Apr 10, 2013 at 11:56:08AM +0100, Mel Gorman wrote:
> During major activity there is likely to be "good" behaviour
> with stalls roughly every 30 seconds roughly corresponding to
> dirty_expire_centiseconds. As you'd expect, the flusher thread is stuck
> when this happens.
>
> 237 ? 00:00:00 flush-8:0
> [<ffffffff811a35b9>] sleep_on_buffer+0x9/0x10
> [<ffffffff811a35ee>] __lock_buffer+0x2e/0x30
> [<ffffffff8123a21f>] do_get_write_access+0x43f/0x4b0

If we're stalling on lock_buffer(), that implies that buffer was being
written, and for some reason it was taking a very long time to
complete.

It might be worthwhile to put a timestamp in struct dm_crypt_io, and
record the time when a particular I/O encryption/decryption is getting
queued to the kcryptd workqueues, and when they finally squirt out.

Something else that might be worth trying is to add WQ_HIGHPRI to the
workqueue flags and see if that makes a difference.

- Ted

2013-04-11 17:04:10

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Wed, Apr 10, 2013 at 09:12:45AM -0400, Theodore Ts'o wrote:
> On Wed, Apr 10, 2013 at 11:56:08AM +0100, Mel Gorman wrote:
> > During major activity there is likely to be "good" behaviour
> > with stalls roughly every 30 seconds roughly corresponding to
> > dirty_expire_centiseconds. As you'd expect, the flusher thread is stuck
> > when this happens.
> >
> > 237 ? 00:00:00 flush-8:0
> > [<ffffffff811a35b9>] sleep_on_buffer+0x9/0x10
> > [<ffffffff811a35ee>] __lock_buffer+0x2e/0x30
> > [<ffffffff8123a21f>] do_get_write_access+0x43f/0x4b0
>
> If we're stalling on lock_buffer(), that implies that buffer was being
> written, and for some reason it was taking a very long time to
> complete.
>

Yes.

> It might be worthwhile to put a timestamp in struct dm_crypt_io, and
> record the time when a particular I/O encryption/decryption is getting
> queued to the kcryptd workqueues, and when they finally squirt out.
>

That somewhat assumes that dm_crypt was at fault which is not unreasonable
but I was skeptical as the workload on dm_crypt was opening a maildir
and mostly reads.

I used a tracepoint in jbd2 to get an idea of what device the buffer_head
was managing and dm did not show up on the list. This is what a trace-cmd
log of the test told me.

flush-8:0-240 [005] 236.655363: jbd2_lock_buffer_stall: dev 8,8 stall_ms 1096
awesome-1364 [005] 290.594396: jbd2_lock_buffer_stall: dev 8,6 stall_ms 7312
gnome-pty-helpe-2256 [005] 290.836952: jbd2_lock_buffer_stall: dev 8,8 stall_ms 7528
flush-8:0-240 [003] 304.012424: jbd2_lock_buffer_stall: dev 8,8 stall_ms 4472
gnome-terminal-2332 [005] 308.290879: jbd2_lock_buffer_stall: dev 8,6 stall_ms 3060
awesome-1364 [006] 308.291318: jbd2_lock_buffer_stall: dev 8,6 stall_ms 3048
flush-8:0-240 [005] 331.525996: jbd2_lock_buffer_stall: dev 8,5 stall_ms 8732
flush-8:0-240 [005] 332.353526: jbd2_lock_buffer_stall: dev 8,5 stall_ms 472
flush-8:0-240 [005] 345.341547: jbd2_lock_buffer_stall: dev 8,5 stall_ms 10024
gnome-terminal-2418 [005] 347.166876: jbd2_lock_buffer_stall: dev 8,6 stall_ms 11852
awesome-1364 [005] 347.167082: jbd2_lock_buffer_stall: dev 8,6 stall_ms 11844
flush-8:0-240 [005] 347.424520: jbd2_lock_buffer_stall: dev 8,5 stall_ms 2012
flush-8:0-240 [005] 347.583752: jbd2_lock_buffer_stall: dev 8,5 stall_ms 156
flush-8:0-240 [005] 390.079682: jbd2_lock_buffer_stall: dev 8,8 stall_ms 396
flush-8:0-240 [002] 407.882385: jbd2_lock_buffer_stall: dev 8,8 stall_ms 12244
flush-8:0-240 [005] 408.003976: jbd2_lock_buffer_stall: dev 8,8 stall_ms 124
gnome-terminal-2610 [005] 413.613365: jbd2_lock_buffer_stall: dev 8,6 stall_ms 3400
awesome-1364 [006] 413.613605: jbd2_lock_buffer_stall: dev 8,6 stall_ms 3736
flush-8:0-240 [002] 430.706475: jbd2_lock_buffer_stall: dev 8,5 stall_ms 9748
flush-8:0-240 [005] 458.188896: jbd2_lock_buffer_stall: dev 8,5 stall_ms 7748
flush-8:0-240 [005] 458.828143: jbd2_lock_buffer_stall: dev 8,5 stall_ms 348
flush-8:0-240 [006] 459.163814: jbd2_lock_buffer_stall: dev 8,5 stall_ms 252
flush-8:0-240 [005] 462.340173: jbd2_lock_buffer_stall: dev 8,5 stall_ms 3160
flush-8:0-240 [005] 469.917705: jbd2_lock_buffer_stall: dev 8,5 stall_ms 6340
flush-8:0-240 [005] 474.434206: jbd2_lock_buffer_stall: dev 8,5 stall_ms 4512
tar-2315 [005] 510.043613: jbd2_lock_buffer_stall: dev 8,5 stall_ms 4316
tclsh-1780 [005] 773.336488: jbd2_lock_buffer_stall: dev 8,5 stall_ms 736
git-3100 [005] 775.933506: jbd2_lock_buffer_stall: dev 8,5 stall_ms 3664
git-4763 [005] 864.093317: jbd2_lock_buffer_stall: dev 8,5 stall_ms 140
flush-8:0-240 [005] 864.242068: jbd2_lock_buffer_stall: dev 8,6 stall_ms 280
git-4763 [005] 864.264157: jbd2_lock_buffer_stall: dev 8,5 stall_ms 148
flush-8:0-240 [005] 865.200004: jbd2_lock_buffer_stall: dev 8,5 stall_ms 464
git-4763 [000] 865.602469: jbd2_lock_buffer_stall: dev 8,5 stall_ms 300
flush-8:0-240 [005] 865.705448: jbd2_lock_buffer_stall: dev 8,5 stall_ms 500
flush-8:0-240 [005] 885.367576: jbd2_lock_buffer_stall: dev 8,8 stall_ms 11024
flush-8:0-240 [005] 895.339697: jbd2_lock_buffer_stall: dev 8,5 stall_ms 120
flush-8:0-240 [005] 895.765488: jbd2_lock_buffer_stall: dev 8,5 stall_ms 424
systemd-journal-265 [005] 915.687201: jbd2_lock_buffer_stall: dev 8,8 stall_ms 14844
flush-8:0-240 [005] 915.690529: jbd2_lock_buffer_stall: dev 8,6 stall_ms 19656
git-5442 [005] 1034.845674: jbd2_lock_buffer_stall: dev 8,5 stall_ms 344
git-5442 [005] 1035.157389: jbd2_lock_buffer_stall: dev 8,5 stall_ms 264
flush-8:0-240 [005] 1035.875478: jbd2_lock_buffer_stall: dev 8,8 stall_ms 1368
flush-8:0-240 [005] 1036.189218: jbd2_lock_buffer_stall: dev 8,8 stall_ms 312
gnome-terminal-5592 [005] 1037.318594: jbd2_lock_buffer_stall: dev 8,6 stall_ms 2628
awesome-1364 [000] 1037.318913: jbd2_lock_buffer_stall: dev 8,6 stall_ms 2632
git-5789 [005] 1076.805405: jbd2_lock_buffer_stall: dev 8,5 stall_ms 184
flush-8:0-240 [005] 1078.401721: jbd2_lock_buffer_stall: dev 8,5 stall_ms 700
flush-8:0-240 [005] 1078.784200: jbd2_lock_buffer_stall: dev 8,5 stall_ms 356
git-5789 [005] 1079.722683: jbd2_lock_buffer_stall: dev 8,5 stall_ms 1452
flush-8:0-240 [005] 1109.928552: jbd2_lock_buffer_stall: dev 8,5 stall_ms 976
flush-8:0-240 [005] 1111.762280: jbd2_lock_buffer_stall: dev 8,5 stall_ms 1832
flush-8:0-240 [005] 1260.197720: jbd2_lock_buffer_stall: dev 8,5 stall_ms 344
flush-8:0-240 [005] 1260.403556: jbd2_lock_buffer_stall: dev 8,5 stall_ms 204
flush-8:0-240 [005] 1260.550904: jbd2_lock_buffer_stall: dev 8,5 stall_ms 108
git-6598 [005] 1260.832948: jbd2_lock_buffer_stall: dev 8,5 stall_ms 1084
flush-8:0-240 [005] 1311.736367: jbd2_lock_buffer_stall: dev 8,5 stall_ms 260
flush-8:0-240 [005] 1313.689297: jbd2_lock_buffer_stall: dev 8,5 stall_ms 412
flush-8:0-240 [005] 1314.230420: jbd2_lock_buffer_stall: dev 8,5 stall_ms 540
git-7022 [006] 1314.241607: jbd2_lock_buffer_stall: dev 8,5 stall_ms 668
flush-8:0-240 [000] 1347.980425: jbd2_lock_buffer_stall: dev 8,5 stall_ms 308
flush-8:0-240 [005] 1348.164598: jbd2_lock_buffer_stall: dev 8,5 stall_ms 104
git-7998 [005] 1547.755328: jbd2_lock_buffer_stall: dev 8,5 stall_ms 304
flush-8:0-240 [006] 1547.764209: jbd2_lock_buffer_stall: dev 8,5 stall_ms 208
flush-8:0-240 [005] 1548.653365: jbd2_lock_buffer_stall: dev 8,5 stall_ms 844
flush-8:0-240 [005] 1549.255022: jbd2_lock_buffer_stall: dev 8,5 stall_ms 460
flush-8:0-240 [005] 1725.036408: jbd2_lock_buffer_stall: dev 8,6 stall_ms 156
git-8743 [005] 1740.492630: jbd2_lock_buffer_stall: dev 8,5 stall_ms 15032
git-8743 [005] 1749.485214: jbd2_lock_buffer_stall: dev 8,5 stall_ms 8648
flush-8:0-240 [005] 1775.937819: jbd2_lock_buffer_stall: dev 8,5 stall_ms 4268
flush-8:0-240 [006] 1776.335682: jbd2_lock_buffer_stall: dev 8,5 stall_ms 336
flush-8:0-240 [006] 1776.446799: jbd2_lock_buffer_stall: dev 8,5 stall_ms 112
flush-8:0-240 [005] 1802.593183: jbd2_lock_buffer_stall: dev 8,6 stall_ms 108
flush-8:0-240 [006] 1802.809237: jbd2_lock_buffer_stall: dev 8,8 stall_ms 208
flush-8:0-240 [005] 2012.041976: jbd2_lock_buffer_stall: dev 8,6 stall_ms 292
tclsh-1778 [005] 2012.055139: jbd2_lock_buffer_stall: dev 8,5 stall_ms 424
latency-output-1933 [002] 2012.055147: jbd2_lock_buffer_stall: dev 8,5 stall_ms 136
git-10209 [005] 2012.074584: jbd2_lock_buffer_stall: dev 8,5 stall_ms 164
flush-8:0-240 [005] 2012.177241: jbd2_lock_buffer_stall: dev 8,5 stall_ms 128
git-10209 [005] 2012.297472: jbd2_lock_buffer_stall: dev 8,5 stall_ms 216
flush-8:0-240 [005] 2012.299828: jbd2_lock_buffer_stall: dev 8,5 stall_ms 120

dm is not obviously at fault there. sda5 is /usr/src (git checkout
running there with some logging), sda6 is /home and sda8 is / . This is
the tracepoint patch used.

---8<---
jbd2: Trace when lock_buffer at the start of a journal write takes a long time

While investigating interactivity problems it was clear that processes
sometimes stall for long periods of times if an attempt is made to lock
a buffer that is already part of a transaction. It would stall in a
trace looking something like

[<ffffffff811a39de>] __lock_buffer+0x2e/0x30
[<ffffffff8123a60f>] do_get_write_access+0x43f/0x4b0
[<ffffffff8123a7cb>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81220f79>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff811f3198>] ext4_reserve_inode_write+0x78/0xa0
[<ffffffff811f3209>] ext4_mark_inode_dirty+0x49/0x220
[<ffffffff811f57d1>] ext4_dirty_inode+0x41/0x60
[<ffffffff8119ac3e>] __mark_inode_dirty+0x4e/0x2d0
[<ffffffff8118b9b9>] update_time+0x79/0xc0
[<ffffffff8118ba98>] file_update_time+0x98/0x100
[<ffffffff81110ffc>] __generic_file_aio_write+0x17c/0x3b0
[<ffffffff811112aa>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ea853>] ext4_file_write+0x83/0xd0
[<ffffffff81172b23>] do_sync_write+0xa3/0xe0
[<ffffffff811731ae>] vfs_write+0xae/0x180
[<ffffffff8117361d>] sys_write+0x4d/0x90
[<ffffffff8159d62d>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

There was a suspicion that dm_crypt might be part responsible so this
patch adds a tracepoint capturing when lock_buffer takes too long
in do_get_write_access() that logs what device is being written and
how long the stall was for.

Signed-off-by: Mel Gorman <[email protected]>
---
fs/jbd2/transaction.c | 8 ++++++++
include/trace/events/jbd2.h | 21 +++++++++++++++++++++
2 files changed, 29 insertions(+)

diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 325bc01..1be0ccb 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -640,6 +640,7 @@ do_get_write_access(handle_t *handle, struct journal_head *jh,
int error;
char *frozen_buffer = NULL;
int need_copy = 0;
+ unsigned long start_lock, time_lock;

if (is_handle_aborted(handle))
return -EROFS;
@@ -655,9 +656,16 @@ repeat:

/* @@@ Need to check for errors here at some point. */

+ start_lock = jiffies;
lock_buffer(bh);
jbd_lock_bh_state(bh);

+ /* If it takes too long to lock the buffer, trace it */
+ time_lock = jbd2_time_diff(start_lock, jiffies);
+ if (time_lock > HZ/10)
+ trace_jbd2_lock_buffer_stall(bh->b_bdev->bd_dev,
+ jiffies_to_msecs(time_lock));
+
/* We now hold the buffer lock so it is safe to query the buffer
* state. Is the buffer dirty?
*
diff --git a/include/trace/events/jbd2.h b/include/trace/events/jbd2.h
index 070df49..c1d1f3e 100644
--- a/include/trace/events/jbd2.h
+++ b/include/trace/events/jbd2.h
@@ -358,6 +358,27 @@ TRACE_EVENT(jbd2_write_superblock,
MINOR(__entry->dev), __entry->write_op)
);

+TRACE_EVENT(jbd2_lock_buffer_stall,
+
+ TP_PROTO(dev_t dev, unsigned long stall_ms),
+
+ TP_ARGS(dev, stall_ms),
+
+ TP_STRUCT__entry(
+ __field( dev_t, dev )
+ __field(unsigned long, stall_ms )
+ ),
+
+ TP_fast_assign(
+ __entry->dev = dev;
+ __entry->stall_ms = stall_ms;
+ ),
+
+ TP_printk("dev %d,%d stall_ms %lu",
+ MAJOR(__entry->dev), MINOR(__entry->dev),
+ __entry->stall_ms)
+);
+
#endif /* _TRACE_JBD2_H */

/* This part must be outside protection */

2013-04-11 18:36:03

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu, Apr 11, 2013 at 06:04:02PM +0100, Mel Gorman wrote:
> > If we're stalling on lock_buffer(), that implies that buffer was being
> > written, and for some reason it was taking a very long time to
> > complete.
> >
>
> Yes.
>
> > It might be worthwhile to put a timestamp in struct dm_crypt_io, and
> > record the time when a particular I/O encryption/decryption is getting
> > queued to the kcryptd workqueues, and when they finally squirt out.
> >
>
> That somewhat assumes that dm_crypt was at fault which is not unreasonable
> but I was skeptical as the workload on dm_crypt was opening a maildir
> and mostly reads.

Hmm... well, I've reviewed all of the places in the ext4 and jbd2
layer where we call lock_buffer(), and with one exception[1] we're not
holding the the bh locked any longer than necessary. There are a few
places where we grab a spinlock or two before we can do what we need
to do and then release the lock'ed buffer head, but the only time we
hold the bh locked for long periods of time is when we submit metadata
blocks for I/O.

[1] There is one exception in ext4_xattr_release_block() where I
believe we should move the call to unlock_buffer(bh) before the call
to ext4_free_blocks(), since we've already elevanted the bh count and
ext4_free_blocks() does not need to have the bh locked. It's not
related to any of the stalls you've repored, though, as near as I can
tell (none of the stack traces include the ext4 xattr code, and this
would only affect external extended attribute blocks).


Could you code which checks the hold time of lock_buffer(), measuing
from when the lock is successfully grabbed, to see if you can see if I
missed some code path in ext4 or jbd2 where the bh is locked and then
there is some call to some function which needs to block for some
random reason? What I'd suggest is putting a timestamp in buffer_head
structure, which is set by lock_buffer once it is successfully grabbed
the lock, and then in unlock_buffer(), if it is held for more than a
second or some such, to dump out the stack trace.

Because at this point, either I'm missing something or I'm beginning
to suspect that your hard drive (or maybe something the block layer?)
is simply taking a long time to service an I/O request. Putting in
this check should be able to very quickly determine what code path
and/or which subsystem we should be focused upon.

Thanks,

- Ted

2013-04-11 21:33:40

by Jan Kara

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu 11-04-13 14:35:12, Ted Tso wrote:
> On Thu, Apr 11, 2013 at 06:04:02PM +0100, Mel Gorman wrote:
> > > If we're stalling on lock_buffer(), that implies that buffer was being
> > > written, and for some reason it was taking a very long time to
> > > complete.
> > >
> >
> > Yes.
> >
> > > It might be worthwhile to put a timestamp in struct dm_crypt_io, and
> > > record the time when a particular I/O encryption/decryption is getting
> > > queued to the kcryptd workqueues, and when they finally squirt out.
> > >
> >
> > That somewhat assumes that dm_crypt was at fault which is not unreasonable
> > but I was skeptical as the workload on dm_crypt was opening a maildir
> > and mostly reads.
>
> Hmm... well, I've reviewed all of the places in the ext4 and jbd2
> layer where we call lock_buffer(), and with one exception[1] we're not
> holding the the bh locked any longer than necessary. There are a few
> places where we grab a spinlock or two before we can do what we need
> to do and then release the lock'ed buffer head, but the only time we
> hold the bh locked for long periods of time is when we submit metadata
> blocks for I/O.
>
> [1] There is one exception in ext4_xattr_release_block() where I
> believe we should move the call to unlock_buffer(bh) before the call
> to ext4_free_blocks(), since we've already elevanted the bh count and
> ext4_free_blocks() does not need to have the bh locked. It's not
> related to any of the stalls you've repored, though, as near as I can
> tell (none of the stack traces include the ext4 xattr code, and this
> would only affect external extended attribute blocks).
>
>
> Could you code which checks the hold time of lock_buffer(), measuing
> from when the lock is successfully grabbed, to see if you can see if I
> missed some code path in ext4 or jbd2 where the bh is locked and then
> there is some call to some function which needs to block for some
> random reason? What I'd suggest is putting a timestamp in buffer_head
> structure, which is set by lock_buffer once it is successfully grabbed
> the lock, and then in unlock_buffer(), if it is held for more than a
> second or some such, to dump out the stack trace.
>
> Because at this point, either I'm missing something or I'm beginning
> to suspect that your hard drive (or maybe something the block layer?)
> is simply taking a long time to service an I/O request. Putting in
> this check should be able to very quickly determine what code path
> and/or which subsystem we should be focused upon.
I think it might be more enlightening if Mel traced which process in
which funclion is holding the buffer lock. I suspect we'll find out that
the flusher thread has submitted the buffer for IO as an async write and
thus it takes a long time to complete in presence of reads which have
higher priority.

Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR

2013-04-12 02:57:56

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu, Apr 11, 2013 at 11:33:35PM +0200, Jan Kara wrote:
> I think it might be more enlightening if Mel traced which process in
> which funclion is holding the buffer lock. I suspect we'll find out that
> the flusher thread has submitted the buffer for IO as an async write and
> thus it takes a long time to complete in presence of reads which have
> higher priority.

That's an interesting theory. If the workload is one which is very
heavy on reads and writes, that could explain the high latency. That
would explain why those of us who are using primarily SSD's are seeing
the problems, because would be reads are nice and fast.

If that is the case, one possible solution that comes to mind would be
to mark buffer_heads that contain metadata with a flag, so that the
flusher thread can write them back at the same priority as reads.

The only problem I can see with this hypothesis is that if this is the
explanation for what Mel and Jiri are seeing, it's something that
would have been around for a long time, and would affect ext3 as well
as ext4. That isn't quite consistent, however, with Mel's observation
that this is a probablem which has gotten worse in relatively
recently.

- Ted

2013-04-12 04:50:53

by Dave Chinner

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu, Apr 11, 2013 at 10:57:08PM -0400, Theodore Ts'o wrote:
> On Thu, Apr 11, 2013 at 11:33:35PM +0200, Jan Kara wrote:
> > I think it might be more enlightening if Mel traced which process in
> > which funclion is holding the buffer lock. I suspect we'll find out that
> > the flusher thread has submitted the buffer for IO as an async write and
> > thus it takes a long time to complete in presence of reads which have
> > higher priority.
>
> That's an interesting theory. If the workload is one which is very
> heavy on reads and writes, that could explain the high latency. That
> would explain why those of us who are using primarily SSD's are seeing
> the problems, because would be reads are nice and fast.
>
> If that is the case, one possible solution that comes to mind would be
> to mark buffer_heads that contain metadata with a flag, so that the
> flusher thread can write them back at the same priority as reads.

Ext4 is already using REQ_META for this purpose.

I'm surprised that no-one has suggested "change the IO elevator"
yet.....

Cheers,

Dave.
--
Dave Chinner
[email protected]

2013-04-12 09:45:38

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu, Apr 11, 2013 at 02:35:12PM -0400, Theodore Ts'o wrote:
> On Thu, Apr 11, 2013 at 06:04:02PM +0100, Mel Gorman wrote:
> > > If we're stalling on lock_buffer(), that implies that buffer was being
> > > written, and for some reason it was taking a very long time to
> > > complete.
> > >
> >
> > Yes.
> >
> > > It might be worthwhile to put a timestamp in struct dm_crypt_io, and
> > > record the time when a particular I/O encryption/decryption is getting
> > > queued to the kcryptd workqueues, and when they finally squirt out.
> > >
> >
> > That somewhat assumes that dm_crypt was at fault which is not unreasonable
> > but I was skeptical as the workload on dm_crypt was opening a maildir
> > and mostly reads.
>
> Hmm... well, I've reviewed all of the places in the ext4 and jbd2
> layer where we call lock_buffer(), and with one exception[1] we're not
> holding the the bh locked any longer than necessary. There are a few
> places where we grab a spinlock or two before we can do what we need
> to do and then release the lock'ed buffer head, but the only time we
> hold the bh locked for long periods of time is when we submit metadata
> blocks for I/O.
>

Yeah, ok. This is not the answer I was hoping for but it's the answer I
expected.

> Could you code which checks the hold time of lock_buffer(), measuing
> from when the lock is successfully grabbed, to see if you can see if I
> missed some code path in ext4 or jbd2 where the bh is locked and then
> there is some call to some function which needs to block for some
> random reason?
>
> What I'd suggest is putting a timestamp in buffer_head
> structure, which is set by lock_buffer once it is successfully grabbed
> the lock, and then in unlock_buffer(), if it is held for more than a
> second or some such, to dump out the stack trace.
>

I can do that but the results might lack meaning. What I could do instead
is use a variation of the page owner tracking patch (current iteration at
https://lkml.org/lkml/2012/12/7/487) to record a stack trace in lock_buffer
and dump it from jbd2/transaction.c if it stalls for too long. I'll report
if I find something useful.

> Because at this point, either I'm missing something or I'm beginning
> to suspect that your hard drive (or maybe something the block layer?)
> is simply taking a long time to service an I/O request.

It could be because the drive is a piece of crap.

--
Mel Gorman
SUSE Labs

2013-04-12 09:47:37

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Thu, Apr 11, 2013 at 10:57:08PM -0400, Theodore Ts'o wrote:
> On Thu, Apr 11, 2013 at 11:33:35PM +0200, Jan Kara wrote:
> > I think it might be more enlightening if Mel traced which process in
> > which funclion is holding the buffer lock. I suspect we'll find out that
> > the flusher thread has submitted the buffer for IO as an async write and
> > thus it takes a long time to complete in presence of reads which have
> > higher priority.
>
> That's an interesting theory. If the workload is one which is very
> heavy on reads and writes, that could explain the high latency. That
> would explain why those of us who are using primarily SSD's are seeing
> the problems, because would be reads are nice and fast.
>
> If that is the case, one possible solution that comes to mind would be
> to mark buffer_heads that contain metadata with a flag, so that the
> flusher thread can write them back at the same priority as reads.
>
> The only problem I can see with this hypothesis is that if this is the
> explanation for what Mel and Jiri are seeing, it's something that
> would have been around for a long time, and would affect ext3 as well
> as ext4. That isn't quite consistent, however, with Mel's observation
> that this is a probablem which has gotten worse in relatively
> recently.
>

According to the tests I've run, multi-second stalls have been a problem for
a while but never really bothered me. I'm not sure why it felt particularly
bad around -rc2 or why it seems to be better now. Maybe I just had my
cranky pants on.

--
Mel Gorman
SUSE Labs

2013-04-12 10:19:15

by Tvrtko Ursulin

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2


Hi all,

On Thursday 11 April 2013 22:57:08 Theodore Ts'o wrote:
> That's an interesting theory. If the workload is one which is very
> heavy on reads and writes, that could explain the high latency. That
> would explain why those of us who are using primarily SSD's are seeing
> the problems, because would be reads are nice and fast.
>
> If that is the case, one possible solution that comes to mind would be
> to mark buffer_heads that contain metadata with a flag, so that the
> flusher thread can write them back at the same priority as reads.
>
> The only problem I can see with this hypothesis is that if this is the
> explanation for what Mel and Jiri are seeing, it's something that
> would have been around for a long time, and would affect ext3 as well
> as ext4. That isn't quite consistent, however, with Mel's observation
> that this is a probablem which has gotten worse in relatively
> recently.

Dropping in as a casual observer and having missed the start of the thread,
risking that I will just muddle the water for you.

I had a similar problem for quite a while with ext4, at least that was my
conclusion since the fix was to migrate one filesystem to xfs which fixed it
for me. Time period when I observed this was between 3.5 and 3.7 kernels.

Situation was I had an ext4 filesystem (on top of LVM, which was on top of MD
RAID 1, which was on top of two mechanical hard drives) which was dedicated to
holding a large SVN check-out. Other filesystems were also ext4 on different
logical volumes (but same spindles).

Symptoms were long stalls of everything (including window management!) on a
relatively heavily loaded desktop (which was KDE). Stalls would last anything
from five to maybe even 30 seconds. Not sure exactly but long enough that you
think the system has actually crashed. I couldn't even switch away to a
different virtual terminal during the stall, nothing.

Eventually I traced it down to kdesvn (subversion client) periodically
refreshing (or something) it's metadata and hence generating some IO on that
dedicated filesystem. That combined with some other desktop activity had an
effect of stalling everything else. I thought it was very weird, but I suppose
KDE and all the rest nowadays do to much IO in everything they do.

Following a hunch I reformatted that filesystem as XFS which fixed the
problem.

I can't reproduce this now to run any tests so I know this is not very helpful
now. But perhaps some of the info will be useful to someone.

Tvrtko

2013-04-12 15:20:43

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Fri, Apr 12, 2013 at 02:50:42PM +1000, Dave Chinner wrote:
> > If that is the case, one possible solution that comes to mind would be
> > to mark buffer_heads that contain metadata with a flag, so that the
> > flusher thread can write them back at the same priority as reads.
>
> Ext4 is already using REQ_META for this purpose.

We're using REQ_META | REQ_PRIO for reads, not writes.

> I'm surprised that no-one has suggested "change the IO elevator"
> yet.....

Well, testing to see if the stalls go away with the noop schedule is a
good thing to try just to validate the theory.

The thing is, we do want to make ext4 work well with cfq, and
prioritizing non-readahead read requests ahead of data writeback does
make sense. The issue is with is that metadata writes going through
the block device could in some cases effectively cause a priority
inversion when what had previously been an asynchronous writeback
starts blocking a foreground, user-visible process.

At least, that's the theory; we should confirm that this is indeed
what is causing the data stalls which Mel is reporting on HDD's before
we start figuring out how to fix this problem.

- Ted

2013-04-13 01:23:47

by Dave Chinner

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Fri, Apr 12, 2013 at 11:19:52AM -0400, Theodore Ts'o wrote:
> On Fri, Apr 12, 2013 at 02:50:42PM +1000, Dave Chinner wrote:
> > > If that is the case, one possible solution that comes to mind would be
> > > to mark buffer_heads that contain metadata with a flag, so that the
> > > flusher thread can write them back at the same priority as reads.
> >
> > Ext4 is already using REQ_META for this purpose.
>
> We're using REQ_META | REQ_PRIO for reads, not writes.
>
> > I'm surprised that no-one has suggested "change the IO elevator"
> > yet.....
>
> Well, testing to see if the stalls go away with the noop schedule is a
> good thing to try just to validate the theory.

Exactly.

> The thing is, we do want to make ext4 work well with cfq, and
> prioritizing non-readahead read requests ahead of data writeback does
> make sense. The issue is with is that metadata writes going through
> the block device could in some cases effectively cause a priority
> inversion when what had previously been an asynchronous writeback
> starts blocking a foreground, user-visible process.

Here's the historic problem with CFQ: it's scheduling algorithms
change from release to release, and so what you tune the filesystem
to for this release is likely to cause different behaviour
in a few releases time.

We've had this problem time and time again with CFQ+XFS, so we
stopped trying to "tune" to a particular elevator long ago. The
best you can do it tag the Io as appropriately as possible (e.g.
metadata with REQ_META, sync IO with ?_SYNC, etc), and then hope CFQ
hasn't been broken since the last release....

> At least, that's the theory; we should confirm that this is indeed
> what is causing the data stalls which Mel is reporting on HDD's before
> we start figuring out how to fix this problem.

*nod*.

Cheers,

Dave.
--
Dave Chinner
[email protected]

2013-04-21 00:06:11

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

As an update to this thread, we brought up this issue at LSF/MM, and
there is a thought that we should be able to solve this problem by
having lock_buffer() check to see if the buffer is locked due to a
write being queued, to have the priority of the write bumped up in the
write queues to resolve the priority inversion. I believe Jeff Moyer
was going to look into this, if I remember correctly.

An alternate solution which I've been playing around adds buffer_head
flags so we can indicate that a buffer contains metadata and/or should
have I/O submitted with the REQ_PRIO flag set.

Adding a buffer_head flag for at least BH_Meta is probably a good
thing, since that way the blktrace will be properly annotated.
Whether we should keep the BH_Prio flag or rely on lock_buffer()
automatically raising the priority is, my feeling is that if
lock_buffer() can do the right thing, we should probably do it via
lock_buffer(). I have a feeling this might be decidedly non-trivial,
though, so perhaps we should just doing via BH flags?

- Ted

2013-04-21 00:07:26

by Theodore Y. Ts'o

[permalink] [raw]
Subject: [PATCH 2/3] buffer: add BH_Prio and BH_Meta flags

Add buffer_head flags so that buffer cache writebacks can be marked
with the the appropriate request flags, so that metadata blocks can be
marked appropriately in blktrace.

Signed-off-by: "Theodore Ts'o" <[email protected]>
---
fs/buffer.c | 5 +++++
include/linux/buffer_head.h | 4 ++++
2 files changed, 9 insertions(+)

diff --git a/fs/buffer.c b/fs/buffer.c
index b4dcb34..a15575c 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -2988,6 +2988,11 @@ int submit_bh(int rw, struct buffer_head * bh)
/* Take care of bh's that straddle the end of the device */
guard_bh_eod(rw, bio, bh);

+ if (buffer_meta(bh))
+ rw |= REQ_META;
+ if (buffer_prio(bh))
+ rw |= REQ_PRIO;
+
bio_get(bio);
submit_bio(rw, bio);

diff --git a/include/linux/buffer_head.h b/include/linux/buffer_head.h
index 5afc4f9..33c0f81 100644
--- a/include/linux/buffer_head.h
+++ b/include/linux/buffer_head.h
@@ -34,6 +34,8 @@ enum bh_state_bits {
BH_Write_EIO, /* I/O error on write */
BH_Unwritten, /* Buffer is allocated on disk but not written */
BH_Quiet, /* Buffer Error Prinks to be quiet */
+ BH_Meta, /* Buffer contains metadata */
+ BH_Prio, /* Buffer should be submitted with REQ_PRIO */

BH_PrivateStart,/* not a state bit, but the first bit available
* for private allocation by other entities
@@ -124,6 +126,8 @@ BUFFER_FNS(Delay, delay)
BUFFER_FNS(Boundary, boundary)
BUFFER_FNS(Write_EIO, write_io_error)
BUFFER_FNS(Unwritten, unwritten)
+BUFFER_FNS(Meta, meta)
+BUFFER_FNS(Prio, prio)

#define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK)

--
1.7.12.rc0.22.gcdd159b

2013-04-21 00:07:33

by Theodore Y. Ts'o

[permalink] [raw]
Subject: [PATCH 3/3] ext4: mark metadata blocks using bh flags

This allows metadata writebacks which are issued via block device
writeback to be sent with the current write request flags.

Signed-off-by: "Theodore Ts'o" <[email protected]>
---
fs/ext4/ext4_jbd2.c | 2 ++
fs/ext4/inode.c | 6 +++++-
2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/fs/ext4/ext4_jbd2.c b/fs/ext4/ext4_jbd2.c
index 0e1dc9e..fd97b81 100644
--- a/fs/ext4/ext4_jbd2.c
+++ b/fs/ext4/ext4_jbd2.c
@@ -215,6 +215,8 @@ int __ext4_handle_dirty_metadata(const char *where, unsigned int line,

might_sleep();

+ mark_buffer_meta(bh);
+ mark_buffer_prio(bh);
if (ext4_handle_valid(handle)) {
err = jbd2_journal_dirty_metadata(handle, bh);
if (err) {
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 62492e9..d7518e2 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1080,10 +1080,14 @@ retry_journal:
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
{
+ int ret;
if (!buffer_mapped(bh) || buffer_freed(bh))
return 0;
set_buffer_uptodate(bh);
- return ext4_handle_dirty_metadata(handle, NULL, bh);
+ ret = ext4_handle_dirty_metadata(handle, NULL, bh);
+ clear_buffer_meta(bh);
+ clear_buffer_prio(bh);
+ return ret;
}

/*
--
1.7.12.rc0.22.gcdd159b

2013-04-21 00:07:25

by Theodore Y. Ts'o

[permalink] [raw]
Subject: [PATCH 1/3] ext4: mark all metadata I/O with REQ_META

As Dave Chinner pointed out at the 2013 LSF/MM workshop, it's
important that metadata I/O requests are marked as such to avoid
priority inversions caused by I/O bandwidth throttling.

Signed-off-by: "Theodore Ts'o" <[email protected]>
---
fs/ext4/balloc.c | 2 +-
fs/ext4/ialloc.c | 2 +-
fs/ext4/mmp.c | 4 ++--
fs/ext4/super.c | 2 +-
4 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index 8dcaea6..d0f13ea 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -441,7 +441,7 @@ ext4_read_block_bitmap_nowait(struct super_block *sb, ext4_group_t block_group)
trace_ext4_read_block_bitmap_load(sb, block_group);
bh->b_end_io = ext4_end_bitmap_read;
get_bh(bh);
- submit_bh(READ, bh);
+ submit_bh(READ | REQ_META | REQ_PRIO, bh);
return bh;
verify:
ext4_validate_block_bitmap(sb, desc, block_group, bh);
diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c
index 18d36d8..00a818d 100644
--- a/fs/ext4/ialloc.c
+++ b/fs/ext4/ialloc.c
@@ -166,7 +166,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
trace_ext4_load_inode_bitmap(sb, block_group);
bh->b_end_io = ext4_end_bitmap_read;
get_bh(bh);
- submit_bh(READ, bh);
+ submit_bh(READ | REQ_META | REQ_PRIO, bh);
wait_on_buffer(bh);
if (!buffer_uptodate(bh)) {
put_bh(bh);
diff --git a/fs/ext4/mmp.c b/fs/ext4/mmp.c
index b3b1f7d..214461e 100644
--- a/fs/ext4/mmp.c
+++ b/fs/ext4/mmp.c
@@ -54,7 +54,7 @@ static int write_mmp_block(struct super_block *sb, struct buffer_head *bh)
lock_buffer(bh);
bh->b_end_io = end_buffer_write_sync;
get_bh(bh);
- submit_bh(WRITE_SYNC, bh);
+ submit_bh(WRITE_SYNC | REQ_META | REQ_PRIO, bh);
wait_on_buffer(bh);
sb_end_write(sb);
if (unlikely(!buffer_uptodate(bh)))
@@ -86,7 +86,7 @@ static int read_mmp_block(struct super_block *sb, struct buffer_head **bh,
get_bh(*bh);
lock_buffer(*bh);
(*bh)->b_end_io = end_buffer_read_sync;
- submit_bh(READ_SYNC, *bh);
+ submit_bh(READ_SYNC | REQ_META | REQ_PRIO, *bh);
wait_on_buffer(*bh);
if (!buffer_uptodate(*bh)) {
brelse(*bh);
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index bfa29ec..dbc7c09 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -4252,7 +4252,7 @@ static journal_t *ext4_get_dev_journal(struct super_block *sb,
goto out_bdev;
}
journal->j_private = sb;
- ll_rw_block(READ, 1, &journal->j_sb_buffer);
+ ll_rw_block(READ | REQ_META | REQ_PRIO, 1, &journal->j_sb_buffer);
wait_on_buffer(journal->j_sb_buffer);
if (!buffer_uptodate(journal->j_sb_buffer)) {
ext4_msg(sb, KERN_ERR, "I/O error on journal device");
--
1.7.12.rc0.22.gcdd159b

2013-04-21 06:09:20

by Jiri Slaby

[permalink] [raw]
Subject: Re: [PATCH 3/3] ext4: mark metadata blocks using bh flags

On 04/21/2013 02:07 AM, Theodore Ts'o wrote:
> This allows metadata writebacks which are issued via block device
> writeback to be sent with the current write request flags.

Hi, where do these come from?
fs/ext4/ext4_jbd2.c: In function ‘__ext4_handle_dirty_metadata’:
fs/ext4/ext4_jbd2.c:218:2: error: implicit declaration of function
‘mark_buffer_meta’ [-Werror=implicit-function-declaration]
fs/ext4/ext4_jbd2.c:219:2: error: implicit declaration of function
‘mark_buffer_prio’ [-Werror=implicit-function-declaration]
cc1: some warnings being treated as errors

> Signed-off-by: "Theodore Ts'o" <[email protected]>
> ---
> fs/ext4/ext4_jbd2.c | 2 ++
> fs/ext4/inode.c | 6 +++++-
> 2 files changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/fs/ext4/ext4_jbd2.c b/fs/ext4/ext4_jbd2.c
> index 0e1dc9e..fd97b81 100644
> --- a/fs/ext4/ext4_jbd2.c
> +++ b/fs/ext4/ext4_jbd2.c
> @@ -215,6 +215,8 @@ int __ext4_handle_dirty_metadata(const char *where, unsigned int line,
>
> might_sleep();
>
> + mark_buffer_meta(bh);
> + mark_buffer_prio(bh);
> if (ext4_handle_valid(handle)) {
> err = jbd2_journal_dirty_metadata(handle, bh);
> if (err) {
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 62492e9..d7518e2 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -1080,10 +1080,14 @@ retry_journal:
> /* For write_end() in data=journal mode */
> static int write_end_fn(handle_t *handle, struct buffer_head *bh)
> {
> + int ret;
> if (!buffer_mapped(bh) || buffer_freed(bh))
> return 0;
> set_buffer_uptodate(bh);
> - return ext4_handle_dirty_metadata(handle, NULL, bh);
> + ret = ext4_handle_dirty_metadata(handle, NULL, bh);
> + clear_buffer_meta(bh);
> + clear_buffer_prio(bh);
> + return ret;
> }
>
> /*
>


--
js
suse labs

2013-04-21 19:56:27

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: [PATCH 3/3] ext4: mark metadata blocks using bh flags

On Sun, Apr 21, 2013 at 08:09:14AM +0200, Jiri Slaby wrote:
> On 04/21/2013 02:07 AM, Theodore Ts'o wrote:
> > This allows metadata writebacks which are issued via block device
> > writeback to be sent with the current write request flags.
>
> Hi, where do these come from?
> fs/ext4/ext4_jbd2.c: In function ‘__ext4_handle_dirty_metadata’:
> fs/ext4/ext4_jbd2.c:218:2: error: implicit declaration of function
> ‘mark_buffer_meta’ [-Werror=implicit-function-declaration]
> fs/ext4/ext4_jbd2.c:219:2: error: implicit declaration of function
> ‘mark_buffer_prio’ [-Werror=implicit-function-declaration]
> cc1: some warnings being treated as errors

They are defined by "[PATCH 2/3] buffer: add BH_Prio and BH_Meta flags" here:

+BUFFER_FNS(Meta, meta)
+BUFFER_FNS(Prio, prio)

When you tried applying this patch, did you try applying all three
patches in the patch series?

- Ted

2013-04-21 20:49:36

by Theodore Y. Ts'o

[permalink] [raw]
Subject: [PATCH 3/3 -v2] ext4: mark metadata blocks using bh flags

Whoops, here's the right version of the patch.

>From 13fca323e9a8b63c08de7a4e05d3c702516b535d Mon Sep 17 00:00:00 2001
From: Theodore Ts'o <[email protected]>
Date: Sun, 21 Apr 2013 16:45:54 -0400
Subject: [PATCH 3/3] ext4: mark metadata blocks using bh flags

This allows metadata writebacks which are issued via block device
writeback to be sent with the current write request flags.

Signed-off-by: "Theodore Ts'o" <[email protected]>
---
fs/ext4/ext4_jbd2.c | 2 ++
fs/ext4/inode.c | 6 +++++-
2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/fs/ext4/ext4_jbd2.c b/fs/ext4/ext4_jbd2.c
index 0e1dc9e..451eb40 100644
--- a/fs/ext4/ext4_jbd2.c
+++ b/fs/ext4/ext4_jbd2.c
@@ -215,6 +215,8 @@ int __ext4_handle_dirty_metadata(const char *where, unsigned int line,

might_sleep();

+ set_buffer_meta(bh);
+ set_buffer_prio(bh);
if (ext4_handle_valid(handle)) {
err = jbd2_journal_dirty_metadata(handle, bh);
if (err) {
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 62492e9..d7518e2 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -1080,10 +1080,14 @@ retry_journal:
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
{
+ int ret;
if (!buffer_mapped(bh) || buffer_freed(bh))
return 0;
set_buffer_uptodate(bh);
- return ext4_handle_dirty_metadata(handle, NULL, bh);
+ ret = ext4_handle_dirty_metadata(handle, NULL, bh);
+ clear_buffer_meta(bh);
+ clear_buffer_prio(bh);
+ return ret;
}

/*
--
1.7.12.rc0.22.gcdd159b

2013-04-22 11:49:01

by Zheng Liu

[permalink] [raw]
Subject: Re: [PATCH 1/3] ext4: mark all metadata I/O with REQ_META

On Sat, Apr 20, 2013 at 08:07:06PM -0400, Theodore Ts'o wrote:
> As Dave Chinner pointed out at the 2013 LSF/MM workshop, it's
> important that metadata I/O requests are marked as such to avoid
> priority inversions caused by I/O bandwidth throttling.
>
> Signed-off-by: "Theodore Ts'o" <[email protected]>

Reviewed-by: Zheng Liu <[email protected]>

Regards,
- Zheng
> ---
> fs/ext4/balloc.c | 2 +-
> fs/ext4/ialloc.c | 2 +-
> fs/ext4/mmp.c | 4 ++--
> fs/ext4/super.c | 2 +-
> 4 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
> index 8dcaea6..d0f13ea 100644
> --- a/fs/ext4/balloc.c
> +++ b/fs/ext4/balloc.c
> @@ -441,7 +441,7 @@ ext4_read_block_bitmap_nowait(struct super_block *sb, ext4_group_t block_group)
> trace_ext4_read_block_bitmap_load(sb, block_group);
> bh->b_end_io = ext4_end_bitmap_read;
> get_bh(bh);
> - submit_bh(READ, bh);
> + submit_bh(READ | REQ_META | REQ_PRIO, bh);
> return bh;
> verify:
> ext4_validate_block_bitmap(sb, desc, block_group, bh);
> diff --git a/fs/ext4/ialloc.c b/fs/ext4/ialloc.c
> index 18d36d8..00a818d 100644
> --- a/fs/ext4/ialloc.c
> +++ b/fs/ext4/ialloc.c
> @@ -166,7 +166,7 @@ ext4_read_inode_bitmap(struct super_block *sb, ext4_group_t block_group)
> trace_ext4_load_inode_bitmap(sb, block_group);
> bh->b_end_io = ext4_end_bitmap_read;
> get_bh(bh);
> - submit_bh(READ, bh);
> + submit_bh(READ | REQ_META | REQ_PRIO, bh);
> wait_on_buffer(bh);
> if (!buffer_uptodate(bh)) {
> put_bh(bh);
> diff --git a/fs/ext4/mmp.c b/fs/ext4/mmp.c
> index b3b1f7d..214461e 100644
> --- a/fs/ext4/mmp.c
> +++ b/fs/ext4/mmp.c
> @@ -54,7 +54,7 @@ static int write_mmp_block(struct super_block *sb, struct buffer_head *bh)
> lock_buffer(bh);
> bh->b_end_io = end_buffer_write_sync;
> get_bh(bh);
> - submit_bh(WRITE_SYNC, bh);
> + submit_bh(WRITE_SYNC | REQ_META | REQ_PRIO, bh);
> wait_on_buffer(bh);
> sb_end_write(sb);
> if (unlikely(!buffer_uptodate(bh)))
> @@ -86,7 +86,7 @@ static int read_mmp_block(struct super_block *sb, struct buffer_head **bh,
> get_bh(*bh);
> lock_buffer(*bh);
> (*bh)->b_end_io = end_buffer_read_sync;
> - submit_bh(READ_SYNC, *bh);
> + submit_bh(READ_SYNC | REQ_META | REQ_PRIO, *bh);
> wait_on_buffer(*bh);
> if (!buffer_uptodate(*bh)) {
> brelse(*bh);
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index bfa29ec..dbc7c09 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -4252,7 +4252,7 @@ static journal_t *ext4_get_dev_journal(struct super_block *sb,
> goto out_bdev;
> }
> journal->j_private = sb;
> - ll_rw_block(READ, 1, &journal->j_sb_buffer);
> + ll_rw_block(READ | REQ_META | REQ_PRIO, 1, &journal->j_sb_buffer);
> wait_on_buffer(journal->j_sb_buffer);
> if (!buffer_uptodate(journal->j_sb_buffer)) {
> ext4_msg(sb, KERN_ERR, "I/O error on journal device");
> --
> 1.7.12.rc0.22.gcdd159b
>
> --
> To unsubscribe, send a message with 'unsubscribe linux-mm' in
> the body to [email protected]. For more info on Linux MM,
> see: http://www.linux-mm.org/ .
> Don't email: <a href=mailto:"[email protected]"> [email protected] </a>

2013-04-22 14:38:58

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

(Adding Jeff Moyer to the cc as I'm told he is interested in the blktrace)

On Fri, Apr 12, 2013 at 11:19:52AM -0400, Theodore Ts'o wrote:
> On Fri, Apr 12, 2013 at 02:50:42PM +1000, Dave Chinner wrote:
> > > If that is the case, one possible solution that comes to mind would be
> > > to mark buffer_heads that contain metadata with a flag, so that the
> > > flusher thread can write them back at the same priority as reads.
> >
> > Ext4 is already using REQ_META for this purpose.
>
> We're using REQ_META | REQ_PRIO for reads, not writes.
>
> > I'm surprised that no-one has suggested "change the IO elevator"
> > yet.....
>
> Well, testing to see if the stalls go away with the noop schedule is a
> good thing to try just to validate the theory.
>

I still haven't tested with a different elevator. While this bug is
relatively high priority for me, there are still are other issues in the way.

TLDR: Flusher writes pages very quickly after processes dirty a buffer. Reads
starve flusher writes.

Now the ugliness and being a windbag.

I collected blktrace and some other logs and they are available at
http://www.csn.ul.ie/~mel/postings/stalls-20130419/log.tar.gz and there
is a lot of stuff in there. The unix time the test started is in the
first line of the file tests-timestamp-bisect . This can help figure out
how far into the test some of the other timestamped logs are

The kernel log with information from the lock_buffer debugging patch is
in dmesg-bisect-gitcheckout . The information in it is race-prone and
cannot be 100% trusted but it's still useful.

iostat is in iostat-bisect-gitcheckout

Here are a few observations I got from the data.

1. flushers appear to acquire buffer locks *very* quickly after an
application writes. Look for lines that look like

"foo failed trylock without holder released 0 ms ago acquired 0 ms ago by bar"

There are a lot of entries like this

jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar
jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar
jbd2 2124 tar failed trylock without holder released 0 ms ago acquired 0 ms ago by 239 flush-8:0
jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar
jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar
jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar
jbd2 239 flush-8:0 failed trylock without holder released 0 ms ago acquired 0 ms ago by 2124 tar

I expected flushers to be writing back the buffers just released in about
5 seconds time, not immediately. It may indicate that when flushers
wake to clean expired inodes that it keeps cleaning inodes as they are
being dirtied.

2. The flush thread can prevent a process making forward progress for
a long time. Take this as an example

jbd2 stalled dev 8,8 for 8168 ms lock holdtime 20692 ms
Last Owner 239 flush-8:0 Acquired Stack
[<ffffffff8100fd8a>] save_stack_trace+0x2a/0x50
[<ffffffff811a3ad6>] set_lock_buffer_owner+0x86/0x90
[<ffffffff811a72ee>] __block_write_full_page+0x16e/0x360
[<ffffffff811a75b3>] block_write_full_page_endio+0xd3/0x110
[<ffffffff811a7600>] block_write_full_page+0x10/0x20
[<ffffffff811aa7f3>] blkdev_writepage+0x13/0x20
[<ffffffff81119352>] __writepage+0x12/0x40
[<ffffffff81119b56>] write_cache_pages+0x206/0x460
[<ffffffff81119df5>] generic_writepages+0x45/0x70
[<ffffffff8111accb>] do_writepages+0x1b/0x30
[<ffffffff81199d60>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119c40a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c656>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c8fb>] wb_writeback+0x27b/0x330
[<ffffffff8119e300>] wb_do_writeback+0x190/0x1d0
[<ffffffff8119e3c3>] bdi_writeback_thread+0x83/0x280
[<ffffffff8106901b>] kthread+0xbb/0xc0
[<ffffffff8159e1fc>] ret_from_fork+0x7c/0xb0
[<ffffffffffffffff>] 0xffffffffffffffff

This part is saying that we locked the buffer due to blkdev_writepage
which I assume must be a metadata update. Based on where we lock the
buffer, the only reason we would leave the buffer unlocked if this
was an asynchronous write request leaving the buffer to be unlocked by
end_buffer_async_write at some time in the future

Last Owner Activity Stack: 239 flush-8:0
[<ffffffff812aee61>] __blkdev_issue_zeroout+0x191/0x1a0
[<ffffffff812aef51>] blkdev_issue_zeroout+0xe1/0xf0
[<ffffffff8121abe9>] ext4_ext_zeroout.isra.30+0x49/0x60
[<ffffffff8121ee47>] ext4_ext_convert_to_initialized+0x227/0x5f0
[<ffffffff8121f8a3>] ext4_ext_handle_uninitialized_extents+0x2f3/0x3a0
[<ffffffff8121ff57>] ext4_ext_map_blocks+0x5d7/0xa00
[<ffffffff811f0715>] ext4_map_blocks+0x2d5/0x470
[<ffffffff811f47da>] mpage_da_map_and_submit+0xba/0x2f0
[<ffffffff811f52e0>] ext4_da_writepages+0x380/0x620
[<ffffffff8111accb>] do_writepages+0x1b/0x30
[<ffffffff81199d60>] __writeback_single_inode+0x40/0x1b0
[<ffffffff8119c40a>] writeback_sb_inodes+0x19a/0x350
[<ffffffff8119c656>] __writeback_inodes_wb+0x96/0xc0
[<ffffffff8119c8fb>] wb_writeback+0x27b/0x330

This part is indicating that at the time a process tried to acquire
the buffer lock that flusher was off doing something else entirely.
That points again to the metadata write being asynchronous.

Current Owner 1829 stap
[<ffffffff8100fd8a>] save_stack_trace+0x2a/0x50
[<ffffffff811a3ad6>] set_lock_buffer_owner+0x86/0x90
[<ffffffff8123a5b2>] do_get_write_access+0xd2/0x800
[<ffffffff8123ae2b>] jbd2_journal_get_write_access+0x2b/0x50
[<ffffffff81221249>] __ext4_journal_get_write_access+0x39/0x80
[<ffffffff81229bca>] ext4_free_blocks+0x36a/0xbe0
[<ffffffff8121c686>] ext4_remove_blocks+0x256/0x2d0
[<ffffffff8121c905>] ext4_ext_rm_leaf+0x205/0x520
[<ffffffff8121e64c>] ext4_ext_remove_space+0x4dc/0x750
[<ffffffff8122051b>] ext4_ext_truncate+0x19b/0x1e0
[<ffffffff811efde5>] ext4_truncate.part.61+0xd5/0xf0
[<ffffffff811f0ee4>] ext4_truncate+0x34/0x90
[<ffffffff811f382d>] ext4_setattr+0x18d/0x640
[<ffffffff8118d3e2>] notify_change+0x1f2/0x3c0
[<ffffffff811716f9>] do_truncate+0x59/0xa0
[<ffffffff8117d3f6>] handle_truncate+0x66/0xa0
[<ffffffff81181576>] do_last+0x626/0x820
[<ffffffff81181823>] path_openat+0xb3/0x4a0
[<ffffffff8118237d>] do_filp_open+0x3d/0xa0
[<ffffffff81172869>] do_sys_open+0xf9/0x1e0
[<ffffffff8117296c>] sys_open+0x1c/0x20
[<ffffffff8159e2ad>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

This is just showing where stap was trying to acquire the buffer lock
truncating data.

3. The blktrace indicates that reads can starve writes from flusher

While there are people that can look at a blktrace and find problems
like they are rain man, I'm more like an ADHD squirrel when looking at
a trace. I wrote a script to look for what unrelated requests completed
while an request got stalled for over a second. It seemed like something
that a tool shoudl already exist for but I didn't find one unless btt
can give the information somehow.

Each delayed request is quite long but here is the first example
discovered by the script

Request 4174 took 1.060828037 to complete
239 W 260608696 [flush-8:0]
Request started time index 4.731862902
Inflight while queued
239 W 260608696 [flush-8:0]
239 W 260608072 [flush-8:0]
239 W 260607872 [flush-8:0]
239 W 260608488 [flush-8:0]
239 W 260608472 [flush-8:0]
239 W 260608568 [flush-8:0]
239 W 260608008 [flush-8:0]
239 W 260607728 [flush-8:0]
239 W 260607112 [flush-8:0]
239 W 260608544 [flush-8:0]
239 W 260622168 [flush-8:0]
239 W 271863816 [flush-8:0]
239 W 260608672 [flush-8:0]
239 W 260607944 [flush-8:0]
239 W 203833687 [flush-8:0]
1676 R 541999743 [watch-inbox-ope]
239 W 260608240 [flush-8:0]
239 W 203851359 [flush-8:0]
239 W 272019768 [flush-8:0]
239 W 260607272 [flush-8:0]
239 W 260607992 [flush-8:0]
239 W 483478791 [flush-8:0]
239 W 260608528 [flush-8:0]
239 W 260607456 [flush-8:0]
239 W 261310704 [flush-8:0]
239 W 260608200 [flush-8:0]
239 W 260607744 [flush-8:0]
239 W 204729015 [flush-8:0]
239 W 204728927 [flush-8:0]
239 W 260608584 [flush-8:0]
239 W 260608352 [flush-8:0]
239 W 270532504 [flush-8:0]
239 W 260608600 [flush-8:0]
239 W 260607152 [flush-8:0]
239 W 260607888 [flush-8:0]
239 W 260607192 [flush-8:0]
239 W 260607568 [flush-8:0]
239 W 260607632 [flush-8:0]
239 W 271831080 [flush-8:0]
239 W 260608312 [flush-8:0]
239 W 260607440 [flush-8:0]
239 W 204729023 [flush-8:0]
239 W 260608056 [flush-8:0]
239 W 272019776 [flush-8:0]
239 W 260608632 [flush-8:0]
239 W 260607704 [flush-8:0]
239 W 271827168 [flush-8:0]
239 W 260607208 [flush-8:0]
239 W 260607384 [flush-8:0]
239 W 260607856 [flush-8:0]
239 W 260607320 [flush-8:0]
239 W 271827160 [flush-8:0]
239 W 260608152 [flush-8:0]
239 W 261271552 [flush-8:0]
239 W 260607168 [flush-8:0]
239 W 260608088 [flush-8:0]
239 W 260607480 [flush-8:0]
239 W 260608424 [flush-8:0]
239 W 260608040 [flush-8:0]
239 W 260608400 [flush-8:0]
239 W 260608224 [flush-8:0]
239 W 260607680 [flush-8:0]
239 W 260607808 [flush-8:0]
239 W 266347440 [flush-8:0]
239 W 260607776 [flush-8:0]
239 W 260607512 [flush-8:0]
239 W 266347280 [flush-8:0]
239 W 260607424 [flush-8:0]
239 W 260607656 [flush-8:0]
239 W 260607976 [flush-8:0]
239 W 260608440 [flush-8:0]
239 W 260608272 [flush-8:0]
239 W 260607536 [flush-8:0]
239 W 260607920 [flush-8:0]
239 W 260608456 [flush-8:0]
Complete since queueing
1676 R 541999743 [watch-inbox-ope]
239 W 203833687 [flush-8:0]
1676 R 541999759 [watch-inbox-ope]
1676 R 541999791 [watch-inbox-ope]
1676 R 541999807 [watch-inbox-ope]
1676 R 541999839 [watch-inbox-ope]
1676 R 541999855 [watch-inbox-ope]
1676 R 542210351 [watch-inbox-ope]
1676 R 542210367 [watch-inbox-ope]
1676 R 541999887 [watch-inbox-ope]
1676 R 541999911 [watch-inbox-ope]
1676 R 541999935 [watch-inbox-ope]
1676 R 541999967 [watch-inbox-ope]
1676 RM 540448791 [watch-inbox-ope]
1676 R 541999983 [watch-inbox-ope]
1676 R 542051791 [watch-inbox-ope]
1676 R 541999999 [watch-inbox-ope]
1676 R 541949839 [watch-inbox-ope]
1676 R 541949871 [watch-inbox-ope]
1676 R 541949903 [watch-inbox-ope]
1676 R 541949935 [watch-inbox-ope]
1676 R 541949887 [watch-inbox-ope]
1676 R 542051823 [watch-inbox-ope]
1676 R 541949967 [watch-inbox-ope]
1676 R 542051839 [watch-inbox-ope]
1676 R 541949999 [watch-inbox-ope]
1676 R 541950015 [watch-inbox-ope]
1676 R 541950031 [watch-inbox-ope]
1676 R 541950047 [watch-inbox-ope]
1676 R 541950063 [watch-inbox-ope]
1676 R 542112079 [watch-inbox-ope]
1676 R 542112095 [watch-inbox-ope]
1676 R 542112111 [watch-inbox-ope]
1676 R 542112127 [watch-inbox-ope]
1676 R 542112847 [watch-inbox-ope]
1676 R 542112863 [watch-inbox-ope]
1676 RM 540461311 [watch-inbox-ope]
1676 RM 540448799 [watch-inbox-ope]
1676 R 542112879 [watch-inbox-ope]
1676 R 541950087 [watch-inbox-ope]
1676 R 541950111 [watch-inbox-ope]
1676 R 542112895 [watch-inbox-ope]
1676 R 541950127 [watch-inbox-ope]
1676 R 541950159 [watch-inbox-ope]
1676 R 541950175 [watch-inbox-ope]
1676 R 541950191 [watch-inbox-ope]
1676 R 541950207 [watch-inbox-ope]
1676 R 541950239 [watch-inbox-ope]
1676 R 541950255 [watch-inbox-ope]
1676 R 541950287 [watch-inbox-ope]
1676 R 541950303 [watch-inbox-ope]
1676 R 541950319 [watch-inbox-ope]
1676 R 542113103 [watch-inbox-ope]
1676 R 541950343 [watch-inbox-ope]
1676 R 541950367 [watch-inbox-ope]
1676 R 541950399 [watch-inbox-ope]
1676 R 542113119 [watch-inbox-ope]
1676 R 542113135 [watch-inbox-ope]
1676 R 541950415 [watch-inbox-ope]
1676 RM 540448815 [watch-inbox-ope]
1676 R 542113151 [watch-inbox-ope]
1676 R 541950447 [watch-inbox-ope]
1676 R 541950463 [watch-inbox-ope]
1676 R 542113743 [watch-inbox-ope]
1676 R 542113759 [watch-inbox-ope]
1676 R 542113775 [watch-inbox-ope]
1676 R 542113791 [watch-inbox-ope]
239 W 203851359 [flush-8:0]
239 W 204729015 [flush-8:0]
239 W 204728927 [flush-8:0]
239 W 204729023 [flush-8:0]
239 W 260569008 [flush-8:0]
1676 R 542145871 [watch-inbox-ope]
1676 R 542145903 [watch-inbox-ope]
1676 R 542145887 [watch-inbox-ope]
1676 R 542154639 [watch-inbox-ope]
1676 R 542154655 [watch-inbox-ope]
1676 R 542154671 [watch-inbox-ope]
1676 R 542154687 [watch-inbox-ope]
1676 R 542154831 [watch-inbox-ope]
1676 R 542154863 [watch-inbox-ope]
1676 R 542157647 [watch-inbox-ope]
1676 R 542157663 [watch-inbox-ope]
1676 R 541950479 [watch-inbox-ope]
1676 R 541950503 [watch-inbox-ope]
1676 R 541950535 [watch-inbox-ope]
1676 R 541950599 [watch-inbox-ope]
1676 R 541950727 [watch-inbox-ope]
1676 R 541950751 [watch-inbox-ope]
1676 R 541950767 [watch-inbox-ope]
1676 RM 540448823 [watch-inbox-ope]
1676 R 541950783 [watch-inbox-ope]
1676 R 541950807 [watch-inbox-ope]
1676 R 541950839 [watch-inbox-ope]
1676 R 541950855 [watch-inbox-ope]
1676 R 541950879 [watch-inbox-ope]
1676 R 541950895 [watch-inbox-ope]
1676 R 541950919 [watch-inbox-ope]
1676 R 541950951 [watch-inbox-ope]
1676 R 541950959 [watch-inbox-ope]
1676 R 541950975 [watch-inbox-ope]
1676 R 541951007 [watch-inbox-ope]
1676 R 541951023 [watch-inbox-ope]
1676 R 541951055 [watch-inbox-ope]
1676 R 541951087 [watch-inbox-ope]
1676 R 541951103 [watch-inbox-ope]
1676 R 541951119 [watch-inbox-ope]
1676 R 541951143 [watch-inbox-ope]
1676 R 541951167 [watch-inbox-ope]
1676 R 542157679 [watch-inbox-ope]
1676 R 542157695 [watch-inbox-ope]
1676 R 541951183 [watch-inbox-ope]
1676 R 541951215 [watch-inbox-ope]
1676 R 541951231 [watch-inbox-ope]
1676 R 542158223 [watch-inbox-ope]
1676 RM 540448831 [watch-inbox-ope]
1676 R 541951247 [watch-inbox-ope]
1676 R 541951271 [watch-inbox-ope]
1676 R 541951295 [watch-inbox-ope]
1676 R 542158239 [watch-inbox-ope]
1676 R 542158255 [watch-inbox-ope]
1676 R 541951311 [watch-inbox-ope]
1676 R 542158271 [watch-inbox-ope]
1676 R 541951343 [watch-inbox-ope]
1676 R 541951359 [watch-inbox-ope]
1676 R 541951391 [watch-inbox-ope]
1676 R 541951407 [watch-inbox-ope]
1676 R 541951423 [watch-inbox-ope]
1676 R 541951439 [watch-inbox-ope]
1676 R 541951471 [watch-inbox-ope]
1676 R 542158607 [watch-inbox-ope]
1676 R 541951487 [watch-inbox-ope]
1676 R 542158639 [watch-inbox-ope]
1676 R 542158655 [watch-inbox-ope]
1676 R 542187215 [watch-inbox-ope]
1676 R 542187231 [watch-inbox-ope]
1676 R 542187247 [watch-inbox-ope]
1676 R 541951503 [watch-inbox-ope]
1676 RM 540448839 [watch-inbox-ope]
1676 R 542187263 [watch-inbox-ope]
1676 R 541951535 [watch-inbox-ope]
1676 R 541951551 [watch-inbox-ope]
1676 R 541951599 [watch-inbox-ope]
1676 R 541951575 [watch-inbox-ope]
1676 R 542190607 [watch-inbox-ope]
239 W 261310704 [flush-8:0]
239 W 266347280 [flush-8:0]
239 W 266347440 [flush-8:0]
1676 R 542190623 [watch-inbox-ope]
1676 R 542190639 [watch-inbox-ope]
1676 R 542190655 [watch-inbox-ope]
1676 R 542193999 [watch-inbox-ope]
1676 R 542194015 [watch-inbox-ope]
1676 R 541951631 [watch-inbox-ope]
1676 R 541951663 [watch-inbox-ope]
1676 R 541951679 [watch-inbox-ope]
1676 R 541951711 [watch-inbox-ope]
1676 R 541951727 [watch-inbox-ope]
1676 R 541951743 [watch-inbox-ope]
1676 R 542194031 [watch-inbox-ope]
1676 R 542194047 [watch-inbox-ope]
1676 R 542197711 [watch-inbox-ope]
1676 RM 540448847 [watch-inbox-ope]
1676 R 541951759 [watch-inbox-ope]
1676 R 541951783 [watch-inbox-ope]
1676 R 541951807 [watch-inbox-ope]
1676 R 542197727 [watch-inbox-ope]
1676 R 542197743 [watch-inbox-ope]
1676 R 542197759 [watch-inbox-ope]
1676 R 541951823 [watch-inbox-ope]
1676 R 541951855 [watch-inbox-ope]
1676 R 541951871 [watch-inbox-ope]
1676 R 541951895 [watch-inbox-ope]
1676 R 541951919 [watch-inbox-ope]
1676 R 541951935 [watch-inbox-ope]
1676 R 541951951 [watch-inbox-ope]
1676 R 541951967 [watch-inbox-ope]
1676 R 541951983 [watch-inbox-ope]
1676 R 542207567 [watch-inbox-ope]
1676 R 542207599 [watch-inbox-ope]
1676 R 542210383 [watch-inbox-ope]
1676 R 542210399 [watch-inbox-ope]
1676 R 542210415 [watch-inbox-ope]
1676 R 542210431 [watch-inbox-ope]
1676 RM 540448855 [watch-inbox-ope]
1676 R 541952015 [watch-inbox-ope]
1676 R 541952047 [watch-inbox-ope]
1676 R 541952063 [watch-inbox-ope]
1676 R 541952079 [watch-inbox-ope]
1676 R 541952103 [watch-inbox-ope]
1676 R 541952127 [watch-inbox-ope]
1676 R 541952159 [watch-inbox-ope]
1676 R 541952175 [watch-inbox-ope]
1676 R 541952207 [watch-inbox-ope]
1676 R 541952223 [watch-inbox-ope]
1676 R 541952255 [watch-inbox-ope]
1676 R 541952303 [watch-inbox-ope]
1676 R 541952319 [watch-inbox-ope]
1676 R 541952335 [watch-inbox-ope]
1676 R 541952351 [watch-inbox-ope]
1676 R 541952383 [watch-inbox-ope]
1676 R 542051855 [watch-inbox-ope]
1676 R 542051871 [watch-inbox-ope]
1676 R 542051887 [watch-inbox-ope]
1676 R 542051903 [watch-inbox-ope]
1676 R 542051919 [watch-inbox-ope]
1676 R 541952391 [watch-inbox-ope]
1676 R 541952415 [watch-inbox-ope]
1676 RM 540448863 [watch-inbox-ope]
1676 R 542051935 [watch-inbox-ope]
1676 R 541952431 [watch-inbox-ope]
1676 R 541952447 [watch-inbox-ope]
1676 R 541952463 [watch-inbox-ope]
1676 R 541952487 [watch-inbox-ope]
1676 R 541952511 [watch-inbox-ope]
1676 R 541952527 [watch-inbox-ope]
1676 R 541952559 [watch-inbox-ope]
1676 R 541952607 [watch-inbox-ope]
1676 R 541952623 [watch-inbox-ope]
1676 R 542051951 [watch-inbox-ope]
1676 R 541952639 [watch-inbox-ope]
1676 R 542112271 [watch-inbox-ope]
239 W 261271552 [flush-8:0]
239 W 270532504 [flush-8:0]
239 W 271827168 [flush-8:0]
239 W 271827160 [flush-8:0]
239 W 271831080 [flush-8:0]
1676 R 542112287 [watch-inbox-ope]
1676 R 542112303 [watch-inbox-ope]
1676 R 542112319 [watch-inbox-ope]
1676 R 542112335 [watch-inbox-ope]
1676 R 542112351 [watch-inbox-ope]
1676 R 542112367 [watch-inbox-ope]
1676 R 542112383 [watch-inbox-ope]
1676 R 542112655 [watch-inbox-ope]
1676 RM 540448871 [watch-inbox-ope]
1676 R 542112671 [watch-inbox-ope]
1676 R 542112687 [watch-inbox-ope]
1676 R 542112703 [watch-inbox-ope]
1676 R 542112719 [watch-inbox-ope]
1676 R 542112735 [watch-inbox-ope]
1676 R 541952655 [watch-inbox-ope]
1676 R 541952687 [watch-inbox-ope]
1676 R 541952703 [watch-inbox-ope]
1676 R 541952735 [watch-inbox-ope]
1676 R 541952751 [watch-inbox-ope]
1676 R 542112751 [watch-inbox-ope]
1676 R 541952767 [watch-inbox-ope]
1676 R 541952783 [watch-inbox-ope]
1676 R 541952799 [watch-inbox-ope]
1676 R 541952815 [watch-inbox-ope]
1676 R 541952831 [watch-inbox-ope]
1676 R 541952863 [watch-inbox-ope]
1676 R 541952879 [watch-inbox-ope]
1676 R 542113807 [watch-inbox-ope]
1676 R 541952903 [watch-inbox-ope]
1676 R 541952935 [watch-inbox-ope]
1676 RM 540448879 [watch-inbox-ope]
1676 R 541952959 [watch-inbox-ope]
1676 R 542113823 [watch-inbox-ope]
1676 R 542113839 [watch-inbox-ope]
1676 R 542113855 [watch-inbox-ope]
1676 R 541952975 [watch-inbox-ope]
1676 R 541952991 [watch-inbox-ope]
1676 R 541953007 [watch-inbox-ope]
1676 R 541953023 [watch-inbox-ope]
1676 R 541953055 [watch-inbox-ope]
1676 R 541953071 [watch-inbox-ope]
1676 R 541953103 [watch-inbox-ope]
1676 R 541953119 [watch-inbox-ope]
1676 R 541953135 [watch-inbox-ope]
1676 R 542113871 [watch-inbox-ope]
1676 R 542113887 [watch-inbox-ope]
1676 R 541953167 [watch-inbox-ope]
1676 R 541953191 [watch-inbox-ope]
1676 R 541953223 [watch-inbox-ope]
1676 R 541953247 [watch-inbox-ope]
1676 R 541953263 [watch-inbox-ope]
1676 R 541953279 [watch-inbox-ope]
1676 RM 540448887 [watch-inbox-ope]
1676 R 541953303 [watch-inbox-ope]
1676 R 541953327 [watch-inbox-ope]
1676 R 542113903 [watch-inbox-ope]
1676 R 542113919 [watch-inbox-ope]
1676 R 541953359 [watch-inbox-ope]
1676 R 541953375 [watch-inbox-ope]
1676 R 541953391 [watch-inbox-ope]
1676 R 541953407 [watch-inbox-ope]
1676 R 542145679 [watch-inbox-ope]
1676 R 542145695 [watch-inbox-ope]
1676 R 542145711 [watch-inbox-ope]
1676 R 542145727 [watch-inbox-ope]
1676 R 542145743 [watch-inbox-ope]
1676 R 541953423 [watch-inbox-ope]
1676 R 542145759 [watch-inbox-ope]
1676 R 541953455 [watch-inbox-ope]
1676 R 541953471 [watch-inbox-ope]
1676 R 542145775 [watch-inbox-ope]
1676 R 542145791 [watch-inbox-ope]
1676 R 541953487 [watch-inbox-ope]
1676 R 541953519 [watch-inbox-ope]
1676 RM 540448895 [watch-inbox-ope]
1676 R 541953535 [watch-inbox-ope]
1676 R 541953551 [watch-inbox-ope]
1676 R 541953567 [watch-inbox-ope]
1676 R 541953599 [watch-inbox-ope]
1676 R 541953615 [watch-inbox-ope]
1676 R 541953631 [watch-inbox-ope]
1676 R 541953647 [watch-inbox-ope]
1676 R 542157455 [watch-inbox-ope]
1676 R 542157471 [watch-inbox-ope]
1676 R 542157487 [watch-inbox-ope]
1676 R 541953671 [watch-inbox-ope]
1676 RA 540386719 [watch-inbox-ope]
1676 RA 540386727 [watch-inbox-ope]
1676 RA 540386735 [watch-inbox-ope]
1676 RA 540386743 [watch-inbox-ope]
1676 RA 540386751 [watch-inbox-ope]
1676 RA 540386759 [watch-inbox-ope]
1676 RA 540386767 [watch-inbox-ope]
1676 RA 540386775 [watch-inbox-ope]
1676 RA 540386783 [watch-inbox-ope]
1676 RA 540386791 [watch-inbox-ope]
1676 RA 540386799 [watch-inbox-ope]
1676 RA 540386807 [watch-inbox-ope]
1676 RA 540386815 [watch-inbox-ope]
1676 RA 540386823 [watch-inbox-ope]
1676 RA 540386831 [watch-inbox-ope]
1676 RA 540386839 [watch-inbox-ope]
1676 RA 540386847 [watch-inbox-ope]
1676 RA 540386855 [watch-inbox-ope]
1676 RA 540386863 [watch-inbox-ope]
1676 RA 540386871 [watch-inbox-ope]
1676 RA 540386879 [watch-inbox-ope]
1676 RA 540386887 [watch-inbox-ope]
1676 RA 540386895 [watch-inbox-ope]
1676 RA 540386903 [watch-inbox-ope]
1676 RA 540386911 [watch-inbox-ope]
1676 RA 540386919 [watch-inbox-ope]
1676 RA 540386927 [watch-inbox-ope]
1676 RA 540386935 [watch-inbox-ope]
1676 RA 540386943 [watch-inbox-ope]
1676 RA 540386951 [watch-inbox-ope]
1676 RA 540386959 [watch-inbox-ope]
1676 RM 540386711 [watch-inbox-ope]
239 W 271863816 [flush-8:0]
239 W 272019768 [flush-8:0]
239 W 272019776 [flush-8:0]
239 W 483478791 [flush-8:0]
239 W 260578312 [flush-8:0]
239 W 260578400 [flush-8:0]
1676 R 541953695 [watch-inbox-ope]
1676 R 541953711 [watch-inbox-ope]
1676 R 542157503 [watch-inbox-ope]
1676 R 541953743 [watch-inbox-ope]
1676 R 541953759 [watch-inbox-ope]
1676 R 541953775 [watch-inbox-ope]
1676 R 542157519 [watch-inbox-ope]
1676 R 541953791 [watch-inbox-ope]
1676 R 542157551 [watch-inbox-ope]
1676 R 541953807 [watch-inbox-ope]
1676 R 541953831 [watch-inbox-ope]
1676 R 541953863 [watch-inbox-ope]
1676 R 541953927 [watch-inbox-ope]
1676 R 541954055 [watch-inbox-ope]
1676 RM 540448903 [watch-inbox-ope]
1676 R 542157567 [watch-inbox-ope]
1676 R 541954127 [watch-inbox-ope]
1676 R 541954143 [watch-inbox-ope]
1676 R 541954159 [watch-inbox-ope]
1676 R 541954183 [watch-inbox-ope]
1676 R 541954207 [watch-inbox-ope]
1676 R 541954223 [watch-inbox-ope]
1676 R 541954239 [watch-inbox-ope]
1676 R 541954255 [watch-inbox-ope]
1676 R 541954271 [watch-inbox-ope]
1676 R 541954287 [watch-inbox-ope]
1676 R 541954319 [watch-inbox-ope]
1676 R 541954335 [watch-inbox-ope]
1676 R 541954351 [watch-inbox-ope]
1676 R 541954367 [watch-inbox-ope]
1676 R 541954391 [watch-inbox-ope]
1676 R 541954415 [watch-inbox-ope]
1676 R 541954431 [watch-inbox-ope]
1676 R 541954455 [watch-inbox-ope]
1676 R 541954479 [watch-inbox-ope]
1676 R 541954495 [watch-inbox-ope]
1676 RM 540456719 [watch-inbox-ope]
239 W 260622168 [flush-8:0]
239 W 260625528 [flush-8:0]
239 W 260625608 [flush-8:0]
239 W 260614368 [flush-8:0]
239 W 260614336 [flush-8:0]
239 W 260614304 [flush-8:0]
239 W 260614280 [flush-8:0]
1676 R 541954511 [watch-inbox-ope]
1676 R 541954527 [watch-inbox-ope]
1676 R 541954543 [watch-inbox-ope]
1676 R 541954567 [watch-inbox-ope]
1676 R 541954599 [watch-inbox-ope]
1676 R 541954607 [watch-inbox-ope]
1676 R 541954623 [watch-inbox-ope]
1676 R 541954655 [watch-inbox-ope]
1676 R 541954671 [watch-inbox-ope]
1676 R 541954687 [watch-inbox-ope]
1676 R 542158351 [watch-inbox-ope]
1676 R 542158367 [watch-inbox-ope]
1676 R 542158383 [watch-inbox-ope]
1676 R 542158399 [watch-inbox-ope]
1676 R 542158415 [watch-inbox-ope]
1676 R 541954703 [watch-inbox-ope]
1676 R 541954727 [watch-inbox-ope]
1676 R 541954751 [watch-inbox-ope]
1676 R 542158431 [watch-inbox-ope]
1676 R 542158447 [watch-inbox-ope]
1676 R 542158463 [watch-inbox-ope]
1676 R 541954767 [watch-inbox-ope]
1676 RM 540456727 [watch-inbox-ope]
1676 R 541954831 [watch-inbox-ope]
1676 R 541954863 [watch-inbox-ope]
1676 R 541954783 [watch-inbox-ope]
1676 R 541954799 [watch-inbox-ope]
1676 R 541954895 [watch-inbox-ope]
1676 R 541954911 [watch-inbox-ope]
1676 R 541954927 [watch-inbox-ope]
1676 R 541954943 [watch-inbox-ope]
1676 R 542158479 [watch-inbox-ope]
1676 R 541954959 [watch-inbox-ope]
1676 R 542158495 [watch-inbox-ope]
1676 R 541954983 [watch-inbox-ope]
1676 R 541955007 [watch-inbox-ope]
1676 R 541955023 [watch-inbox-ope]
1676 R 541955047 [watch-inbox-ope]
1676 R 541955071 [watch-inbox-ope]
1676 R 541955087 [watch-inbox-ope]
1676 R 541955119 [watch-inbox-ope]
1676 R 541955183 [watch-inbox-ope]
239 W 260607112 [flush-8:0]
239 W 260607152 [flush-8:0]
239 W 260607168 [flush-8:0]
239 W 260607192 [flush-8:0]
239 W 260607208 [flush-8:0]
239 W 260607272 [flush-8:0]
239 W 260607320 [flush-8:0]
239 W 260607384 [flush-8:0]
239 W 260607424 [flush-8:0]
239 W 260607440 [flush-8:0]
239 W 260607456 [flush-8:0]
239 W 260607480 [flush-8:0]
239 W 260607512 [flush-8:0]
239 W 260607536 [flush-8:0]
239 W 260607568 [flush-8:0]
239 W 260607632 [flush-8:0]
239 W 260607656 [flush-8:0]
239 W 260607680 [flush-8:0]
239 W 260607704 [flush-8:0]
239 W 260607728 [flush-8:0]
239 W 260607744 [flush-8:0]
239 W 260607776 [flush-8:0]
239 W 260607808 [flush-8:0]
239 W 260607856 [flush-8:0]
239 W 260607872 [flush-8:0]
239 W 260607888 [flush-8:0]
239 W 260607920 [flush-8:0]
239 W 260607944 [flush-8:0]
239 W 260607976 [flush-8:0]
239 W 260607992 [flush-8:0]
239 W 260608008 [flush-8:0]
239 W 260608040 [flush-8:0]
239 W 260608056 [flush-8:0]
239 W 260608072 [flush-8:0]
239 W 260608088 [flush-8:0]
239 W 260608152 [flush-8:0]
239 W 260608200 [flush-8:0]
239 W 260608224 [flush-8:0]
239 W 260608240 [flush-8:0]
239 W 260608272 [flush-8:0]
239 W 260608312 [flush-8:0]
239 W 260608352 [flush-8:0]
239 W 260608400 [flush-8:0]
239 W 260608424 [flush-8:0]
239 W 260608440 [flush-8:0]
1676 R 541955311 [watch-inbox-ope]
1676 R 541955327 [watch-inbox-ope]
1676 R 542158511 [watch-inbox-ope]
1676 R 542158543 [watch-inbox-ope]
1676 R 542158559 [watch-inbox-ope]
1676 RM 540456735 [watch-inbox-ope]
1676 R 542158575 [watch-inbox-ope]
1676 R 542158591 [watch-inbox-ope]
1676 R 541955343 [watch-inbox-ope]
1676 R 541955359 [watch-inbox-ope]
1676 R 541955375 [watch-inbox-ope]
1676 R 541955391 [watch-inbox-ope]
1676 R 541955423 [watch-inbox-ope]
1676 R 541955439 [watch-inbox-ope]
1676 R 542187279 [watch-inbox-ope]
239 W 260608456 [flush-8:0]
239 W 260608472 [flush-8:0]
239 W 260608488 [flush-8:0]
239 W 260608528 [flush-8:0]
239 W 260608544 [flush-8:0]
239 W 260608568 [flush-8:0]
239 W 260608584 [flush-8:0]
239 W 260608600 [flush-8:0]
239 W 260608632 [flush-8:0]
239 W 260608672 [flush-8:0]
1676 R 542187295 [watch-inbox-ope]
1676 R 542187311 [watch-inbox-ope]
1676 R 541955463 [watch-inbox-ope]
1676 R 541955503 [watch-inbox-ope]
1676 R 541955487 [watch-inbox-ope]
1676 R 542187327 [watch-inbox-ope]
1676 R 541955535 [watch-inbox-ope]
1676 R 541955551 [watch-inbox-ope]
1676 R 541955567 [watch-inbox-ope]
1676 R 541955583 [watch-inbox-ope]
1676 R 541955615 [watch-inbox-ope]
1676 R 541955655 [watch-inbox-ope]
1676 RM 540456743 [watch-inbox-ope]
1676 R 541955679 [watch-inbox-ope]
1676 R 541955695 [watch-inbox-ope]
1676 R 541955711 [watch-inbox-ope]
1676 R 542187343 [watch-inbox-ope]
1676 R 542187359 [watch-inbox-ope]
1676 R 542187375 [watch-inbox-ope]
1676 R 542187391 [watch-inbox-ope]
1676 R 542190479 [watch-inbox-ope]
1676 R 541955727 [watch-inbox-ope]
1676 R 541955751 [watch-inbox-ope]
1676 R 541955775 [watch-inbox-ope]
1676 R 542190495 [watch-inbox-ope]
1676 R 542190511 [watch-inbox-ope]
1676 R 542190527 [watch-inbox-ope]
1676 R 541955791 [watch-inbox-ope]
1676 R 541955823 [watch-inbox-ope]
1676 R 541955839 [watch-inbox-ope]
1676 R 541955855 [watch-inbox-ope]
1676 R 541955879 [watch-inbox-ope]
1676 R 541955903 [watch-inbox-ope]
1676 R 541955919 [watch-inbox-ope]
1676 RM 540456751 [watch-inbox-ope]
1676 R 541955943 [watch-inbox-ope]
1676 R 541955967 [watch-inbox-ope]
1676 R 542190543 [watch-inbox-ope]
1676 R 542190559 [watch-inbox-ope]
1676 R 542190575 [watch-inbox-ope]
1676 R 542190591 [watch-inbox-ope]
1676 R 542193807 [watch-inbox-ope]
1676 R 541955983 [watch-inbox-ope]
1676 R 541956015 [watch-inbox-ope]
1676 R 541956031 [watch-inbox-ope]
1676 R 541956047 [watch-inbox-ope]
1676 R 541956079 [watch-inbox-ope]
1676 R 541956095 [watch-inbox-ope]
1676 R 542193839 [watch-inbox-ope]
1676 R 542193855 [watch-inbox-ope]
1676 R 542193871 [watch-inbox-ope]
1676 R 541956111 [watch-inbox-ope]
1676 R 541956143 [watch-inbox-ope]
1676 R 541956207 [watch-inbox-ope]
1676 R 541956255 [watch-inbox-ope]
1676 R 542193887 [watch-inbox-ope]
1676 R 541956271 [watch-inbox-ope]
1676 R 541956287 [watch-inbox-ope]
1676 R 541956335 [watch-inbox-ope]
1676 RM 540456759 [watch-inbox-ope]
1676 R 542193903 [watch-inbox-ope]
1676 R 541956319 [watch-inbox-ope]
1676 R 541956367 [watch-inbox-ope]
1676 R 541956399 [watch-inbox-ope]
1676 R 541956415 [watch-inbox-ope]
1676 R 541956431 [watch-inbox-ope]
1676 R 541956447 [watch-inbox-ope]
1676 R 541956479 [watch-inbox-ope]
1676 R 542197775 [watch-inbox-ope]
1676 R 542197791 [watch-inbox-ope]
1676 R 542197807 [watch-inbox-ope]
1676 R 542197823 [watch-inbox-ope]
1676 R 542197839 [watch-inbox-ope]

I recognise that the output will have a WTF reaction but the key
observations to me are

a) a single write request from flusher took over a second to complete
b) at the time it was queued, it was mostly other writes that were in
the queue at the same time
c) The write request and the parallel writes were all asynchronous write
requests
D) at the time the request completed, there were a LARGE number of
other requested queued and completed at the same time.

Of the requests queued and completed in the meantime the breakdown was

22 RM
31 RA
82 W
445 R

If I'm reading this correctly, it is saying that 22 reads were merged (RM),
31 reads were remapped to another device (RA) which is probably reads from
the dm-crypt partition, 82 were writes (W) which is not far off the number
of writes that were in the queue and 445 were other reads. The delay was
dominated by reads that were queued after the write request and completed
before it.

There are lots of other example but here is part of one from much later
that starts with.

Request 27128 took 7.536721619 to complete
239 W 188973663 [flush-8:0]

That's saying that the 27128th request in the trace took over 7 seconds
to complete and was an asynchronous write from flusher. The contents of
the queue are displayed at that time and the breakdown of requests is

23 WS
86 RM
124 RA
442 W
1931 R

7 seconds later when it was completed the breakdown of completed
requests was

25 WS
114 RM
155 RA
408 W
2457 R

In combination, that confirms for me that asynchronous writes from flush
are being starved by reads. When a process requires a buffer that is locked
by that asynchronous write from flusher, it stalls.

> The thing is, we do want to make ext4 work well with cfq, and
> prioritizing non-readahead read requests ahead of data writeback does
> make sense. The issue is with is that metadata writes going through
> the block device could in some cases effectively cause a priority
> inversion when what had previously been an asynchronous writeback
> starts blocking a foreground, user-visible process.
>
> At least, that's the theory;

I *think* the data more or less confirms the theory but it'd be nice if
someone else double checked in case I'm seeing what I want to see
instead of what is actually there.

--
Mel Gorman
SUSE Labs

2013-04-22 22:44:03

by Jeff Moyer

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

Mel Gorman <[email protected]> writes:

> (Adding Jeff Moyer to the cc as I'm told he is interested in the blktrace)

Thanks. I've got a few comments and corrections for you below.

> TLDR: Flusher writes pages very quickly after processes dirty a buffer. Reads
> starve flusher writes.
[snip]

> 3. The blktrace indicates that reads can starve writes from flusher
>
> While there are people that can look at a blktrace and find problems
> like they are rain man, I'm more like an ADHD squirrel when looking at
> a trace. I wrote a script to look for what unrelated requests completed
> while an request got stalled for over a second. It seemed like something
> that a tool shoudl already exist for but I didn't find one unless btt
> can give the information somehow.

Care to share that script?

[snip]

> I recognise that the output will have a WTF reaction but the key
> observations to me are
>
> a) a single write request from flusher took over a second to complete
> b) at the time it was queued, it was mostly other writes that were in
> the queue at the same time
> c) The write request and the parallel writes were all asynchronous write
> requests
> D) at the time the request completed, there were a LARGE number of
> other requested queued and completed at the same time.
>
> Of the requests queued and completed in the meantime the breakdown was
>
> 22 RM
> 31 RA
> 82 W
> 445 R
>
> If I'm reading this correctly, it is saying that 22 reads were merged (RM),
> 31 reads were remapped to another device (RA) which is probably reads from
> the dm-crypt partition, 82 were writes (W) which is not far off the number
> of writes that were in the queue and 445 were other reads. The delay was
> dominated by reads that were queued after the write request and completed
> before it.

RM == Read Meta
RA == Read Ahead (remapping, by the way, does not happen across
devices, just into partitions)
W and R you understood correctly.

> That's saying that the 27128th request in the trace took over 7 seconds
> to complete and was an asynchronous write from flusher. The contents of
> the queue are displayed at that time and the breakdown of requests is
>
> 23 WS [JEM: write sync]
> 86 RM [JEM: Read Meta]
> 124 RA [JEM: Read Ahead]
> 442 W
> 1931 R
>
> 7 seconds later when it was completed the breakdown of completed
> requests was
>
> 25 WS
> 114 RM
> 155 RA
> 408 W
> 2457 R
>
> In combination, that confirms for me that asynchronous writes from flush
> are being starved by reads. When a process requires a buffer that is locked
> by that asynchronous write from flusher, it stalls.
>
>> The thing is, we do want to make ext4 work well with cfq, and
>> prioritizing non-readahead read requests ahead of data writeback does
>> make sense. The issue is with is that metadata writes going through
>> the block device could in some cases effectively cause a priority
>> inversion when what had previously been an asynchronous writeback
>> starts blocking a foreground, user-visible process.
>>
>> At least, that's the theory;
>
> I *think* the data more or less confirms the theory but it'd be nice if
> someone else double checked in case I'm seeing what I want to see
> instead of what is actually there.

Looks sane. You can also see a lot of "preempt"s in the blkparse
output, which indicates exactly what you're saying. Any sync request
gets priority over the async requests.

I'll also note that even though your I/O is going all over the place
(D2C is pretty bad, 14ms), most of the time is spent waiting for a
struct request allocation or between Queue and Merge:

==================== All Devices ====================

ALL MIN AVG MAX N
--------------- ------------- ------------- ------------- -----------

Q2Q 0.000000001 0.000992259 8.898375882 2300861
Q2G 0.000000843 10.193261239 2064.079501935 1016463 <====
G2I 0.000000461 0.000044702 3.237065090 1015803
Q2M 0.000000101 8.203147238 2064.079367557 1311662
I2D 0.000002012 1.476824812 2064.089774419 1014890
M2D 0.000003283 6.994306138 283.573348664 1284872
D2C 0.000061889 0.014438316 0.857811758 2291996
Q2C 0.000072284 13.363007244 2064.092228625 2292191

==================== Device Overhead ====================

DEV | Q2G G2I Q2M I2D D2C
---------- | --------- --------- --------- --------- ---------
( 8, 0) | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%
---------- | --------- --------- --------- --------- ---------
Overall | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%

I'm not sure I believe that max value. 2064 seconds seems a bit high.
Also, Q2M should not be anywhere near that big, so more investigation is
required there. A quick look over the data doesn't show any such delays
(making me question the tools), but I'll write some code tomorrow to
verify the btt output.

Jan, if I were to come up with a way of promoting a particular async
queue to the front of the line, where would I put such a call in the
ext4/jbd2 code to be effective?

Mel, can you reproduce this at will? Do you have a reproducer that I
could run so I'm not constantly bugging you?

Cheers,
Jeff

2013-04-23 00:03:44

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Mon, Apr 22, 2013 at 06:42:23PM -0400, Jeff Moyer wrote:
>
> Jan, if I were to come up with a way of promoting a particular async
> queue to the front of the line, where would I put such a call in the
> ext4/jbd2 code to be effective?

Well, I thought we had discussed trying to bump a pending I/O
automatically when there was an attempt to call lock_buffer() on the
bh? That would be ideal, because we could keep the async writeback
low priority until someone is trying to wait upon it, at which point
obviously it should no longer be considered an async write call.

Failing that, this is something I've been toying with.... what do you
think?

http://patchwork.ozlabs.org/patch/238192/
http://patchwork.ozlabs.org/patch/238257/

(The first patch in the series just makes sure that allocation bitmap
reads are marked with the META/PRIO flags. It's not strictly speaking
related to the problem discussed here, but for completeness:
http://patchwork.ozlabs.org/patch/238193/)

- Ted

2013-04-23 09:31:11

by Jan Kara

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Mon 22-04-13 18:42:23, Jeff Moyer wrote:
> Jan, if I were to come up with a way of promoting a particular async
> queue to the front of the line, where would I put such a call in the
> ext4/jbd2 code to be effective?
As Ted wrote the simplies might be to put his directly in
__lock_buffer(). Something like:

diff --git a/fs/buffer.c b/fs/buffer.c
index b4dcb34..e026a3e 100644
--- a/fs/buffer.c
+++ b/fs/buffer.c
@@ -69,6 +69,12 @@ static int sleep_on_buffer(void *word)

void __lock_buffer(struct buffer_head *bh)
{
+ /*
+ * Likely under async writeback? Tell io scheduler we are
+ * now waiting for the IO...
+ */
+ if (PageWriteback(bh->b_page))
+ io_now_sync(bh->b_bdev, bh->b_blocknr);
wait_on_bit_lock(&bh->b_state, BH_Lock, sleep_on_buffer,
TASK_UNINTERRUPTIBLE);
}

Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR

2013-04-23 14:01:42

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Mon, Apr 22, 2013 at 06:42:23PM -0400, Jeff Moyer wrote:
> > 3. The blktrace indicates that reads can starve writes from flusher
> >
> > While there are people that can look at a blktrace and find problems
> > like they are rain man, I'm more like an ADHD squirrel when looking at
> > a trace. I wrote a script to look for what unrelated requests completed
> > while an request got stalled for over a second. It seemed like something
> > that a tool shoudl already exist for but I didn't find one unless btt
> > can give the information somehow.
>
> Care to share that script?
>

I would have preferred not to because it is an ugly hatchet job churned
out in a few minutes. It's written in perl and uses the text output from
blkparse making it slow. It uses an excessive amount of memory because I was
taking shortcuts so is resource heavy. It ignores most of the information
from blkparse and so there are gaps in what it reports. Even though it's
dogshit, it was useful in this particular case so I added it to mmtests
anyway. Be aware that it takes ages to run and you might want to break
the blkparse output into pieces

It's used something like

blkparse -i blktrace-log > blkparse-log
cat blkparse-log | $PATH_TO_MMTESTS/subreport/blktrace-queue-watch.pl

> [snip]
>
> > I recognise that the output will have a WTF reaction but the key
> > observations to me are
> >
> > a) a single write request from flusher took over a second to complete
> > b) at the time it was queued, it was mostly other writes that were in
> > the queue at the same time
> > c) The write request and the parallel writes were all asynchronous write
> > requests
> > D) at the time the request completed, there were a LARGE number of
> > other requested queued and completed at the same time.
> >
> > Of the requests queued and completed in the meantime the breakdown was
> >
> > 22 RM
> > 31 RA
> > 82 W
> > 445 R
> >
> > If I'm reading this correctly, it is saying that 22 reads were merged (RM),
> > 31 reads were remapped to another device (RA) which is probably reads from
> > the dm-crypt partition, 82 were writes (W) which is not far off the number
> > of writes that were in the queue and 445 were other reads. The delay was
> > dominated by reads that were queued after the write request and completed
> > before it.
>
> RM == Read Meta
> RA == Read Ahead (remapping, by the way, does not happen across
> devices, just into partitions)
> W and R you understood correctly.
>

Thanks for those corrections. I misread the meaning of the action
identifiers section of the blkparse manual. I should have double checked
the source.

> >> <SNIP>
> >> The thing is, we do want to make ext4 work well with cfq, and
> >> prioritizing non-readahead read requests ahead of data writeback does
> >> make sense. The issue is with is that metadata writes going through
> >> the block device could in some cases effectively cause a priority
> >> inversion when what had previously been an asynchronous writeback
> >> starts blocking a foreground, user-visible process.
> >>
> >> At least, that's the theory;
> >
> > I *think* the data more or less confirms the theory but it'd be nice if
> > someone else double checked in case I'm seeing what I want to see
> > instead of what is actually there.
>
> Looks sane. You can also see a lot of "preempt"s in the blkparse
> output, which indicates exactly what you're saying. Any sync request
> gets priority over the async requests.
>

Good to know.

> I'll also note that even though your I/O is going all over the place
> (D2C is pretty bad, 14ms), most of the time is spent waiting for a
> struct request allocation or between Queue and Merge:
>
> ==================== All Devices ====================
>
> ALL MIN AVG MAX N
> --------------- ------------- ------------- ------------- -----------
>
> Q2Q 0.000000001 0.000992259 8.898375882 2300861
> Q2G 0.000000843 10.193261239 2064.079501935 1016463 <====

This is not normally my sandbox so do you mind spelling this out?

IIUC, the time to allocate the struct request from the slab cache is just a
small portion of this time. The bulk of the time is spent in get_request()
waiting for congestion to clear on the request list for either the sync or
async queue. Once a process goes to sleep on that waitqueue, it has to wait
until enough requests on that queue have been serviced before it gets woken
again at which point it gets priority access to prevent further starvation.
This is the Queue To Get Reqiest (Q2G) delay. What we may be seeing here
is that the async queue was congested and on average, we are waiting for
10 seconds for it to clear. The maximum value may be bogus for reasons
explained later.

Is that accurate?

> G2I 0.000000461 0.000044702 3.237065090 1015803
> Q2M 0.000000101 8.203147238 2064.079367557 1311662
> I2D 0.000002012 1.476824812 2064.089774419 1014890
> M2D 0.000003283 6.994306138 283.573348664 1284872
> D2C 0.000061889 0.014438316 0.857811758 2291996
> Q2C 0.000072284 13.363007244 2064.092228625 2292191
>
> ==================== Device Overhead ====================
>
> DEV | Q2G G2I Q2M I2D D2C
> ---------- | --------- --------- --------- --------- ---------
> ( 8, 0) | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%
> ---------- | --------- --------- --------- --------- ---------
> Overall | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%
>
> I'm not sure I believe that max value. 2064 seconds seems a bit high.

It is so I looked closer at the timestamps and there is an one hour
correction about 4400 seconds into the test. Daylight savings time kicked
in on March 31st and the machine is rarely rebooted until this test case
came along. It looks like there is a timezone or time misconfiguration
on the laptop that starts the machine with the wrong time. NTP must have
corrected the time which skewed the readings in that window severely :(

Normally on my test machines these services are disabled to avoid
exactly this sort of problem.

> Also, Q2M should not be anywhere near that big, so more investigation is
> required there. A quick look over the data doesn't show any such delays
> (making me question the tools), but I'll write some code tomorrow to
> verify the btt output.
>

It might be a single set of readings during a time correction that
screwed it.

> Jan, if I were to come up with a way of promoting a particular async
> queue to the front of the line, where would I put such a call in the
> ext4/jbd2 code to be effective?
>
> Mel, can you reproduce this at will? Do you have a reproducer that I
> could run so I'm not constantly bugging you?
>

I can reproduce it at will. Due to the nature of the test, the test
results are variable and unfortunately it is one of the tricker mmtest
configurations to setup.

1. Get access to a webserver
2. Close mmtests to your test machine
git clone https://github.com/gormanm/mmtests.git
3. Edit shellpacks/common-config.sh and set WEBROOT to a webserver path
4. Create a tar.gz of a large git tree and place it at $WEBROOT/linux-2.6.tar.gz
Alternatively place a compressed git tree anywhere and edit
configs/config-global-dhp__io-multiple-source-latency
and update GITCHECKOUT_SOURCETAR
5. Create a tar.gz of a large maildir directory and place it at
$WEBROOT/$WEBROOT/maildir.tar.gz
Alternatively, use an existing maildir folder and set
MONITOR_INBOX_OPEN_MAILDIR in
configs/config-global-dhp__io-multiple-source-latency

It's awkward but it's not like there are standard benchmarks lying around
and it seemed the best way to reproduce the problems I typically see early
in the lifetime of a system or when running a git checkout when the tree
has not been used in a few hours. Run the actual test with

./run-mmtests.sh --config configs/config-global-dhp__io-multiple-source-latency --run-monitor test-name-of-your-choice

Results will be in work/log. You'll need to run this as root so it
can run blktrace and so it can drop_caches between git checkouts
(to force disk IO). If systemtap craps out on you, then edit
configs/config-global-dhp__io-multiple-source-latency and remove dstate
from MONITORS_GZIP

If you have trouble getting this running, ping me on IRC.

Thanks.

--
Mel Gorman
SUSE Labs

2013-04-23 15:33:11

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Sat, Apr 20, 2013 at 08:05:22PM -0400, Theodore Ts'o wrote:
> An alternate solution which I've been playing around adds buffer_head
> flags so we can indicate that a buffer contains metadata and/or should
> have I/O submitted with the REQ_PRIO flag set.
>

I beefed up the reporting slightly and tested the patches comparing
3.9-rc6 vanilla with your patches. The full report with graphs are at

http://www.csn.ul.ie/~mel/postings/ext4tag-20130423/report.html

3.9.0-rc6 3.9.0-rc6
vanilla ext4tag
User min 0.00 ( 0.00%) 0.00 ( 0.00%)
User mean nan ( nan%) nan ( nan%)
User stddev nan ( nan%) nan ( nan%)
User max 0.00 ( 0.00%) 0.00 ( 0.00%)
User range 0.00 ( 0.00%) 0.00 ( 0.00%)
System min 9.14 ( 0.00%) 9.13 ( 0.11%)
System mean 9.60 ( 0.00%) 9.73 ( -1.33%)
System stddev 0.39 ( 0.00%) 0.94 (-142.69%)
System max 10.31 ( 0.00%) 11.58 (-12.32%)
System range 1.17 ( 0.00%) 2.45 (-109.40%)
Elapsed min 665.54 ( 0.00%) 612.25 ( 8.01%)
Elapsed mean 775.35 ( 0.00%) 688.01 ( 11.26%)
Elapsed stddev 69.11 ( 0.00%) 58.22 ( 15.75%)
Elapsed max 858.40 ( 0.00%) 773.06 ( 9.94%)
Elapsed range 192.86 ( 0.00%) 160.81 ( 16.62%)
CPU min 3.00 ( 0.00%) 3.00 ( 0.00%)
CPU mean 3.60 ( 0.00%) 4.20 (-16.67%)
CPU stddev 0.49 ( 0.00%) 0.75 (-52.75%)
CPU max 4.00 ( 0.00%) 5.00 (-25.00%)
CPU range 1.00 ( 0.00%) 2.00 (-100.00%)

The patches appear to improve the git checkout times slightly but this
test is quite variable.

The vmstat figures report some reclaim activity but if you look at the graphs
further down you will see that the bulk of the kswapd reclaim scan and
steal activity is at the start of the test when it's downloading and
untarring a git tree to work with. (I also note that the mouse-over
graph for direct reclaim efficiency is broken but it's not important
right now).

>From iostat

3.9.0-rc6 3.9.0-rc6
vanilla ext4tag
Mean dm-0-avgqz 1.18 1.19
Mean dm-0-await 17.30 16.50
Mean dm-0-r_await 17.30 16.50
Mean dm-0-w_await 0.94 0.48
Mean sda-avgqz 650.29 719.81
Mean sda-await 2501.33 2597.23
Mean sda-r_await 30.01 24.91
Mean sda-w_await 11228.80 11120.64
Max dm-0-avgqz 12.30 10.14
Max dm-0-await 42.65 52.23
Max dm-0-r_await 42.65 52.23
Max dm-0-w_await 541.00 263.83
Max sda-avgqz 3811.93 3375.11
Max sda-await 7178.61 7170.44
Max sda-r_await 384.37 297.85
Max sda-w_await 51353.93 50338.25

There are no really obvious massive advantages to me there and if you look
at the graphs for the avgqs, await etc over time, the patched kernel are
not obviously better. The Wait CPU usage looks roughly the same too.

On the more positive side, the dstate systemtap monitor script tells me
that all processes were stalled for less time -- 9575 seconds versus
10910. The most severe event to stall on is sleep_on_buffer() as a
result of ext4_bread.

Vanilla kernel 3325677 ms stalled with 57 events
Patched kernel 2411471 ms stalled with 42 events

That's a pretty big drop but it gets bad again for the second worst stall --
wait_on_page_bit as a result of generic_file_buffered_write.

Vanilla kernel 1336064 ms stalled with 109 events
Patched kernel 2338781 ms stalled with 164 events

So conceptually the patches make sense but the first set of tests do
not indicate that they'll fix the problem and the stall times do not
indicate that interactivity will be any better. I'll still apply them
and boot them on my main work machine and see how they "feel" this
evening.

--
Mel Gorman
SUSE Labs

2013-04-23 15:51:09

by Theodore Y. Ts'o

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 23, 2013 at 04:33:05PM +0100, Mel Gorman wrote:
> That's a pretty big drop but it gets bad again for the second worst stall --
> wait_on_page_bit as a result of generic_file_buffered_write.
>
> Vanilla kernel 1336064 ms stalled with 109 events
> Patched kernel 2338781 ms stalled with 164 events

Do you have the stack trace for this stall? I'm wondering if this is
caused by the waiting for stable pages in write_begin() , or something
else.

If it is blocking caused by stable page writeback that's interesting,
since it would imply that something in your workload is trying to
write to a page that has already been modified (i.e., appending to a
log file, or updating a database file). Does that make sense given
what your workload might be running?

- Ted

2013-04-23 16:13:25

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Tue, Apr 23, 2013 at 11:50:19AM -0400, Theodore Ts'o wrote:
> On Tue, Apr 23, 2013 at 04:33:05PM +0100, Mel Gorman wrote:
> > That's a pretty big drop but it gets bad again for the second worst stall --
> > wait_on_page_bit as a result of generic_file_buffered_write.
> >
> > Vanilla kernel 1336064 ms stalled with 109 events
> > Patched kernel 2338781 ms stalled with 164 events
>
> Do you have the stack trace for this stall? I'm wondering if this is
> caused by the waiting for stable pages in write_begin() , or something
> else.
>

[<ffffffff81110238>] wait_on_page_bit+0x78/0x80
[<ffffffff815af294>] kretprobe_trampoline+0x0/0x4c
[<ffffffff81110e84>] generic_file_buffered_write+0x114/0x2a0
[<ffffffff81111ccd>] __generic_file_aio_write+0x1bd/0x3c0
[<ffffffff81111f4a>] generic_file_aio_write+0x7a/0xf0
[<ffffffff811ee639>] ext4_file_write+0x99/0x420
[<ffffffff81174d87>] do_sync_write+0xa7/0xe0
[<ffffffff81175447>] vfs_write+0xa7/0x180
[<ffffffff811758cd>] sys_write+0x4d/0x90
[<ffffffff815b3eed>] system_call_fastpath+0x1a/0x1f
[<ffffffffffffffff>] 0xffffffffffffffff

The processes that stalled in this particular trace are wget, latency-output,
tar and tclsh. Most of these are sequential writers except for tar which
is both a sequential reader and sequential writers.

> If it is blocking caused by stable page writeback that's interesting,
> since it would imply that something in your workload is trying to
> write to a page that has already been modified (i.e., appending to a
> log file, or updating a database file). Does that make sense given
> what your workload might be running?
>

I doubt it is stable write consider the type of processes that are running. I
would expect the bulk of the activity to be sequential readers or writers
of multiple files. The summarised report from the raw data is now at

http://www.csn.ul.ie/~mel/postings/ext4tag-20130423/dstate-summary-vanilla
http://www.csn.ul.ie/~mel/postings/ext4tag-20130423/dstate-summary-ext4tag

It's an aside but the worst of the stalls are incurred by systemd-tmpfile
which were not a deliberate part of the test and yet another thing that
I would not have caught unless I was running tests on my laptop. Looking
closer at that thing, the default configuration is to run the service 15
minutes after boot and after that it runs once a day. It looks like the
bulk of the scanning would be in /var/tmp/ looking at systemds own files
(over 3000 of them) which I'm a little amused by.

My normal test machines would not hit this because they are not systemd
based but the existance of thing thing is worth noting. Any IO-based tests
run on systemd-based distributions may give different results depending
on whether this service triggered during the test or not.

--
Mel Gorman
SUSE Labs

2013-04-24 19:09:53

by Jeff Moyer

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

Mel Gorman <[email protected]> writes:

>> I'll also note that even though your I/O is going all over the place
>> (D2C is pretty bad, 14ms), most of the time is spent waiting for a
>> struct request allocation or between Queue and Merge:
>>
>> ==================== All Devices ====================
>>
>> ALL MIN AVG MAX N
>> --------------- ------------- ------------- ------------- -----------
>>
>> Q2Q 0.000000001 0.000992259 8.898375882 2300861
>> Q2G 0.000000843 10.193261239 2064.079501935 1016463 <====
>
> This is not normally my sandbox so do you mind spelling this out?
>
> IIUC, the time to allocate the struct request from the slab cache is just a
> small portion of this time. The bulk of the time is spent in get_request()
> waiting for congestion to clear on the request list for either the sync or
> async queue. Once a process goes to sleep on that waitqueue, it has to wait
> until enough requests on that queue have been serviced before it gets woken
> again at which point it gets priority access to prevent further starvation.
> This is the Queue To Get Reqiest (Q2G) delay. What we may be seeing here
> is that the async queue was congested and on average, we are waiting for
> 10 seconds for it to clear. The maximum value may be bogus for reasons
> explained later.
>
> Is that accurate?

Yes, without getting into excruciating detail.

>> G2I 0.000000461 0.000044702 3.237065090 1015803
>> Q2M 0.000000101 8.203147238 2064.079367557 1311662
>> I2D 0.000002012 1.476824812 2064.089774419 1014890
>> M2D 0.000003283 6.994306138 283.573348664 1284872
>> D2C 0.000061889 0.014438316 0.857811758 2291996
>> Q2C 0.000072284 13.363007244 2064.092228625 2292191
>>
>> ==================== Device Overhead ====================
>>
>> DEV | Q2G G2I Q2M I2D D2C
>> ---------- | --------- --------- --------- --------- ---------
>> ( 8, 0) | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%
>> ---------- | --------- --------- --------- --------- ---------
>> Overall | 33.8259% 0.0001% 35.1275% 4.8932% 0.1080%
>>
>> I'm not sure I believe that max value. 2064 seconds seems a bit high.
>
> It is so I looked closer at the timestamps and there is an one hour
> correction about 4400 seconds into the test. Daylight savings time kicked
> in on March 31st and the machine is rarely rebooted until this test case
> came along. It looks like there is a timezone or time misconfiguration
> on the laptop that starts the machine with the wrong time. NTP must have
> corrected the time which skewed the readings in that window severely :(

Not sure I'm buying that argument, as there are no gaps in the blkparse
output. The logging is not done using wallclock time. I still haven't
had sufficient time to dig into these numbers.

>> Also, Q2M should not be anywhere near that big, so more investigation is
>> required there. A quick look over the data doesn't show any such delays
>> (making me question the tools), but I'll write some code tomorrow to
>> verify the btt output.
>>
>
> It might be a single set of readings during a time correction that
> screwed it.

Again, I don't think so.

> I can reproduce it at will. Due to the nature of the test, the test
> results are variable and unfortunately it is one of the tricker mmtest
> configurations to setup.
>
> 1. Get access to a webserver
> 2. Close mmtests to your test machine
> git clone https://github.com/gormanm/mmtests.git
> 3. Edit shellpacks/common-config.sh and set WEBROOT to a webserver path
> 4. Create a tar.gz of a large git tree and place it at $WEBROOT/linux-2.6.tar.gz
> Alternatively place a compressed git tree anywhere and edit
> configs/config-global-dhp__io-multiple-source-latency
> and update GITCHECKOUT_SOURCETAR
> 5. Create a tar.gz of a large maildir directory and place it at
> $WEBROOT/$WEBROOT/maildir.tar.gz
> Alternatively, use an existing maildir folder and set
> MONITOR_INBOX_OPEN_MAILDIR in
> configs/config-global-dhp__io-multiple-source-latency
>
> It's awkward but it's not like there are standard benchmarks lying around
> and it seemed the best way to reproduce the problems I typically see early
> in the lifetime of a system or when running a git checkout when the tree
> has not been used in a few hours. Run the actual test with
>
> ./run-mmtests.sh --config configs/config-global-dhp__io-multiple-source-latency --run-monitor test-name-of-your-choice
>
> Results will be in work/log. You'll need to run this as root so it
> can run blktrace and so it can drop_caches between git checkouts
> (to force disk IO). If systemtap craps out on you, then edit
> configs/config-global-dhp__io-multiple-source-latency and remove dstate
> from MONITORS_GZIP

And how do I determine whether I've hit the problem?

> If you have trouble getting this running, ping me on IRC.

Yes, I'm having issues getting things to go, but you didn't provide me a
time zone, an irc server or a nick to help me find you. Was that
intentional? ;-)

Cheers,
Jeff

2013-04-25 12:21:18

by Mel Gorman

[permalink] [raw]
Subject: Re: Excessive stall times on ext4 in 3.9-rc2

On Wed, Apr 24, 2013 at 03:09:13PM -0400, Jeff Moyer wrote:
> Mel Gorman <[email protected]> writes:
>
> >> I'll also note that even though your I/O is going all over the place
> >> (D2C is pretty bad, 14ms), most of the time is spent waiting for a
> >> struct request allocation or between Queue and Merge:
> >>
> >> ==================== All Devices ====================
> >>
> >> ALL MIN AVG MAX N
> >> --------------- ------------- ------------- ------------- -----------
> >>
> >> Q2Q 0.000000001 0.000992259 8.898375882 2300861
> >> Q2G 0.000000843 10.193261239 2064.079501935 1016463 <====
> >
> > This is not normally my sandbox so do you mind spelling this out?
> >
> > IIUC, the time to allocate the struct request from the slab cache is just a
> > small portion of this time. The bulk of the time is spent in get_request()
> > waiting for congestion to clear on the request list for either the sync or
> > async queue. Once a process goes to sleep on that waitqueue, it has to wait
> > until enough requests on that queue have been serviced before it gets woken
> > again at which point it gets priority access to prevent further starvation.
> > This is the Queue To Get Reqiest (Q2G) delay. What we may be seeing here
> > is that the async queue was congested and on average, we are waiting for
> > 10 seconds for it to clear. The maximum value may be bogus for reasons
> > explained later.
> >
> > Is that accurate?
>
> Yes, without getting into excruciating detail.


Good enough, thanks.

> >> I'm not sure I believe that max value. 2064 seconds seems a bit high.
> >
> > It is so I looked closer at the timestamps and there is an one hour
> > correction about 4400 seconds into the test. Daylight savings time kicked
> > in on March 31st and the machine is rarely rebooted until this test case
> > came along. It looks like there is a timezone or time misconfiguration
> > on the laptop that starts the machine with the wrong time. NTP must have
> > corrected the time which skewed the readings in that window severely :(
>
> Not sure I'm buying that argument, as there are no gaps in the blkparse
> output. The logging is not done using wallclock time. I still haven't
> had sufficient time to dig into these numbers.
>

Ok.

> > It's awkward but it's not like there are standard benchmarks lying around
> > and it seemed the best way to reproduce the problems I typically see early
> > in the lifetime of a system or when running a git checkout when the tree
> > has not been used in a few hours. Run the actual test with
> >
> > ./run-mmtests.sh --config configs/config-global-dhp__io-multiple-source-latency --run-monitor test-name-of-your-choice
> >
> > Results will be in work/log. You'll need to run this as root so it
> > can run blktrace and so it can drop_caches between git checkouts
> > (to force disk IO). If systemtap craps out on you, then edit
> > configs/config-global-dhp__io-multiple-source-latency and remove dstate
> > from MONITORS_GZIP
>
> And how do I determine whether I've hit the problem?
>

If systemtap is available then

cat work/log/dstate-TESTNAME-gitcheckout | subreport/stap-dstate-frequency

will give you a report on the worst stalls and the stack traces when those
stalls occurred. If the stalls are 10+ seconds then you've certainly hit
the problem.

Alternatively

cd work/log
../../compare-kernels.sh

Look at the average time it takes to run the git checkout. Is it
abnormally high in comparison to if there was no parallel IO? If you do
not know what the normal time is, run with

./run-mmtests.sh --config configs/config-global-dhp__io-multiple-source-latency --no-monitor test-name-no-monitor

The monitors are what's opening the maildir and generating the parallel
IO. If there is an excessive difference between the git checkout times,
then you've hit the problem.

Furthermore, look at the await times. If they do not appear in the
compare-kernels.sh output then

../../bin/compare-mmtests.pl -d . -b gitcheckout -n test-name-of-your-choice --print-monitor iostat

And look at the await times. Are they very high? If so, you hit the
problem. If you want a better look at the await figures over time,
either look at the iostat file or you can graph it with

../../bin/graph-mmtests.sh -d . -b gitcheckout -n test-name-of-your-choice --print-monitor iostat --sub-heading sda-await

where sda-await should be substituted for whatever the disk is that
you're running the test one.


> > If you have trouble getting this running, ping me on IRC.
>
> Yes, I'm having issues getting things to go, but you didn't provide me a
> time zone, an irc server or a nick to help me find you. Was that
> intentional? ;-)
>

Not consciously :) . I'm in the IST timezone (GMT+1), OFTC IRC network
and #mm channel. If it's my evening I'm not always responsive so send me
the error output on email and I'll try fixing any problems that way.

--
Mel Gorman
SUSE Labs