linux-kernel - Re: [Regression] Filesystem I/O is CPU-bound in rc7 and rc8

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <m33a12b8op.fsf@lugabout.jhcloos.org>
Date:	Mon, 15 Feb 2010 14:27:42 -0500
From:	James Cloos <cloos@...loos.com>
To:	Chris Mason <chris.mason@...cle.com>
Cc:	linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
	linux-ext4@...r.kernel.org, linux-btrfs@...r.kernel.org,
	jfs-discussion@...ts.sourceforge.net
Subject: Re: [Regression] Filesystem I/O is CPU-bound in rc7 and rc8

>>>>> "Chris" == Chris Mason <chris.mason@...cle.com> writes:

>> Sometime between rc6 and rc7 all filesystem I/O started using 100% CPU,
>> usually on the order of 60% sys, 40% user.

Chris> (Even better would be running perf record <your dd command line>, if
Chris> you're willing to setup perf)

The box has 512MB (the chipset's max).  (I did mention p3m, therefore OLD! :)

:; mount
/dev/root on / type jfs (rw,noatime)
/dev/sda3 on /var type ext4 (rw,noatime,barrier=1,data=ordered)
/dev/sdb2 on /media/bay type btrfs (rw,noatime,noacl)

: I only vaguely remember installing perf....
:; perf version
perf version 0.0.1.PERF

:; perf record dd if=/dev/sda4 of=/dev/null bs=4M count=16
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 14.7279 s, 4.6 MB/s
[ perf record: Captured and wrote 0.803 MB perf1.data (~35093 samples) ]

:; perf record dd if=/dev/sdb2 of=/dev/null bs=4M count=16
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 7.31642 s, 9.2 MB/s
[ perf record: Captured and wrote 0.938 MB perf2.data (~40980 samples) ]

:; perf record dd if=/dev/sdc2 of=/dev/null bs=4M count=16
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 60.9204 s, 1.1 MB/s
[ perf record: Captured and wrote 3.309 MB perf3.data (~144590 samples) ]

(I then mounted:
/dev/sdc2 on /media/usb type ext4 (rw,relatime,barrier=1,data=ordered)
)

:; perf record dd if=/dev/zero of=ZERO bs=4M count=16 # sda4
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 10.0027 s, 6.7 MB/s
[ perf record: Captured and wrote 1.634 MB perf4.data (~71386 samples) ]

:; perf record dd if=/dev/zero of=/var/tmp/ZERO bs=4M count=16 # sda3
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 1.712 s, 39.2 MB/s
[ perf record: Captured and wrote 0.364 MB perf5.data (~15915 samples) ]

:; perf record dd if=/dev/zero of=/media/bay/ZERO bs=4M count=16
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 4.51859 s, 14.9 MB/s
[ perf record: Captured and wrote 0.611 MB perf6.data (~26675 samples) ]

:; perf record dd if=/dev/zero of=/media/usb/ZERO bs=4M count=16
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 8.49166 s, 7.9 MB/s
[ perf record: Captured and wrote 1.523 MB perf.data (~66556 samples) ]

:; for ij in /media/usb/ZERO /media/bay/ZERO /var/tmp/ZERO /root/ZERO;do perf record dd of=/dev/null if=${ij} bs=4M count=16; mv -v perf.data perf$[kl++].data; done
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 49.1844 s, 1.4 MB/s
[ perf record: Captured and wrote 0.569 MB perf.data (~24858 samples) ]
`perf.data' -> `perf8.data'
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 10.0016 s, 6.7 MB/s
[ perf record: Captured and wrote 0.948 MB perf.data (~41438 samples) ]
`perf.data' -> `perf9.data'
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 5.72855 s, 11.7 MB/s
[ perf record: Captured and wrote 0.580 MB perf.data (~25355 samples) ]
`perf.data' -> `perf10.data'
16+0 records in
16+0 records out
67108864 bytes (67 MB) copied, 7.32789 s, 9.2 MB/s
[ perf record: Captured and wrote 0.899 MB perf.data (~39278 samples) ]
`perf.data' -> `perf11.data'


View attachment "perf1.report" of type "text/plain" (29694 bytes)

View attachment "perf2.report" of type "text/plain" (31320 bytes)

View attachment "perf3.report" of type "text/plain" (36989 bytes)

View attachment "perf4.report" of type "text/plain" (29539 bytes)

View attachment "perf5.report" of type "text/plain" (26764 bytes)

View attachment "perf6.report" of type "text/plain" (32231 bytes)

View attachment "perf7.report" of type "text/plain" (33242 bytes)

View attachment "perf8.report" of type "text/plain" (24418 bytes)

View attachment "perf9.report" of type "text/plain" (33507 bytes)

View attachment "perf10.report" of type "text/plain" (28650 bytes)

View attachment "perf11.report" of type "text/plain" (33524 bytes)


-JimC
-- 
James Cloos <cloos@...loos.com>         OpenPGP: 1024D/ED7DAEA6