fio.git
9 years agoFio 1.42 fio-1.42
Jens Axboe [Wed, 14 Jul 2010 23:39:52 +0000 (17:39 -0600)]
Fio 1.42

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoNo need to use specific flag for io_u fill length
Radha Ramachandran [Wed, 14 Jul 2010 23:39:05 +0000 (17:39 -0600)]
No need to use specific flag for io_u fill length

The fill length already provides that information.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoReally fix broken option length check
Jens Axboe [Wed, 14 Jul 2010 22:27:02 +0000 (16:27 -0600)]
Really fix broken option length check

Commit 7b4cb1320928bdea2360cf36fc45c83700052d43 matched the
string passed in, but that doesn't work for postfix values
appended.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoTurn io_u filled variable into a flag
Jens Axboe [Wed, 14 Jul 2010 06:42:03 +0000 (08:42 +0200)]
Turn io_u filled variable into a flag

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoReuse filled pattern
Radha Ramachandran [Wed, 14 Jul 2010 06:36:07 +0000 (08:36 +0200)]
Reuse filled pattern

I made changes to fio so we wld re-use the already populated io_u
buffer (when there is a non-random pattern) during writes. That way
only the header will be re-calculated for every I/O. This way the
buffer wld get populated in the beginning and as long as the
subsequent ios using the same io_u structure are writes and have same
or less block size, it wld get re-used. If any of the subsequent i/o
is a read or has a block size greater than the pre-filled one, then
the buffer is invalidated and will be re-filled at the next write.

Reason for this risky change: (Performance)
I tested this change on a tmpfs(with no swap backing), with the
following config file:
[sscan_write]
filename=/mytmpfs/datafile.tmp
rw=write
bs=64k
size=3G
ioengine=libaio
iodepth=1024
iodepth_low=512
runtime=10800
bwavgtime=5000
thread=1
do_verify=0
verify=meta
verify_pattern=0x55aaa55a
verify_interval=4k
continue_on_error=1

fio-1-41-6 gave 306MB/s and the new change had a performance of 1546MB/s

Side effects/Risks:
There is a risk with this fix, that if the buffer gets corrupted then
the subsequent writes will also be corrupt. I think for both
sequential writes and random writes (with verify, where the I/O log is
replayed) we shld be able to find the first I/O that started with the
corruption and if the buffer is getting corrupted, there are other
issues here.

Testing:
I have tested this fix with sequential write(verify)/random read write
mix combination(with verify).

I think I have taken care of most of the case, but please let me know
if there is anything I have missed. I have attached the patch along
with this email. I think the performance improvement outweighs the
risk associated with the fix. But I will let you decide if you wld
like to pick it up.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.6 fio-1.41.6
Jens Axboe [Fri, 9 Jul 2010 11:48:14 +0000 (13:48 +0200)]
Fio 1.41.6

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoOnly flag sync_file_range() support for newer Linux versions
Jens Axboe [Fri, 9 Jul 2010 11:47:32 +0000 (13:47 +0200)]
Only flag sync_file_range() support for newer Linux versions

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoUpdate documentation on new terse format and add version number
Jens Axboe [Wed, 30 Jun 2010 13:22:21 +0000 (15:22 +0200)]
Update documentation on new terse format and add version number

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix startdelay option with s/m/h/d postfix
Jens Axboe [Tue, 29 Jun 2010 08:40:30 +0000 (10:40 +0200)]
Fix startdelay option with s/m/h/d postfix

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agocrc32c-intel: replace SIGILL approach with cpuid
Jens Axboe [Tue, 29 Jun 2010 08:07:13 +0000 (10:07 +0200)]
crc32c-intel: replace SIGILL approach with cpuid

Some operating systems log illegal instructions, lets try and
be a bit nicer and use cpuid to identify the presence of
SSE4.2 which is needed to support the hw crc32c instruction.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agofix posix_fallocate() return value usage
Greg Edwards [Fri, 25 Jun 2010 15:25:22 +0000 (09:25 -0600)]
fix posix_fallocate() return value usage

posix_fallocate() returns the error number on failure.  Fix the return value
usage.

Signed-off-by: Greg Edwards <greg.edwards@hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agogracefully handle full /tmp file system
Greg Edwards [Fri, 25 Jun 2010 15:24:19 +0000 (09:24 -0600)]
gracefully handle full /tmp file system

If /tmp happens to be full, fio gives you a very unhelpful error:

# ./fio rand-write-256k-q256.job
Bus error

Use posix_fallocate() to gracefully handle this condition.

Signed-off-by: Greg Edwards <greg.edwards@hp.com>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.5 fio-1.41.5
Jens Axboe [Fri, 25 Jun 2010 12:34:14 +0000 (14:34 +0200)]
Fio 1.41.5

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoMerge branch 'master' of ssh://router/data/git/fio
Jens Axboe [Fri, 25 Jun 2010 12:34:08 +0000 (14:34 +0200)]
Merge branch 'master' of ssh://router/data/git/fio

9 years agoRevert "fix 1000 vs kb_base confusion in show_run_stats."
Jens Axboe [Fri, 25 Jun 2010 12:33:37 +0000 (14:33 +0200)]
Revert "fix 1000 vs kb_base confusion in show_run_stats."

This reverts commit 1f53e9d84573b1f9b9a23ad3ac4ef3d312d59c78.

The conversion is from msec to sec, so the 1000 factor
was correct.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoDocumentation update
Jens Axboe [Fri, 25 Jun 2010 09:32:48 +0000 (11:32 +0200)]
Documentation update

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.4 fio-1.41.4
Jens Axboe [Thu, 24 Jun 2010 08:39:02 +0000 (10:39 +0200)]
Fio 1.41.4

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd chardev size getting
Jens Axboe [Thu, 24 Jun 2010 08:27:22 +0000 (10:27 +0200)]
Add chardev size getting

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd total latency log
Jens Axboe [Thu, 24 Jun 2010 07:59:34 +0000 (09:59 +0200)]
Add total latency log

We log submission and completion latencies, also log total latencies.
This makes it easier to graph and make histograms, if you care only
about the complete latency.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix FreeBSD warning
Jens Axboe [Thu, 24 Jun 2010 07:36:11 +0000 (09:36 +0200)]
Fix FreeBSD warning

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd real blockdev_size() for FreeBSD
Jens Axboe [Thu, 24 Jun 2010 07:25:15 +0000 (09:25 +0200)]
Add real blockdev_size() for FreeBSD

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoMake fill_device detect bdev sizes too
Jens Axboe [Thu, 24 Jun 2010 07:22:41 +0000 (09:22 +0200)]
Make fill_device detect bdev sizes too

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix man page documentation for verify_pattern
Jens Axboe [Thu, 24 Jun 2010 06:37:13 +0000 (08:37 +0200)]
Fix man page documentation for verify_pattern

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAllow use of KiB etc to give explicit base 10 multiplies
Jens Axboe [Wed, 23 Jun 2010 20:24:07 +0000 (22:24 +0200)]
Allow use of KiB etc to give explicit base 10 multiplies

Useful for disk drives, where capacity is given in base 10 units.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agofix 1000 vs kb_base confusion in show_run_stats.
YAMAMOTO Takashi [Wed, 23 Jun 2010 12:46:35 +0000 (14:46 +0200)]
fix 1000 vs kb_base confusion in show_run_stats.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoposixaio: return any ready event in ->getevents()
Jens Axboe [Tue, 22 Jun 2010 08:52:13 +0000 (10:52 +0200)]
posixaio: return any ready event in ->getevents()

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd off64_t to FreeBSD
Jens Axboe [Tue, 22 Jun 2010 08:30:11 +0000 (10:30 +0200)]
Add off64_t to FreeBSD

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.3 fio-1.41.3
Jens Axboe [Mon, 21 Jun 2010 19:08:47 +0000 (21:08 +0200)]
Fio 1.41.3

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoSpeedup verify random fills by 10-15x
Jens Axboe [Mon, 21 Jun 2010 18:39:38 +0000 (20:39 +0200)]
Speedup verify random fills by 10-15x

Move the pseudo-random helper into lib/rand.c and use that
from the verify populate as well.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoDocumentation update
Jens Axboe [Mon, 21 Jun 2010 13:22:56 +0000 (15:22 +0200)]
Documentation update

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAuto-detect missing hw support for crc32c and fallback to software
Jens Axboe [Mon, 21 Jun 2010 13:08:17 +0000 (15:08 +0200)]
Auto-detect missing hw support for crc32c and fallback to software

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoMissing new line
Jens Axboe [Mon, 21 Jun 2010 12:36:42 +0000 (14:36 +0200)]
Missing new line

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoCleanup up SIGILL for crc32c-intel
Jens Axboe [Mon, 21 Jun 2010 12:36:11 +0000 (14:36 +0200)]
Cleanup up SIGILL for crc32c-intel

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd documentation for verify_backlog and verify_backlog_batch
Jens Axboe [Mon, 21 Jun 2010 10:53:26 +0000 (12:53 +0200)]
Add documentation for verify_backlog and verify_backlog_batch

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.2 fio-1.41.2
Jens Axboe [Mon, 21 Jun 2010 10:41:06 +0000 (12:41 +0200)]
Fio 1.41.2

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix disk util for slaves
Jens Axboe [Mon, 21 Jun 2010 10:40:53 +0000 (12:40 +0200)]
Fix disk util for slaves

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoDo diskutil check earlier
Jens Axboe [Mon, 21 Jun 2010 10:35:02 +0000 (12:35 +0200)]
Do diskutil check earlier

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41.1 fio-1.41.1
Jens Axboe [Mon, 21 Jun 2010 07:02:19 +0000 (09:02 +0200)]
Fio 1.41.1

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix a bug with multiple files in the same io_piece rbtree
Jens Axboe [Mon, 21 Jun 2010 07:01:33 +0000 (09:01 +0200)]
Fix a bug with multiple files in the same io_piece rbtree

Sort by file first, then sort by offset. This avoids false
positives on IO aliases, when it's really for a different file.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.41 fio-1.41
Jens Axboe [Fri, 18 Jun 2010 12:49:46 +0000 (14:49 +0200)]
Fio 1.41

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoVerify fatal fixups
Jens Axboe [Fri, 18 Jun 2010 12:48:43 +0000 (14:48 +0200)]
Verify fatal fixups

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoError reporting fixups for meta/pattern verifies
Jens Axboe [Fri, 18 Jun 2010 12:46:06 +0000 (14:46 +0200)]
Error reporting fixups for meta/pattern verifies

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoInclude file name in offset failure
Jens Axboe [Fri, 18 Jun 2010 12:32:21 +0000 (14:32 +0200)]
Include file name in offset failure

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoUnified verify failure error reporting
Jens Axboe [Fri, 18 Jun 2010 12:29:52 +0000 (14:29 +0200)]
Unified verify failure error reporting

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoparser: match on length of argument, not potential value
Jens Axboe [Fri, 18 Jun 2010 12:28:53 +0000 (14:28 +0200)]
parser: match on length of argument, not potential value

Otherwise we match subset of a string, that doesn't always
work well.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFirst step in unified crc error logging
Jens Axboe [Fri, 18 Jun 2010 11:44:11 +0000 (13:44 +0200)]
First step in unified crc error logging

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.40 fio-1.40
Jens Axboe [Fri, 18 Jun 2010 07:24:48 +0000 (09:24 +0200)]
Fio 1.40

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoRevert "Only apply verify interval to meta verifier, which uses it"
Jens Axboe [Fri, 18 Jun 2010 07:22:56 +0000 (09:22 +0200)]
Revert "Only apply verify interval to meta verifier, which uses it"

This reverts commit efcc4586d3c8974118fd66d53a09ffd739d77d11.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFix bad int truncate in verify
Jens Axboe [Fri, 18 Jun 2010 07:21:45 +0000 (09:21 +0200)]
Fix bad int truncate in verify

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoEnsure we never overlap with existing blocks
Jens Axboe [Thu, 17 Jun 2010 19:27:11 +0000 (21:27 +0200)]
Ensure we never overlap with existing blocks

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoOnly apply verify interval to meta verifier, which uses it
Jens Axboe [Thu, 17 Jun 2010 18:51:54 +0000 (20:51 +0200)]
Only apply verify interval to meta verifier, which uses it

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agomd5: fix bug in hash size
Jens Axboe [Thu, 17 Jun 2010 18:37:58 +0000 (20:37 +0200)]
md5: fix bug in hash size

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoDefault verify backlog batch to verify backlog setting if not given
Jens Axboe [Tue, 15 Jun 2010 12:27:28 +0000 (14:27 +0200)]
Default verify backlog batch to verify backlog setting if not given

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoAdd some support for a verify backlog
Jens Axboe [Tue, 15 Jun 2010 12:25:36 +0000 (14:25 +0200)]
Add some support for a verify backlog

Instead of writing everything and then verifying everything,
allow a job to specify incremental verify stages. This can
help reduce memory consumption of fio, since we don't have
to store a huge number of verify entries to be processed
when the write phase is complete.

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agofix a weird check of the return value of aio_cancel.
YAMAMOTO Takashi [Fri, 11 Jun 2010 07:21:33 +0000 (09:21 +0200)]
fix a weird check of the return value of aio_cancel.

i don't know what this 1 means in the first place but at least
it isn't portable.  instead, check if the i/o has already
completed. (AIO_ALLDONE)

Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoUpdate email addresses
Jens Axboe [Wed, 9 Jun 2010 07:49:38 +0000 (09:49 +0200)]
Update email addresses

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoinclude libgen.h for basename
YAMAMOTO Takashi [Wed, 9 Jun 2010 07:44:53 +0000 (09:44 +0200)]
include libgen.h for basename

Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.39 fio-1.39
Jens Axboe [Thu, 3 Jun 2010 08:29:03 +0000 (10:29 +0200)]
Fio 1.39

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoremove unnecessary bash-ism
YAMAMOTO Takashi [Wed, 2 Jun 2010 06:24:03 +0000 (08:24 +0200)]
remove unnecessary bash-ism

the following patch removes unnecesary dependency on bash.

Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMerge branch 'master' of ssh://router/data/git/fio
Jens Axboe [Tue, 1 Jun 2010 12:03:53 +0000 (14:03 +0200)]
Merge branch 'master' of ssh://router/data/git/fio

9 years agoFix undeclared mkdir() on Fedora 13
Jens Axboe [Tue, 1 Jun 2010 12:03:40 +0000 (14:03 +0200)]
Fix undeclared mkdir() on Fedora 13

Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
9 years agoFio 1.39-rc1 fio-1.39-rc1
Jens Axboe [Sat, 22 May 2010 18:46:05 +0000 (20:46 +0200)]
Fio 1.39-rc1

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix problem with --section matching
Jens Axboe [Sat, 22 May 2010 18:43:11 +0000 (20:43 +0200)]
Fix problem with --section matching

Strip the ']' at the end before doing section matching, otherwise
the strcmp() fails.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd missing os/os-netbsd.h file
Jens Axboe [Fri, 21 May 2010 05:14:04 +0000 (07:14 +0200)]
Add missing os/os-netbsd.h file

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoInclude latest Makefile changes in Makefile.NetBSD
YAMAMOTO Takashi [Thu, 20 May 2010 09:20:42 +0000 (11:20 +0200)]
Include latest Makefile changes in Makefile.NetBSD

Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix blkparse typo
Jens Axboe [Mon, 17 May 2010 17:51:45 +0000 (19:51 +0200)]
Fix blkparse typo

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoImprove ETA for fill_device based job files
Jens Axboe [Mon, 17 May 2010 10:29:57 +0000 (12:29 +0200)]
Improve ETA for fill_device based job files

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agofio netbsd support
YAMAMOTO Takashi [Wed, 12 May 2010 13:06:46 +0000 (15:06 +0200)]
fio netbsd support

Preliminary NetBSD support.

Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix typo in flist_sort.c
Jens Axboe [Tue, 4 May 2010 12:20:21 +0000 (14:20 +0200)]
Fix typo in flist_sort.c

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd flist sorting
Jens Axboe [Tue, 4 May 2010 11:06:47 +0000 (13:06 +0200)]
Add flist sorting

Imported from the kernel implementation.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoadd link to new list archives
Erik Inge Bolsø [Wed, 14 Apr 2010 08:13:57 +0000 (10:13 +0200)]
add link to new list archives

Signed-off-by: Erik Inge Bolsø <knan-fio@anduin.net>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix spelling error
Jens Axboe [Tue, 13 Apr 2010 08:39:14 +0000 (10:39 +0200)]
Fix spelling error

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMake sure that alias names are also valid on the command line
Jens Axboe [Tue, 13 Apr 2010 08:38:16 +0000 (10:38 +0200)]
Make sure that alias names are also valid on the command line

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMake the 'closest option' print a bit better
Jens Axboe [Thu, 1 Apr 2010 17:56:23 +0000 (19:56 +0200)]
Make the 'closest option' print a bit better

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd an option "cgroup_nodelete" to not delete cgroups after job completion
Vivek Goyal [Wed, 31 Mar 2010 20:55:15 +0000 (22:55 +0200)]
Add an option "cgroup_nodelete" to not delete cgroups after job completion

o Add an option cgroup_nodelete to not remove cgroups created by fio after
  the job completion. This can help a user in inspecting various cgroup
  files after fio job completion.

Signed-off-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agosfree(NULL) is ok
Jens Axboe [Wed, 31 Mar 2010 18:05:59 +0000 (20:05 +0200)]
sfree(NULL) is ok

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoOnly use true random generator for initial seed
Jens Axboe [Thu, 25 Mar 2010 22:31:50 +0000 (23:31 +0100)]
Only use true random generator for initial seed

Than hokus pokus the rest. End result is still OK random, and
it doesn't compress (which is the real test here).

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd documentation for the random number generator
Jens Axboe [Thu, 25 Mar 2010 22:13:32 +0000 (23:13 +0100)]
Add documentation for the random number generator

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years ago"Expand" random number from 32-bit to 64-bit when necessary
Jens Axboe [Thu, 25 Mar 2010 22:12:04 +0000 (23:12 +0100)]
"Expand" random number from 32-bit to 64-bit when necessary

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoRemember to prime the random generator
Jens Axboe [Thu, 25 Mar 2010 22:07:35 +0000 (23:07 +0100)]
Remember to prime the random generator

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMerge branch 'master' of ssh://brick.kernel.dk/data/git/fio
Jens Axboe [Thu, 25 Mar 2010 22:03:41 +0000 (23:03 +0100)]
Merge branch 'master' of ssh://brick.kernel.dk/data/git/fio

9 years agoFirst step at speeding up io_u rand refill
Jens Axboe [Thu, 25 Mar 2010 22:03:18 +0000 (23:03 +0100)]
First step at speeding up io_u rand refill

Makes it 3x faster here.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoKeep the global command line bw/lat logs separate from per job logs
Jens Axboe [Thu, 25 Mar 2010 09:42:49 +0000 (10:42 +0100)]
Keep the global command line bw/lat logs separate from per job logs

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix typo in bandwidth log command line option
Jens Axboe [Thu, 25 Mar 2010 09:38:07 +0000 (10:38 +0100)]
Fix typo in bandwidth log command line option

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoRemove default setting of clocksource
Jens Axboe [Wed, 24 Mar 2010 12:44:34 +0000 (13:44 +0100)]
Remove default setting of clocksource

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd IA64 support for CPU clock
Jens Axboe [Wed, 24 Mar 2010 12:35:28 +0000 (13:35 +0100)]
Add IA64 support for CPU clock

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMerge branch 'master' of ssh://router/data/git/fio
Jens Axboe [Wed, 24 Mar 2010 12:24:05 +0000 (13:24 +0100)]
Merge branch 'master' of ssh://router/data/git/fio

9 years agoAdd support for specific clock sources
Jens Axboe [Wed, 24 Mar 2010 12:23:53 +0000 (13:23 +0100)]
Add support for specific clock sources

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFio 1.38 fio-1.38
Jens Axboe [Tue, 23 Mar 2010 07:29:38 +0000 (08:29 +0100)]
Fio 1.38

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMake ARMv7 build and work
Jens Axboe [Sat, 20 Mar 2010 20:30:36 +0000 (21:30 +0100)]
Make ARMv7 build and work

Contributed by debian bug #573027

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=573027

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoOSX should use off_t for off64_t
Jens Axboe [Fri, 19 Mar 2010 20:31:01 +0000 (21:31 +0100)]
OSX should use off_t for off64_t

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoOnly define str_sfr_cb() if FIO_HAVE_SYNC_FILE_RANGE is set
Jens Axboe [Fri, 19 Mar 2010 18:17:15 +0000 (19:17 +0100)]
Only define str_sfr_cb() if FIO_HAVE_SYNC_FILE_RANGE is set

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years ago__dprint() fixup
Jens Axboe [Fri, 19 Mar 2010 18:09:37 +0000 (19:09 +0100)]
__dprint() fixup

Only define __dprint() function when FIO_INC_DEBUG is set. And
don't check debug type once, not both in dprint() and __dprint().

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoAdd log_valist()
Jens Axboe [Fri, 19 Mar 2010 15:50:37 +0000 (16:50 +0100)]
Add log_valist()

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoUpdate other OS makefiles
Jens Axboe [Fri, 19 Mar 2010 15:43:22 +0000 (16:43 +0100)]
Update other OS makefiles

They lack profile.o and debug.o

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoMake the dprint() processing out-of-line
Jens Axboe [Fri, 19 Mar 2010 15:41:52 +0000 (16:41 +0100)]
Make the dprint() processing out-of-line

Instead of having the big macro inlined everywhere, only
inline the mask check and put the rest out-of-line. This reduces
the size of fio with 4% here, and speeds it up.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoOptimize __get_io_u() for better code foot print
Jens Axboe [Fri, 19 Mar 2010 09:33:39 +0000 (10:33 +0100)]
Optimize __get_io_u() for better code foot print

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoFix bad sign on td_verror()
Jens Axboe [Fri, 19 Mar 2010 07:48:05 +0000 (08:48 +0100)]
Fix bad sign on td_verror()

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoDon't call getpid() in dprint() unless we are going to use it
Jens Axboe [Fri, 19 Mar 2010 07:08:06 +0000 (08:08 +0100)]
Don't call getpid() in dprint() unless we are going to use it

Reported-by: David Andersen <dga@cs.cmu.edu>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
9 years agoUse 'o' consistently in fixup_options()
Jens Axboe [Thu, 18 Mar 2010 19:45:43 +0000 (20:45 +0100)]
Use 'o' consistently in fixup_options()

It points to &td->o, and it's used interchangably in that function.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>