Jens Axboe [Fri, 10 Oct 2008 11:49:50 +0000 (13:49 +0200)]
Fio 1.22
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 7 Oct 2008 09:33:21 +0000 (11:33 +0200)]
Fio 1.22-rc3
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 7 Oct 2008 09:33:00 +0000 (11:33 +0200)]
Warning fixes
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Aaron Carroll [Tue, 7 Oct 2008 09:25:38 +0000 (11:25 +0200)]
Document environment variable expansion
Add environment variable expansion documentation to HOWTO, and fix
a few nearby typo's while we're at it.
Signed-off-by: Aaron Carroll <aaronc@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Aaron Carroll [Tue, 7 Oct 2008 09:25:20 +0000 (11:25 +0200)]
Add environment-variable substitution to config options
Strings of the form ${VARNAME} in config option strings are substituted with
the value of the environment variable VARNAME. Only the right hand side of
an option assignment undergoes substitution. If VARNAME is empty or
undefined, the empty string is substituted.
Signed-off-by: Aaron Carroll <aaronc@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 1 Oct 2008 07:04:28 +0000 (09:04 +0200)]
Update README for new mailing list info
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 30 Sep 2008 06:39:17 +0000 (08:39 +0200)]
Fix missing initialization of merge/io counts
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shehjar Tikoo [Tue, 30 Sep 2008 01:11:35 +0000 (11:11 +1000)]
Aggregate slave device stats into the stats for
But do so by adding new fields in the stats line.
We dont want the existing output format to get clobbered.
The new fields in master device line which contain
aggregated stats start with "aggr".
Signed-off-by: Shehjar Tikoo <shehjart@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shehjar Tikoo [Tue, 30 Sep 2008 00:35:44 +0000 (10:35 +1000)]
Add support to display slave device stats for software RAIDs.
When running tests on a device created using mdadm, the relevant
disk statistics must also come from the individual
disks that constitute the RAID, because the software RAID device
is just a virtual device.
This patch goes through the 'slaves' directory in the software
RAID device's directory in /sys/block, gets a list of all the
slave devices and creates a master-slave relationship between
the virtual device and the actual disks.
The individual slave disks' stats are displayed along with the
master's disk stats at the end of the test.
Signed-off-by: Shehjar Tikoo <shehjart@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Sat, 27 Sep 2008 10:33:53 +0000 (12:33 +0200)]
HOWTO: add comment on terse and --minimal
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 18 Sep 2008 01:01:34 +0000 (18:01 -0700)]
Fio 1.22-rc2
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Aaron Carroll [Fri, 12 Sep 2008 08:57:13 +0000 (10:57 +0200)]
Don't compile crc32c-intel on non-x86 platforms
crc32c_intel() doesn't make sense on platforms without SSE, so alias it
to crc32c() on such machines.
Signed-off-by: Aaron Carroll <aaronc@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 11 Sep 2008 08:17:26 +0000 (10:17 +0200)]
Update ramp_time
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 11 Sep 2008 07:27:15 +0000 (09:27 +0200)]
Update options
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 10 Sep 2008 07:48:11 +0000 (09:48 +0200)]
Fio 1.22-rc1
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 10 Sep 2008 07:46:16 +0000 (09:46 +0200)]
Add ramp_time option
Sometimes it's useful to let a job settle for a little while
before taking any measurements on latency and throughput, since
the initial rate on eg a write workload may be much higher than
the longer sustained rate.
So add a ramp_time option that allows the user to specify a lead
ramp time that must have passed before fio takes any performance
numbers into account.
Suggested by "Jenkins, Lee" <Lee.Jenkins@hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shaozhi Shawn Ye [Wed, 10 Sep 2008 07:09:37 +0000 (09:09 +0200)]
eta: fix bugs withe fill_device and exitall_on_terminate
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 10 Sep 2008 07:07:55 +0000 (09:07 +0200)]
filehash: check for NULL ->file_name in hash lookup
Original patch from Shaozhi Shawn Ye <yeshao@google.com>, modified
by me to continue the lookup and just skip the file instead of
returning NULL.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shaozhi Shawn Ye [Wed, 10 Sep 2008 07:05:17 +0000 (09:05 +0200)]
filehash: Filehash should be removed after the file is closed
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shaozhi Shawn Ye [Wed, 10 Sep 2008 07:04:18 +0000 (09:04 +0200)]
filesetup: When terse_output is set, The 'laying out' message should not be displayed.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Shaozhi Shawn Ye [Wed, 10 Sep 2008 07:02:51 +0000 (09:02 +0200)]
smalloc: Remove read/write lock race condition which will result in segmentation fault
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Daniel Rall [Wed, 10 Sep 2008 06:56:01 +0000 (08:56 +0200)]
Allow fio headers to be included by a C++ build
The attached patch allows fio 1.17.2 header files to be included by a
C++ build without generating compile errors due to C-ism.
Specifically, due to:
- use of the keyword "new" as a parameter name
- a cast of a return value to void *, when the function signature
returns a different type
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 4 Aug 2008 19:43:55 +0000 (21:43 +0200)]
Add SIGILL signal handler
If crc32c-intel is used and the opcode isn't supported by the CPU,
we'll display a nice warning instead of just dying.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 4 Aug 2008 13:40:35 +0000 (15:40 +0200)]
Remember to add crc32c_intel option check
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 4 Aug 2008 13:35:26 +0000 (15:35 +0200)]
Add intel hardware assisted crc32c support
Not tested, since I don't have the hardware.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 31 Jul 2008 17:55:02 +0000 (19:55 +0200)]
Detect and complain on smalloc failures
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 31 Jul 2008 17:50:54 +0000 (19:50 +0200)]
smalloc: default to 128 pools instead of 4
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Aaron Carroll [Wed, 25 Jun 2008 07:20:37 +0000 (09:20 +0200)]
Prevent filetype disappearing
fio_file->filetype is set in add_file(), but for duplicated files
fio-file->file_name is already set so add_file() is not called.
This patch copies the filetype at the same time as file_name so
it remains valid for dup'd files.
Signed-off-by: Aaron Carroll <aaronc@gelato.unsw.edu.au>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Fri, 13 Jun 2008 08:27:05 +0000 (10:27 +0200)]
Add uninitialized_var() to silence bogus compiler warnings
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Gurudas Pai [Fri, 13 Jun 2008 06:40:11 +0000 (08:40 +0200)]
eta overflow fix
Sometime I get eta's like "eta 4294967286m:4294967266s" , just an overflow
issue. The following patch fixes the issue.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Fri, 13 Jun 2008 06:38:40 +0000 (08:38 +0200)]
Revert "Add spinlocks"
This reverts commit
69ebbd397e80b23ec4281c99e9f7242a089cc771.
Jens Axboe [Fri, 13 Jun 2008 06:38:35 +0000 (08:38 +0200)]
Revert "Add spinlock wrapper helpers"
This reverts commit
8b4a7c860d004ea8dffc11f6232124b574fe6cd7.
Jens Axboe [Fri, 13 Jun 2008 06:38:31 +0000 (08:38 +0200)]
Revert "smalloc: convert to spinlocks"
This reverts commit
3f1959101ed6a935ff3e2aec0fe7f5828f23fbcb.
Jens Axboe [Fri, 13 Jun 2008 06:38:25 +0000 (08:38 +0200)]
Revert "Convert file hash lock to spinlocks"
This reverts commit
0ec7b3c9a0926e6377bd145206bb9f6a70be610d.
Jens Axboe [Fri, 13 Jun 2008 06:38:20 +0000 (08:38 +0200)]
Revert "Remove unused define"
This reverts commit
edf2c6cf5adc82ee43c81e033cf2215eb9ffdb16.
Jens Axboe [Thu, 12 Jun 2008 18:05:33 +0000 (20:05 +0200)]
Remove unused define
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 12 Jun 2008 09:53:57 +0000 (11:53 +0200)]
Convert file hash lock to spinlocks
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 12 Jun 2008 09:50:47 +0000 (11:50 +0200)]
smalloc: convert to spinlocks
Makes the allocator 10-20% faster.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 12 Jun 2008 09:49:46 +0000 (11:49 +0200)]
Add spinlock wrapper helpers
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 12 Jun 2008 09:26:26 +0000 (11:26 +0200)]
Add spinlocks
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Carl Henrik Lunde [Thu, 12 Jun 2008 07:17:46 +0000 (09:17 +0200)]
fio.1: Correct terse (--minimal) output - page faults
Major and minor page faults were added in commit
e7823a941468e
Signed-off-by: Carl Henrik Lunde <chlunde@ping.uio.no>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 18:46:29 +0000 (20:46 +0200)]
Merge branch 'master' of ssh://git.kernel.dk/data/git/fio
Jens Axboe [Wed, 11 Jun 2008 18:46:19 +0000 (20:46 +0200)]
Add crc32c checksum
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 18:44:57 +0000 (20:44 +0200)]
crc32 is uint32_t, not unsigned long
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 13:37:32 +0000 (15:37 +0200)]
Improve bsrange buflen generations
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 11:18:30 +0000 (13:18 +0200)]
Don't include headers in depend target
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 09:47:56 +0000 (11:47 +0200)]
Clear num_opts early
If we error out, it could be garbage.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 11 Jun 2008 08:21:43 +0000 (10:21 +0200)]
Older Linux versions do not have O_NOATIME
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 10 Jun 2008 18:26:06 +0000 (20:26 +0200)]
Remember to free **opts as well
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Tue, 10 Jun 2008 17:46:23 +0000 (19:46 +0200)]
Add option priorities
For instance, filename= must come after directory= or things will
go wrong. So add a priority value to each option, so that the parser
will take that into account.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 9 Jun 2008 17:25:49 +0000 (19:25 +0200)]
splice: if splicing fails with EINVAL, it's most likely missing fs support
So tell the user in more direct terms.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Fri, 6 Jun 2008 10:21:38 +0000 (12:21 +0200)]
Fio 1.21
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Fri, 6 Jun 2008 10:08:04 +0000 (12:08 +0200)]
Linux sysfs change broke fio disk lookup
/sys/block/<dev> is now a sym link to the real path, which
broke our disk lookup. Fix it.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 23:17:03 +0000 (01:17 +0200)]
ETA fix
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 09:59:42 +0000 (11:59 +0200)]
Fio 1.21-rc8
Should be the last -rc, not aware of any pending issues.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 09:53:08 +0000 (11:53 +0200)]
HOWTO spelling
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 09:48:22 +0000 (11:48 +0200)]
Help entry typos
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 07:03:30 +0000 (09:03 +0200)]
Add iodepth_batch_complete control
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Thu, 5 Jun 2008 07:03:05 +0000 (09:03 +0200)]
Don't print 100% done, when we don't have a time estimate yet
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 18:37:18 +0000 (20:37 +0200)]
MAP_ANON requires fd == -1
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 18:13:04 +0000 (20:13 +0200)]
Add OS agnostic RAND_MAX
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 17:57:47 +0000 (19:57 +0200)]
Compile warning for non-signal driven solaris asynch io
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 17:55:58 +0000 (19:55 +0200)]
Solaris: support for proper random functions
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 13:14:24 +0000 (15:14 +0200)]
Add directio support to Solaris
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 13:13:02 +0000 (15:13 +0200)]
We need to grab and mark the file open before jumping to an error path
The error path will close the file.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 12:49:27 +0000 (14:49 +0200)]
Kill useless undef's
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 12:48:29 +0000 (14:48 +0200)]
Add missing splice syscall numbers for alpha and s390
All archs now verified as compiling.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 12:31:25 +0000 (14:31 +0200)]
Add write barriers
And enable them in solarisaio, then it should be safe to use the
signal driven IO "interrupts".
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 08:47:08 +0000 (10:47 +0200)]
Fio 1.21-rc7
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Wed, 4 Jun 2008 08:35:25 +0000 (10:35 +0200)]
Missing time store in rwmix switch
Caused missing display of either read or write direction for a mixed
workload.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 14:15:59 +0000 (16:15 +0200)]
Base FreeBSD Makefile on Linux, but insist on gmake usage
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 13:57:45 +0000 (15:57 +0200)]
Makefile update
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 13:02:46 +0000 (15:02 +0200)]
Remove comment wrt sigaction() usage, it's deprecated
We already use sigaction(), so no difference between *BSD/Solaris
and Linux.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 13:01:16 +0000 (15:01 +0200)]
solarisaio: support signal driven async IO
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 12:18:23 +0000 (14:18 +0200)]
solarisaio: aiowait() can return (void *) -1 for error
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 11:31:26 +0000 (13:31 +0200)]
Move fls.h in with the other lib helpers
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 11:28:21 +0000 (13:28 +0200)]
Include solarisaio in README
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 11:23:34 +0000 (13:23 +0200)]
solarisaio: IO depth constrained by MAXASYNCHIO
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:55:14 +0000 (12:55 +0200)]
Fio 1.21-rc6
Stupid Linux compile error in -rc5
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:55:03 +0000 (12:55 +0200)]
Missed fls() -> __fls() conversion
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:52:56 +0000 (12:52 +0200)]
Fio 1.21-rc5
Getting close to 1.21, all three supported platforms (Linux, Solaris,
and FreeBSD) now compile and work fine.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:51:42 +0000 (12:51 +0200)]
FreeBSD Makefile update
Don't ask...
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:50:10 +0000 (12:50 +0200)]
mutex: only regard less-than-zero return as error
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:30:06 +0000 (12:30 +0200)]
net engine: missing includes
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:28:02 +0000 (12:28 +0200)]
posixaio: not all platforms have aio_fsync()
FreeBSD is (again) the odd man out...
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:25:12 +0000 (12:25 +0200)]
posixaio: need fcntl.h
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:23:40 +0000 (12:23 +0200)]
Guard posix_fallocate() behind OS define
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:22:30 +0000 (12:22 +0200)]
Rename fls() to __fls()
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:19:50 +0000 (12:19 +0200)]
Not all platforms have ENODATA
So use EIO instead.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 10:19:19 +0000 (12:19 +0200)]
Rename list_* function and file to flist_ to avoid conflict with FreeBSD
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 09:59:30 +0000 (11:59 +0200)]
Add Solaris native aio engine
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 08:37:36 +0000 (10:37 +0200)]
Don't make posix_fallocate() set td->verror
It's supposed to be an error log, not stop the job.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 08:17:12 +0000 (10:17 +0200)]
Fio 1.21-rc4
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 08:15:48 +0000 (10:15 +0200)]
Do at least one block if randommap fails us
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 08:09:43 +0000 (10:09 +0200)]
Inc block in mark_random_map()
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 07:59:32 +0000 (09:59 +0200)]
smalloc: use optimized ffz()
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 07:57:41 +0000 (09:57 +0200)]
Be sure to use internal __ffs, not glibc variant
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 07:53:05 +0000 (09:53 +0200)]
Speedup mark_random_map()
Do each int at the time, not each bit.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 07:43:29 +0000 (09:43 +0200)]
x86-64: enable arch ffz again
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Jens Axboe [Mon, 2 Jun 2008 07:41:51 +0000 (09:41 +0200)]
x86-64: ffz need only search the first 32-bits
So use bsfl, not bsfq.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>