All of lore.kernel.org
 help / color / mirror / Atom feed
* Fio on solaris
@ 2009-11-19  9:39 Tom
  2009-11-19 20:14 ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Tom @ 2009-11-19  9:39 UTC (permalink / raw)
  To: fio

Hello!

I've a problem to make that fio for solaris:

> gmake -f Makefile.solaris
gcc -o gettime.o -c -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE
-D_FILE_OFFSET_BITS=64 -DFIO_INC_DEBUG gettime.c
In file included from gettime.c:11:
hash.h:27:2: #error Define GOLDEN_RATIO_PRIME for your wordsize.
In file included from gettime.c:11:
hash.h: In function `hash_long':
hash.h:51: error: `GOLDEN_RATIO_PRIME' undeclared (first use in this
function)
hash.h:51: error: (Each undeclared identifier is reported only once
hash.h:51: error: for each function it appears in.)
hash.h:55: error: `__WORDSIZE' undeclared (first use in this function)
gmake: *** [gettime.o] Error 1

I've no idea what that means, any help?


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-19  9:39 Fio on solaris Tom
@ 2009-11-19 20:14 ` Jens Axboe
  2009-11-23 12:03   ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Jens Axboe @ 2009-11-19 20:14 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Thu, Nov 19 2009, Tom wrote:
> Hello!
> 
> I've a problem to make that fio for solaris:
> 
> > gmake -f Makefile.solaris
> gcc -o gettime.o -c -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE
> -D_FILE_OFFSET_BITS=64 -DFIO_INC_DEBUG gettime.c
> In file included from gettime.c:11:
> hash.h:27:2: #error Define GOLDEN_RATIO_PRIME for your wordsize.
> In file included from gettime.c:11:
> hash.h: In function `hash_long':
> hash.h:51: error: `GOLDEN_RATIO_PRIME' undeclared (first use in this
> function)
> hash.h:51: error: (Each undeclared identifier is reported only once
> hash.h:51: error: for each function it appears in.)
> hash.h:55: error: `__WORDSIZE' undeclared (first use in this function)
> gmake: *** [gettime.o] Error 1
> 
> I've no idea what that means, any help?

Probably my define shuffling, I'll check and fix that up.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-19 20:14 ` Jens Axboe
@ 2009-11-23 12:03   ` Jens Axboe
  2009-11-23 12:13     ` Tom
  0 siblings, 1 reply; 11+ messages in thread
From: Jens Axboe @ 2009-11-23 12:03 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Thu, Nov 19 2009, Jens Axboe wrote:
> On Thu, Nov 19 2009, Tom wrote:
> > Hello!
> > 
> > I've a problem to make that fio for solaris:
> > 
> > > gmake -f Makefile.solaris
> > gcc -o gettime.o -c -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE
> > -D_FILE_OFFSET_BITS=64 -DFIO_INC_DEBUG gettime.c
> > In file included from gettime.c:11:
> > hash.h:27:2: #error Define GOLDEN_RATIO_PRIME for your wordsize.
> > In file included from gettime.c:11:
> > hash.h: In function `hash_long':
> > hash.h:51: error: `GOLDEN_RATIO_PRIME' undeclared (first use in this
> > function)
> > hash.h:51: error: (Each undeclared identifier is reported only once
> > hash.h:51: error: for each function it appears in.)
> > hash.h:55: error: `__WORDSIZE' undeclared (first use in this function)
> > gmake: *** [gettime.o] Error 1
> > 
> > I've no idea what that means, any help?
> 
> Probably my define shuffling, I'll check and fix that up.

Re-pull (or download latest snapshot), it's fixed. I just did a check
compile on Solaris.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 12:03   ` Jens Axboe
@ 2009-11-23 12:13     ` Tom
  2009-11-23 12:14       ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Tom @ 2009-11-23 12:13 UTC (permalink / raw)
  To: Jens Axboe; +Cc: Tom, fio

Hi Jens,
thanks for that fast answer and job.
Where will I find the new Version, I checked http://brick.kernel.dk/snaps
but the latest file is
"fio-git-20091113222504.tar.gz        13-Nov-2009"

Thanks
Tom

> On Thu, Nov 19 2009, Jens Axboe wrote:
>> On Thu, Nov 19 2009, Tom wrote:
>> > Hello!
>> >
>> > I've a problem to make that fio for solaris:
>> >
>> > > gmake -f Makefile.solaris
>> > gcc -o gettime.o -c -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE
>> > -D_FILE_OFFSET_BITS=64 -DFIO_INC_DEBUG gettime.c
>> > In file included from gettime.c:11:
>> > hash.h:27:2: #error Define GOLDEN_RATIO_PRIME for your wordsize.
>> > In file included from gettime.c:11:
>> > hash.h: In function `hash_long':
>> > hash.h:51: error: `GOLDEN_RATIO_PRIME' undeclared (first use in this
>> > function)
>> > hash.h:51: error: (Each undeclared identifier is reported only once
>> > hash.h:51: error: for each function it appears in.)
>> > hash.h:55: error: `__WORDSIZE' undeclared (first use in this function)
>> > gmake: *** [gettime.o] Error 1
>> >
>> > I've no idea what that means, any help?
>>
>> Probably my define shuffling, I'll check and fix that up.
>
> Re-pull (or download latest snapshot), it's fixed. I just did a check
> compile on Solaris.
>
> --
> Jens Axboe
>
>





^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 12:13     ` Tom
@ 2009-11-23 12:14       ` Jens Axboe
  2009-11-23 12:51         ` Tom
  0 siblings, 1 reply; 11+ messages in thread
From: Jens Axboe @ 2009-11-23 12:14 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Mon, Nov 23 2009, Tom wrote:
> Hi Jens,
> thanks for that fast answer and job.
> Where will I find the new Version, I checked http://brick.kernel.dk/snaps
> but the latest file is
> "fio-git-20091113222504.tar.gz        13-Nov-2009"

Check again, I ran the script by hand (it runs once every hour when a
new commit is made).

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 12:14       ` Jens Axboe
@ 2009-11-23 12:51         ` Tom
  2009-11-23 12:59           ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Tom @ 2009-11-23 12:51 UTC (permalink / raw)
  To: Jens Axboe; +Cc: Tom, fio

sorry to say that, but it won't run:

> gmake -f Makefile.solaris
gcc -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64
-DFIO_INC_DEBUG -o fio gettime.o fio.o ioengines.o init.o stat.o log.o
time.o filesetup.o eta.o verify.o memory.o io_u.o parse.o mutex.o
options.o rbtree.o fifo.o smalloc.o filehash.o lib/strsep.o crc/crc7.o
crc/crc16.o crc/crc32.o crc/crc32c.o crc/crc32c-intel.o crc/crc64.o
crc/sha1.o crc/sha256.o crc/sha512.o crc/md5.o engines/cpu.o
engines/mmap.o engines/posixaio.o engines/sync.o engines/null.o
engines/net.o engines/solarisaio.o  -lpthread -lm -ldl -laio -lrt -lnsl
-lsocket
Undefined                       first referenced
 symbol                             in file
posix_fallocate                     filesetup.o
posix_memalign                      fio.o
ld: fatal: Symbol referencing errors. No output written to fio
collect2: ld returned 1 exit status
gmake: *** [fio] Error 1


What I have here:
> uname -a
SunOS servername 5.10 Generic_141415-07 i86pc i386 i86pc

Tom
> On Mon, Nov 23 2009, Tom wrote:
>> Hi Jens,
>> thanks for that fast answer and job.
>> Where will I find the new Version, I checked
>> http://brick.kernel.dk/snaps
>> but the latest file is
>> "fio-git-20091113222504.tar.gz        13-Nov-2009"
>
> Check again, I ran the script by hand (it runs once every hour when a
> new commit is made).
>
> --
> Jens Axboe
>
>






^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 12:51         ` Tom
@ 2009-11-23 12:59           ` Jens Axboe
  2009-11-23 13:21             ` Tom
  0 siblings, 1 reply; 11+ messages in thread
From: Jens Axboe @ 2009-11-23 12:59 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Mon, Nov 23 2009, Tom wrote:
> sorry to say that, but it won't run:
> 
> > gmake -f Makefile.solaris
> gcc -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64
> -DFIO_INC_DEBUG -o fio gettime.o fio.o ioengines.o init.o stat.o log.o
> time.o filesetup.o eta.o verify.o memory.o io_u.o parse.o mutex.o
> options.o rbtree.o fifo.o smalloc.o filehash.o lib/strsep.o crc/crc7.o
> crc/crc16.o crc/crc32.o crc/crc32c.o crc/crc32c-intel.o crc/crc64.o
> crc/sha1.o crc/sha256.o crc/sha512.o crc/md5.o engines/cpu.o
> engines/mmap.o engines/posixaio.o engines/sync.o engines/null.o
> engines/net.o engines/solarisaio.o  -lpthread -lm -ldl -laio -lrt -lnsl
> -lsocket
> Undefined                       first referenced
>  symbol                             in file
> posix_fallocate                     filesetup.o
> posix_memalign                      fio.o
> ld: fatal: Symbol referencing errors. No output written to fio
> collect2: ld returned 1 exit status
> gmake: *** [fio] Error 1
> 
> 
> What I have here:
> > uname -a
> SunOS servername 5.10 Generic_141415-07 i86pc i386 i86pc

So apparently that version doesn't have posix_fallocate or
posix_memalign, that's a bit odd. What I tested on:

$ uname -a
SunOS unknown 5.11 snv_117 sun4v sparc sun4v

It builds and links fine for me. If you have the time, can you double
check if it's hiding on your system in some other lib that isn't linked
in? I take it you don't get any compile warnings on missing
declarations, it's just the linker error?

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 12:59           ` Jens Axboe
@ 2009-11-23 13:21             ` Tom
  2009-11-23 14:19               ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Tom @ 2009-11-23 13:21 UTC (permalink / raw)
  To: Jens Axboe; +Cc: Tom, fio

I search both on my server and found nothing, so I ask grandma google and
I found this:

Don't use the POSIX_MEMALIGN function on solaris.

on
http://www.ruby-forum.com/topic/165346

in a Website which is not our theme, but interessting.

I try to find it with grep:

# grep memalign /usr/include/stdlib.h
extern void *memalign(size_t, size_t);
extern void *memalign();

# grep fallocate /usr/include/stdlib.h
# grep fallocate /usr/include/*


Tom

> On Mon, Nov 23 2009, Tom wrote:
>> sorry to say that, but it won't run:
>>
>> > gmake -f Makefile.solaris
>> gcc -Wall -O2 -g -D_GNU_SOURCE -D_LARGEFILE_SOURCE
>> -D_FILE_OFFSET_BITS=64
>> -DFIO_INC_DEBUG -o fio gettime.o fio.o ioengines.o init.o stat.o log.o
>> time.o filesetup.o eta.o verify.o memory.o io_u.o parse.o mutex.o
>> options.o rbtree.o fifo.o smalloc.o filehash.o lib/strsep.o crc/crc7.o
>> crc/crc16.o crc/crc32.o crc/crc32c.o crc/crc32c-intel.o crc/crc64.o
>> crc/sha1.o crc/sha256.o crc/sha512.o crc/md5.o engines/cpu.o
>> engines/mmap.o engines/posixaio.o engines/sync.o engines/null.o
>> engines/net.o engines/solarisaio.o  -lpthread -lm -ldl -laio -lrt -lnsl
>> -lsocket
>> Undefined                       first referenced
>>  symbol                             in file
>> posix_fallocate                     filesetup.o
>> posix_memalign                      fio.o
>> ld: fatal: Symbol referencing errors. No output written to fio
>> collect2: ld returned 1 exit status
>> gmake: *** [fio] Error 1
>>
>>
>> What I have here:
>> > uname -a
>> SunOS servername 5.10 Generic_141415-07 i86pc i386 i86pc
>
> So apparently that version doesn't have posix_fallocate or
> posix_memalign, that's a bit odd. What I tested on:
>
> $ uname -a
> SunOS unknown 5.11 snv_117 sun4v sparc sun4v
>
> It builds and links fine for me. If you have the time, can you double
> check if it's hiding on your system in some other lib that isn't linked
> in? I take it you don't get any compile warnings on missing
> declarations, it's just the linker error?
>
> --
> Jens Axboe
>
>






^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 13:21             ` Tom
@ 2009-11-23 14:19               ` Jens Axboe
  2009-11-23 15:50                 ` Tom
  0 siblings, 1 reply; 11+ messages in thread
From: Jens Axboe @ 2009-11-23 14:19 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Mon, Nov 23 2009, Tom wrote:
> I search both on my server and found nothing, so I ask grandma google and
> I found this:
> 
> Don't use the POSIX_MEMALIGN function on solaris.
> 
> on
> http://www.ruby-forum.com/topic/165346
> 
> in a Website which is not our theme, but interessting.
> 
> I try to find it with grep:
> 
> # grep memalign /usr/include/stdlib.h
> extern void *memalign(size_t, size_t);
> extern void *memalign();
> 
> # grep fallocate /usr/include/stdlib.h
> # grep fallocate /usr/include/*

Can you try the newest snapshot? I added weak directives for those two
helpers.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 14:19               ` Jens Axboe
@ 2009-11-23 15:50                 ` Tom
  2009-11-23 16:24                   ` Jens Axboe
  0 siblings, 1 reply; 11+ messages in thread
From: Tom @ 2009-11-23 15:50 UTC (permalink / raw)
  To: Jens Axboe; +Cc: Tom, fio

Thanks Jens!
now it's possible to compile that, but (yes, we've a "but"), when I call:

> fio four_threads.fio
fio: engine libaio not loadable
fio: failed to load engine libaio
fio: file:ioengines.c:98, func=dlopen, error=ld.so.1: fio: fatal: libaio:
open failed: No such file or directory

I search for that libaio and found it

> ll    /usr/lib/libaio.so*
lrwxrwxrwx   1 root     root          21 Sep  4 10:28 /usr/lib/libaio.so
-> ../../lib/libaio.so.1
lrwxrwxrwx   1 root     root          21 Sep  4 10:28 /usr/lib/libaio.so.1
-> ../../lib/libaio.so.1

> ll /lib/libaio.so*
lrwxrwxrwx   1 root     root          11 Sep  4 10:34 /lib/libaio.so ->
libaio.so.1
-rwxr-xr-x   1 root     bin        46452 Feb 12  2009 /lib/libaio.so.1

but that on isn't the right one?

What kind of libaio have you installed at your SUN where ist it from? We
have looked around. For Linux we found tonns of libaio and libaio-devel,
but nothing for SUN.

Tom
> On Mon, Nov 23 2009, Tom wrote:
>> I search both on my server and found nothing, so I ask grandma google
>> and
>> I found this:
>>
>> Don't use the POSIX_MEMALIGN function on solaris.
>>
>> on
>> http://www.ruby-forum.com/topic/165346
>>
>> in a Website which is not our theme, but interessting.
>>
>> I try to find it with grep:
>>
>> # grep memalign /usr/include/stdlib.h
>> extern void *memalign(size_t, size_t);
>> extern void *memalign();
>>
>> # grep fallocate /usr/include/stdlib.h
>> # grep fallocate /usr/include/*
>
> Can you try the newest snapshot? I added weak directives for those two
> helpers.
>
> --
> Jens Axboe
>
>






^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: Fio on solaris
  2009-11-23 15:50                 ` Tom
@ 2009-11-23 16:24                   ` Jens Axboe
  0 siblings, 0 replies; 11+ messages in thread
From: Jens Axboe @ 2009-11-23 16:24 UTC (permalink / raw)
  To: Tom; +Cc: fio

On Mon, Nov 23 2009, Tom wrote:
> Thanks Jens!
> now it's possible to compile that, but (yes, we've a "but"), when I call:
> 
> > fio four_threads.fio
> fio: engine libaio not loadable
> fio: failed to load engine libaio
> fio: file:ioengines.c:98, func=dlopen, error=ld.so.1: fio: fatal: libaio:
> open failed: No such file or directory
> 
> I search for that libaio and found it
> 
> > ll    /usr/lib/libaio.so*
> lrwxrwxrwx   1 root     root          21 Sep  4 10:28 /usr/lib/libaio.so
> -> ../../lib/libaio.so.1
> lrwxrwxrwx   1 root     root          21 Sep  4 10:28 /usr/lib/libaio.so.1
> -> ../../lib/libaio.so.1
> 
> > ll /lib/libaio.so*
> lrwxrwxrwx   1 root     root          11 Sep  4 10:34 /lib/libaio.so ->
> libaio.so.1
> -rwxr-xr-x   1 root     bin        46452 Feb 12  2009 /lib/libaio.so.1
> 
> but that on isn't the right one?
> 
> What kind of libaio have you installed at your SUN where ist it from? We
> have looked around. For Linux we found tonns of libaio and libaio-devel,
> but nothing for SUN.

The libaio engine is Linux specific. On Solaris, the available async
engines are posixaio and the native solarisaio.

-- 
Jens Axboe


^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2009-11-23 16:24 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2009-11-19  9:39 Fio on solaris Tom
2009-11-19 20:14 ` Jens Axboe
2009-11-23 12:03   ` Jens Axboe
2009-11-23 12:13     ` Tom
2009-11-23 12:14       ` Jens Axboe
2009-11-23 12:51         ` Tom
2009-11-23 12:59           ` Jens Axboe
2009-11-23 13:21             ` Tom
2009-11-23 14:19               ` Jens Axboe
2009-11-23 15:50                 ` Tom
2009-11-23 16:24                   ` Jens Axboe

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.