Bug #20394 for BSD-Resource: Wrong Values Returned on Mac OS X?

Mon Jul 10 11:26:10 2006 david [...] kineticode.com - Ticket created

CC:	Dave Rolsky <autarch [...] urth.org>
Subject:	Wrong Values Returned on Mac OS X?
Date:	Mon, 10 Jul 2006 08:25:33 -0700
To:	bug-bsd-resource [...] rt.cpan.org
From:	David Wheeler <david [...] kineticode.com>

Hi Jarkko, In testing a new version of Apache::SizeLimit, a bunch of tests failed. Dave Rolsky, the new maintainer of that module, asked me to run: geertz% perl -MBSD::Resource -le 'print for (BSD::Resource::getrusage ())[2,3]' 0 0 It seems, however, that these values are unexpected. So is it the case that rusage() works differently on Mac OS X than on other BSDs? Or could this be a bug in BSD::Resource? Versions: Mac OS X 10.4.7 geertz% pv BSD::Resource 1.28 geertz% perl -v This is perl, v5.8.8 built for darwin-2level

Sat Oct 06 16:19:15 2007 JHI [...] cpan.org - Correspondence added

Sorry, missed this ticket... but doesn't help much that I know noticed it since BSD::Resource doesn't do much beyond what the underlying system supports: if your system returns zeros from getrusage(), that's what BSD::Resource will return from getrusage. (There is one known problem in Mac OS X in this area, but it's for setrlimit() and max number of processes.) FWIW, I'm not seeing the problem in my Mac OS X, and yours is the only one, I'm afraid. Maybe there's something different about the system you are running it in, or the user you are running it as, or something like that?

Sat Oct 06 16:19:17 2007 The RT System itself - Status changed from 'new' to 'open'

Sat Oct 06 23:01:22 2007 david [...] kineticode.com - Correspondence added

Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Sat, 6 Oct 2007 20:01:00 -0700
To:	bug-BSD-Resource [...] rt.cpan.org
From:	"David E. Wheeler" <david [...] kineticode.com>

On Oct 6, 2007, at 13:19, Jarkko_Hietaniemi via RT wrote: Show quoted text

> Sorry, missed this ticket... but doesn't help much that I know > noticed it since BSD::Resource > doesn't do much beyond what the underlying system supports: if your > system returns zeros > from getrusage(), that's what BSD::Resource will return from > getrusage. > > (There is one known problem in Mac OS X in this area, but it's for > setrlimit() and max number of > processes.) > > FWIW, I'm not seeing the problem in my Mac OS X, and yours is the > only one, I'm afraid. Maybe > there's something different about the system you are running it in, > or the user you are running > it as, or something like that? >

Odd. I'm now using a completely different box, and I still get the same result: boas% perl -MBSD::Resource -le 'print for (BSD::Resource::getrusage()) [2,3]' 0 0 Is there some simple test I can do to see what the system actually returns? Versions: Mac OS X 10.4.10 boas% pv BSD::Resource 1.28 boas% perl -v This is perl, v5.8.8 built for darwin-2level Thanks, David

Sun Oct 07 08:14:35 2007 jhi [...] iki.fi - Correspondence added

Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Sun, 07 Oct 2007 08:14:08 -0400
To:	bug-BSD-Resource [...] rt.cpan.org
From:	Jarkko Hietaniemi <jhi [...] iki.fi>

Show quoted text

> 0 > > Is there some simple test I can do to see what the system actually > returns? >

#include <sys/resource.h> #include <stdio.h> #include <errno.h> #include <string.h> int main() { struct rusage ru; int got = getrusage(RUSAGE_SELF, &ru); int err = errno; printf("getrusage = %d\n", got); printf("errno = %d (%s)\n", err, strerror(err)); #define PRINTF_RU(t) printf("%-14s = %9ld\n", "ru.ru_"#t, ru.ru_##t) PRINTF_RU(maxrss); PRINTF_RU(ixrss); return 0; } But let's take a step back: what is it that you *expect* to be returned? When I said my system works I meant: "it returns zeros, too". In many systems the getrusage() is not that well maintained, that is, the various kernel statistics are *not* returned by getrusage(). (And most often this means the fields are and stay zero.) Show quoted text

> Versions: > > Mac OS X 10.4.10 > > boas% pv BSD::Resource > 1.28 > boas% perl -v > > This is perl, v5.8.8 built for darwin-2level > > Thanks, > > David > >

Sun Oct 07 20:17:57 2007 david [...] kineticode.com - Correspondence added

CC:	Dave Rolsky <autarch [...] urth.org>
Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Sun, 7 Oct 2007 17:17:09 -0700
To:	bug-BSD-Resource [...] rt.cpan.org
From:	"David E. Wheeler" <david [...] kineticode.com>

On Oct 7, 2007, at 05:14, jhi@iki.fi via RT wrote: Show quoted text

> <URL: http://rt.cpan.org/Ticket/Display.html?id=20394 >

Show quoted text

>> Is there some simple test I can do to see what the system actually >> returns? >>

> > #include <sys/resource.h> > #include <stdio.h> > #include <errno.h> > #include <string.h> > int main() { > struct rusage ru; > int got = getrusage(RUSAGE_SELF, &ru); > int err = errno; > printf("getrusage = %d\n", got); > printf("errno = %d (%s)\n", err, strerror(err)); > #define PRINTF_RU(t) printf("%-14s = %9ld\n", "ru.ru_"#t, ru.ru_##t) > PRINTF_RU(maxrss); > PRINTF_RU(ixrss); > return 0; > } > > But let's take a step back: what is it that you *expect* to be > returned? > When I said my system works I meant: "it returns zeros, too". In many > systems the getrusage() is not that well maintained, that is, the > various kernel statistics are *not* returned by getrusage(). (And > most > often this means the fields are and stay zero.)

I dunno. Dave? This was related to something with Apache::SizeLimit, and it was Rolsky who seemed to think that the 0s were a bug. Best, David

Sun Oct 07 21:07:20 2007 jhi [...] iki.fi - Correspondence added

Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Sun, 07 Oct 2007 21:06:46 -0400
To:	bug-BSD-Resource [...] rt.cpan.org
From:	Jarkko Hietaniemi <jhi [...] iki.fi>

Show quoted text

>> returns?

>>>

>> #include <sys/resource.h> >> #include <stdio.h> >> #include <errno.h> >> #include <string.h> >> int main() { >> struct rusage ru; >> int got = getrusage(RUSAGE_SELF, &ru); >> int err = errno; >> printf("getrusage = %d\n", got); >> printf("errno = %d (%s)\n", err, strerror(err)); >> #define PRINTF_RU(t) printf("%-14s = %9ld\n", "ru.ru_"#t, ru.ru_##t) >> PRINTF_RU(maxrss); >> PRINTF_RU(ixrss); >> return 0; >> } >> >> But let's take a step back: what is it that you *expect* to be >> returned? >> When I said my system works I meant: "it returns zeros, too". In many >> systems the getrusage() is not that well maintained, that is, the >> various kernel statistics are *not* returned by getrusage(). (And >> most >> often this means the fields are and stay zero.)

> > I dunno. Dave? This was related to something with Apache::SizeLimit, > and it was Rolsky who seemed to think that the 0s were a bug.

Tried the above in four different Unixes, only one of them reported non-zero (for maxrss), ixrss stayed zero in all of them. Show quoted text

> Best, > > David > > >

Sun Oct 07 21:52:38 2007 autarch [...] urth.org - Correspondence added

CC:	bug-BSD-Resource [...] rt.cpan.org
Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Sun, 7 Oct 2007 20:52:10 -0500 (CDT)
To:	"David E. Wheeler" <david [...] kineticode.com>
From:	Dave Rolsky <autarch [...] urth.org>

On Sun, 7 Oct 2007, David E. Wheeler wrote: Show quoted text

>> But let's take a step back: what is it that you *expect* to be returned? >> When I said my system works I meant: "it returns zeros, too". In many >> systems the getrusage() is not that well maintained, that is, the >> various kernel statistics are *not* returned by getrusage(). (And most >> often this means the fields are and stay zero.)

> > I dunno. Dave? This was related to something with Apache::SizeLimit, and it > was Rolsky who seemed to think that the 0s were a bug.

Well, Apache::SizeLimit uses BSD::Resource on systems where that's the right thing to use. If BSD::Resource can't return any useful values about memory usage, I'll leave it to the BSD::Resource maintainer to declare it not a bug ;) That may just mean that the Apache::SizeLimit tests need adjusting, but ultimately it means Apache::SizeLimit probably won't work on the given platform(s). -dave

Wed Oct 10 14:53:29 2007 david [...] kineticode.com - Correspondence added

Subject:	Re: [rt.cpan.org #20394] Wrong Values Returned on Mac OS X?
Date:	Wed, 10 Oct 2007 11:49:32 -0700
To:	bug-BSD-Resource [...] rt.cpan.org
From:	"David E. Wheeler" <david [...] kineticode.com>

On Oct 7, 2007, at 18:07, jhi@iki.fi via RT wrote: Show quoted text

>> I dunno. Dave? This was related to something with Apache::SizeLimit, >> and it was Rolsky who seemed to think that the 0s were a bug.

> > Tried the above in four different Unixes, only one of them reported > non-zero (for maxrss), ixrss stayed zero in all of them.

Heya Jarkko, id you hear back from Dave directly? Thanks, David

Sat Nov 29 13:49:04 2008 JHI [...] cpan.org - Status changed from 'open' to 'stalled'

Thu Sep 09 18:32:04 2010 dwheeler [...] cpan.org - Correspondence added

On Wed Oct 10 14:53:29 2007, david@kineticode.com wrote: Show quoted text

> On Oct 7, 2007, at 18:07, jhi@iki.fi via RT wrote: >

> >> I dunno. Dave? This was related to something with Apache::SizeLimit, > >> and it was Rolsky who seemed to think that the 0s were a bug.

> > > > Tried the above in four different Unixes, only one of them reported > > non-zero (for maxrss), ixrss stayed zero in all of them.

> > Heya Jarkko, id you hear back from Dave directly?

FWIW, I just ran perl -MBSD::Resource -le 'print for (BSD::Resource::getrusage ())[2,3]' 3039232 0 So does that mean that the issue has been fixed somewhere? Perhaps on a newer version of OS X (I'm running 10.6.4 now)? David

Thu Sep 09 18:32:05 2010 The RT System itself - Status changed from 'stalled' to 'open'

Sat Jun 22 18:41:56 2013 JHI [...] cpan.org - Correspondence added

I think this has been irrevocably fixed in OS X.

Sat Jun 22 18:41:57 2013 JHI [...] cpan.org - Status changed from 'open' to 'resolved'