perl 5.16 tests...is shm working right?

perl 5.16 tests...is shm working right?

[Isaac Dunham]

[-- Attachment #1: Type: text/plain, Size: 575 bytes --]

hello,
I built perl 5.16.0 (defaults to ansi c89, but needs
SIG_BLOCK...used -D_GNU_SOURCE -std=gnu99, but _POSIX_C_SOURCE +
-std=c99 is probably enough)
Still using gcc-3.4 (with -Os)

10 tests failed, including both shm tests. 
I had at least 500 MB of RAM completely unused, so I
wouldn't expect any issues there.
So I'm wondering if the shm support is fully working, or if that's just
perl. (the log from ./perl harness is attached as a bzip2, if anyone
wants a look--it's nearly 200k uncompressed)

On the up side, almost all (99.54%) tests passed.

Thanks,
Isaac Dunham

[-- Attachment #2: perlt.bz2 --]
[-- Type: application/x-bzip, Size: 18634 bytes --]

Re: perl 5.16 tests...is shm working right?

[Rich Felker]

On Sun, Jun 17, 2012 at 07:28:54AM -0700, Isaac Dunham wrote:
> hello,
> I built perl 5.16.0 (defaults to ansi c89, but needs
> SIG_BLOCK...used -D_GNU_SOURCE -std=gnu99, but _POSIX_C_SOURCE +
> -std=c99 is probably enough)
> Still using gcc-3.4 (with -Os)
> 
> 10 tests failed, including both shm tests. 
> I had at least 500 MB of RAM completely unused, so I
> wouldn't expect any issues there.
> So I'm wondering if the shm support is fully working, or if that's just

For the record, this is legacy sysv shm, not modern POSIX shm. Are you
on a 32- or 64-bit machine? I'm guessing some of the structures and
padding might be messed up on 64-bit, although I thought we looked
into and fixed that a while back.

Rich

Re: perl 5.16 tests...is shm working right?

[Isaac Dunham]

On Sun, 17 Jun 2012 10:44:51 -0400
Rich Felker <dalias@aerifal.cx> wrote:

> On Sun, Jun 17, 2012 at 07:28:54AM -0700, Isaac Dunham wrote:
> > hello,
> > I built perl 5.16.0 (defaults to ansi c89, but needs
> > SIG_BLOCK...used -D_GNU_SOURCE -std=gnu99, but _POSIX_C_SOURCE +
> > -std=c99 is probably enough)
> > Still using gcc-3.4 (with -Os)
> > 
> > 10 tests failed, including both shm tests. 
> > So I'm wondering if the shm support is fully working, or if that's
> > just
> 
> For the record, this is legacy sysv shm, not modern POSIX shm. Are you
> on a 32- or 64-bit machine? I'm guessing some of the structures and
> padding might be messed up on 64-bit, although I thought we looked
> into and fixed that a while back.
Atom/32-bit x86.
I have only one computer out of 3 that supports 64-bit and it has 3 GB
RAM (until recently, 1 GB), so I haven't bothered trying 64-bit (partly
so I can share kernels and binaries, partly thanks to pointer bloat,
partly because I use dosemu with vm86 on them, and partly just
inertia).

Isaac Dunham

Re: perl 5.16 tests...is shm working right?

[Rich Felker]

On Sun, Jun 17, 2012 at 09:00:27AM -0700, Isaac Dunham wrote:
> On Sun, 17 Jun 2012 10:44:51 -0400
> Rich Felker <dalias@aerifal.cx> wrote:
> 
> > On Sun, Jun 17, 2012 at 07:28:54AM -0700, Isaac Dunham wrote:
> > > hello,
> > > I built perl 5.16.0 (defaults to ansi c89, but needs
> > > SIG_BLOCK...used -D_GNU_SOURCE -std=gnu99, but _POSIX_C_SOURCE +
> > > -std=c99 is probably enough)
> > > Still using gcc-3.4 (with -Os)
> > > 
> > > 10 tests failed, including both shm tests. 
> > > So I'm wondering if the shm support is fully working, or if that's
> > > just
> > 
> > For the record, this is legacy sysv shm, not modern POSIX shm. Are you
> > on a 32- or 64-bit machine? I'm guessing some of the structures and
> > padding might be messed up on 64-bit, although I thought we looked
> > into and fixed that a while back.
> Atom/32-bit x86.
> I have only one computer out of 3 that supports 64-bit and it has 3 GB
> RAM (until recently, 1 GB), so I haven't bothered trying 64-bit (partly
> so I can share kernels and binaries, partly thanks to pointer bloat,
> partly because I use dosemu with vm86 on them, and partly just
> inertia).

OK. Have you tried strace on the tests?

Rich