*** SPAM LEVEL 4.447 *** Re: Dynamic linker behaviour difference between Linux, Hurd and FreeBSD

[Simon Richter]

Hi,

On 03.05.2013 17:17, Carlos O'Donell wrote:

> Can you produce a small self-contained test case that shows the
> expected versus observed behaviour?

Sure, attached.

I'm not entirely confident about expected behaviour -- the behaviour
observed on Hurd and BSD can be correct, and Linux the outlier.

When libdl calls back into pthread_mutex_lock when looking up the
pthread_mutex_lock function, the abort() is triggered.

I'd be interested if there is a way around the problem, ideally more
portable than doing ELF symbol lookups on hands and knees, or whether it
is indeed wrong to use pthread_mutex_lock from libdl while the program
is still being initialized.

   Simon
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bad.bz2
Type: application/octet-stream
Size: 3448 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-help/attachments/20130503/a37f3a74/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: good.bz2
Type: application/octet-stream
Size: 3120 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-help/attachments/20130503/a37f3a74/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: preload.tar.gz
Type: application/gzip
Size: 586 bytes
Desc: not available
URL: <http://sourceware.org/pipermail/libc-help/attachments/20130503/a37f3a74/attachment.gz>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 380 bytes
Desc: OpenPGP digital signature
URL: <http://sourceware.org/pipermail/libc-help/attachments/20130503/a37f3a74/attachment.sig>