Re: dlopen() and libpthread

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Mon, Oct 22, 2012 at 09:58:50AM +0000, Emmanuel Dreyfus wrote:
> On Mon, Oct 22, 2012 at 08:40:27AM +0000, Emmanuel Dreyfus wrote:
> > I also bet that Linux behavior must allow dlopen() of libpthread followed
> > by a thread startup.
> 
> I confirm that: I crafted a test program that dlopen a DSO which calls
> pthread_create, it works very well on Linux.
> 
> We cannot blame OpenSC and PCSC-lite for assuming this is possible. No
> standard backs this limitation of our pthread implementation.

It is not a limitation of only our pthread implementation. In fact, most
systems do have this limitation if they care about the performance of
non-threaded applications. The only way to magically fix this is by
making the stubs in libc more expensive, which I object for the sake of
badly written applications. The only reason I can find in PCSC-lite for
using threads is to do background interactions with the card. That
somehow just begs for using a daemon to interact with the hardware...

Joerg