Re: A new pthread patch

[hjl@lucon.org (H.J. Lu)]

> 
> hjl@lucon.org (H.J. Lu) writes:
> 
> > How can you guarantee those real-time signals are not used for
> > other purposes? We have __libc_allocate_rtsig () in <signal.h>.
> > I can image someone wants to call it to get a free signal.
> 
> First, this function is not meant to be called by user code.  The
> comments clearly state this.  And: I've removed the allocation
> function now from the installed header.

That is exactly where the problem is. gdb HAS to call
__libc_allocate_rtsig () to debug threads in glibc 2.1.

> 
> Second, even if the user would allocate signals using this function
> nothing bad would happen.  The whole purpose of this function is to
> prevent this.

Have you looked at how gdb 4.17.0.10 works with glibc 2.1? If not,
please do take a look at linuxthreads_new_objfile () in lnx-thread.c
before you may any more comments on gdb and thread.


> 
> > unless you can
> > provide me a patch for gdb which is as reliable as gdb 4.17.0.11.
> 
> You have not even described the problem.  How shall I know what you
> want to fix.
> 

gdb 4.17.0.10 HAS to call __libc_allocate_rtsig () to debug threads in
glibc 2.1. If you can remove __libc_allocate_rtsig from gdb, I will be
happy.

H.J.