Research Assistant University of Kansas glibc problem

[penumarthy sreenivas]

Hi!
   I am Research Assistant at University of Kansas. I
developed a user space thread library for LINUX based
systems and used makecontext, swapcontext functions
for creating and switching contexts. I have a small
doubt w.r.t signal handling. I observed the
implementation of swapcontext (assembly file
glibc-2.2.4 version: sysdeps/unix/sysv/linux/i386/
swapcontext.S. I  developed code on LINUX for pentium
intel processor, so this is the file I looked at. I
observed that we save the current signal mask and
install new one somewhere in this assembly code (by
making sigprocmask system call). My question is this
might not give us completely correct behavior when
handling signals for threads. For instance if we
delivered some signals from tty interface and current
thread blocked them. So if the scheduler is called and
the new thread has not blocked them, then it should be
delivered once the new thread starts running. But in
current case the signal is delivered on old threads
stack as new threads mask is applied before it starts
running. So I wanted to know if it is correct to say
that this is undesirable behavior wrt swapcontext
usage for threads.
Thanks a lot,
Regards,
Sunil.


=====
Who you are speaks so loudly I can't hear what you're saying.
To be yourself in a world that is constantly trying to make you something else 
is the greatest accomplishment.
The glory of friendship is not the outstretched hand, nor the kindly smile nor 
the joy of companionship; it is the spiritual inspiration that comes to one 
when he discovers that someone else believes in him and is willing to trust him.
A friend is a person with whom I may be sincere. Before him, I may think aloud
          Ralph Waldo Emerson
Arise,Awake, Stop not till goal is reached.
          Swami Vivekananda

__________________________________________________
Do you Yahoo!?
Yahoo! Tax Center - File online, calculators, forms, and more
http://tax.yahoo.com