hi

In function sigaddset (sigset_t *set,int sogno) signal set has a type sigset_t, but macro sigmask() returns signal set as an int number.

I assume sigset_t is just an integer?But still,why are there two different type names used for signal set?

Couldn't sigset_t on some system also be a struct ?

thank you

...sigmask obsolete by sigprocmask...

check : http://www.die.net/doc/linux/man/man2/sigmask.2.html

hi

I need to know sigmask() cos some older apps still use it and cos I can retrieve an int mask value for individual signal.

sigprocmask() uses sigset_t type for signal mask and if you look at my previous question you will see that I have no idea what kind of type sigset_t is ,nor do I know how to extract mask value for individual signal and see what it is

thank you

you can check sigset_t
at : http://fxr.watson.org/fxr/source/sys/_sigset.h#L57

and you can control the individual values with sigismember()

..take your time and read all the man pages concerning the posix signal lib.

sigset_t is an unsigned long...
atleast that is how sigset_t is typedef'd as in sys/signal.h

MACRO doesnot return anything as it is not a function but plain code replacement and it uses sigset_t (check out the sigmask macro definition in signal.h) ...
so there is no discrepancy here ...

this is so because there are 32 signals and hence each bit representing one signal ..(on a 64 bit architecture also ... i am sure some provisions are made to keep it working seamlessly)

on a 32 bit system (except windows) int is same size as long...(4 bytes)

hope it helped

This is not really rellated to the this subject,but I noticed that many of types defined in header files have their name prefixed with '_'. But man pages always ommit '_' from their name.

For example,in signal.h there is type definition for _signalhandler_t,but in "man signal" you will notice name signalhandler_t with '_' ommited.

What is going on?

thank you for helping me and for the links