> I think the problem here is that there aren't two free signals available on
> each platform. You would have to use a 64 bit integer so you can have a bit
> for each signal. This may not be too hard to add, but probably has a
> performance impact on all the 32 bit linuxes. But then again, maybe its not
> much.
Hmm, I didn't think of that. But if there really is no another way to
implement this, then this must be done sooner or later. Right?
I don't think that adding another integer would introduce a major
performance hit, but on the other hand, I don't know what I'm talking
about :).
Also I've heard that some level of posix introduces custom signals, how
are they going to be handled? I don't know anything more about this
matter..
> There may also be binary compatibility problems to solve too.
I guess that for ia32 platform there is no other way to 'blend' the code
than to create a new system call, which accepts a 64-bit bitmask. And
change libc too. Or maybe glibc2 already uses 64-bit masks, I don't know.
_____________________________________________________________________
/ __// /__ ____ __ http://www.inside.org/~flux/\ \
/ /_ / // // /\ \/ / \ /
/_/ /_/ \___/ /_/\_\@inside.org \/