[PATCH/RFC future] nptl: reinstate SIGCANCEL/SIGSETXID safeguards

Rich Felker dalias at aerifal.cx
Tue May 10 21:38:16 UTC 2011


On Tue, May 10, 2011 at 11:06:40PM +0200, Peter Mazinger wrote:
> > 2) How much of the observed glibc behavior should we try to emulate in
> > uClibc?
> 
> I am for only one implementation (that should go only into libc),
> else you have unpredictable/different results depending on which
> linking order you choose libc/libpthread

I agree, and after further consideration, sigfillset, sigaddset, and
sigaction/signal (for the sig arg, not the sa_mask) are the right
places in which to place the restrictions. Anything else (like
checking and patching up at each sigprocmask call, for instance, or
fixing the sa_mask at sigaction time) results in additional bloat and
runtime cost with little/no benefit.

Rich


More information about the uClibc mailing list