On Tue, Mar 25, 2008 at 08:39:01AM +0000, Bastian Blank wrote:
> On Tue, Mar 25, 2008 at 09:26:51AM +0100, Pierre Habouzit wrote:
> > Problem is that memcpy/memmove/memset probably generate rep stos; in
> > the end, I believe memset/memcpy/memmove to be async signal safe, and
> > those are inlined fully in many cases.
>
> Please show the sections of POSIX docu or the referenced documentation
> which lists them as safe.
I'm not saying _POSIX_ I'm saying in the glibc, because it's used in
places like sigemptyset/sigfillset like aurélien told you.
And fwiw, unless I'm mistaken, code like the following generates
memsets and string ops in the compiler:
some_type foo = { 0 };
And I see absolutely no valid reason to invalidate such code in signal
handlers. I don't say standards, I say common sense. I'm sure you'll
find many applications in Debian doing that. Though probably the types
are small enough to be in the cases where the memset are compiled as
unrolled loops of 0 assignments.
--
·O· Pierre Habouzit
··O madcoder@debian.org
OOO http://www.madism.org
Attachment:
pgpo4IBL9Z_ya.pgp
Description: PGP signature