Re: RFC: 'more signals' patch, 2.1.33

Ulrich Drepper (drepper@myware.rz.uni-karlsruhe.de)
15 Apr 1997 22:49:43 +0200


Richard Henderson <richard@stommel.tamu.edu> writes:

> Symbol versioning will not help. As I indicated, the issue is that
> there is not room to store the entire state of the signal.
>
> Thus the application, using an increased sigset_t size, sets up an
> sa_mask with bit 1500 set. A library, having only room for 1024
> bits, is not able to save the entire set, and so only the low 1024
> are saved and restored.

But the application itself will (should) never call the kernel
directly. Everything goes to the libc. And it is easy to filter out
signal numbers which are higher as the ones which are used for
compiling the libc.

I don't think we should spend a thought on situation where someone
really uses the kernel directly and so could introduce situations
which cannot be controlled by the libc.

-- Uli
---------------. drepper@cygnus.com ,-. Rubensstrasse 5
Ulrich Drepper \ ,-------------------' \ 76149 Karlsruhe/Germany
Cygnus Solutions `--' drepper@gnu.ai.mit.edu `------------------------