Jason R Thorpe <thorpej@wasabisystems.com> writes:

> I would like to propose the following names:
> 
> 	__sigtramp		legacy signal trampoline like we
> 				currently set up in the kernel on
> 				the top of the user stack
> 
> 	__sigtramp_siginfo	trampoline used for SA_SIGINFO signals
> 
> When we support RT signals, we can have __rtsigtramp versions of these
> names.

RT signals aren't different from ordinary SA_SIGINFO signals once you
get to the point of the delivery trampoline. The differences are in
how the application requests them, that the kernel queues them, and
that there are some ordering constraints on delivery. I don't think
additional entry points will be necessary.

> The other thing on my list: libsyscall has to go.  Some other imposition
> mechanism for cancellation points needs to be found.  However, this only
> blocks the libpthread bits, not the kernel bits.

I'd like to make it go, but a big concern I have is that this not
add overhead to non-threaded programs. Mechanisms like creating a
dispatch table which the syscall wrappers consult at runtime add quite
a bit of overhead (function setup, PIC setup, and dispatch through a
pointer) even when threads aren't in use.

        - Nathan