linuxthreads tst-cancel4.c et al.

[Jakub Jelinek]

On Mon, Apr 11, 2005 at 01:42:30PM -0700, Roland McGrath wrote:
> I haven't read over those tests.  Is the need just for some syscall that
> will block?  Or is it testing that write returns a proper short write count?
> If just the former, then there are easy ways to block.  e.g., a pipe read
> with noone writing.

That part of the test is testing write syscall cancellation, so we need
a write syscall to whatever file descriptor that will block.
A pipe read with noone writing doesn't help us here, that's read syscall
blocking and that is tested in another part of the test.

	Jakub