Martin Husemann a écrit :
On Mon, Jan 18, 2010 at 03:39:50PM +0100, Martin Husemann wrote:Heh, ok. So this behaviour is explained by libpthread optimizing for the single CPU case - but we still need to find out if the test case is buggy or not.Actually in the "already used" case I was thinking of the error would be EINVAL instead of ENOMEM. Could you try to run the test case in /usr/tests/kernel/t_lwpctl please? (just run the program manually, it only outputs very few lines)
I'm building new userland. Do you want output with UP or MP kernel ? JKB