Re: [PATCH] Fix CPU bitmask truncation

[William Lee Irwin III]

William Lee Irwin III <[e-mail]> writes:
|> ===== include/linux/init_task.h 1.19 vs edited =====
|> --- 1.19/include/linux/init_task.h	Sun Sep 29 07:02:55 2002
|> +++ edited/include/linux/init_task.h	Fri Dec 20 02:22:04 2002
|> @@ -63,7 +63,7 @@
|>  	.prio		= MAX_PRIO-20,					\
|>  	.static_prio	= MAX_PRIO-20,					\
|>  	.policy		= SCHED_NORMAL,					\
|> -	.cpus_allowed	= -1,						\
|> +	.cpus_allowed	= ~0UL,						\

On Fri, Dec 20, 2002 at 01:17:24PM +0100, Andreas Schwab wrote:
> This is useless.  Assigning -1 to any unsigned type is garanteed to give
> you all bits one, and with two's complement this also holds for any signed
> type.

Not so on all gcc versions. The rest of the world can figure out what
to do about the versions that do not.

Bill