This is definite 2.6.17 material...
As of rc3-mm1, inet_init, which schedules, is called before the UML timer_init,
which sets up the timer. The result is the interval timers being manipulated
before the appropriate signal handlers are established, causing unhandled
timers.
This is fixed by making timer_init be called earlier.
Signed-off-by: Jeff Dike <[email protected]>
Index: linux-2.6.17-mm/arch/um/kernel/time_kern.c
===================================================================
--- linux-2.6.17-mm.orig/arch/um/kernel/time_kern.c 2006-03-23 16:40:20.000000000 -0500
+++ linux-2.6.17-mm/arch/um/kernel/time_kern.c 2006-05-01 17:34:41.000000000 -0400
@@ -209,4 +209,4 @@ int __init timer_init(void)
return(0);
}
-__initcall(timer_init);
+arch_initcall(timer_init);
Jeff Dike <[email protected]> wrote:
>
> This is definite 2.6.17 material...
>
> As of rc3-mm1, inet_init, which schedules, is called before the UML timer_init,
> which sets up the timer. The result is the interval timers being manipulated
> before the appropriate signal handlers are established, causing unhandled
> timers.
>
Which means nobody's tested uml against the last couple of -mm's. Bad.
On Mon, May 01, 2006 at 02:58:37PM -0700, Andrew Morton wrote:
> Which means nobody's tested uml against the last couple of -mm's. Bad.
Yeah.
I didn't check rc2-mm1 because it came out after rc3, and I missed
rc1-mm3. But rc1-mm2 was OK.
Jeff