On Thu, Dec 02, 2004 at 05:09:24PM +0100, Manuel Bouyer wrote:
> Hi,
> I have a U5 with a current kernel from today, hung is what I think is the
> sleep forever bug. Stack trace from ddb:
> db> tr
> sab_intr(1d71c80, 0, e0017ed0, 5, 1392060, c559bb71) at netbsd:sab_intr+0xcc
> ?(0, 0, 140ccb8, 0, ffffffff, ffffffff) at 0x1008fc4
> cc_microtime(5d60cc0, 0, 140ccb8, 0, 0, 41af347c) at netbsd:cc_microtime+0x8c
> mi_switch(5d60cc0, 0, 0, 0, 0, 0) at netbsd:mi_switch+0x150
> ltsleep(612a820, 128, 0, 0, 0, 187db44) at netbsd:ltsleep+0x3b4
> sigsuspend1(5547450, 14f0400, ffffffff, 6158000, 0, 0) at netbsd:sigsuspend1+0xe
> c
> sys___sigsuspend14(5d60cc0, 615bdd8, 615bdd0, 6, 615bf10, 200) at netbsd:sys___s
> igsuspend14+0x28
> syscall(615bed0, 126, 0, 6, 4023ecb4, 615bdd8) at netbsd:syscall+0xe0
> ?(ffffccb8, 70000, ffffccb8, 5, ffffcc34, c559bb71) at 0x1008cb8
> 
> I can keep the box in this state until tomorow. Let me know if there is
> something else you want to know from ddb.

Well, no. I managed to panic it from ddb :(

I set a breakpoint on cc_microtime(), thinking that one possible cause
could be that it's looping here at splclock. It wasn't looping in there,
but was called several time, before I got an alignement fault in ddb.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--