On Thu, 10 May 2007 15:00:05 -0700
Jeremy Fitzhardinge [off-list ref] wrote:

Herbert Xu wrote:

quoted

[NET] link_watch: Move link watch list into net_device

These days the link watch mechanism is an integral part of the
network subsystem as it manages the carrier status.  So it now
makes sense to allocate some memory for it in net_device rather
than allocating it on demand.

I think there's a problem with one of these two patches.  I've been
noticing that one of my events/X threads has been going into a spin for
about 5 mins after boot.  I added some debugging to
kernel/workqueue.c:run_workqueue, since its that loop which seems to be
spinning due to list corruption.

When I look to see if that loop has iterated for more than 100 times in
one go (which seems unlikely),  I get this:

BUG: cpu 3, count=101 list screwup on c04babe4, func c03217e8
func=linkwatch_event+0x0/0x2a
 [<c0109173>] show_trace_log_lvl+0x1a/0x30
 [<c0109c7f>] show_trace+0x12/0x14
 [<c0109d0c>] dump_stack+0x16/0x18
 [<c0137c25>] run_workqueue+0x97/0x18c
 [<c01386a4>] worker_thread+0xe5/0xf5
 [<c013afe9>] kthread+0x3b/0x62
 [<c0108d47>] kernel_thread_helper+0x7/0x10
 =======================


I wonder if the problem is that the linkwatch_work is being rescheduled
when its already been scheduled, or something like that?

Five minutes after boot is when jiffies wraps.  Are you sure it's
a list-screwup rather than a jiffy-wrap screwup?