kdb is behaving odd of late while switching
cpus. Don't know when it started happening,
but I'm seeing this as of today. I have a completely
uptodate tree ...
On a running kernel, I entered kdb with a <cntrl>-a,
then trying to switch cpus hangs the system:
----------
Entering kdb (0xc02d8000) on processor 0 due to Keyboard Entry
[0]kdb> bt
EBP EIP Function(args)
0xc010730d default_idle+0x2d
kernel .text 0xc0100000 0xc01072e0 0xc0107314
0xc0107352 cpu_idle+0x3e
kernel .text 0xc0100000 0xc0107314 0xc0107368
[0]kdb> cpu 1
Debugger re-entered, allowing event to proceed
wait_on_irq, CPU 1:
irq: 1 [ 1 0 ]
bh: 1 [ 0 1 ]
Stack dumps:
CPU 0: <unknown>
CPU 1:c117bebc c0212413 00000001 00000000 00000020 c010c46d c0212428 c11b1380
00000003 00000001 c035b320 c018bdf4 00000004 c035f090 c117a000 c035f580
c01257c8 00000000 00000020 c117a000 c035f580 c030b4d0 00000020 c0123a3e
Call Trace: [<c0212413>] [<c010c46d>] [<c0212428>] [<c018bdf4>] [<c01257c8>]
[<c0123a3e>]
[<c0123945>]
[<c012376a>] [<c010ca11>] [<c01072e0>] [<c01072e0>] [<c010a5a0>]
[<c01072e0>]
[<c01072e0>] [<c0100018>]
[<c010730d>] [<c0107352>] [<c010a5a0>]
-----------------
The wait_on_irq symptom seems to repeat a few times
after long pauses, but I have seen the system hang
witouth any message at all.
Obviously, this is a big impediment to development work ...
--
--------------------------------------------------------------------------
Rajagopal Ananthanarayanan ("ananth")
Member Technical Staff, SGI.
--------------------------------------------------------------------------
|