* unkillable process in R state, 2.6.6
@ 2004-05-20 6:41 carbonated beverage
2004-05-20 12:36 ` Ben Collins
` (4 more replies)
0 siblings, 5 replies; 6+ messages in thread
From: carbonated beverage @ 2004-05-20 6:41 UTC (permalink / raw)
To: sparclinux
Noticed any site I go through my proxy chain was not reachable, and poked
about. squid sits in the 'R' state forever, and I can't shut it down with
kill -KILL | -STOP | -ILL | -FPU | -SEGV
The first site I go to through it works fine. I can browse pages off it
and follow links within that domain. When I go to another site through
squid, it hits the 'R' state and sits there forever. I can't even attach
with gdb/strace to find out what code it's looping on.
It's the binary out of debian/testing for sparc, using a bk pull of 2.6.6.
Anyone have ideas how I can go about debugging this to see what's going on?
It's an Ultra 5 (Ultra IIi/333MHz), gcc 3.3.3, binutils 2.14.90.0.7,
make 3.80, following possibly relevant flags set:
CONFIG_IOSCHED_AS=y
CONFIG_CC_OPTIMIZE_FOR_SIZE=y
CONFIG_PREEMPT=y
Previous kernel, 2.6.5 from a bk pull, worked fine.
-- DN
Daniel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: unkillable process in R state, 2.6.6
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
@ 2004-05-20 12:36 ` Ben Collins
2004-05-21 11:44 ` carbonated beverage
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Ben Collins @ 2004-05-20 12:36 UTC (permalink / raw)
To: sparclinux
> CONFIG_CC_OPTIMIZE_FOR_SIZE=y
> CONFIG_PREEMPT=y
Try getting rid of these two just to see what happens.
--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: unkillable process in R state, 2.6.6
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
2004-05-20 12:36 ` Ben Collins
@ 2004-05-21 11:44 ` carbonated beverage
2004-05-21 15:08 ` Ben Collins
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: carbonated beverage @ 2004-05-21 11:44 UTC (permalink / raw)
To: sparclinux
On Thu, May 20, 2004 at 08:36:26AM -0400, Ben Collins wrote:
> > CONFIG_CC_OPTIMIZE_FOR_SIZE=y
> > CONFIG_PREEMPT=y
>
> Try getting rid of these two just to see what happens.
Tried without, and still happens. I repulled the bk tree before doing
so about 22:30 PDT, and booted with a kernel built with the above two
turned off -- system froze.
Is there a way to enable a watchdog oopser? Or some other way of finding
out where the kernel's hanging?
-- DN
Daniel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: unkillable process in R state, 2.6.6
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
2004-05-20 12:36 ` Ben Collins
2004-05-21 11:44 ` carbonated beverage
@ 2004-05-21 15:08 ` Ben Collins
2004-05-23 22:15 ` carbonated beverage
2004-05-27 16:00 ` Meelis Roos
4 siblings, 0 replies; 6+ messages in thread
From: Ben Collins @ 2004-05-21 15:08 UTC (permalink / raw)
To: sparclinux
On Fri, May 21, 2004 at 04:44:44AM -0700, carbonated beverage wrote:
> On Thu, May 20, 2004 at 08:36:26AM -0400, Ben Collins wrote:
> > > CONFIG_CC_OPTIMIZE_FOR_SIZE=y
> > > CONFIG_PREEMPT=y
> >
> > Try getting rid of these two just to see what happens.
>
> Tried without, and still happens. I repulled the bk tree before doing
> so about 22:30 PDT, and booted with a kernel built with the above two
> turned off -- system froze.
>
> Is there a way to enable a watchdog oopser? Or some other way of finding
> out where the kernel's hanging?
Try enabling some debug options, like spinlock debug and such.
--
Debian - http://www.debian.org/
Linux 1394 - http://www.linux1394.org/
Subversion - http://subversion.tigris.org/
WatchGuard - http://www.watchguard.com/
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: unkillable process in R state, 2.6.6
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
` (2 preceding siblings ...)
2004-05-21 15:08 ` Ben Collins
@ 2004-05-23 22:15 ` carbonated beverage
2004-05-27 16:00 ` Meelis Roos
4 siblings, 0 replies; 6+ messages in thread
From: carbonated beverage @ 2004-05-23 22:15 UTC (permalink / raw)
To: sparclinux
On Fri, May 21, 2004 at 11:08:05AM -0400, Ben Collins wrote:
> Try enabling some debug options, like spinlock debug and such.
Enabled the following:
CONFIG_CC_OPTIMIZE_FOR_SIZE=y
CONFIG_PREEMPT=y
CONFIG_DEBUG_KERNEL=y
CONFIG_DEBUG_SLAB=y
CONFIG_MAGIC_SYSRQ=y
CONFIG_DEBUG_SPINLOCK=y
CONFIG_DEBUG_SPINLOCK_SLEEP=y
CONFIG_STACK_DEBUG=y
CONFIG_MCOUNT=y
CONFIG_FRAME_POINTER=y
And it doesn't boot.
After the Loading... at... message, it just sits there once the spinner goes
away. The disk is pretty loud when accessed, so I know it's not even hitting
the disk.
I can't even break out into the OpenPROM prompt.
Ideas?
-- DN
Daniel
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: unkillable process in R state, 2.6.6
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
` (3 preceding siblings ...)
2004-05-23 22:15 ` carbonated beverage
@ 2004-05-27 16:00 ` Meelis Roos
4 siblings, 0 replies; 6+ messages in thread
From: Meelis Roos @ 2004-05-27 16:00 UTC (permalink / raw)
To: sparclinux
>> Try enabling some debug options, like spinlock debug and such.
cb>
cb> Enabled the following:
cb>
cb> CONFIG_CC_OPTIMIZE_FOR_SIZE=y
cb> CONFIG_PREEMPT=y
cb> CONFIG_DEBUG_KERNEL=y
cb> CONFIG_DEBUG_SLAB=y
cb> CONFIG_MAGIC_SYSRQ=y
cb> CONFIG_DEBUG_SPINLOCK=y
cb> CONFIG_DEBUG_SPINLOCK_SLEEP=y
cb> CONFIG_STACK_DEBUG=y
cb> CONFIG_MCOUNT=y
cb> CONFIG_FRAME_POINTER=y
cb>
cb>
cb> And it doesn't boot.
Same here on an UE2. Turned on spinlock debugging, gained no info.
Turned on stack debug:
CONFIG_DEBUG_BUGVERBOSE=y
CONFIG_DEBUG_DCFLUSH=y
# CONFIG_DEBUG_INFO is not set
-# CONFIG_STACK_DEBUG is not set
+CONFIG_STACK_DEBUG=y
# CONFIG_DEBUG_BOOTMEM is not set
+CONFIG_MCOUNT=y
+CONFIG_FRAME_POINTER=y
This causes lots of warning about undefined module symbols _mcount but I
tried the kernel anyway. This kernel does not load at all - the last
thing I see is SILO telling that it's loading the kernel.
--
Meelis Roos
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2004-05-27 16:00 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-05-20 6:41 unkillable process in R state, 2.6.6 carbonated beverage
2004-05-20 12:36 ` Ben Collins
2004-05-21 11:44 ` carbonated beverage
2004-05-21 15:08 ` Ben Collins
2004-05-23 22:15 ` carbonated beverage
2004-05-27 16:00 ` Meelis Roos
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.