Reply-To: akpm@linux-foundation.org

(switched to email.  Please respond via emailed reply-to-all, not via the
bugzilla web interface).

On Sat, 22 Mar 2008 06:53:40 -0700 (PDT) bugme-daemon@bugzilla.kernel.org wrote:

> http://bugzilla.kernel.org/show_bug.cgi?id=10309
> 
>            Summary: BUG: soft lockup - CPU#0 stuck
>            Product: Platform Specific/Hardware
>            Version: 2.5
>      KernelVersion: 2.6.24.3
>           Platform: All
>         OS/Version: Linux
>               Tree: Mainline
>             Status: NEW
>           Severity: normal
>           Priority: P1
>          Component: SPARC64
>         AssignedTo: platform_sparc64@kernel-bugs.osdl.org
>         ReportedBy: arnova@eld.physics.leidenuniv.nl
> 
> 
> Earliest failing kernel version: 2.6.24.3
> Distribution: Debian 4.0/Etch Sparc64
> Hardware Environment: Sparc64
> Problem Description:
> During boot of my Sparc64 with a vanille 2.6.24.3 kernel the kernel gets
> stuck.
> After while it continues allowing normal use of the machine. The message
> shown
> in the kernel logs looks like this:
> 
> BUG: soft lockup - CPU#0 stuck for 11s! [ifconfig:2408]
> TSTATE: 0000004480009603 TPC: 0000000010013390 TNPC: 0000000010013394 Y:
> 00000000    Not tainted
> TPC: <gem_interrupt+0x14/0xec [sungem]>
> g0: 0000000000009000 g1: 0800000000000001 g2: 0000000000000100 g3:
> 0000000000000400
> g4: fffff8003e61c060 g5: 0000000000000020 g6: fffff8003e7e8000 g7:
> 0000000000000000
> o0: 0000000000000001 o1: fffff8003d090670 o2: 0000000000000001 o3:
> 000001fe0000f078
> o4: 7fffffffffffffff o5: 0000000080000000 sp: fffff8003e7ea341 ret_pc:
> 00000000100133bc
> RPC: <gem_interrupt+0x40/0xec [sungem]>
> l0: fffff8003d090670 l1: 0000000000821400 l2: fffff8003e7eaca0 l3:
> 0000000000000400
> l4: 0000000000000000 l5: 0000000000000005 l6: 0000000000000000 l7:
> 0000000000000008
> i0: 0000000000000009 i1: fffff8003d090620 i2: 000000001c2245fa i3:
> 000000000000000c
> i4: 7fffffffffffffff i5: 0000000000000000 i6: fffff8003e7ea401 i7:
> 000000000047e974
> I7: <handle_IRQ_event+0x34/0x74>
> 
> It appartently has to do with the network (ifconfig) although I'm not 100%. I
> tried changing several settings in the kernel (to many to write down here)
> but
> all give the same result.
> 
> Steps to reproduce:
> - Build a 2.6.24.3 for Sparc64
> - Boot a SUN Sparc64 machine with this kernel
> - System gets stuck during boot but after a while it continues its boot.
> 

I expect it would be useful if you tell us the latest version of the kernel
on which this didn't happen.  ie: what kernel version were you running before
you "up"graded to 2.6.24?

Thanks.

Sorry, I forgot to mention that. The last kernel that didn't have this problem was Debian's 2.6.18-6-sparc64 stock kernel.

Just build and tried a 2.6.19.7 vanilla kernel, and this kernel also does NOT suffer from this issue. I will now try a 2.6.23.17 vanilla kernel, and see what this does....

I can confirm that this issue also doesn't exist in 2.6.23.17. Just tested with a vanilla kernel on my test system and the problem does NOT occur. The obvious conclusion is that this problem got introduced in 2.6.24 (post 2.6.23)...

This issue still exists on 2.6.26.2. 

So it is in 2.6.26.7.

And in 2.6.28-rc7

The bad commit seems to be
commit bea3348eef27e6044b6161fd04c3152215f96411
Author: Stephen Hemminger <shemminger@linux-foundation.org>
Date:   Wed Oct 3 16:41:36 2007 -0700

    [NET]: Make NAPI polling independent of struct net_device objects.

Will try to debug further

It's still present in 2.6.28-rc9 and I was not able to debug more than the commit above - git did not want to revert it :(

Still present in 2.6.29-rc3 (or more precisely: linux-2.6.git at eda58a85ec3fc05855a26654d97a2b53f0e715b9).

Fine, it's fixed now.

Commit: 71822faa3bc0af5dbf5e333a2d085f1ed7cd809f
sungem: Soft lockup in sungem on Netra AC200 when switching interface up