This is a Sun T1000 (6 cores / 24 threads) running a vanilla 2.6.18
kernel. Everthing seems to be working OK, but this message appeared in
the kernel log:
BUG: soft lockup detected on CPU#3!
Call Trace:
[000000000043143c] smp_percpu_timer_interrupt+0xd4/0x144
[00000000004109d4] tl0_irq14+0x1c/0x20
[00000000005262e0] p1275_cmd+0x34c/0x354
[0000000000525990] prom_putchar+0x2c/0x34
[0000000000526320] prom_write+0x38/0x4c
[0000000000526378] prom_printf+0x44/0x4c
[0000000000540e50] promcon_putcs+0x2fc/0x304
[0000000000581198] do_con_write+0x1c34/0x1c6c
[0000000000581220] con_write+0x18/0x2c
[0000000000573198] write_chan+0x2e4/0x36c
[000000000056ff08] tty_write+0x154/0x1e0
[00000000004921e8] vfs_write+0x78/0x124
[00000000004926d8] sys_write+0x34/0x60
[0000000000406c14] linux_sparc_syscall+0x3c/0x44
[0000000000101d74] 0x101d7c
Hope thats meaningful to someone :)
Anything I can do, let me know
Andrew Walrond
BTW Googling for "BUG: soft lockup detected on CPU" returns *lots* of
hits, but no resolutions that I could see...
> [0000000000525990] prom_putchar+0x2c/0x34
I wonder; could this be a timeout during boot because the prom console
is hardware limited to 9600baud and its buffer is full ??
Andrew
From: [email protected]
Date: Mon, 16 Oct 2006 14:11:27 +0000
> This is a Sun T1000 (6 cores / 24 threads) running a vanilla 2.6.18
> kernel. Everthing seems to be working OK, but this message appeared in
> the kernel log:
Turn of CONFIG_PROM_CONSOLE.
From: [email protected]
Date: Mon, 16 Oct 2006 16:41:24 +0000
>
> > [0000000000525990] prom_putchar+0x2c/0x34
>
> I wonder; could this be a timeout during boot because the prom console
> is hardware limited to 9600baud and its buffer is full ??
PROM console is just slower than anything else for whatever
reason.
Even though PROM console drives the same output, using the
native CONFIG_SERIAL_SUNHV is much faster and does not generate
the timeouts.
That's why I told the original poster to simply disable
CONFIG_PROM_CONSOLE, it should never be used.
>> > [0000000000525990] prom_putchar+0x2c/0x34
>>
>> I wonder; could this be a timeout during boot because the prom console
>> is hardware limited to 9600baud and its buffer is full ??
>
>PROM console is just slower than anything else for whatever
>reason.
>
>Even though PROM console drives the same output, using the
>native CONFIG_SERIAL_SUNHV is much faster and does not generate
>the timeouts.
>
>That's why I told the original poster to simply disable
>CONFIG_PROM_CONSOLE, it should never be used.
I have not seen this soft lockup so far, though I run a 2.6.16, most
likely using CONFIG_PROM_CONSOLE (redirected to ttya by prom) because
the machine is not a SUN4V (which SUNHV seems to be for).
-`J'
--
From: Jan Engelhardt <[email protected]>
Date: Tue, 17 Oct 2006 00:02:53 +0200 (MEST)
> I have not seen this soft lockup so far, though I run a 2.6.16, most
> likely using CONFIG_PROM_CONSOLE (redirected to ttya by prom) because
> the machine is not a SUN4V (which SUNHV seems to be for).
You could be using one of the other serial drivers.
Check the boot messages and your kernel config.
>> I have not seen this soft lockup so far, though I run a 2.6.16, most
>> likely using CONFIG_PROM_CONSOLE (redirected to ttya by prom) because
>> the machine is not a SUN4V (which SUNHV seems to be for).
>
>You could be using one of the other serial drivers.
>Check the boot messages and your kernel config.
Thanks for the hint. I am still a bit puzzled why there are so many
serial ports detected even though there is only ttya and ttyb on the
back:
Console: switching to mono PROM 80x24
su0 at 0x000001fff13062f8 (irq = 5,7ea) is a 16550A
su1 at 0x000001fff13083f8 (irq = 9,7e9) is a 16550A
ttyS0 at MMIO 0x1fff1400000 (irq = 7901920) is a SAB82532 V3.2
Console: ttyS0 (SAB82532)
ttyS1 at MMIO 0x1fff1400040 (irq = 7901920) is a SAB82532 V3.2
ttyS2 at MMIO 0x1fff1200000 (irq = 7901664) is a SAB82532 V3.2
ttyS3 at MMIO 0x1fff1200040 (irq = 7901664) is a SAB82532 V3.2
One of them may actually be the mouse though currently no [SUN4]
keyboard (and hence no mouse) is attached.
-`J'
--
From: Jan Engelhardt <[email protected]>
Date: Tue, 17 Oct 2006 00:56:26 +0200 (MEST)
>
> >> I have not seen this soft lockup so far, though I run a 2.6.16, most
> >> likely using CONFIG_PROM_CONSOLE (redirected to ttya by prom) because
> >> the machine is not a SUN4V (which SUNHV seems to be for).
> >
> >You could be using one of the other serial drivers.
> >Check the boot messages and your kernel config.
>
> Thanks for the hint. I am still a bit puzzled why there are so many
> serial ports detected even though there is only ttya and ttyb on the
> back:
What kind of system is this?
The two SU serial ports are usually for keyboard and mouse.
If you have 4 SAB ports, I'm guessing this is an Ultra250.
The 3rd and 4th SAB port are usually used for RSC on the
Ultra250 machines, the 1st and 2nd for normal ttya and ttyb
serial.
>What kind of system is this?
Yes it's Ultra/E250.
>The two SU serial ports are usually for keyboard and mouse.
>
>If you have 4 SAB ports, I'm guessing this is an Ultra250.
>The 3rd and 4th SAB port are usually used for RSC on the
>Ultra250 machines, the 1st and 2nd for normal ttya and ttyb
>serial.
ttya... backside. ttyb... backside. RSC... backside in a PCI slot, that
makes three. Where is the 4th?
-`J'
--
From: Jan Engelhardt <[email protected]>
Date: Tue, 17 Oct 2006 18:53:48 +0200 (MEST)
> ttya... backside. ttyb... backside. RSC... backside in a PCI slot, that
> makes three. Where is the 4th?
Each SAB chip provides two ports at a time, therefore I guess the 4th
is wired down to NULL but it is there.