Re: overlaping printk

[Tigran Aivazian]

It happens to me often (on SMP) too. Originally I thought that printk is
broken but having looked at the code I see that it has sufficient locking
in place to prevent this from happening (on printk side). However, the
fact that it only happens on a serial console (especially at low baud
rates like 9600)  points in the direction of the serial driver.

Kind regards
Tigran

On Wed, 19 May 2004, Ricky Beam wrote:

> Can anyone explain why the kernel does this on a serial console:
> 
> C<PU0 >C2P:U M 3a:ch Mianech iCnhee cCk heExckc epEtxcieopnt: i0o0n0: 0000000000
> 0000000000000040
> 00C4P
> U 2C:P UE 3IP:: E cI0P:1 0c10fa108 1fEFa8L AEGSF:L AG00S0:0 00200460
> 0246    e
> ax:     e 0ax00: 0000000000 0eb00x:  ebfx7f: a6f070f0a4 0ec00x:  e0c2x:a7 a02bea
> 7e aebedxe : efdx7f: a6f700f0a4
> 000     e
> si      e: sif7: faf76f0a0040 e0d0i e:d ci0: 10c01f107e1f e7be p:eb 0p:00 000000
> 00000 e00sp e:s fp:7 faf77ffab45
> fb4*
> ** **Ba*n kBa 0nk:  00:00 0000000000000000000000000000[00[0000000000000000000000
> 0000000]00] a at t 00000000000000000000000000000000
> 
> ***** * BaBankn k1 1: :0 000000000000000000000000000000g0eneral protection fault
> : 0000 [#1]
> ....
> 
> That's two MCE's being printed at the exact same time. (CPU2 & 3 are a single
> P4 Xeon.)  It makes it real damn hard to debug when the errors are printed
> like that.
> 
> The GPF was printed correctly:
> CPU:    3
> EIP:    0060:[<c010c0a7>]    Not tainted
> EFLAGS: 00010282   (2.6.6-SMP-rc3+BK@xxxxxx)
> EIP is at intel_machine_check+0x12b/0x364
> eax: 0000001f   ebx: 00000004   ecx: 00000407   edx: 00006f52
> esi: 00000407   edi: 00000000   ebp: 00000000   esp: f7fa5f04
> ds: 007b   es: 007b   ss: 0068
> Process swapper (pid: 0, threadinfo=f7fa4000 task=f7fa85c0)
> Stack: c0304002 00000001 00000000 00000000 f7fa5fb4 0000000f c0120adc 00000003
>        00000001 00000000 00000004 00000001 00000000 f7fa4000 02a7abee f7fa4000
>        f7fa4000 c0101f7e 00000000 f7fa5fb4 00000246 c0101fa8 00000046 c03cf108
> Call Trace:
>  [<c0120adc>] run_timer_softirq+0x110/0x180
>  [<c0101f7e>] default_idle+0x0/0x2d
>  [<c0101fa8>] default_idle+0x2a/0x2d
>  [<c010bf7c>] intel_machine_check+0x0/0x364
>  [<c01049b5>] error_code+0x2d/0x38
>  [<c0101f7e>] default_idle+0x0/0x2d
>  [<c0101fa8>] default_idle+0x2a/0x2d
>  [<c010201c>] cpu_idle+0x37/0x40
>  [<c0119558>] printk+0x172/0x1a8
>  [<c03b2981>] print_cpu_info+0x86/0xd2
> 
> Code: 0f 32 81 c3 02 04 00 00 89 44 24 08 89 54 24 04 c7 04 24 f7
> 
> I wouldn't trust it as it was on the MCE'ing processor.
> 
> Note: The answer to "which kernel" is ALL of them. (even the bastard stepchild
> redhat 2.4 kernel will do it.)
> 
> It goes on to have a fit unblanking a never blanked VT.  Stupid kernel!
> 
> --Ricky
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/