Re: [BUG]linux-2.4.24 with k8 numa support panic when init scsi

[Andi Kleen]

lepton <lepton@xxxxxxxxxxxxxxxxxxxxxx> writes:

> I have do some test on weekend. The result is:
>
> 1. Compiling kernel with gcc 3.2:
>    2.4.20 2.4.21: can boot, can mount reiserfs filesystem
>    2.4.22: can boot, can mount reiserfs filesystem, but will panic
>    when reboot. It is perhaps because of "reboot=triple" ? 

Possible.

>    2.4.23: panic when init scsi, like before.

It could be a problem with the SCSI driver. I don't have any other reports
of this.

>    2.4.24: can boot, can mount reiserfs filesystem, but will panic when
>    reboot.

> 2. Compiling kernel with gcc 3.3
>    2.4.20: can not compile.... 
>    2.4.21: can boot, can mount reiserfs filesystem
>    2.4.22: can boot, can mount reiserfs filesystem, but will panic when
>    reboot.
>    2.4.23: panic when init scsi, like before
>    2.4.24: panic when init scsi, like before
>
> 3. when panic, reboot=bios or reboot=triple both can not work.

Of course they don't help for SCSI panics, just possible for reboot
problems.

>    2.4.24 changes a little from 2.4.23, so it is strange system will
>    panic in 2.4.23 and don't panic in 2.4.24 when using gcc 3.2
>    Perhaps there is some random error?

Yes, it looks very strange. I don't have a good explanation. Especially
since the behaviour changes with the compiler. It's probably some 
random memory corruption somewhere.

I will look into the reboot crashes with non standard options. 

Regarding the original reboot issues it seems to be related to C stepping 
CPUs (we never had any with B stepping).

Anyways, the reboot problem is probably not fatal for you so I guess
you can just use 2.4.24 compiled with gcc 3.2 for now.

-Andi

-
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/