[LAU] RE: 2.6.21-rt7 Oopses

[David Baron]

>Also does it. Did not see any on bootup (rt4 had loads) but I have them on 
>logchecks. Stuff like:
>May 28 17:12:54 d_baron kernel:  [<c01faab9>] 
debug_smp_processor_id+0xa9/0xb0
>May 28 17:12:54 d_baron kernel: .. [<c01faa5f>] .... 
> ******

>May 28 17:12:54 d_baron kernel: BUG: using smp_processor_id() in preemptible 
>[00000000] code: IRQ-9/1114
>May 28 17:12:54 d_baron kernel: caller is icmp_push_reply+0x25/0x190
>May 28 17:12:54 d_baron kernel:  [<c02e5105>] icmp_push_reply+0x25/0x190
>May 28 17:12:54 d_baron kernel:  [<c02e5b36>] icmp_send+0x2c6/0x3f0
>May 28 17:12:54 d_baron kernel:  [<e1392755>] _nv005692rm+0x75/0xa0 [nvidia]
>May 28 17:12:54 d_baron kernel:  [<c0142b02>] add_preempt_count+0x12/0xe0
>May 28 17:12:54 d_baron kernel:  [<e130ff82>] _nv004775rm+0x1e/0x48 [nvidia]
>May 28 17:12:54 d_baron kernel:  [<c030b0a4>] rt_read_unlock+0x24/0x50
>May 28 17:12:54 d_baron kernel:  [<e0b0d75b>] ipt_do_table+0x21b/0x350 
>[ip_tables]
>May 28 17:12:54 d_baron kernel:  [<e130478c>] _nv007481rm+0x38/0x60 [nvidia]
>THese are all over the place. They apparently all touch on smp. Maybe this 
>kernel should be build without SMP (multiprocessor) support (Debian kernels 
>have this by default now!).

And obviously involves the nvidia (their older 9631 propietary) driver. What 
ip_tables is doing here? Maybe the IRQ of the agp and the network card are 
being shared?



-rt1 did not have any of these problems.

dmesg has all of them (without the times) if anyone working on this is 
interested.

BTW, kernel seems to be working OK otherwize.