[Click] driver crash

[Eddie Kohler]

I bet this is the problem Qinghua Ye recently reported.  When he sends the patch 
to the list, please apply it & see if that helps!

Eddie


rchertov at purdue.edu wrote:
> Quoting Eddie Kohler <kohler at cs.ucla.edu>:
> 
> 
>>Roman,
>>
>>A ksymoops would be extremely helpful!
> 
> 
> 
> I noticed that sometimes there is no crash but nothing is received on the device
> until the machine is restarted.  
> 
> ksymoops output
> 
> Unable to handle kernel NULL pointer dereference at virtual address 00000080
> f8829281
> *pde = 00000000
> Oops: 0000
> CPU:    0
> EIP:    0010:[<f8829281>]    Not tainted
> Using defaults from ksymoops -t elf32-i386 -a i386
> EFLAGS: 00010246
> eax: 00000002   ebx: 00000047   ecx: 00000000   edx: 00000040
> esi: f708d470   edi: 00000000   ebp: f6869e10   esp: f6869dd8
> ds: 0018   es: 0018   ss: 0018
> Process kclick (pid: 975, stackpage=f6869000)
> Stack: c0244a14 00035846 00000282 00000000 c030d21c 00000001 00000040 00000000
>        f88ff58c f7731800 f7731aa8 00000000 f77319a0 00000040 f6869e40 f8828daf
>        f77319a0 f6869e30 00000040 00000287 00001b30 00000000 00000001 f77318c4
> Call Trace:    [<c0244a14>] [<f8828daf>] [<c024a066>] [<c0125d19>] [<c010af19>]
>   [<c010da78>] [<f882a3db>] [<f89b45c5>] [<f89b1d89>] [<f89dd862>] [<f89760e0>]
>   [<f89d2670>] [<c010752e>] [<f89d25ec>]
> Code: 8b 8f 80 00 00 00 83 ea 04 85 c9 8b 47 10 0f 85 82 02 00 00
>  
> 
>>>EIP; f8829281 <[e1000]e1000_clean_rx_irq+9d/340>   <=====
> 
> Trace; c0244a14 <alloc_skb+c4/1d0>
> Trace; f8828daf <[e1000]e1000_clean+43/e4>
> Trace; c024a066 <net_rx_action+a6/160>
> Trace; c0125d19 <do_softirq+d9/e0>
> Trace; c010af19 <do_IRQ+f9/120>
> Trace; c010da78 <call_do_IRQ+5/d>
> Trace; f882a3db <[e1000]e1000_rx_poll+4b/2ec>
> Trace; f89b45c5 <END_OF_CODE+60aae/????>
> Trace; f89b1d89 <END_OF_CODE+5e272/????>
> Trace; f89dd862 <END_OF_CODE+89d4b/????>
> Trace; f89760e0 <END_OF_CODE+225c9/????>
> Trace; f89d2670 <END_OF_CODE+7eb59/????>
> Trace; c010752e <arch_kernel_thread+2e/40>
> Trace; f89d25ec <END_OF_CODE+7ead5/????>
> Code;  f8829281 <[e1000]e1000_clean_rx_irq+9d/340>
> 00000000 <_EIP>:
> Code;  f8829281 <[e1000]e1000_clean_rx_irq+9d/340>   <=====
>    0:   8b 8f 80 00 00 00         mov    0x80(%edi),%ecx   <=====
> Code;  f8829287 <[e1000]e1000_clean_rx_irq+a3/340>
>    6:   83 ea 04                  sub    $0x4,%edx
> Code;  f882928a <[e1000]e1000_clean_rx_irq+a6/340>
>    9:   85 c9                     test   %ecx,%ecx
> Code;  f882928c <[e1000]e1000_clean_rx_irq+a8/340>
>    b:   8b 47 10                  mov    0x10(%edi),%eax
> Code;  f882928f <[e1000]e1000_clean_rx_irq+ab/340>
>    e:   0f 85 82 02 00 00         jne    296 <_EIP+0x296>
>  
>  <0>Kernel panic: Aiee, killing interrupt handler!
>  
> 1 warning issued.  Results may not be reliable.
> 
> 
> Roman
> 
> 
>>
>>On Jul 6, 2005, at 5:29 PM, rchertov at purdue.edu wrote:
>>
>>
>>>This is on the 2.4.26 SMP kernel using the latest CVS click SMP  
>>>build.  The
>>>driver is the e1000-5.x.  I usually get this after I send a high  
> 
> 
> 
>>
>>On Jul 6, 2005, at 5:29 PM, rchertov at purdue.edu wrote:
>>
>>
>>>This is on the 2.4.26 SMP kernel using the latest CVS click SMP  
>>>build.  The
>>>driver is the e1000-5.x.  I usually get this after I send a high  
>>>rate of UDP
>>>packets with a 10 byte payload from my own packet generator (user  
>>>land).  I am
>>>going to try to use the 1.4.3 click as it seemed to crash less  
>>>frequently.
>>>
>>>
>>>This is the script that I run on the node that crashes.  The  
>>>machine runs on 2
>>>Xenon 2.8 Ghz CPUs with hyperthreading (linux reports 4 cpus) and  
>>>the NICs are
>>>pci-66 Intel Pro 1000
>>>
>>>PollDevice(eth1)
>>>        -> Queue(200)
>>>        -> ToDevice(eth2);
>>>
>>>PollDevice(eth2)
>>>        -> Queue(200)
>>>        -> ToDevice(eth1);
>>>
>>>
>>>
>>>Unable to handle kernel NULL pointer dereference at virtual address  
>>>00000080
>>> printing eip:
>>>f8829281
>>>*pde = 00000000
>>>Oops: 0000
>>>CPU:    0
>>>EIP:    0010:[<f8829281>]    Not tainted
>>>EFLAGS: 00010246
>>>eax: 00000002   ebx: 00000047   ecx: 00000000   edx: 00000040
>>>esi: f708d470   edi: 00000000   ebp: f6869e10   esp: f6869dd8
>>>ds: 0018   es: 0018   ss: 0018
>>>Process kclick (pid: 975, stackpage=f6869000)
>>>Stack: c0244a14 00035846 00000282 00000000 c030d21c 00000001  
>>>00000040 00000000
>>>       f88ff58c f7731800 f7731aa8 00000000 f77319a0 00000040  
>>>f6869e40 f8828daf
>>>       f77319a0 f6869e30 00000040 00000287 00001b30 00000000  
>>>00000001 f77318c4
>>>Call Trace:    [<c0244a14>] [<f8828daf>] [<c024a066>] [<c0125d19>]  
>>>[<c010af19>]
>>>  [<c010da78>] [<f882a3db>] [<f89b45c5>] [<f89b1d89>] [<f89dd862>]  
>>>[<f89760e0>]
>>>  [<f89d2670>] [<c010752e>] [<f89d25ec>]
>>>
>>>Code: 8b 8f 80 00 00 00 83 ea 04 85 c9 8b 47 10 0f 85 82 02 00 00
>>> <0>Kernel panic: Aiee, killing interrupt handler!
>>>In interrupt handler - not syncing
>>>
>>>
>>>I don't suppose there is any quick magic fix laying about? :)
>>>
>>>Roman
>>>
>>>_______________________________________________
>>>click mailing list
>>>click at amsterdam.lcs.mit.edu
>>>https://amsterdam.lcs.mit.edu/mailman/listinfo/click
>>>
>>
>>
> 
> 
>