lkml.org 
[lkml]   [2013]   [Dec]   [2]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: NETDEV WATCHDOG: eth0 (e1000e): transmit queue 0 timed out
> Intel maintains newer drivers out-of-tree at
> http://sourceforge.net/projects/e1000/, and it's possible this is some
> bug that has already been fixed. The current version there looks like
> e1000e-2.5.4, released 2013-09-05.
>
> Possible similar report: http://sourceforge.net/p/e1000/bugs/367/ (no
> real data there).

I've looked through the existing bug reports and version changelogs
and didn't see anything that seemed very relevant.

I was able to debug the e1000e object file and get the specific code
that's bugging out after the interface is unexpectedly reset:
--------
(gdb) l *e1000_clean_rx_irq+0x101
0x19d81 is in e1000_clean_rx_irq
(drivers/net/ethernet/intel/e1000e/netdev.c:933).
928 rmb(); /* read descriptor and rx_buffer_info
after status DD */
929
930 skb = buffer_info->skb;
931 buffer_info->skb = NULL;
932
933 prefetch(skb->data - NET_IP_ALIGN);
934
935 i++;
936 if (i == rx_ring->count)
937 i = 0;
--------

The above code is from kernel version 3.9.2 and e1000e driver version
2.2.14-k. Should there be a check here to see if skb is NULL? I
checked the latest e1000e release (2.5.4) and there is no check there
either (near netdev.c:994).


\
 
 \ /
  Last update: 2013-12-03 00:21    [W:0.095 / U:0.048 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site