Re: Correction: 2.2.15-SMP, eepro100, elusive boot failure

From: Andrey Savochkin (saw@saw.sw.com.sg)
Date: Tue May 23 2000 - 02:41:04 EST


Hello,

On Tue, May 23, 2000 at 09:21:07AM +0200, Christoph Rohland wrote:
> kumon@flab.fujitsu.co.jp writes:
> > Yes, when it produced the message: "card reports no RX buffers", only
> > the interface couldn't be used. The other NIC could be used
> >
> > In the system, RH6.1 is installed. so, gcc is 2.91 as shown below.
>
> I have the same problems on 2.3.99-pre9-3 during stress test (Also RH

No, they look like completely different issues.
In your case it was a clear out-of-memory situation, the driver was unable to
allocate buffers to receive network packets and reported about it ("can't
fill rx buffer"). When the pool of the allocated buffers was exhausted, the
receiving operations were stopped and the card signalled it ("card reports no
resources"). There was no network-related problems here. The driver did its
best. It may only be a problem if the network module didn't recovered when
the system got enough memory back (i.e. when you stopped your stress test and
'free' showed a lot of free memory).

The original reports were about "card reports no resources" messages at the
boot time with a lot of free memory.

> 6.1 based, egcs-2.91.66 19990314/Linux (egcs-1.1.2 release). And sorry
> gcc 2.7.2.3 is way too old). It never happend til yesterday and I did
> the same tests for month now.

Well, gcc 2.7.2.3 is still the official compiler for bug reports.
However, it doesn't apply to your case because there is no bug here (at least
not in network module).

>
> Under heavy memory stress testing I got:
>
> May 22 15:33:12 localhost kernel: eth0: can't fill rx buffer (force 0)!
> May 22 15:33:16 localhost kernel: VM: killing process ipctst
> May 22 15:33:40 localhost last message repeated 2 times
> May 22 15:49:33 localhost last message repeated 4 times
> May 22 15:58:46 localhost kernel: VM: killing process ipctst
> May 22 16:43:18 localhost kernel: eth0: can't fill rx buffer (force 0)!
> May 22 16:43:21 localhost kernel: eth0: can't fill rx buffer (force 0)!
> May 22 16:43:24 localhost kernel: eth0: can't fill rx buffer (force 1)!
> May 22 16:43:45 localhost kernel: eth0: card reports no resources.
> May 22 16:44:10 localhost last message repeated 6 times
[snip]

Best regards
                                        Andrey V.
                                        Savochkin

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



This archive was generated by hypermail 2b29 : Tue May 23 2000 - 21:00:23 EST