[eepro100] eth[01]: card reports no resources

Christopher A. Stevens stevensca@navair.navy.mil
Mon Apr 22 17:28:01 2002


On Mon, 22 Apr 2002, Donald Becker wrote:

> As the regular list readers know, I only support unmodified drivers.
> But this is likely caused by the system configuration, and will be the
> same with a driver variants.

Oops, sorry, my bad.  Didn't realize it was a variant.

> 
> > > Has the kernel source been modified, or the VM parameters changed?
> >
> > I don't think the virtual memory size had changed, since only the
> > O.S. was upgraded and the original disk partitions are the same.  As for
> > any other VM parameters or kernel modifications, I will have to forward
> > this to RAIDZONE technical support and ask.
> 
> They should have tuned the VM parameters for the application.
> Mention "VM bdflush tuning"
> 
> The key parameters are controlled at /proc/sys/vm/bdflush
> 
> 

My first email to RAIDZONE tech support a few months ago produced no results, but they must know more about this now.  They did not respond with the kernel parameters, but told me to change the eepro100 driver module to a new one based on drivers from Intel.  So I'll try that and see if it fixes my problem. 

However, in case you were still interested in debug information, I provided a dump of the /proc/sys/vm/ directory.  I will have both versions of the kernal modules on the system if you decide to follow this up further, and have any fixes to try.

./pagebuf/max_dio_pages
256
./pagebuf/flush_age
15000
./pagebuf/flush_int
1000
./max-readahead
31
./min-readahead
3
./page-cluster
3
./pagetable_cache
25	50
./kswapd
512	32	8
./overcommit_memory
0
./bdflush
10	0	0	0	100	3000	60	0	0

Thanks,
Chris Stevens