HEADS UP: call for nve(4) users to test a patch

[alan bryan]

Well, I tried it and it's still broken.  I patched and
re-compiled last night (FreeBSD 6 beta 1).  The only
change is that it didn't go to 64 before dying, it now
died at 63 (whatever these numbers stand for).

For example from my demsg:
nve0: device timeout (62)
nve0: link state changed to DOWN
nve0: link state changed to UP
nve0: device timeout (63)
nve0: link state changed to DOWN
nve0: link state changed to UP

> ping 10.0.0.1
PING 10.0.0.1 (10.0.0.1): 56 data bytes
ping: sendto: No buffer space available
ping: sendto: No buffer space available
(10.0.0.1 is the gateway)

It lasted about 12 hours while slowly counting up to
63 before dying.

nForce 4 from the onboard nve in the Shuttle SN25P
small form factor PC.

If you need anything else just let me know.

Thanks for trying!

--Alan




--- Maxime Henrion <mux at FreeBSD.org> wrote:

> 	Hi,
> 
> 
> 
> If any of you are using an nve(4) card and are
> experiencing "device
> timeout" errors, I'd like you to test a very simple
> patch.  One person
> already reported success with it, but I'd like to
> see more reports
> before committing and hopefully MFC'ing it in time
> for 6.0-RELEASE.
> 
> This patch just reduces the size of the TX ring by
> one.  Many NIC chips
> in existence today have such bugs and require
> similar fixes, so I'm not
> really surprised.  It also seems Linux's forcedeth
> driver does such a
> thing, but it's hard to tell because it uses an
> entirely different API
> than us.
> 
> Thanks,
> Maxime
> > ? nve.patch
> Index: if_nvereg.h
>
===================================================================
> RCS file: /home/ncvs/src/sys/dev/nve/if_nvereg.h,v
> retrieving revision 1.3
> diff -u -r1.3 if_nvereg.h
> --- if_nvereg.h	10 Jun 2005 16:49:12 -0000	1.3
> +++ if_nvereg.h	12 Sep 2005 17:21:21 -0000
> @@ -49,9 +49,9 @@
>  
>  #define	NV_RID		0x10
>  
> -#define	TX_RING_SIZE	64
> +#define	TX_RING_SIZE	63
>  #define	RX_RING_SIZE	64
> -#define	NV_MAX_FRAGS	63
> +#define	NV_MAX_FRAGS	62
>  
>  #define	FCS_LEN 4
>  
> > _______________________________________________
> freebsd-current at freebsd.org mailing list
>
http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to
"freebsd-current-unsubscribe at freebsd.org"



		
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com