status of via velocity in 2.6.9

status of via velocity in 2.6.9

[Johan]

How 'working' are the via velocity drivers in 2.6.9?

For better or worse, the cheapest gigabit card I could find has the 
vt6122 chip, which seems to want the velocity drivers. (*)

Unfortunately, while they (the driver and card, that is) seem at first 
to work fine, auto negotiating a gigabit connection with my hub, the 
network stops working after 5 ish minutes (could be function of bytes 
tx'ed as well, I guess). restarting the network (appart from a kernel 
upgrade, the box is redhat fc2) fixes the problem... for another 5 minutes.

Is this known behavior?

Thanks

Johan

(*) The card's box advertizes linux compatibility with RH 7.3 (2.4.18-3 
or later), which makes me wonder whether another driver may work better. 
  2.4.18-3 would seem to predate the via-velocity driver.

Re: status of via velocity in 2.6.9

[Francois Romieu]

Johan <johan@ccs.neu.edu> :
> How 'working' are the via velocity drivers in 2.6.9 ?

Apart the fact that the driver does not play nice with the in-kernel
vlan API, I'd say that this is the first PR since the last fixes.

[network stops from time to time]
> Is this known behavior ?

Afaik, not on 2.6.9.

Can you describe the bug and the traffic ?
Any NETDEV watchdog messages or such ?

--
Ueimor

Re: status of via velocity in 2.6.9

[Johan]

1. thanks for the quick reply.

2. thanks for cc:ing me directly; I forgot to ask that when I posted

3. I would love to provide any debugging output, but don't know how. 
Please be more specific.  A pointer to the proper FAQ will suffice.

There is is no discernable error message: at least not to console or to 
/var/log/messages.  It just stops sending or receiving.

Re: status of via velocity in 2.6.9

[Francois Romieu]

Johan <johan@ccs.neu.edu> :
[...]
> 3. I would love to provide any debugging output, but don't know how. 
> Please be more specific.  A pointer to the proper FAQ will suffice.
> 
> There is is no discernable error message: at least not to console or to 
> /var/log/messages.  It just stops sending or receiving.

See REPORTING-BUGS for some hints. ethtool output before/after problem can
be added. It surprises me that the TX watchdog does not trigger if the device
stops working. Anything like a cron which could take place during the
5 minutes ? ifconfig and /proc/interrupts does not evolve any more once the
traffic is stopped ?

Cc: on netdev@oss.sgi.com is welcome.

--
Ueimor

Re: status of via velocity in 2.6.9

[Lenar Lõhmus]

Hi,

At least Abit AV8 onboard velocity doesn't work with 2.6.9. With earlier 
kernels,
the machine just locked up after ifconfig up. With 2.6.9, it doesn't 
lock up, but it doesn't
work either (data seems to go to black hole or sth). But there seem to 
be some success reports
too with this kernel.

Lenar

Johan wrote:

> How 'working' are the via velocity drivers in 2.6.9?
>
> For better or worse, the cheapest gigabit card I could find has the 
> vt6122 chip, which seems to want the velocity drivers. (*)
>
> Unfortunately, while they (the driver and card, that is) seem at first 
> to work fine, auto negotiating a gigabit connection with my hub, the 
> network stops working after 5 ish minutes (could be function of bytes 
> tx'ed as well, I guess). restarting the network (appart from a kernel 
> upgrade, the box is redhat fc2) fixes the problem... for another 5 
> minutes.
>
> Is this known behavior?
>
> Thanks
>
> Johan
>
> (*) The card's box advertizes linux compatibility with RH 7.3 
> (2.4.18-3 or later), which makes me wonder whether another driver may 
> work better.  2.4.18-3 would seem to predate the via-velocity driver.
>
>
>
> -
> To unsubscribe from this list: send the line "unsubscribe 
> linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

Re: status of via velocity in 2.6.9

[Francois Romieu]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset=unknown-8bit, Size: 609 bytes --]

Lenar Lõhmus <lenar@vision.ee> :
[...]
> the machine just locked up after ifconfig up. With 2.6.9, it doesn't 
> lock up, but it doesn't work either (data seems to go to black hole
> or sth). But there seem to be some success reports too with this kernel.

Can you check if the computer hosts the latest bios from its vendor and
if booting with "acpi=off" makes a difference ?

The content of /proc/interrupts after a known number of TX packets could
give some hint (use ping or such and correlate ifconfig output with
/proc/interrupts).

Please direct/Cc: further messages to netdev@oss.sgi.com.

--
Ueimor

Re: status of via velocity in 2.6.9

[Alan Cox]

On Maw, 2004-12-07 at 00:07, Johan wrote:
> How 'working' are the via velocity drivers in 2.6.9?

They should be pretty solid. The original VIA code worked pretty well,
the kernel merge had a few glitches but in 2.6.9* it seems rock solid
with the bits Francois applied.

> Unfortunately, while they (the driver and card, that is) seem at first 
> to work fine, auto negotiating a gigabit connection with my hub, the 
> network stops working after 5 ish minutes (could be function of bytes 
> tx'ed as well, I guess). restarting the network (appart from a kernel 
> upgrade, the box is redhat fc2) fixes the problem... for another 5 minutes.
> 
> Is this known behavior?

I've seen this with an Edimax el cheapo gigabit switch, but it was also
doing it with some other cards so I assumed it was random cheap junk.
Since I swapped that for a slightly better one it's behaved reliably.

> (*) The card's box advertizes linux compatibility with RH 7.3 (2.4.18-3 
> or later), which makes me wonder whether another driver may work better. 
>   2.4.18-3 would seem to predate the via-velocity driver.

There is a vendor provided driver for older systems (and in fact VIA
networking wrote and contributed the code that was cleaned up to go into
the kernel too)

Alan