Via-Rhine dirve in 2.4.21-pre7

Via-Rhine dirve in 2.4.21-pre7

[Mark Syms]

Roger,

Just tried the via-rhine driver in 2.4.21-pre7 and it appears to exhibit
the problem Erik Hensema reported with the VT6103 onboard network card
on my Chaintech 7VJL Athlon board whereby you get NETDEV Watchdog
transmit timeout errors unless you turn local io-apic off in which case
it all works fine.

Please cc me directly as I'm not subscribed to the mailing list.

Thank you,

	Mark Syms

Re: Via-Rhine dirve in 2.4.21-pre7

[Roger Luethi]

On Mon, 07 Apr 2003 10:10:37 +0100, Mark Syms wrote:
> Just tried the via-rhine driver in 2.4.21-pre7 and it appears to exhibit
> the problem Erik Hensema reported with the VT6103 onboard network card
> on my Chaintech 7VJL Athlon board whereby you get NETDEV Watchdog
> transmit timeout errors unless you turn local io-apic off in which case
> it all works fine.

How did previous versions fare?

Re: Via-Rhine dirve in 2.4.21-pre7

[Mark Syms]

[-- Attachment #1: Type: text/plain, Size: 1059 bytes --]

Roger,

Just saw your response on the mailing list archive. Sorry I haven't
responded earlier but as I said below I'm not subscribed to the list.

The reason I was trying 2.4.21-pre7 is that I have been unable to get
the network card working on any kernel version if local IO-APIC is
enabled and I was hoping that the changes you had submitted for the
21pre series might help but they did not.

If you want me to try anything let me know or send me a patch (directly
please).

Sorry not to be more help.

	Mark.

On Mon, 2003-04-07 at 10:10, Mark Syms wrote:
> Roger,
> 
> Just tried the via-rhine driver in 2.4.21-pre7 and it appears to exhibit
> the problem Erik Hensema reported with the VT6103 onboard network card
> on my Chaintech 7VJL Athlon board whereby you get NETDEV Watchdog
> transmit timeout errors unless you turn local io-apic off in which case
> it all works fine.
> 
> Please cc me directly as I'm not subscribed to the mailing list.
> 
> Thank you,
> 
> 	Mark Syms
-- 
Mark Syms <mark@marksyms.demon.co.uk>

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Via-Rhine dirve in 2.4.21-pre7

[Jeff Garzik]

On Thu, Apr 17, 2003 at 06:53:50PM +0100, Mark Syms wrote:
> The reason I was trying 2.4.21-pre7 is that I have been unable to get
> the network card working on any kernel version if local IO-APIC is
> enabled and I was hoping that the changes you had submitted for the
> 21pre series might help but they did not.

Until further notice, please do not attempt to use Via IOAPIC support.

This has nothing to do with via-rhine, and remains an open issue.

	Jeff

Re: Via-Rhine dirve in 2.4.21-pre7

[Roger Luethi]

On Thu, 17 Apr 2003 18:53:50 +0100, Mark Syms wrote:
> Just saw your response on the mailing list archive. Sorry I haven't
> responded earlier but as I said below I'm not subscribed to the list.

Your email bounced. I would have told you, but your email bounced :-P.
(list cced in case it happens again)

Roger

Re: Via-Rhine dirve in 2.4.21-pre7

[Marc Wilson]

On Thu, Apr 17, 2003 at 01:59:24PM -0400, Jeff Garzik wrote:
> Until further notice, please do not attempt to use Via IOAPIC support.
> This has nothing to do with via-rhine, and remains an open issue.

I thought I'd read everything about the Rhine and Via in my lurking on
lkml, but I must have missed something.  This box here is an Abit KD7-RAID
(Via KT400 + Rhine II) and I have IOAPIC enabled and am using Roger's latest
driver with 2.4.21-pre6 with great success.

rei $ uname -a
Linux rei 2.4.21-pre6-preempt #1 Mon Mar 31 17:05:40 PST 2003 i686 unknown unknown GNU/Linux

rei $ cat /var/log/dmesg | egrep -i "via|rhine"
PCI: Using IRQ router VIA [1106/3177] at 00:11.0
PCI: Via IRQ fixup for 00:10.0, from 10 to 5
PCI: Via IRQ fixup for 00:10.2, from 11 to 5
VP_IDE: VIA vt8235 (rev 00) IDE UDMA133 controller on pci00:11.1
agpgart: Detected Via Apollo Pro KT400 chipset
via-rhine.c:v1.10-LK1.1.17  March-1-2003  Written by Donald Becker
  http://www.scyld.com/network/via-rhine.html
eth0: VIA VT6102 Rhine-II at 0xd800, 00:50:8d:45:af:b7, IRQ 23.

rei $ grep APIC /usr/src/linux/.config
CONFIG_X86_GOOD_APIC=y
CONFIG_X86_UP_APIC=y
CONFIG_X86_UP_IOAPIC=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_X86_IO_APIC=y

What am I missing here?  It sounds like it's dangerous to use, but it
doesn't look like it's broken.

-- 
 Marc Wilson |     One big pile is better than two little piles.  --
 msw@cox.net |     Arlo Guthrie

Re: Via-Rhine dirve in 2.4.21-pre7

[Jeff Garzik]

Marc Wilson wrote:
> On Thu, Apr 17, 2003 at 01:59:24PM -0400, Jeff Garzik wrote:
> 
>>Until further notice, please do not attempt to use Via IOAPIC support.
>>This has nothing to do with via-rhine, and remains an open issue.
> 
> 
> I thought I'd read everything about the Rhine and Via in my lurking on
> lkml, but I must have missed something.  This box here is an Abit KD7-RAID
> (Via KT400 + Rhine II) and I have IOAPIC enabled and am using Roger's latest
> driver with 2.4.21-pre6 with great success.
[...]
> eth0: VIA VT6102 Rhine-II at 0xd800, 00:50:8d:45:af:b7, IRQ 23.
[...]
> What am I missing here?  It sounds like it's dangerous to use, but it
> doesn't look like it's broken.


If it works for you, great!

It doesn't work at all for some people, and it works until you hit high 
loads, for others.  Besides all the bug reports I've seen, Alan Cox has 
noted it's an open issue as well.  He may have more info on the APIC 
details.

Basically there are a lot of "I boot with 'noapic', and suddenly 
everything works" reports.

	Jeff