Re: RTL-Poll Error and <sleep> for ever

[Dirk von Suchodoletz <dirk@goe.net>]

On Sun, 16 Jan 2000, Marty Connor wrote:

> On 1/16/2000 1:59 PM dirk@goe.net Dirk von Suchodoletz wrote:
> >The machine got it dhcp answer (IP, network etc.) and then tried to get
> >its boot image but without success (other machines got it from the same
> >server so tftpd should be OK). After a reboot all was fine ...
> 
> Could you try loading from a floppy to see if this has anything to do 
> with the code being in BIOS flash.

The last things I tried (tcpdump, new code (4.4.1) with -DCONGESTED was
from floppy I made for this purpurse...
 
> Can you also try setting the RAM speed slower (70ns?) in your BIOS 
> config?  I am wondering if the BUS speed is too fast for these PCI cards, 
> or if this is some PCI 2.x compatibility bug.  See also if there is a 
> setting for BUS wait states or something that will slow things down a bit 
> for the card.  What speed (CPU/BUS) are these machines, btw?  The driver 
> code does have some timing related considerations.

I ran it on various machines: AMD K6 -233 with 66Mhz FSB (3,5*) or on
better 233er with 2,5*100Mhz FSB. I set the options to slowest as
possible, tried it on Celeron machines with 50Mhz FSB ... I would say it
has nothing to do with it, because the cards are running fine on
overclocked celerons with 83Mhz FSB, that means PCI at 41,5Mhz.

> Did DEBUG-RX yield anything more specific on the nature of the 
> Poll-Error?  I note in the Changelog for rtl8139.c there is mention of 
> receive code bugs being fixed. I can't reproduce the problem, so it may 
> be the card, the server, the BIOS, the driver, the LAN, or something we 
> haven't thought of.  We might have to insert a "sleep(2);" call after the 
> error message so it doesn't scroll by too fast.

I will try this option next run. I have forgotten it :-(

> The fact that a reboot seems to fix things suggests that the card might 
> not be fully initialized before the receive is attempted or something 
> timing related. The main obvious difference I see right now is that 
> you're running code from BIOS Flash and I'm doing ROMs and floppies.  I 
> should get one of these flashable motherboards so I can test this sort of 
> thing.

I wouldn't say it has anything to do with the location of code. I tried it
with the code on EPROM, with the code on floppy and of course in the BIOS
flash rom, but I get the exactly same behavior :-((

These are not special motherboards. You could try it with every new board,
but dont take AMI-Bios... Every newer board I got, I had no problems with
to add the code to the main BIOS. It's getting much easier because the
size of flash is increasing, so I dont have to remove (i.e. cpucode or ncr
scsi) anything from it.

Thank you!

Ciao,
	Dirk


===========================================================================
This Mail was sent to netboot mailing list by:
Dirk von Suchodoletz <dirk@goe.net>
To get help about this list, send a mail with 'help' as the only string in
it's body to majordomo@baghira.han.de. If you have problems with this list,
send a mail to netboot-owner@baghira.han.de.