Hi, I´ve the same situtaion.
> I am having problems with large transfers. Small stuff,
> interactive, web, etc works fine, but large transfers halt after 200K
> - 6MB. I have tried scp and ftp transfers from the 2.2.11 to a
> 2.1.115 and a 2.0.36 box, started from either end comes to a halt,
> there is still network activity, and the 2.2.11 box can still talk to
> anyone but that transfer is dead, no more data moves ever.
Here, NFS stops after a while with 0.89H 75% of my cards are affected,
like yours.
> At first I though it was an issue with 2.2.11 and v91 (I have
> to run 91, the new 21143 cards I have require it), but after screwing
> around for hours with it I dug up an older 21140 card that runs fine
> with 89H, it moved more data (6MB compared to < 1MB), but still came
> to a halt.
Maybe the de4x5 driver is also an option.
> I don't know anything else to try. I do with that 91 would
> get put into the current kernel though better than 75% of the cards
> around here will not function with 89H. A dump of traffic between the
> 2.2.11/89H host (borg) and the 2.1.115 host (news) during one of the
> halts is below.
As said, a test with 0.89H gave same results here: On 7 from 9
Systems, a network install of a default Linux system was not possible.
NFS halted here. 0.90 worked well on all these systems.
In case you don´t have 0.90, it is in 2.0.37 and I´should have diff for
2.2.11 at ftp://ftp.suse.com/pub/people/bk/kernel/for-2.2.11/tulip-0.90.diff
within 24 hours.
Also, there is a 0.91g tulip.c, which is quite good hidden in the second
part of a link on the tulip development page.
http://cesdis.gsfc.nasa.gov/linux/drivers/test/tulip.c
There are also two secret 0.90 versions:
http://cesdis1.gsfc.nasa.gov/linux/drivers/test/tulip.c-90q
http://cesdis1.gsfc.nasa.gov/linux/drivers/test/tulip.c-90z
Please tell me your results.
- Bernd
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/