realtek 8139too driver problem with kernel 2.6.13.5

hi,

i notice a problem with the 8139too driver (version: 0.9.27) and kernel
2.6.13.5 on a DFI mainboard:

loading and using the driver usually works fine. but when i try to copy
a file > 1 MB from a NFS-mounted server the copy hangs and i have to
reboot the system !
after reboot i detect that only 708608 bytes of the requested file have
been copied.

when using kernel 2.4.22 with the contained 8139too driver on the very
same mainboard everything works fine !

does anyone know the reason for that ?
is there already a fix or work around for this problem ?
what should i do ?

thanks for help
peter

peter schb wrote:
> hi,
>
> i notice a problem with the 8139too driver (version: 0.9.27) and kernel
> 2.6.13.5 on a DFI mainboard:
>
> loading and using the driver usually works fine. but when i try to copy
> a file > 1 MB from a NFS-mounted server the copy hangs and i have to
> reboot the system !
> after reboot i detect that only 708608 bytes of the requested file have
> been copied.
>
> when using kernel 2.4.22 with the contained 8139too driver on the very
> same mainboard everything works fine !
>
> does anyone know the reason for that ?
> is there already a fix or work around for this problem ?
> what should i do ?

There have been a number of changes to 8139too.c over the last few
months. The most critical was to fix a deadlock, which sounds like the
problem you are having. To get this fix, you would need to download
v2.6.17.y of the kernel.

Larry

On 27 Jul 2006 07:14:26 -0700, peter schb <(E-Mail Removed)> wrote:
> hi,
>
> i notice a problem with the 8139too driver (version: 0.9.27) and kernel
> 2.6.13.5 on a DFI mainboard:
>
> loading and using the driver usually works fine. but when i try to copy
> a file > 1 MB from a NFS-mounted server the copy hangs and i have to
> reboot the system !
> after reboot i detect that only 708608 bytes of the requested file have
> been copied.

I have had problems using DLink DFE530TX+ with 8139too in various Linux
versions (the nic goes into a frenzy flickering LED with heavy packet
loss). My quick and simple solution was to slow it down with "mii-tool -A
10baseT-FD,10baseT-HD eth0" (or whatever). Not sure if it is a nic or
module problem, but just seems to be a problem autonegotioating speed
(full-duplex works fine).

peter schb wrote:

>
> does anyone know the reason for that ?
> is there already a fix or work around for this problem ?
> what should i do ?
>

NFS still hanging after all these years? I remember this happening
circa 1998. Not exactly a 'fix' but check out the 'soft' mount option
in the man page. At least you won't hang forever.