CONFIG_8139TOO_PIO needed with RT patches

[Adam Swift]

Hi all,

I'm encountering a strange problem on the 2.6.24.7-rt14 kernel. I have  
a machine that uses the 8139too driver for its ethernet. This works  
fine at normal and high throughput with CONFIG_8139TOO_PIO both off  
and on, with both the stock Fedora 8 kernel and an unpatched 2.6.24.7  
kernel. However, on the Planet CCRMA realtime kernel, and a vanilla  
2.6.24.7-rt14, with CONFIG_8139TOO_PIO off the machine locks up at  
somewhat high throughput. I can trigger this by piping a large file  
over a netcat connection. I tested it at lower throughputs, and it  
seems to be fine at <128KiB/s but dies at >1MiB/s.

When I say locks up, I mean completely- ctrl-alt-del does nothing,  
magic sysrq does nothing, and num lock does not even respond. The  
keyboard lights do not flash to indicate kernel panic, and nothing  
gets into /var/log/messages or dmesg.

I understand that CONFIG_8139TOO_PIO is provided to cater for broken  
motherboards, which mine may be. However doesn't a difference between  
the behaviour on the RT and normal kernel indicate a problem somewhere  
with the RT kernel?

The kernel config that triggers it (with 2.6.24.7-rt14) can be found  
at http://omnitude.net/vike/rt.config , lspci -vv output at  
http://omnitude.net/vike/lspcivv . Hardware info:
Motherboard: Asrock P4I65G, using the onboard graphics
CPU: Intel Pentium IV 2.8GHz
RAM: 1GB of DDR
Plus a generic PCI Firewire card which has the VIA chip listed in the  
lspci output.

Please CC me as I'm not on the list.

Thanks,
Adam Swift

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/