21143 - Tx hung - some more info

[Vojtech Pavlik]

Hi!

I've enabled more debugging to shed more light on this. With debug level
greater than 4, too many messages are printed and the bug doesn't manifest
itself due to the system slowdown by the printk()'s.

With debug level equal to 4 it prints these messages:

eht0: Too much work during an interrupt, csr5=0xf03b0040
eth0: Re-enabling interrupts, 0xf06988c5
eth0: Too much work during an interrupt, csr5=0xf06988c5
eth0: Transmit timeout using MII device [repeats a couple times]
eth0: 21143 autonegotiation status 0x000000c6, MII
eth0: MII status 782d, Link partner report 41e1
eth0: Transmit timeout using MII device [repeats a many times]

And doesn't manage to send a packet.

If, inspired by an earlier mail on the mailing list about a similar problem,
I bump up the max work in interrupt to 100 (what I'm causing by doing this?
is that a risk?), I can't trigger this hang anymore.

Vojtech