3c59x v0.40 WARNING -- still screwed

Chris Evans (chris@ferret.lmh.ox.ac.uk)
Thu, 15 May 1997 15:53:42 +0100 (BST)

I hear rumours of this going into 2.0.31. IT does _not_ fix my problems
here, which were "transmitter access conflicts". Note that I was able to
negate the effects of these with a carefully placed dev->tbusy = 0 etc.,
on top of 3c59x.c v0.30-all.

With unpatched v0.40 the ethernet transmit died in less that 12 hours..

May 15 01:15:39 ferret kernel: 3c900.c:v0.40 4/16/97 becker@cesdis.gsfc.nasa.gov
May 15 01:15:39 ferret kernel: loading device 'eth0'...
May 15 01:15:39 ferret kernel: eth0: 3Com 3c590 Vortex 10Mbps at 0x6000, 00:a0:24:6c:18:dc, IRQ 10
May 15 01:15:39 ferret kernel: 8K byte-wide RAM 1:1 Rx:Tx split, autoselect/10baseT interface.
May 15 01:15:39 ferret kernel: eth0: Overriding PCI latency timer (CFLT) setting of 32, new value is 248.

.....

May 15 12:12:11 ferret kernel: eth0: transmit timed out, tx_status 88 status e000.
May 15 12:12:11 ferret kernel: eth0: Transmitter encountered 16 collisions -- network network cable problem?
May 15 12:12:11 ferret kernel: Flags; bus-master 0, full 0; dirty 0 current 0.
May 15 12:12:11 ferret kernel: Down list ffffffff vs. 01d39210.
May 15 12:12:11 ferret kernel: 0: 01d39210 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 1: 01d39220 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 2: 01d39230 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 3: 01d39240 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 4: 01d39250 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 5: 01d39260 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 6: 01d39270 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 7: 01d39280 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 8: 01d39290 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 9: 01d392a0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 10: 01d392b0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 11: 01d392c0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 12: 01d392d0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 13: 01d392e0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 14: 01d392f0 length 00000000 status 00000000
May 15 12:12:11 ferret kernel: 15: 01d39300 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: eth0: transmit timed out, tx_status 00 status e000.
May 15 12:12:12 ferret kernel: Flags; bus-master 0, full 0; dirty 0 current 0.
May 15 12:12:12 ferret kernel: Down list ffffffff vs. 01d39210.
May 15 12:12:12 ferret kernel: 0: 01d39210 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 1: 01d39220 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 2: 01d39230 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 3: 01d39240 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 4: 01d39250 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 5: 01d39260 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 6: 01d39270 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 7: 01d39280 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 8: 01d39290 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 9: 01d392a0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 10: 01d392b0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 11: 01d392c0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 12: 01d392d0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 13: 01d392e0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 14: 01d392f0 length 00000000 status 00000000
May 15 12:12:12 ferret kernel: 15: 01d39300 length 00000000 status 00000000

[repeat lots]

Interestingly enough the system appeared to recover from this loop an
couple of hours later (unsure if ethernet Tx functions were resumed
though), but have since died.

I am now running on v0.30-all again, with custom patch. Alas I only have a
binary module, with the "dev_kfree_skb missing mem-leak". Anyone got the
source to v0.30-all please???

Chris