Failure to detect card ins/rem with pd6729

[Mikhail Degtiarev (Michael)]

> Can you check whether the count increments in /proc/interrupts
> when you do the "ping" command?
> 
> and no "interrupts dropped" message in your system log?

 Sure Komuro, there they are in kernel log:

Sep 17 23:51:37 (none) kernel: eth0: 3Com 3c562, io 0x300, irq 9, hw_addr 00:60:97:FF:B7:08
Sep 17 23:51:37 (none) kernel:   8K FIFO split 5:3 Rx:Tx, auto xcvr
Sep 17 23:51:37 (none) kernel: ttyS03 at port 0x02e8 (irq = 9) is a 16550A
Sep 18 00:02:31 (none) kernel: eth0: switched to 10baseT port
Sep 18 00:03:10 (none) kernel: eth0: interrupt(s) dropped!
Sep 18 00:04:07 (none) last message repeated 3 times
Sep 18 00:05:10 (none) last message repeated 6 times
Sep 18 00:05:58 (none) last message repeated 6 times

 "message repeated" appeares after shuffling some files to and fro.
Also seems like there is a throughput drop roughly at half maybe.
Well, and /proc/interrupts before:
 
           CPU0
  0:    1076973          XT-PIC  timer
  1:      14348          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  4:     322858          XT-PIC  serial
  5:          1          XT-PIC  soundblaster
  9:          0          XT-PIC  3c589_cs
 12:     135774          XT-PIC  PS/2 Mouse
 14:       7778          XT-PIC  ide0
NMI:          0
ERR:          1

and after some network traffic:

           CPU0
  0:    1178150          XT-PIC  timer
  1:      15090          XT-PIC  keyboard
  2:          0          XT-PIC  cascade
  5:          1          XT-PIC  soundblaster
  9:          0          XT-PIC  3c589_cs
 12:     143255          XT-PIC  PS/2 Mouse
 14:       8178          XT-PIC  ide0
NMI:          0
ERR:          1

 same after card ins/rem, no increment in irq 9 count whatsoever.
detected ok though, as i said before, due to poll_interval.
So, not all is a charm as i thought at first.
Supplying i82365 with other parms and without the poll_interval makes no good also.