Possible fxp(4) problem in -CURRENT

Hi.

Situation:
Single-purpose machines only serving http requests for static content,
running 6.0-BETA5 and now 6.0-RC1. Hardware is IBM eServer x330 (dual
P3, 1GB RAM, 2 * SCSI disk, 2 * Intel fxp(4) network interface). In
addition to base system there is running boa, cdpd, qmail a net-snmp.
Kernel is custom, config and dmesg are attached. Machines are under
quite heavy network load.

Problem:
After some time of serving requests the ethernet interface in the
machine stops communicating on the wire. It does not respond to any
packets (ping, http, nfs, ssh) and vmstat(8) shows stopped interrupt
counter for that interface. Machine is running just fine, one can log in
localy and even configure second interface to communicate on network
through it. It seems to be the same problem as described in PR
amd64/82425 with one exception - there is no 'device timeout' in logs.
In fact there is nothing suspicious in logs when the system reaches
such state. This leads me to that it also may be close to PR kern/87418
- interface just stops receiving (and/or sending) data. The period in
which this happens varies a lot - some machines run for two weeks with
no problem, some stop that way twice a day. It may be related to
network traffic volume. We have tried to disable acpi(4) and turn on
polling(4) but with no effect.

Questions:
Is this known problem (extensive googling returned nothing similar)? Is
there any way I can help debug the problem? When I encounter the same
problem again what data would be valuable for debugging?

Dan