jmarin@embedtronics.fi wrote:

> I can confirm this - RX is the problem and transmitting packets during stall
> does not re-enable reception.

Okay. Maybe we should check nfe_rxeof() especially around error paths
(I wonder what "FIXME" magic vaules mean ;-p) and which interrupt
request (RXERR, RX, RX_NOBUF) causes the stall.

> > If it's caused by some race condition, how about the attached one?
> Should it be applied on top of your previous patch or to clean NetBSD
> sources?

It can be applied independently, but prefer with the previous one.
---
Izumi Tsutsui