recent stability problems with fxp driver

[John Birrell]

On Fri, Sep 12, 2003 at 01:51:56PM -0700, John Polstra wrote:
> The problem is real, at least on some hardware.  I had to give up on
> using the two integrated fxp devices on my Dell 1550 -- which is a
> real bummer, since it's a 1U box that only has two PCI slots.  With
> the latest -stable driver, I couldn't fetch a 560 MB file from
> another machine on the LAN using FTP without killing the fxp device.

I'm suffering what seem like different problems with the fxp driver. In RELENG_4
the driver works well on two different embedded boards that I have. On one
of those boards I regularly transfer ~500 MB files [via HTTP 8-)] and the
driver appears stable.

As soon as I try to run -current on one of those boards, the kernel hangs
in fxp's attach, just after the IRQ resource is allocated, and possibly just
when the driver tries to reset the board. I haven't been able to debug this.
I wonder if the device is creating an interrupt storm.

I have a hard time believing that the BIOS is behaving any differently when
booting a RELENG_4 kernel compared to a current one. Certainly the probe
output for fxp is the same.

When I try the same -current kernel on the other embedded board, it works.

The boards are very different. One has a Cyrix processor and the other is
an AMD Elan SC520. They both have a General Software bios, but obviously
the different processors make that largely irrelevant.

What makes my situation even more confusing is that the -current kernel
that hangs attaching fxp, was actually netbooted over the device that it
hangs attaching. I'm tempted to conclude that it isn't rogue hardware - just
a situation that isn't catered for FreeBSD. I wish I knew how to debug it.

-- 
John Birrell