[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: mec0 dies under NFS load
-----BEGIN PGP SIGNED MESSAGE-----
On Feb 27, 2008, at 15:25, Stephen M. Rumble wrote:
Quoting Martin Husemann <martin%duskware.de@localhost>:
When I do a "build.sh build sets" on another machine and then copy
binary sets via NFS, my machine silently resets. The same happens
doing a "installworld".
Copying the sets via scp works though.
I haven't yet tested other archs, but my gut feeling blames
Please prove me wrong.
This rings a bell... look at rev 1.12 of if_mec.c:
"sprinkle delay() and mec_mii_wait() in MII access functions
with this I no longer get spontaneous reboots during network
Perhaps macallan@ has more wisdom for you. I don't understand why
if_mec would have been resetting the system like that. Is 1.12 a
fix or a dirty hack?
I think the idea came from the OpenBSD driver and no, I don't have
any idea why or how mec would reset the whole machine either but
that's what it used to do. So, the commit was the result of me poking
around and finding something that made it work for me. My mec's been
pretty much reliable ever since. Not the fastest NIC ever but it works.
It may be interesting to know if a PCI NIC causes similar trouble,
but you have no riser ;)
I've got an ex in my O2, never used it since tsutsui-san committed
something that made mec go fast enough for me ( now I get 4MB/s
sustained, before it would eventually slow down to something like 8kB/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (Darwin)
-----END PGP SIGNATURE-----
Main Index |
Thread Index |