Date: Thu, 17 Nov 2005 10:58:24 -0800 (PST) From: Matthew Dillon <dillon@apollo.backplane.com> To: John Baldwin <jhb@freebsd.org> Cc: alan bryan <alanbryan1234@yahoo.com>, Alexander Leidinger <Alexander@leidinger.net>, freebsd-current@freebsd.org, FreeBSD current mailing list <current@freebsd.org>, "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net> Subject: Re: [PATCH] nve(4) locking cleanup Message-ID: <200511171858.jAHIwOoc006061@apollo.backplane.com> References: <20051117010651.97608.qmail@web50303.mail.yahoo.com> <Pine.BSF.4.53.0511171134430.80787@e0-0.zab2.int.zabbadoz.net> <20051117143322.lvz347pzkcg480co@netchild.homeip.net> <200511171135.53903.jhb@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
A side note on the work I did on nv (nve) for DragonFly.
I spent a very, very long time trying to fix the watchdog problems. I
was able to greatly reduce instances of the watchdog timeout, but it
still occurs under heavy load.
Even worse, I can wedge the hardware to the point where a reboot or
soft power-down does not fix the problem! I have to physically unplug
the power cable from the box to make the device work again. It only
takes ~30 min - 3 hours of testing to get the ethernet into this
hardware-wedged state.
At this point I believe that the remaining problems are entirely within
Nvidia's nvnet object module. I don't think there is anything we can do
about it short of NVidia coming out with an update (which isn't likely).
Now, linux *has* a native implementation of this driver that does not
use the Nvidia module, and I have gotten reports that it does not suffer
from the same problems.
I think the only way to solve the watchdog problems that nv (nve) has
is to throw it in the trash and do a fresh port from the native driver
in the linux tree.
-Matt
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200511171858.jAHIwOoc006061>