Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 24 Sep 1998 01:31:57 -0400
From:      Adam McDougall <mcdougall@ameritech.net>
To:        current@FreeBSD.ORG
Subject:   Re: options DPT_LOST_IRQ
Message-ID:  <3609D94D.8397CF49@ameritech.net>
References:  <199809240354.VAA03623@narnia.plutotech.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Justin T. Gibbs wrote:
> 
> In article <3609A685.A9B65B3A@ameritech.net> you wrote:
> > Does CAM still perform what is supposed to when using this kernel
> > option? Because it doesn't show any evidence of it in Boot: -v and the
> > unwanted behavior of not using this option is showing up again, even
> > though its still in my kernel.
> 
> The CAM driver does not currently honor this option.  It could easily
> be added again, but it would be nice to know why it is necessary.  Is
> the symptom simply a timeout?
> 
The symptom (at least on my system) is that the HD totally freezes up
with the light on, and the dpt sits there merrily idle, while freebsd
processes start losing sanity because the disk cannot be accessed (eg.
any interaction with the system which would trigger a HD access would
freeze up that portion of the system..)   I know there is a 'irq
pending' led on the dpt which may indicate the condition has occurred
for some people, but mine didnt happen this way, and adding this kernel
option ceased the described lockups.  I have a 2144UW, a Atlas II, and a
Mtech R534 motherboard.  

#  DPT_LOST_IRQ             When enabled, will try, once per second, to
catch
#                           any interrupt that got lost.  Seems to help
in some
#                           DPT-firmware/Motherboard combinations. 
Minimal
#                           cost, great benefit.
^^ from LINT

would hate to lose that great benefit :P



> One thing I noticed that the Linux driver does in it's interrupt
> handler that FreeBSD doesn't, is to watch for the busy bit in the
> aux status register.  Is this important?  (I don't have any docs...)
> Could it be that we inadvertantly clear an interrupt before the
> status packet write is complete?
> 
> If the problem is in the FreeBSD interrupt code, we should find out
> for sure.  I may not have a fast enough adapter (PM3224A) to make
> the failure occur here, under a PCI bus analyzer, to determine one
> way or the other.
> 
> > Thanks in advance..
> 
> --
> Justin

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3609D94D.8397CF49>