6.0-CURRENT SNAP004 hangs on amr (patch)

Mike Tancsa mike at sentex.net
Wed Aug 24 16:08:53 GMT 2005

At 01:37 PM 23/08/2005, Joao Barros wrote:
>On 8/23/05, Eric Anderson <anderson at centtech.com> wrote:
> > Joao Barros wrote:
> > > On 8/23/05, Mike Tancsa <mike at sentex.net> wrote:
> > >
> > >>At 11:50 AM 23/08/2005, Joao Barros wrote:
> > >>
> > >>>Hi,
> > >>>
> > >>>I have a spare controller, exact same model I'm trying to use with
> > >>>FreeBSD 6 and that I'm willing to donate if needed / requested.
> > >>>I live in Portugal so shipping to Europe would be preferable.
> > >>>
> > >>>I'm available for any questions concerning this matter.
> > >>
> > >>I dont think its that controller.  I am able to lock the machine in
> > >>SMP mode without the AMR by generating a high interrupt load either
> > >>on the NIC or onboard SCSI controller or on a crypto card.
> > >>
> > >>         ---Mike
> > >
> > >
> > > Well, for me sums up to this:
> > > FreeBSD 5 with amr: ok
> > > FreeBSD 6 with amr: doesn't boot
> > >
> > > And when I mean 'same controller' I mean what I have PERC 3/DC aka
> > > LSILogic MegaRAID 1600 Elite.
> >
> > When you boot verbosely, do you see a bunch of retries?  If you do, let
> > it sit until the retries hits 16k.  Maybe it's the same cam bug I'm
> > seeing (I have a patch to try soon).
> >
> > Eric
>Actually no as you can verify in my posted dmesgs here:
>Thanks for your 'eye' on this but I don't seem to be affected by that
>bug your seeing.

I dont think its the AMR card.  I tried a OpenBSD snapshot on it, and 
the same tests that crash FreeBSD work fine on OpenBSD.

Something like the cryptotest with multiple threads would quickly 
lockup the box (i.e. no VGA output nothing).  Its not the HiFn card 
per se, it seems anything that generates a high interrupt load will 
lockup the machine.  I found generating a lot of network traffic 
would lock it up as well.  This doesnt seem to be the case on OpenBSD 
nor Linux.

dmesges from various OSes are at
including the one from OpenBSD.


I will pop the AMR card back in for good measure too.

Comparing the dmesgs, both Linux and OpenBSD seem to squak about the 
ISA bus/ATA controller and USB being disabled.  But FreeBSD seems to 
try and do something with it that causes it to blow up.  By blow up, 
I mean shut the machine down.


More information about the freebsd-current mailing list