Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 27 Apr 2010 01:49:55 +0530
From:      "C. Jayachandran" <c.jayachandran@gmail.com>
To:        Juli Mallett <jmallett@freebsd.org>
Cc:        freebsd-mips@freebsd.org
Subject:   Re: Crashes with the current -HEAD
Message-ID:  <n2r98a59be81004261319q711684f9gaa0cade95f9f21b@mail.gmail.com>
In-Reply-To: <l2keaa228be1004261249sad10dfao9ad91186d97af5ea@mail.gmail.com>
References:  <g2v98a59be81004230638q91b48b47i99ec64a137790bbc@mail.gmail.com> <q2jeaa228be1004231125hed2bbe9fj6680b29fdd63511a@mail.gmail.com> <t2g98a59be81004261236pf417a26cye108cb2ccb823d41@mail.gmail.com> <l2keaa228be1004261249sad10dfao9ad91186d97af5ea@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Apr 27, 2010 at 1:19 AM, Juli Mallett <jmallett@freebsd.org> wrote:
> On Mon, Apr 26, 2010 at 12:36, C. Jayachandran <c.jayachandran@gmail.com>=
 wrote:
>> I checked a few versions, =A0r206714 does not have the crash, r206717
>> has. I haven't looked at the changes yet, but if you have any
>> suggestions let me know.
>
> There are two things there that could be problematic.
>
> The first is that interrupt enable/disable has been changed. =A0I don't
> know anything about hazards on RMI systems, but if there's an
> interrupt disable hazard, you may want to change CLEAR_STATUS and
> DO_AST macros (in exception.S and asm.h respectively) to use many more
> nops or ehbs and see if that makes the problem go away.

XLR does not need any hazards for this.

> The second is the changes to the lmem pages. =A0If you reduce MAXMEM so
> that all of your memory is addressable within KSEG0 and KSEG1, you
> won't use lmem at all. =A0If that makes the problem go away, it suggests
> there is trouble in that area.
>
> I'm glad to do whatever I can, but I've been unable to get any RMI
> hardware, so all I can do is suggest you look at those two areas your
> try to narrow down to the specific change that caused the problem.


>From the crash it looks like the  PTE/PDE entires of some user
processes are getting corrupted. May be this is due to the LMEM
changes - will look at the code....

Regards,
JC.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?n2r98a59be81004261319q711684f9gaa0cade95f9f21b>