[PATCH] agp(4) for ULi M1689/nVidia nForce3

Jung-uk Kim jkim at FreeBSD.org
Wed Sep 21 08:44:05 PDT 2005


On Wednesday 21 September 2005 07:49 am, Adam Gregoire wrote:
> Well as it stands now I was fooling around with some options in my
> xorg.conf, namely experimenting with enabling agpfastwrite, which
> caused some kind of issue(blank screen) and heavy cpu(at least)
> usage as passive cooling kicked in after a few seconds. Anyways I
> was unable to kill X ala ctrk-alt-backspace and trying to switch to
> vt0 would not work, nor would ctrl-alt-del to bring the machine
> down nicely. So after I killed the power and rebooted I got the
> panic back that I had before. .. the machine probes the gart but
> fails the device_attach of agp0 returning 6. Further along the boot
> as it tries to probe drm the panic happens; all I was really able
> to get at this time was:
> agp_get_info+0x87: movq 0(%r13),%rdi
>
> So after I got this, I tried to reboot again and the same thing
> happens. So I decided to reset my BIOS to defaults to see if that
> is what was actually making the error go away before(I thought
> flashing my BIOS had something to do with it originally), to no
> avail. What to do next ... I decided to boot off a windows drive I
> have, got to the login prompt and told the machine to reboot, let
> it boot back into FBSD and viola no panic(Interesting). To see if I
> was just getting lucky I rebooted FBSD and again no panic. Okay so
> at this point I provoked another system crash via enabling
> agpfastwrite option, killed the power, reboot in FBSD and got the
> panic again, rebooted FBSD and yet again a panic. I then boot into
> windows hit the login screen and have the system reboot again into
> FBSD and the machine came up with agp loaded fine and no panic. I
> did this routine once more and had the same results.
>
> One thing I will try to do later is to provoke getting the machine
> in a state where it will repeatably panic and at that point I will
> power the system down and re-seat my video card to see if it has
> the same effect as booting into windows and shutting down.
>
> If anyone has any ideas, questions or patches for me to try, get
> back to me and I will be more then happy test out any theories or
> patches you may have.

AGPFastWrite is known to cause system instability for many chipsets.  
There's no measurable performance gain even if it worked.

Jung-uk Kim

> P.S. I apologize for my wording if I am not making sense here, I
> have been up a long time.
>
> On Tue, 2005-09-20 at 19:59 +0000, Ganael Laplanche wrote:
> > Hi,
> >
> > I tried the patch without drm support in the kernel. Again, I
> > crashed my machine, this time at X startup. I have no more panic
> > at boot time, neither do I have agp0 detected. I don't know if
> > it's related, but I updated my BIOS to the latest revision as
> > suggested by Adam Gregoire...
> >
> > Regards,
> >
> > Gana챘l LAPLANCHE
> > ganael.laplanche at martymac.com
> > http://www.martymac.com
> > Tel : (+33)6.84.03.57.24.
> >
> > ---------- Original Message -----------
> > From: Jung-uk Kim <jkim at FreeBSD.org>
> > To: "Ganael Laplanche" <ganael.laplanche at martymac.com>
> > Cc: freebsd-amd64 at FreeBSD.org, "K.Wieland" <wieland.k at gmail.com>
> > Sent: Mon, 19 Sep 2005 17:13:40 -0400
> > Subject: Re: [PATCH] agp(4) for ULi M1689/nVidia nForce3
> >
> > > On Monday 19 September 2005 04:56 pm, Ganael Laplanche wrote:
> > > > Hi all,
> > > >
> > > > I've just tested the patch, but unfortunately, it didn't work
> > > > for me :( Kernel panic at bootup, with (nearly) this message
> > > > :
> > > >
> > > > [...]
> > > > drm0: <ATI Radeon if R250 9000>
> > > > Fatal trap 12 : page fault while in kernel mode
> > > > [...]
> > > >
> > > > ... and the system reboots :(
> > >
> > > That's really strange.
> > >
> > > > Let me know if you want me to make further tests...
> > >
> > > You don't have to compile DRM in the kernel.  Xorg server will
> > > automatically load it when it's needed.  Please remove them
> > > from the configuration.  I just want to see if agp(4) works
> > > without drm.
> > >
> > > Thanks,
> > >
> > > Jung-uk Kim
> >
> > ------- End of Original Message -------
> >
> > _______________________________________________
> > freebsd-amd64 at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-amd64
> > To unsubscribe, send any mail to
> > "freebsd-amd64-unsubscribe at freebsd.org"


More information about the freebsd-amd64 mailing list