[ltp] diagnosing a freeze on T60p with ubuntu hardy

[Michael Karcher]

--=-EFRA7e4SGhauhdOBIXx+
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable


Am Donnerstag, den 15.05.2008, 08:50 -0400 schrieb Alex Deucher:
> On Thu, May 15, 2008 at 7:59 AM, Phil Shotton <phil@stairbridge.org.uk> w=
rote:
> > I'm using the radeon driver, not the proprietary one. Interestingly if =
I use
> > the default video output mode (xv) with mplayer I don't get the problem=
.
> > It's only if I use the gl2 driver, which suggests the problem may be in=
 mesa
> > rather than the radeon driver? (not sure - I don't understand much abou=
t the
> > various video components).
> >
>=20
> Unless you are using the experimental r5xx 3D driver, it sounds like
> an xserver or mesa problem as there is no 3D support for r5xx hardware
> by default.

This should not be a mesa problem (but could be a problem *exposed* by
mesa). If there is no 3D driver, mesa does not talk to hardware (it does
not even have permission to in a standard Unix system), so it can not
cause hard locks like this one. The symptoms sound like interrupt
processing is blocked, which can mean one of three things: The processor
is not doing anything at all, it is stuck in some loop with interrupts
disabled or stuck in a hard interrupt (where interrupts are disabled,
too). The former could be caused by defective hardware (communication
between processor and chipset broke down), whereas the other two points
would point towards the kernel. Still, the X Server could be a cause of
the problem too, as it (opposed to mesa) *has* the possibility to mess
up the video card, which in turn has the possibility to cause lots of
interrupts keeping the kernel from doing anything else in extreme cases.

When graphics card were still PCI, a buggy 3D driver could lock the PCI
bus, halting processing immediately as no further memory access from the
processor was possible, but in that case, sound would have stopped too.
Does anybody know how a blocked PCIe port (for example by a crashed
graphics chip) is handled nowadays?

Regards,
  Michael Karcher

--=-EFRA7e4SGhauhdOBIXx+
Content-Type: application/pgp-signature; name=signature.asc
Content-Description: Dies ist ein digital signierter Nachrichtenteil

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQBILDZlzhek2R7EicoRAo4SAJ41N7221L+zakgtwIXEseb2dKGHsQCgkGhH
58jgiGFyuNbGLDkcPJFDJp0=
=+8AX
-----END PGP SIGNATURE-----

--=-EFRA7e4SGhauhdOBIXx+--