[fglrx] built for DRI 4.x.x, not 5.x.x

[Mark Shuttleworth]

I'm trying to get the fglrx driver to work on my desktop system. The driver
claims to offer full support for the card, but I'm getting MESA GL rendering
instead, and it appears to be loading bits from Xorg instead of from ATI. I'll
attach the latest xorg.conf, and log.

[Mark Shuttleworth (sabdfl)]

Created an attachment (id=3262)
xorg.conf

[Mark Shuttleworth (sabdfl)]

Created an attachment (id=3263)
log file

[Daniel Stone (daniels)]

'(II) Loading sub module "radeon"
(II) LoadModule: "radeon"
(II) Loading /usr/X11R6/lib/modules/drivers/radeon_drv.o
(II) Module radeon: vendor="X.Org Foundation"
 compiled for 6.8.2, module version = 4.0.1
 Module class: X.Org Video Driver
 ABI class: X.Org Video Driver, version 0.7'

Try using fglrx instead of radeon ...

[Mark Shuttleworth (sabdfl)]

Check out the attached xorg.conf:

> Section "Device"
> Identifier "ATI Graphics Adapter"
> Driver "fglrx"

[Daniel Stone (daniels)]

Hm.

(==) Using config file: "/home/mark/xorg.conf"

This is what you want?

[Daniel Stone (daniels)]

I'll be fixing this in breezy soon.

[Daniel Stone (daniels)]

*** Bug 21132 has been marked as a duplicate of this bug. ***

[joe (joerg-unglaub)]

> *** Bug 21132 has been marked as a duplicate of this bug. ***

That line

(EE) fglrx(0): R200DRIScreenInit failed (DRI version = 5.0.0, expected 4.1.0).
Disabling DRI.

is the evil line fglrx and the dri-libs from xorg

it bugs me

[Daniel Stone (daniels)]

*** Bug 21321 has been marked as a duplicate of this bug. ***

[Sebastian Latz (sebastian-latz)]

(In reply to comment #8)
> > *** Bug 21132 has been marked as a duplicate of this bug. ***
>
> That line
>
> (EE) fglrx(0): R200DRIScreenInit failed (DRI version = 5.0.0, expected 4.1.0).
> Disabling DRI.
>
> is the evil line fglrx and the dri-libs from xorg
>
> it bugs me

For me it works, if I change the file /usr/X11R6/lib/modules/extensions/libdri.a
with the one of Hoary Hedgehogs version. This makes my fglrx driver run with
dri! Unfortunately the contrast is very high then. I don't have a clue if its
caused by the "old" libdri.a file or the fglrx xorg-module. Its the one that
comes with ubuntu.

[Sebastian Latz (sebastian-latz)]

(In reply to comment #10)
> (In reply to comment #8)
> > > *** Bug 21132 has been marked as a duplicate of this bug. ***
> >
> > That line
> >
> > (EE) fglrx(0): R200DRIScreenInit failed (DRI version = 5.0.0, expected 4.1.0).
> > Disabling DRI.
> >
> > is the evil line fglrx and the dri-libs from xorg
> >
> > it bugs me
>
>
> For me it works, if I change the file /usr/X11R6/lib/modules/extensions/libdri.a
> with the one of Hoary Hedgehogs version. This makes my fglrx driver run with
> dri! Unfortunately the contrast is very high then. I don't have a clue if its
> caused by the "old" libdri.a file or the fglrx xorg-module. Its the one that
> comes with ubuntu.

Sorry for my unclear epression :-) I mean the fglrx driver that comes with ubuntu.

[joe (joerg-unglaub)]

> > For me it works, if I change the file /usr/X11R6/lib/modules/extensions/libdri.a
> > with the one of Hoary Hedgehogs version. This makes my fglrx driver run with
> > dri! Unfortunately the contrast is very high then. I don't have a clue if its
> > caused by the "old" libdri.a file or the fglrx xorg-module. Its the one that
> > comes with ubuntu.

Worked for me too. without strange sideeffects.

So fglrx has to be updated to use new dri-api rebuild? glue? ATI itself?
has to be done before breezy release.

[Sebastian Latz (sebastian-latz)]

(In reply to comment #12)
> > > For me it works, if I change the file
/usr/X11R6/lib/modules/extensions/libdri.a
> > > with the one of Hoary Hedgehogs version. This makes my fglrx driver run with
> > > dri! Unfortunately the contrast is very high then. I don't have a clue if its
> > > caused by the "old" libdri.a file or the fglrx xorg-module. Its the one that
> > > comes with ubuntu.
>
> Worked for me too. without strange sideeffects.
>
> So fglrx has to be updated to use new dri-api rebuild? glue? ATI itself?
> has to be done before breezy release.

No, I Think it's ATI duty to change something. The log message says it clearly:
"DRI version = 5.0.0, expected 4.1.0"

I had some similiar problems with my builtin SiS card on my laptop and Debian
Sarge (at that time SID). It always switched back to mesa software rendering
*without* some clear info messages in x-syslog. Debian provided a
backward-compatibility package with some mesa/dri things from older x-server
versions. But I think such a solution should be a definitive optional package,
because drivers shipped with x-org 6.8.2 (ati precisely) do *not* work with the
older libdri.a on my machine! If I do not uncommet the Load "dri" line in
xorg.conf, than I only can see strange deferred objects on screen ... But again,
I don't have some possibilities of comparison here so that I can't really prove
it generically.

[joe (joerg-unglaub)]

ok that means make to packages one with new and one with old libdri.a make
mesa-dri depend one them with "OR" with preference for the new one
and make fglrx depend on the old libdri
if someone installs fglrx-driver the old replaces the new and it works ... hopefully

[Mark Shuttleworth (sabdfl)]

It's getting uncomfortably close to release. My understanding was that ATI were
working on a release that would address this? Daniel, are we still expecting
them to deliver that? We need two weeks testing before release, so the time is
tight for us to put that in Breezy. Also, there are definitely issues with
8.16.20. Is ATI planning a new code drop?

[Daniel Stone (daniels)]

i have that now, and am just fooling around with the linker now to get that into
a format that our loader expects. 8.18 is entering beta soon, but certainly
won't be out before breezy, and 8.16 is a dead end now (no more development).

[Andrew Hutchings (linuxjedi)]

(In reply to comment #16)
> i have that now, and am just fooling around with the linker now to get that into
> a format that our loader expects. 8.18 is entering beta soon, but certainly
> won't be out before breezy, and 8.16 is a dead end now (no more development).

I take it that it is not as simple as just changing the code to detect for the
new DRI version?
Is there anything we can do to help?

[Fabio Massimo Di Nitto (fabbione)]

Fixed in linux-restricted-modules-2.6.12 (2.6.12.3-1) upload.

[Andrew Hutchings (linuxjedi)]

Created an attachment (id=3822)
Xorg log with new fglrx driver

This is with the new linux-restricted and xorg-fglrx packages. Cannot find any
modes. Worked fine before (but without DRI), any ideas?

[Andrew Hutchings (linuxjedi)]

Ignore me, stupid bug in my xorg.conf. Works perfectly

[Allison Karlitskaya (desrt)]

This might be an unrelated issue, but with the latest linux-restricted-modules
(2.6.12.4-2) DRI still isn't working for me

In xorg.log:

(II) fglrx(0): [drm] DRM interface version 1.0
(II) fglrx(0): [drm] created "fglrx" driver at busid "PCI:1:0:0"
(II) fglrx(0): [drm] added 8192 byte SAREA at 0xf8ef9000
(II) fglrx(0): [drm] mapped SAREA 0xf8ef9000 to 0xb7aa8000
(II) fglrx(0): [drm] framebuffer handle = 0xe0000000
(II) fglrx(0): [drm] added 1 reserved context for kernel
(II) fglrx(0): DRIScreenInit done
(II) fglrx(0): Kernel Module Version Information:
(II) fglrx(0): Name: fglrx
(II) fglrx(0): Version: 8.16.20
(II) fglrx(0): Date: Aug 16 2005
(II) fglrx(0): Desc: ATI FireGL DRM kernel module
(II) fglrx(0): Kernel Module version matches driver.
(II) fglrx(0): Kernel Module Build Time Information:
(II) fglrx(0): Build-Kernel UTS_RELEASE: 2.6.12-8-686-smp
(II) fglrx(0): Build-Kernel MODVERSIONS: no
(II) fglrx(0): Build-Kernel __SMP__: no
(II) fglrx(0): Build-Kernel PAGE_SIZE: 0x1000
(II) fglrx(0): [drm] register handle = 0xfe9f0000
(EE) fglrx(0): [agp] unable to acquire AGP, error "xf86_EINVAL"
(EE) fglrx(0): cannot init AGP
(II) fglrx(0): [drm] removed 1 reserved context for kernel
(II) fglrx(0): [drm] unmapping 8192 bytes of SAREA 0xf8ef9000 at 0xb7aa8000
(WW) fglrx(0): ***********************************************
(WW) fglrx(0): * DRI initialization failed! *
(WW) fglrx(0): * (maybe driver kernel module missing or bad) *
(WW) fglrx(0): * 2D acceleraton available (MMIO) *
(WW) fglrx(0): * no 3D acceleration available *
(WW) fglrx(0): ********************************************* *

and in the kernel log:
[4294702.481000] mtrr: type mismatch for e0000000,8000000 old: write-back new:
write-combining
[4294702.481000] [fglrx:firegl_addmap] *ERROR* mtrr allocation failed (-22)
[4294702.481000] [fglrx] Kernel AGP support doesn't provide agplock functionality.
[4294702.481000] [fglrx] AGP detected, AgpState = 0x1f004a1b (hardware caps of
chipset)
[4294702.481000] mtrr: type mismatch for f8000000,4000000 old: write-back new:
write-combining
[4294702.481000] [fglrx:firegl_unlock] *ERROR* Process 6927 using kernel context 0

Should I open a new bug instead?

[Trent Lloyd (lathiat)]

I don't have this problem, my Xorg.0.log shows all DRI working properly etc, but
i still have no dri (glxinfo says no, doesn't actually work)

[Matt Zimmerman (mdz)]

There seem to be a few different issues conflated here. Is the original problem
fixed, or not? Any other issues should be filed separately.

[Daniel Stone (daniels)]

yes, the original issue is fixed