2.6.7-rc2: no more AGP?

PATCH In review Colin Leroy 5 participants 6 messages Jun 4, 2004 – Jun 5, 2004 Char and misc drivers

Share & tools

Colin Leroy

author

2004-06-04 15:48 UTC #0 lore raw

Hi,

just a lousy bugreport... I noticed that agpgart doesn't work anymore on 2.6.7-rc2. Xorg reports that AGP isn't supported, and dmesg doesn't show the
agpgart: Putting AGP V2 device at 0000:00:0b.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:00:10.0 into 4x mode

It only shows
Linux agpgart interface v0.100 (c) Dave Jones
agpgart: Detected Apple UniNorth 2 chipset
agpgart: Maximum main memory to use for agp memory: 565M
agpgart: configuring for size idx: 4
agpgart: AGP aperture is 16M @ 0x0

Using 2.6.6, it works fine.

hth,

Signature · 1 line

Colin

Benjamin Herrenschmidt

2004-06-04 16:18 UTC #1 lore raw

Quoted 8 lines · Colin Leroy

Hi,
just a lousy bugreport... I noticed that agpgart doesn't work anymore on 2.6.7-rc2. Xorg reports that AGP isn't supported, and dmesg doesn't show the
agpgart: Putting AGP V2 device at 0000:00:0b.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:00:10.0 into 4x mode

Quoted 6 lines

It only shows
Linux agpgart interface v0.100 (c) Dave Jones
agpgart: Detected Apple UniNorth 2 chipset
agpgart: Maximum main memory to use for agp memory: 565M
agpgart: configuring for size idx: 4
agpgart: AGP aperture is 16M @ 0x0

Right, something seems broken. I'm also having problems with USB sleep & wakeup and with cpufreq. Argh, I've been away from ppc32 for too long !

I'll try to spare some time this week-end. It might be too late for 2.6.7 though =P

Ben.

Paul Mackerras

2004-06-05 03:35 UTC #2 lore raw

Quoted 16 lines · Benjamin Herrenschmidt

Quoted 8 lines · Colin Leroy
Hi,
just a lousy bugreport... I noticed that agpgart doesn't work anymore on 2.6.7-rc2. Xorg reports that AGP isn't supported, and dmesg doesn't show the
agpgart: Putting AGP V2 device at 0000:00:0b.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:00:10.0 into 4x mode
Quoted 6 lines
It only shows
Linux agpgart interface v0.100 (c) Dave Jones
agpgart: Detected Apple UniNorth 2 chipset
agpgart: Maximum main memory to use for agp memory: 565M
agpgart: configuring for size idx: 4
agpgart: AGP aperture is 16M @ 0x0

Quoted 3 lines

Right, something seems broken. I'm also having problems with USB sleep & wakeup and with cpufreq. Argh, I've been away from ppc32 for too long !

You need this patch. Michel Daenzer tells me that the cant_use_aperture check isn't needed.

Paul.

drivers/char/drm/drm_agpsupport.h +0 -2

diff -urN linux-2.5/drivers/char/drm/drm_agpsupport.h pmac-2.5/drivers/char/drm/drm_agpsupport.h
@@ -109,8 +109,6 @@
 		return -EBUSY;
 	if (!drm_agp->acquire)
 		return -EINVAL;
-	if ( dev->agp->cant_use_aperture )
-		return -EINVAL;
 	if ((retcode = drm_agp->acquire()))
 		return retcode;
 	dev->agp->acquired = 1;

Michel Dänzer

2004-06-04 16:22 UTC #3 lore raw

In reply to Colin Leroy

Quoted 7 lines · Colin Leroy

just a lousy bugreport... I noticed that agpgart doesn't work anymore on 2.6.7-rc2. Xorg reports that AGP isn't supported, and dmesg doesn't show the
agpgart: Putting AGP V2 device at 0000:00:0b.0 into 4x mode
agpgart: Putting AGP V2 device at 0000:00:10.0 into 4x mode

Quoted 8 lines

It only shows
Linux agpgart interface v0.100 (c) Dave Jones
agpgart: Detected Apple UniNorth 2 chipset
agpgart: Maximum main memory to use for agp memory: 565M
agpgart: configuring for size idx: 4
agpgart: AGP aperture is 16M @ 0x0
Using 2.6.6, it works fine.

Paulus brought this up on IRC, it seems to be a bad DRM merge: The code

#ifndef VMAP_4_ARGS
       if ( dev->agp->cant_use_aperture )
               return -EINVAL;

#endif

in DRM(agp_acquire) should be removed altogether in a 2.6 kernel because its vmap() takes 4 arguments; however, only the guards seem to have been removed, which causes this function to erroneously fail if the AGP aperture can't be directly accessed by the CPU.

Signature · 2 lines

Earthling Michel Dänzer      |     Debian (powerpc), X and DRI developer
Libre software enthusiast    |   http://svcs.affero.net/rm.php?r=daenzer

Colin Leroy

author

2004-06-04 19:39 UTC #4 lore raw

PATCH 1/1 2.6.7-rc2: fix agpgart

On 04 Jun 2004 at 18h06, Michel Dänzer wrote:

Hi,

Quoted 6 lines

Paulus brought this up on IRC, it seems to be a bad DRM merge: The code
#ifndef VMAP_4_ARGS
       if ( dev->agp->cant_use_aperture )
               return -EINVAL;

Quoted 6 lines

#endif
in DRM(agp_acquire) should be removed altogether in a 2.6 kernel because its vmap() takes 4 arguments; however, only the guards seem to have been removed, which causes this function to erroneously fail if the AGP aperture can't be directly accessed by the CPU.

Looks like. Removing it fixes the problem. Here's the patch...

Signed-off-by Colin Leroy <[email protected]>

drivers/char/drm/drm_agpsupport.h +0 -2

@@ -109,8 +109,6 @@
 		return -EBUSY;
 	if (!drm_agp->acquire)
 		return -EINVAL;
-	if ( dev->agp->cant_use_aperture )
-		return -EINVAL;
 	if ((retcode = drm_agp->acquire()))
 		return retcode;
 	dev->agp->acquired = 1;

Dave Airlie

2004-06-05 08:23 UTC #5 lore raw

Re: [PATCH] 2.6.7-rc2: fix agpgart

Quoted 9 lines

Quoted 5 lines
in DRM(agp_acquire) should be removed altogether in a 2.6 kernel because its vmap() takes 4 arguments; however, only the guards seem to have been removed, which causes this function to erroneously fail if the AGP aperture can't be directly accessed by the CPU.
Looks like. Removing it fixes the problem. Here's the patch...
Signed-off-by Colin Leroy <[email protected]>

okay I've put this into the DRM bk tree, thanks guys, I'll push it to Linus ASAP..

Dave.

Signature · 3 lines

David Airlie, Software Engineer
http://www.skynet.ie/~airlied / airlied at skynet.ie
pam_smb / Linux DECstation / Linux VAX / ILUG person