2006-03-30 05:52:41

by Dave Airlie

[permalink] [raw]
Subject: [git pull] DRM changes for 2.6.17


Hi Linus,
Can you please pull the 'drm-patches' branch from
git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6.git

This contains the changes for the DRM I'd like in 2.6.17, mainly a new
memory mapper and a safer r300 method for enabling pci ids... they now
require a new Xorg driver..

It also finally makes the DRM use the correct PCI DMA interfaces...

Dave.

drmP.h | 16 ++---
drm_bufs.c | 20 +++----
drm_dma.c | 4 -
drm_memory.c | 59 --------------------
drm_memory_debug.h | 70 ------------------------
drm_pci.c | 29 ++++++++--
drm_pciids.h | 116 +++++++++++++++++++++++++++-------------
i915_dma.c | 2
i915_irq.c | 2
r300_cmdbuf.c | 86 +++++++++++++++++++++++++++---
r300_reg.h | 39 +++++++++++--
radeon_cp.c | 151 +++++++++++++++++++++++++++++++++++++----------------
radeon_drm.h | 5 +
radeon_drv.h | 25 +++++---
radeon_state.c | 107 ++++++++++++++++---------------------
sis_mm.c | 2
16 files changed, 414 insertions(+), 319 deletions(-)

commit 55eb061326765b2d0489387cfb3fc7dbd244f917
Author: Dave Airlie <[email protected]>
Date: Wed Mar 29 08:16:12 2006 +1000

drm: remove drm_{alloc,free}_pages

drm_alloc_pages and drm_free_pages can now be removed.

Signed-off-by: Adrian Bunk <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

commit d2b58b58838159b2afdc624f74b208e8bd3c029e
Author: Dave Airlie <[email protected]>
Date: Wed Mar 29 08:12:52 2006 +1000

drm: sis fix compile warning

Prevent a gcc warning in the SIS DRM driver. offset is a unsigned int and
the printk wants a long.

Signed-off-by: Jon Mason <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

commit f3dd5c37382472a8b245ad791ed768771594e60c
Author: Dave Airlie <[email protected]>
Date: Sat Mar 25 18:09:46 2006 +1100

drm: add new radeon PCI ids..

This adds all the r300 and r400 PCI ids from DRM CVS, it also
makes these cards only initialise when the new xorg driver is
used, as otherwise the DRM can cause lockups.

Signed-off-by: Dave Airlie <[email protected]>

commit 6e5fca53c72c95da92c092411c7ec81e926af632
Author: Dave Airlie <[email protected]>
Date: Mon Mar 20 18:34:29 2006 +1100

drm: read breadcrumb in IRQ handler

From: Keith Whitwell <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

commit e7f947b908921a661efcbc08ce42d7de67691b07
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 20:28:19 2006 +1100

drm: fixup i915 breadcrumb read/write

Some minor issues in the i915 breadcrumb code.

Signed-off-by: Dave Airlie <[email protected]>

commit 2fab58d1a18c752887c2b9f6ee7b6997ced4a77a
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 20:15:41 2006 +1100

drm: remove pointless checks in radeon_state

If these were valid checks, we'd have already oopsed several
lines above where we were already dereferencing them.

DA: these used to be valid but other changes made them unnecessary.

Coverity: 776,777,778
Signed-off-by: Dave Jones <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

commit f15e92d702bba47f451bd5f63e170252c6e1bba2
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 20:12:23 2006 +1100

drm: fixup improper cast.

Signed-off-by: Dave Airlie <[email protected]>

commit dfab11542fbecd8539c092fe36155909b4812f73
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 20:01:37 2006 +1100

drm: rationalise some pci ids

This is the start of some work from Roland Scheidegger to align
the X DDX pci ids and the drm ones, however we don't want to put
r300 ids in the kernel just yet, they destabilise a few machines.

From: Roland Scheidegger (via DRM CVS)
Signed-off-by: Dave Airlie <[email protected]>

commit ee4621f011750a6eff9f56631e12ab7fd9503aaa
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 19:45:26 2006 +1100

drm: Add general-purpose packet for manipulating scratch registers (r300)

From: Aapo Tahkola (via DRM CVS)
Signed-off-by: Dave Airlie <[email protected]>

commit d5ea702f1e8e3edeea6b673a58281bf99f3dbec5
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 19:37:55 2006 +1100

drm: rework radeon memory map (radeon 1.23)

This code reworks the radeon memory map so it works better
for newer r300 chips and for a lot of older PCI chips.

It really requires a new X driver in order to take advantage of this code.

From: Ben Herrenschmidt <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>

commit 45f17100bfd18c99d6479e94598f4e533bbe30d8
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 19:12:10 2006 +1100

drm: update r300 register names

Update some of the DRM register names from DRM CVS

Signed-off-by: Dave Airlie <[email protected]>

commit ddf19b973be5a96d77c8467f657fe5bd7d126e0f
Author: Dave Airlie <[email protected]>
Date: Sun Mar 19 18:56:12 2006 +1100

drm: fixup PCI DMA support

This patch makes the PCI support use the correct Linux interfaces finally.
Tested in DRM CVS on PCI MGA card.

Signed-off-by: Dave Airlie <[email protected]>


2006-03-30 20:32:42

by Linus Torvalds

[permalink] [raw]
Subject: Re: [git pull] DRM changes for 2.6.17



On Thu, 30 Mar 2006, Dave Airlie wrote:
>
> This contains the changes for the DRM I'd like in 2.6.17, mainly a new memory
> mapper and a safer r300 method for enabling pci ids... they now require a new
> Xorg driver..

Can you describe what that means more?

Does it mean that people with old X versions will lose hw acceleration?
For which chips?

Linus

2006-03-30 21:59:21

by Dave Airlie

[permalink] [raw]
Subject: Re: [git pull] DRM changes for 2.6.17

> Can you describe what that means more?
>
> Does it mean that people with old X versions will lose hw acceleration?
> For which chips?

No it means people with old X versions won't try to enable hw accel on
cards that their X.org doesn't suppport...

The X.org drivers with respect to r300 drivers are highly experimental and
enabled DRI on r300 by default before they should, when I added the r300
PCI IDs as I tried last time, lots of people crashed and you backed out
the changes... so now the kernel isn't going to to trigger those problems,
as all of the new r300 class cards require using a new Xorg driver to
enable DRI..

Its the only way I can think off to get the r300 PCI ids into the kernel
and not break current systems... there is nothing I can do in the DRM to
fix the Xorg DDX stupidity..

Dave.

--
David Airlie, Software Engineer
http://www.skynet.ie/~airlied / airlied at skynet.ie
Linux kernel - DRI, VAX / pam_smb / ILUG

2006-03-30 22:27:21

by Linus Torvalds

[permalink] [raw]
Subject: Re: [git pull] DRM changes for 2.6.17



On Thu, 30 Mar 2006, Dave Airlie wrote:
>
> No it means people with old X versions won't try to enable hw accel on cards
> that their X.org doesn't suppport...
>
> The X.org drivers with respect to r300 drivers are highly experimental and
> enabled DRI on r300 by default before they should, when I added the r300 PCI
> IDs as I tried last time, lots of people crashed and you backed out the
> changes... so now the kernel isn't going to to trigger those problems, as all
> of the new r300 class cards require using a new Xorg driver to enable DRI..
>
> Its the only way I can think off to get the r300 PCI ids into the kernel and
> not break current systems... there is nothing I can do in the DRM to fix the
> Xorg DDX stupidity..

Ok, thanks. Sounds good. Pulled,

Linus