Skip to contentSkip to conversation

i915

GIT PULL Chris Wilson 7 participants 10 messages Sep 6, 2010 – Sep 7, 2010 1 message not in the archive

lore
Share & tools
lore
#0 lore raw

Hi Linus,

in order to reduce the burden upon Eric and to try and feed patches upstream in a more timely manner I volunteered to take over maintaining the upstream i915 branches.

This batch of fixes include the timing fixes for the fallout from the conversion of intel_wait_for_vblank() and the error checking for SDVO modesetting. It also includes a [double dose] of module unload oopsen fixes from Daniel Vetter, more work by Jesse on bringing eDP machines to life and some of my pending fixes from bugzilla.

The following changes since commit 2bfc96a127bc1cc94d26bfaa40159966064f9c8c:

Linux 2.6.36-rc3 (2010-08-29 08:36:04 -0700)

are available in the git repository at:
git://anongit.freedesktop.org/~ickle/drm-intel drm-intel-fixes

Chris Wilson (22):
agp/intel: Promote warning about failure to setup flush to error. drm/i915: Drop the msleep parameter to wait_for()
drm/i915: Avoid using msleep under kdb and wait_for() drm/i915: Include a generation number in the device info drm/i915: Fix offset page-flips on i965+
drm/i915: Re-use set_base_atomic to share setting of the display registers drm/i915/crt: Flush register prior to waiting for vblank. drm/i915/dp: Boost timeout for enabling transcoder to 100ms drm/i915/sdvo: Guess the DDC bus in absence of VBIOS drm/i915/tv: Flush register writes before sleeping. drm/i915/dp: Really try 5 times before giving up.
drm/i915: Allocate the PCI resource for the MCHBAR
drm/i915: Rename i915_opregion.c to intel_opregion.c drm/i915: Use the VBT from OpRegion when available (v3) drm/i915: Addin-offset is an unreliable indicator of LVDS presence (v2) drm/i915: Tightly scope intel_encoder to prevent invalid use drm/i915: Ironlake page-flipping is per-plane not per-pipe drm/i915: Avoid use of uninitialised values when disabling panel-fitter Merge remote branch 'jbarnes/edp-testing' into drm-intel-fixes drm/i915/dp: Flush the PLL register write before udelay() drm/i915/dp: Remove unused variables
drm/i915: Clear the vblank status bit before polling for the next vblank

Dan Carpenter (2):
i915: return -EFAULT if copy_to_user fails
i915: return -EFAULT if copy_to_user fails

Daniel Vetter (17):
drm/i915: unload: fix intel dp encoder cleanup drm/i915: unload: fix error_work races drm/i915: unload: fix hotplug_work races drm/i915: unload: don't leak error state drm/i915: unload: fix idle_timer/idle_work races drm/i915: unload: fix unpin_work related races drm/i915: unload: ensure that gem is idle drm/i915: unload: fix retire_work races drm/i915: overlay on gen2 can't address above 1G drm/i915: unload: fix intel dp encoder cleanup drm/i915: unload: fix error_work races drm/i915: unload: fix hotplug_work races drm/i915: unload: don't leak error state drm/i915: unload: fix idle_timer/idle_work races drm/i915: unload: fix unpin_work related races drm/i915: unload: ensure that gem is idle drm/i915: unload: fix retire_work races

Jesse Barnes (8):
drm/i915: fix eDP detection
drm/i915: fix vblank wait test condition
drm/i915: use 125MHz reference clock for PCH attached eDP drm/i915: use VDD AUX for panel power around detection and in prepare drm/i915: split DP link training across panel power sequencing drm/i915: don't change VDD AUX status in panel power functions drm/i915: make sure VDD AUX power has time to settle drm/i915: make sure panel is sequenced off when starting a mode set

Pavel Roskin (1):
i915: revert some checks added by commit 32aad86f

Pekka Enberg (1):
i915: Fix spurious TV detection after 9d0498a2bf + 9559fcdbff

Simon Farnsworth (1):
drm/i915: Avoid pageflipping freeze when we miss the flip prepare interrupt

Sitsofe Wheeler (1):
drm/i915: Revert extra intel_wait_for_vblank to prevent stalls.

 drivers/char/agp/intel-gtt.c                       |    8 +-
 drivers/gpu/drm/i915/Makefile                      |    2 +-
 drivers/gpu/drm/i915/i915_debugfs.c                |   71 +++++
 drivers/gpu/drm/i915/i915_dma.c                    |   71 ++++--
 drivers/gpu/drm/i915/i915_drv.c                    |   68 ++---
 drivers/gpu/drm/i915/i915_drv.h                    |   54 ++---
 drivers/gpu/drm/i915/i915_gem.c                    |    5 +-
 drivers/gpu/drm/i915/i915_irq.c                    |   63 ++++-
 drivers/gpu/drm/i915/intel_bios.c                  |   73 +++--
 drivers/gpu/drm/i915/intel_crt.c                   |    5 +-
 drivers/gpu/drm/i915/intel_display.c               |  305 ++++++++++----------
 drivers/gpu/drm/i915/intel_dp.c                    |  240 ++++++++++------
 drivers/gpu/drm/i915/intel_drv.h                   |   17 +-
 drivers/gpu/drm/i915/intel_lvds.c                  |   50 ++--
 .../drm/i915/{i915_opregion.c => intel_opregion.c} |  120 ++++----
 drivers/gpu/drm/i915/intel_sdvo.c                  |   69 ++++--
 drivers/gpu/drm/i915/intel_tv.c                    |   14 +-
 17 files changed, 745 insertions(+), 490 deletions(-)
 rename drivers/gpu/drm/i915/{i915_opregion.c => intel_opregion.c} (92%)
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#1 lore raw
Quoted 12 lines · Chris Wilson

Hi Linus,

in order to reduce the burden upon Eric and to try and feed patches upstream in a more timely manner I volunteered to take over maintaining the upstream i915 branches.

This batch of fixes include the timing fixes for the fallout from the conversion of intel_wait_for_vblank() and the error checking for SDVO modesetting. It also includes a [double dose] of module unload oopsen fixes from Daniel Vetter, more work by Jesse on bringing eDP machines to life and some of my pending fixes from bugzilla.

Quoted 6 lines

The following changes since commit 2bfc96a127bc1cc94d26bfaa40159966064f9c8c:

Linux 2.6.36-rc3 (2010-08-29 08:36:04 -0700)

are available in the git repository at:
git://anongit.freedesktop.org/~ickle/drm-intel drm-intel-fixes

Sadly, these still do not include a fix for the build failure¹ that occurs if CONFIG_DEBUG_FS is not set:

,----

|   CC [M]  drivers/gpu/drm/i915/intel_overlay.o
| drivers/gpu/drm/i915/intel_overlay.c: In function 'intel_overlay_print_error_state':
| drivers/gpu/drm/i915/intel_overlay.c:1467: error: implicit declaration of function 'seq_printf'
| make[4]: *** [drivers/gpu/drm/i915/intel_overlay.o] Error 1

`----

Regards,
Sven

¹ http://lists.freedesktop.org/archives/intel-gfx/2010-August/007822.html

#2 lore raw
Quoted 8 lines · Sven Joachim

Sadly, these still do not include a fix for the build failure¹ that occurs if CONFIG_DEBUG_FS is not set:

,----

|   CC [M]  drivers/gpu/drm/i915/intel_overlay.o
| drivers/gpu/drm/i915/intel_overlay.c: In function 'intel_overlay_print_error_state':
| drivers/gpu/drm/i915/intel_overlay.c:1467: error: implicit declaration of function 'seq_printf'
| make[4]: *** [drivers/gpu/drm/i915/intel_overlay.o] Error 1

Andrew Morton picked up a patch to fix that, hence why I left it out.

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#3 lore raw
Quoted 11 lines · Chris Wilson
Quoted 8 lines · Sven Joachim

Sadly, these still do not include a fix for the build failureš that occurs if CONFIG_DEBUG_FS is not set:

,----

|   CC [M]  drivers/gpu/drm/i915/intel_overlay.o
| drivers/gpu/drm/i915/intel_overlay.c: In function 'intel_overlay_print_error_state':
| drivers/gpu/drm/i915/intel_overlay.c:1467: error: implicit declaration of function 'seq_printf'
| make[4]: *** [drivers/gpu/drm/i915/intel_overlay.o] Error 1

Andrew Morton picked up a patch to fix that, hence why I left it out.

You shouldn't. Andrew automatically drops patches that have been taken by maintainers and it's better if fixed are merged through the right tree.

Thanks,
Rafael

#4 lore raw
In reply to Chris Wilson
Quoted 5 lines · Chris Wilson

in order to reduce the burden upon Eric and to try and feed patches upstream in a more timely manner I volunteered to take over maintaining the upstream i915 branches.

Chris, could you help to check my sandybridge fixes? One is address mask fix found by Takashi, which should have been applied but I can't find it anywhere upstream. And others are PTE cache control fixes.

Please check 'for-ickle' branch at http://cgit.freedesktop.org/~zhen/drm-intel/

thanks.

Signature · 3 lines
Open Source Technology Center, Intel ltd.

$gpg --keyserver wwwkeys.pgp.net --recv-keys 4D781827

_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw
Quoted 5 lines · Zhenyu Wang

Chris, could you help to check my sandybridge fixes? One is address mask fix found by Takashi, which should have been applied but I can't find it anywhere upstream. And others are PTE cache control fixes.

Please check 'for-ickle' branch at http://cgit.freedesktop.org/~zhen/drm-intel/

I'm currently turning the gtt code upside down. iirc you've mentioned that snb needs to communicate new stuff between gem and the gtt code. But there's nothing in your patches (besides the introduction of intel-gtt.h which I've put in a different directory in my patch series, mea culpa). Can you please elaborate into what snb needs so that I can take this into account when designing the new gem->gtt interface? I have neither hw now docs at hand ...

Thanks, Daniel

btw: The current state of my gtt rework is available at

http://cgit.freedesktop.org/~danvet/drm/

Signature · 3 lines
Daniel Vetter
Mail: [email protected]
Mobile: +41 (0)79 365 57 48
#6 lore raw
In reply to Chris Wilson

On Mon, Sep 06, 2010 at 09:30:23 +0100, Chris Wilson wrote:

[...]

Quoted 3 lines

This batch of fixes include the timing fixes for the fallout from the conversion of intel_wait_for_vblank() and the error checking for SDVO modesetting. It also includes a [double dose] of module unload oopsen

Thanks, this branch fixes all my problems I reported in the thread
http://lkml.org/lkml/2010/8/26/56 (wrong screen resolution due to buggy TV out detection, no signal on monitor after resume from suspend to RAM).

Regards,
Tino

A message that is not in the archive (sent off-list or not yet received): [email protected]

#8 lore raw

Hi Linus,

I've gone over the set of patches and removed any that are not a regression, hang, compilation or other stability fix. I have left one in from Zhenyu that adds new PCI-IDs for Sandybridge.

I've dropped the eDP patches from Jesse as the machine should still be usable in spite of the blank panel. And they are not yet a complete set of patches, there are still failing eDP machines.

Daniel requested that his module unload patches be dropped as although they fix oopses, they should not impact ordinary users.

The following changes since commit 2bfc96a127bc1cc94d26bfaa40159966064f9c8c:

Linux 2.6.36-rc3 (2010-08-29 08:36:04 -0700)

are available in the git repository at:
git://anongit.freedesktop.org/~ickle/drm-intel drm-intel-fixes

Andrew Morton (1):
drivers/gpu/drm/i915/intel_overlay.c needs seq_file.h

Chris Wilson (13):
Revert "drm/i915: Unreference object not handle on creation" drm/i915: Clear the vblank status bit before polling for the next vblank agp/intel: Promote warning about failure to setup flush to error. drm/i915: Include a generation number in the device info drm/i915: Fix offset page-flips on i965+
drm/i915: Re-use set_base_atomic to share setting of the display registers drm/i915/dp: Boost timeout for enabling transcoder to 100ms drm/i915/sdvo: Restore guess of the DDC bus in absence of VBIOS drm/i915/dp: Really try 5 times before giving up.
drm/i915: Allocate the PCI resource for the MCHBAR
drm/i915: Tightly scope intel_encoder to prevent invalid use drm/i915: Avoid use of uninitialised values when disabling panel-fitter drm/i915: Prevent double dpms on

Dan Carpenter (2):
i915: return -EFAULT if copy_to_user fails
i915: return -EFAULT if copy_to_user fails

Daniel Vetter (1):
drm/i915: overlay on gen2 can't address above 1G

Pavel Roskin (1):
i915: revert some checks added by commit 32aad86f

Pekka Enberg (1):
i915: Fix spurious TV detection after 9d0498a2bf + 9559fcdbff

Simon Farnsworth (1):
drm/i915: Avoid pageflipping freeze when we miss the flip prepare interrupt

Zhenyu Wang (5):
agp/intel: fix physical address mask bits for sandybridge agp/intel: use #ifdef idiom for intel-agp.h agp/intel: Fix cache control for Sandybridge drm/i915: Enable MI_FLUSH on Sandybridge intel_agp,i915: Add more sandybridge graphics device ids

 drivers/char/agp/intel-agp.c            |   16 ++-
 drivers/char/agp/intel-agp.h            |   22 ++-
 drivers/char/agp/intel-gtt.c            |   66 ++++++---
 drivers/gpu/drm/i915/i915_debugfs.c     |   50 +++++++
 drivers/gpu/drm/i915/i915_dma.c         |   36 +++--
 drivers/gpu/drm/i915/i915_drv.c         |   65 ++++----
 drivers/gpu/drm/i915/i915_drv.h         |   27 +---
 drivers/gpu/drm/i915/i915_gem.c         |   11 +-
 drivers/gpu/drm/i915/i915_irq.c         |   51 ++++++-
 drivers/gpu/drm/i915/i915_reg.h         |    1 +
 drivers/gpu/drm/i915/intel_display.c    |  244 ++++++++++++++-----------------
 drivers/gpu/drm/i915/intel_dp.c         |   58 ++++----
 drivers/gpu/drm/i915/intel_drv.h        |   10 ++
 drivers/gpu/drm/i915/intel_overlay.c    |    2 +
 drivers/gpu/drm/i915/intel_ringbuffer.c |    8 +-
 drivers/gpu/drm/i915/intel_sdvo.c       |   69 +++++++--
 drivers/gpu/drm/i915/intel_tv.c         |    9 +-
 include/linux/intel-gtt.h               |   20 +++
 18 files changed, 479 insertions(+), 286 deletions(-)
 create mode 100644 include/linux/intel-gtt.h
Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre
#9 lore raw
Quoted 6 lines · Chris Wilson

Hi Linus,

I've gone over the set of patches and removed any that are not a regression, hang, compilation or other stability fix. I have left one in from Zhenyu that adds new PCI-IDs for Sandybridge.

Hi Chris,

Without reverting ce171780, graphics don't come back after resuming on my x201s. (drm/i915: Enable RC6 on Ironlake.) Several users bisected failing graphics on resume to this commit, and I've confirmed that reverting it does indeed fix the problem.

regards, Kyle.

#10 lore raw
Quoted 6 lines · Kyle McMartin

Hi Chris,

Without reverting ce171780, graphics don't come back after resuming on my x201s. (drm/i915: Enable RC6 on Ironlake.) Several users bisected failing graphics on resume to this commit, and I've confirmed that reverting it does indeed fix the problem.

Reverted. Thanks.
-Chris

Signature · 1 line
Chris Wilson, Intel Open Source Technology Centre