Hi all,
Today's linux-next merge of the drm-intel tree got a conflict in:
include/drm/drmP.h
between commit:
c4e68a583202 ("drm: Introduce DRM_DEV_* log messages")
from the drm tree and commit:
30b0da8d556e ("drm: extra printk() wrapper macros")
from the drm-intel tree.
I fixed it up
== Series Details ==
Series: SKL/KBL watermark fixes
URL : https://patchwork.freedesktop.org/series/12082/
State : warning
== Summary ==
Series 12082v1 SKL/KBL watermark fixes
http://patchwork.freedesktop.org/api/1.0/series/12082/revisions/1/mbox/
Test kms_pipe_crc_basic:
Subgroup
== Series Details ==
Series: Enable upfront link training on DDI platforms (rev8)
URL : https://patchwork.freedesktop.org/series/10821/
State : warning
== Summary ==
Series 10821v8 Enable upfront link training on DDI platforms
Bspec says:
"The mailbox response data may not account for memory read latency.
If the mailbox response data for level 0 is 0us, add 2 microseconds
to the result for each valid level."
This means we should only do the +2 in case wm[0] == 0, not always.
So split the sanitizing
According to BSpec, it's the "core CPUs" that need the code, which
means SKL and KBL, but not BXT.
I don't have a KBL to test this patch on it.
Signed-off-by: Paulo Zanoni
---
drivers/gpu/drm/i915/intel_pm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff
The plan is to introduce intel_has_sagv() and then use it to discover
which platforms actually support it.
I thought about keeping the functions with their current skl names,
but found two problems: (i) skl_has_sagv() would become a very
confusing name, and (ii) intel_atomic_commit_tail() doesn't
And use it to move knowledge about the SAGV-supporting platforms from
the callers to the SAGV code.
We'll add more platforms to intel_has_sagv(), so IMHO it makes more
sense to move all this to a single function instead of patching all
the callers every time we add SAGV support to a new platform.
During watermarks calculations, this value is used in 3 different
places. Only one of them was not using a hardcoded 4. Move the code up
so everybody can benefit from the actual value.
This should only help on situations with Y tiling + 90/270 rotation +
1 or 2 bpp or NV12.
Signed-off-by: Paulo
We forgot the "res_blocks += y_tile_minimum" that's described on step
V of our documentation.
Again, this should only affect the Y tiling cases.
Signed-off-by: Paulo Zanoni
---
drivers/gpu/drm/i915/intel_pm.c | 12 +++-
1 file changed, 7 insertions(+), 5
The confusing thing is that plane_blocks_per_line is listed as part of
the method 2 calculation but is also used for other things. We
calculated it in two different places and different ways: one inside
skl_wm_method2() and the other inside skl_compute_plane_wm(). The
skl_wm_method2()
Hi
This series is the result of me comparing the code against BSpec. It
doesn't solve any particular problem I was seeing, but there's enough
changes that this code could potentially change the watermarks values
for most machines. Maybe this will fix somebody's underrun? Even if it
doesn't, we
This should affect linear and X tiled planes on really small htotal
cases. It doesn't seem to be a very feasible case, but let's implement
it since it's on the specification and it's better to have it and
never need than not have it and realize we needed it.
Signed-off-by: Paulo Zanoni
From: Jim Bride
Add upfront link training to intel_dp_mst_mode_valid() so that we know
topology constraints before we validate the legality of modes to be
checked.
Call the function that loops through the link rates and lane counts
starting from highest supported link
From: Dhinakaran Pandiyan
Wrap the max. vswing check in a separate function.
This makes the clock recovery phase of DP link training cleaner
v2:
Fixed the Compiler warning (Mika Kahola)
Signed-off-by: Dhinakaran Pandiyan
From: Durgadoss R
To support USB type C alternate DP mode, the display driver needs to
know the number of lanes required by the DP panel as well as number
of lanes that can be supported by the type-C cable. Sometimes, the
type-C cable may limit the bandwidth even if Panel
From: Dhinakaran Pandiyan
This function cleans up clock recovery loop in link training compliant
tp Dp Spec 1.2. It tries the clock recovery 5 times for the same voltage
or until max voltage swing is reached and removes the additional non
compliant retries. This
According to the DisplayPort Spec, in case of Clock Recovery failure
the link training sequence should fall back to the lower link rate
followed by lower lane count until CR succeeds.
On CR success, the sequence proceeds with Channel EQ.
In case of Channel EQ failures, it should fallback to
lower
On Wed, Aug 31, 2016 at 05:20:26PM -0700, Dhinakaran Pandiyan wrote:
> From: Libin Yang
>
> (This patch is developed by Dave Airlie originally)
>
> This patch adds support for DP MST audio in i915.
>
> Enable audio codec when DP MST is enabled
On Tue, Sep 06, 2016 at 12:58:29PM +0300, Mika Kahola wrote:
> On Fri, 2016-09-02 at 11:05 +0300, Mika Kahola wrote:
> > +1 for this cleanup
> >
> > Reviewed-by: Mika Kahola
> Received couple of compiler warnings to be cleaned up
>
>
On 06/09/16 12:21, Dave Gordon wrote:
> On 04/09/16 19:58, Nicolas Iooss wrote:
>> When building the kernel with clang and some warning flags, the compiler
>> reports that the return value of dcs_get_backlight() may be
>> uninitialized:
>>
>>
On Tue, Sep 06, 2016 at 06:39:12PM +, Srivatsa, Anusha wrote:
> Sending to the list again since Ville's review comment didn't hit the
> mailing list last time.
>
> Regards,
> Anusha
>
> -Original Message-
> From: Ville Syrjälä [mailto:ville.syrj...@linux.intel.com]
> Sent: Monday,
Sending to the list again since Ville's review comment didn't hit the mailing
list last time.
Regards,
Anusha
-Original Message-
From: Ville Syrjälä [mailto:ville.syrj...@linux.intel.com]
Sent: Monday, September 5, 2016 4:39 AM
To: Srivatsa, Anusha
Cc:
On Fri, Sep 02, 2016 at 04:00:20PM +0300, Mika Kahola wrote:
> On Thu, 2016-09-01 at 15:08 -0700, Manasi Navare wrote:
> > According to the DisplayPort Spec, in case of Clock Recovery failure
> > the link training sequence should fall back to the lower link rate
> > followed by lower lane count
On Fri, Sep 02, 2016 at 03:49:02PM +0300, David Weinehall wrote:
> On Thu, Sep 01, 2016 at 03:08:16PM -0700, Manasi Navare wrote:
> > According to the DisplayPort Spec, in case of Clock Recovery failure
> > the link training sequence should fall back to the lower link rate
> > followed by lower
On Fri, Sep 02, 2016 at 03:03:22PM +0300, David Weinehall wrote:
> On Thu, Sep 01, 2016 at 03:08:16PM -0700, Manasi Navare wrote:
> > According to the DisplayPort Spec, in case of Clock Recovery failure
> > the link training sequence should fall back to the lower link rate
> > followed by lower
Hi,
recently I started seeing occasional short screen flicker to black
associated with dmesg entries
[ 583.020853] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU
pipe A FIFO underrun
[13512.114363] [drm:intel_cpu_fifo_underrun_irq_handler [i915]] *ERROR* CPU
pipe A FIFO underrun
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Just pure code movement, cleanup and polish will happen in later
> patches.
>
> v2: Don't forget all the ioctl! To extract those cleanly I decided to
> put check_src_coords into drm_framebuffer.c (and give it a
>
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Some were still left in drm_crtc.h. Also include drm_edid.h in the
> rst files.
>
> Signed-off-by: Daniel Vetter
Reviewed-by: Sean Paul
> ---
>
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Now that there's less stuff in there I noticed that I overlooked them.
> Sprinkle some docs over them while at it.
>
> Signed-off-by: Daniel Vetter
Reviewed-by: Sean Paul
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Again move it from the unmaintainable csv into DOC free-form overview
> sections.
>
> Cc: Lionel Landwerlin
> Signed-off-by: Daniel Vetter
A few more comment
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Try to spec a bit more precisely how they all fit together, now that
> at least the code is for all the additional properties is in one
> place.
>
> Also remove the entries for the standardized properties from the
>
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> Imo zpos, rotatation, blending eq (once we have it) and all that
> should be in drm_blend.c, since those are all about how exactly the
> pixels are rendered onto the CRTC's visible area. Also noticed that
> one
On Wed, Aug 31, 2016 at 12:09 PM, Daniel Vetter wrote:
> We removed it in
>
> commit 6ab10b76ff6252bd9be0849c40f5865e39a29961
> Author: Daniel Vetter
> Date: Fri Aug 12 22:48:45 2016 +0200
>
> drm/kms: Nuke dirty_info property
>
>
On 06/09/16 16:33, Goel, Akash wrote:
On 9/6/2016 6:47 PM, Tvrtko Ursulin wrote:
Hi,
On 06/09/16 11:43, akash.g...@intel.com wrote:
From: Akash Goel
This patch provides a test utility which helps capture GuC firmware
logs and
then dump them to file.
The logs are
On 9/6/2016 6:47 PM, Tvrtko Ursulin wrote:
Hi,
On 06/09/16 11:43, akash.g...@intel.com wrote:
From: Akash Goel
This patch provides a test utility which helps capture GuC firmware
logs and
then dump them to file.
The logs are pulled from a debugfs file
== Series Details ==
Series: drm/i915: Remove 64b mmio write vfuncs (rev2)
URL : https://patchwork.freedesktop.org/series/12029/
State : failure
== Summary ==
Series 12029v2 drm/i915: Remove 64b mmio write vfuncs
http://patchwork.freedesktop.org/api/1.0/series/12029/revisions/2/mbox/
Test
The benchmark was failing with:
gem_busy.c:158:8: error: implicit declaration of function 'intel_gen'
is invalid in C99 [-Werror,-Wimplicit-function-declaration]
gen = intel_gen(intel_get_drm_devid(fd));
The root cause was due to the local lib directory not being specified
in
We don't have safe 64-bit mmio writes as they are really split into
2x32-bit writes. This tearing is dangerous as the hardware *will*
operate on the intermediate value, requiring great care when assigning.
(See, for example, i965_write_fence_reg.) As such we don't currently use
them and strongly
On Sat, Aug 27, 2016 at 02:33:25PM +0100, Matthew Auld wrote:
> Currently it's entirely possible to go through the link training step
> without first determining the lane_count, which is silly since we end up
> doing a bunch of aux transfers of size = 0, as highlighted by
> WARN_ON(!msg->buffer !=
On Tue, 06 Sep 2016, Jani Nikula wrote:
> On Tue, 06 Sep 2016, Zhenyu Wang wrote:
>> [ Unknown signature status ]
>> On 2016.09.06 11:33:57 +0300, Jani Nikula wrote:
>>> On Tue, 06 Sep 2016, Chris Wilson wrote:
>>>
On Tue, 06 Sep 2016, Zhenyu Wang wrote:
> [ Unknown signature status ]
> On 2016.09.06 11:33:57 +0300, Jani Nikula wrote:
>> On Tue, 06 Sep 2016, Chris Wilson wrote:
>> > On Tue, Sep 06, 2016 at 12:04:10PM +0800, Zhenyu Wang wrote:
>> >> Hi,
>>
On Fri, Aug 26, 2016 at 1:50 PM, Dave Gordon wrote:
> We had only DRM_INFO() and DRM_ERROR(), whereas the underlying printk()
> provides several other useful intermediate levels such as NOTICE and
> WARNING. So this patch fills out the set by providing simple macros for
Chris Wilson writes:
> On Tue, Sep 06, 2016 at 02:11:47PM +0300, Mika Kuoppala wrote:
>>
>> Resending my r-b...
>
> It's not enough, even retrying the reset a few times, we still
> eventually get a timeout.
>
> This is just the usual
> 10: 0x
> 20:
Hi,
On 06/09/16 11:43, akash.g...@intel.com wrote:
From: Akash Goel
This patch provides a test utility which helps capture GuC firmware logs and
then dump them to file.
The logs are pulled from a debugfs file '/sys/kernel/debug/dri/guc_log' and
stored into a file
On Tue, Sep 06, 2016 at 12:37:33PM +0100, Dave Gordon wrote:
> On 22/08/16 09:03, Chris Wilson wrote:
> >The obj->dirty bit is a companion to the obj->active bits that were
> >moved to the obj->flags bitmask. Since we also update this bit inside
> >the i915_vma_move_to_active() hotpath, we can
On Wed, Aug 24, 2016 at 12:22:57AM -0700, Dhinakaran Pandiyan wrote:
> Storing the port enum in intel_encoder makes it convenient to know the
> port attached to an encoder. Moving the port information up from
> intel_digital_port to intel_encoder avoids unecessary intel_digital_port
> access and
On 22/08/16 09:03, Chris Wilson wrote:
The obj->dirty bit is a companion to the obj->active bits that were
moved to the obj->flags bitmask. Since we also update this bit inside
the i915_vma_move_to_active() hotpath, we can aide gcc by also moving
the obj->dirty bit to obj->flags bitmask.
On Tue, Sep 06, 2016 at 02:11:47PM +0300, Mika Kuoppala wrote:
>
> Resending my r-b...
It's not enough, even retrying the reset a few times, we still
eventually get a timeout.
This is just the usual
10: 0x
20: goto 10
hanging batch
>
> Chris Wilson
Imre Deak writes:
> From: Ben Widawsky
>
> Consolidate the instdone logic so we can get a bit fancier. This patch also
> removes the duplicated print of INSTDONE[0].
>
> v2: (Imre)
> - Rebased on top of hangcheck INSTDONE changes.
> - Move all
Resending my r-b...
Chris Wilson writes:
> If at first we don't succeed, try again.
>
> Running the reset and recovery routines in a loop ends in a "reset
> request timeout" with a mtbf of an hour on Braswell. This is eerily
> similar to the unrecoverable reset
== Series Details ==
Series: add the generic H.264 decoder settings controls
URL : https://patchwork.freedesktop.org/series/12045/
State : warning
== Summary ==
Series 12045v1 add the generic H.264 decoder settings controls
From: Akash Goel
This patch provides a test utility which helps capture GuC firmware logs and
then dump them to file.
The logs are pulled from a debugfs file '/sys/kernel/debug/dri/guc_log' and
stored into a file '/tmp/guc_log_dump.dat', the name of the output file can
be
On 04/09/16 19:58, Nicolas Iooss wrote:
When building the kernel with clang and some warning flags, the compiler
reports that the return value of dcs_get_backlight() may be
uninitialized:
drivers/gpu/drm/i915/intel_dsi_dcs_backlight.c:53:2: error: variable
'data' is used uninitialized
== Series Details ==
Series: series starting with [1/1] drm/i915/dsi: silence a warning about
uninitialized return value
URL : https://patchwork.freedesktop.org/series/12044/
State : failure
== Summary ==
Series 12044v1 Series without cover letter
On 2016-09-06 11:25, Jani Nikula wrote:
On Tue, 06 Sep 2016, li...@eikelenboom.it wrote:
L.S.,
Since one of the last 4.8 RC's i'm getting the warning below when
booting on my sandybridge based thinkpad.
From what it seems the machine still works fine though.
What does 'lspci -nns 2' say for
On Fri, 2016-09-02 at 11:05 +0300, Mika Kahola wrote:
> +1 for this cleanup
>
> Reviewed-by: Mika Kahola
Received couple of compiler warnings to be cleaned up
drivers/gpu/drm/i915/intel_dp_link_training.c: In function
‘intel_dp_link_max_vswing_reached’:
The generic decoder settings for H.264. It is modified from
the VA-API. Adding the extra data required by the Rockchip.
Signed-off-by: Randy Li
---
include/uapi/linux/videodev2.h | 103 +
1 file changed, 103 insertions(+)
diff --git
This is not done yet. The rockchip VA-API driver[1] still need a third part
library to pre-parse the nalu data. Maybe after the third part library
free version[2] had done, it would be clear that we else filed we may need.
Those structures comes from VA-API SPCE. But still not enough to driver
a
These two controls would be used to set the H.264 codec settings
for decoder.
Signed-off-by: Randy Li
---
drivers/media/v4l2-core/v4l2-ctrls.c | 2 ++
include/uapi/linux/v4l2-controls.h | 2 ++
2 files changed, 4 insertions(+)
diff --git
When building the kernel with clang and some warning flags, the compiler
reports that the return value of dcs_get_backlight() may be
uninitialized:
drivers/gpu/drm/i915/intel_dsi_dcs_backlight.c:53:2: error: variable
'data' is used uninitialized whenever 'for' loop exits because its
Chris Wilson writes:
> Let's avoid mixing sealing the hardware commands for the request and
> adding the request to the software tracking.
>
> Signed-off-by: Chris Wilson
Reviewed-by: Mika Kuoppala
> ---
>
Chris Wilson writes:
> Since we have a cooperative mode now with a direct reset, we can avoid
> the contention on struct_mutex and instead try then sleep on the
> I915_RESET_IN_PROGRESS bit. If the mutex is held and that bit is
> cleared, all is fine. Otherwise, we
Signed-off-by: Juha-Pekka Heikkila
---
tests/gem_pipe_control_store_loop.c | 2 +-
tests/kms_cursor_crc.c | 3 ++-
tools/intel_error_decode.c | 2 +-
tools/intel_stepping.c | 40 ++---
Usage of these was removed on earlier patch.
Signed-off-by: Juha-Pekka Heikkila
---
lib/intel_chipset.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/intel_chipset.h b/lib/intel_chipset.h
index 1003266..ff903bf 100644
--- a/lib/intel_chipset.h
+++
> Meelis, what does 'lspci -nns 2' say for you?
00:02.0 VGA compatible controller [0300]: Intel Corporation 2nd Generation Core
Processor Family Integrated Graphics Controller [8086:0102] (rev 09)
>
> BR,
> Jani.
>
> On Tue, 06 Sep 2016, Jani Nikula wrote:
> >
I had interest to take out PCIIDs from intel_chipset.h in favor of
i915_pciids.h, maybe some seasoned igt veteran can say what's the correct way.
There are lot of magic numbers here and there but I'm not happy adding more
magics into the code. Still seems most of these defines were used just once
On Tue, 06 Sep 2016, li...@eikelenboom.it wrote:
> L.S.,
>
> Since one of the last 4.8 RC's i'm getting the warning below when
> booting on my sandybridge based thinkpad.
> From what it seems the machine still works fine though.
What does 'lspci -nns 2' say for you?
BR,
Jani.
>
> --
> Sander
Chris Wilson writes:
> In the next patch we want to handle reset directly by a locked waiter in
> order to avoid issues with returning before the reset is handled. To
> handle the reset, we must first know whether we hold the struct_mutex.
> If we do not hold the
Meelis, what does 'lspci -nns 2' say for you?
BR,
Jani.
On Tue, 06 Sep 2016, Jani Nikula wrote:
> Lyude, this is due to
>
> commit 87660502f1a4d51fb043e89a45d30c9917787c22
> Author: Lyude
> Date: Wed Aug 17 15:55:53 2016 -0400
>
>
Lyude, this is due to
commit 87660502f1a4d51fb043e89a45d30c9917787c22
Author: Lyude
Date: Wed Aug 17 15:55:53 2016 -0400
drm/i915/gen6+: Interpret mailbox error flags
and on its way to stable.
BR,
Jani.
On Mon, 29 Aug 2016, Meelis Roos wrote:
>
Lyude, this is due to
commit 87660502f1a4d51fb043e89a45d30c9917787c22
Author: Lyude
Date: Wed Aug 17 15:55:53 2016 -0400
drm/i915/gen6+: Interpret mailbox error flags
and on its way to stable.
BR,
Jani.
On Tue, 06 Sep 2016, li...@eikelenboom.it wrote:
> L.S.,
>
>
On 2016.09.06 11:33:57 +0300, Jani Nikula wrote:
> On Tue, 06 Sep 2016, Chris Wilson wrote:
> > On Tue, Sep 06, 2016 at 12:04:10PM +0800, Zhenyu Wang wrote:
> >> Hi,
> >>
> >> Here're two patches for GVT-g guest to fix run against future GVT-g
> >> host driver, which
On Tue, 06 Sep 2016, Chris Wilson wrote:
> On Tue, Sep 06, 2016 at 12:04:10PM +0800, Zhenyu Wang wrote:
>> Hi,
>>
>> Here're two patches for GVT-g guest to fix run against future GVT-g
>> host driver, which try to ensure 4.8 will be ready to use for VM.
>>
>> thanks.
== Series Details ==
Series: drm/i915: Don't wait for a spinlock inside error capture
URL : https://patchwork.freedesktop.org/series/12033/
State : success
== Summary ==
Series 12033v1 drm/i915: Don't wait for a spinlock inside error capture
L.S.,
Since one of the last 4.8 RC's i'm getting the warning below when
booting on my sandybridge based thinkpad.
From what it seems the machine still works fine though.
--
Sander
[2.846265] [ cut here ]
[2.846280] WARNING: CPU: 0 PID: 4 at
On Tue, Sep 06, 2016 at 10:40:17AM +0300, Joonas Lahtinen wrote:
> On ti, 2016-09-06 at 07:19 +0100, Chris Wilson wrote:
> > @@ -3725,7 +3723,6 @@ int intel_freq_opcode(struct drm_i915_private
> > *dev_priv, int val);
> > * act upon the intermediate value, possibly leading to corruption and
> >
On ti, 2016-09-06 at 12:04 +0800, Zhenyu Wang wrote:
> From: Zhi Wang
>
> Disable 48bit full PPGTT on vGPU too for now.
>
> Signed-off-by: Zhi Wang
> Signed-off-by: Zhenyu Wang
Reviewed-by: Joonas Lahtinen
On ti, 2016-09-06 at 12:04 +0800, Zhenyu Wang wrote:
> From: Ping Gao
>
> vGPU capability is handled by GVT-g host driver, not needed to
> put extra HW check for vGPU detection. And we'll actually support
> vGPU from BDW.
>
> Signed-off-by: Ping Gao
On ti, 2016-09-06 at 08:38 +0100, Chris Wilson wrote:
> If we can't grab the breadcrumb's spinlock, possibly due to a driver
> deadlock inside the waiters, ignore them. Like hangcheck, error
> capturing must work no matter how the driver/GPU dies.
>
> Signed-off-by: Chris Wilson
On ti, 2016-09-06 at 07:19 +0100, Chris Wilson wrote:
> @@ -3725,7 +3723,6 @@ int intel_freq_opcode(struct drm_i915_private
> *dev_priv, int val);
> * act upon the intermediate value, possibly leading to corruption and
> * machine death. You have been warned.
> */
I'd update the comment,
If we can't grab the breadcrumb's spinlock, possibly due to a driver
deadlock inside the waiters, ignore them. Like hangcheck, error
capturing must work no matter how the driver/GPU dies.
Signed-off-by: Chris Wilson
---
drivers/gpu/drm/i915/i915_gpu_error.c | 25
On Mon, 05 Sep 2016, Tvrtko Ursulin wrote:
> On 05/09/16 16:06, Tvrtko Ursulin wrote:
>> On 12/08/16 13:06, Dave Gordon wrote:
>>> On 11/08/16 18:35, Patchwork wrote:
== Series Details ==
Series: Reclassify messages from GuC loader/submission (rev3)
== Series Details ==
Series: drm/i915: Remove 64b mmio write vfuncs
URL : https://patchwork.freedesktop.org/series/12029/
State : failure
== Summary ==
Series 12029v1 drm/i915: Remove 64b mmio write vfuncs
http://patchwork.freedesktop.org/api/1.0/series/12029/revisions/1/mbox/
Test
On Tue, Sep 06, 2016 at 12:04:10PM +0800, Zhenyu Wang wrote:
> Hi,
>
> Here're two patches for GVT-g guest to fix run against future GVT-g
> host driver, which try to ensure 4.8 will be ready to use for VM.
>
> thanks.
>
> Ping Gao (1):
> drm/i915: enable vGPU detection for all
>
> Zhi Wang
On Tue, Sep 06, 2016 at 10:54:14AM +0530, Praveen Paneri wrote:
> Decoupled MMIO is an alternative way to access forcewake domain
> registers, which requires less cycles and avoids frequent software
> forcewake.
How about when forcewake is already held? You'll note that we still
require
We don't have safe 64-bit mmio writes as they are really split into
2x32-bit writes. This tearing is dangerous as the hardware *will*
operate on the intermediate value, requiring great care when assigning.
(See, for example, i965_write_fence_reg.) As such we don't currently use
them and strongly
86 matches
Mail list logo