On 13/03/17 01:56, Oscar Mateo wrote:
On 03/13/2017 04:46 AM, Chris Wilson wrote:
On Fri, Mar 10, 2017 at 08:28:55AM -0800, Oscar Mateo wrote:
Doorbell release flow requires that we wait for GEN8_DRB_VALID bit to go
to zero after updating db_status before we call the GuC to release the
2017년 03월 14일 22:28에 Daniel Vetter 이(가) 쓴 글:
> On Mon, Mar 13, 2017 at 03:18:05PM -0400, Sean Paul wrote:
>> On Wed, Mar 08, 2017 at 03:12:53PM +0100, Daniel Vetter wrote:
>>> Again no apparent explanation for the split except hysterical raisins.
>>> Merging them also makes it a bit more obviuos
According to BSpec, "The CD clock frequency must be at least twice the
frequency of the Azalia BCLK." and BCLK is configured to 96 MHz by
default. This check is needed because BXT and GLK support cdclk
frequencies less than 192 MHz.
v2: Include other Gen9 platforms too for completeness.(Paulo)
If link training at a link rate optimal for a particular
mode fails during modeset's atomic commit phase, then we
let the modeset complete and then retry. We save the link rate
value at which link training failed, update the link status property
to "BAD" and use a lower link rate to prune the
On Tue, Mar 14, 2017 at 04:31:58PM +, Tvrtko Ursulin wrote:
>
> On 14/03/2017 09:34, Chris Wilson wrote:
> >It turns out that we may want to restore the original
> >engine->submit_request (and engine->schedule) callbacks from more than
> >just the guc <-> execlists transition. Move this to a
On Tue, 2017-03-14 at 17:47 -0300, Paulo Zanoni wrote:
> Em Ter, 2017-03-07 às 16:12 -0800, Dhinakaran Pandiyan escreveu:
> > According to BSpec, "The CD clock frequency must be at least twice
> > the
> > frequency of the Azalia BCLK." and BCLK is configured to 96 MHz by
> > default. This check is
From: Michal Wajdeczko
Manual pointer manipulation is error prone. Let compiler calculate
right offsets for us in case we need to change ads layout.
v2: don't call it object (Chris)
v3: restyle offset assignments (Chris)
v4: stylistic reductions
Signed-off-by:
Em Ter, 2017-03-07 às 16:12 -0800, Dhinakaran Pandiyan escreveu:
> According to BSpec, "The CD clock frequency must be at least twice
> the
> frequency of the Azalia BCLK." and BCLK is configured to 96 MHz by
> default. This check is needed because BXT and GLK support cdclk
> frequencies less than
Em Ter, 2017-03-07 às 16:12 -0800, Dhinakaran Pandiyan escreveu:
> Implement the DP-Audio cdclk restriction for GLK, similar to what is
> implemented for BDW and other GEN9 platforms. The max. pixel clock
> adjustment for GLK, however factors in the 2 pixels per clock output
> that
> GLK
== Series Details ==
Series: series starting with [1/2] drm/i915: Extract intel_wm_plane_visible()
URL : https://patchwork.freedesktop.org/series/21226/
State : success
== Summary ==
Series 21226v1 Series without cover letter
From: "Navare, Manasi D"
Display stream compression is supported on DP 1.4 DP
devices. This patch adds the corersponding DPCD
register definitions for DSC.
v3:
* Add some SHIFTS and MASKS for uniformity (Jani Nikula)
v2:
* Rebased on drm-tip
Signed-off-by: Manasi
== Series Details ==
Series: GuC Scrub vol. 1 (rev13)
URL : https://patchwork.freedesktop.org/series/16856/
State : success
== Summary ==
Series 16856v13 GuC Scrub vol. 1
https://patchwork.freedesktop.org/api/1.0/series/16856/revisions/13/mbox/
Test gem_exec_suspend:
Subgroup
== Series Details ==
Series: drm/i915/guc: Use formalized struct definition for ads object (rev3)
URL : https://patchwork.freedesktop.org/series/20826/
State : success
== Summary ==
Series 20826v3 drm/i915/guc: Use formalized struct definition for ads object
From: Ville Syrjälä
Add a command for extracting various tags (eg. Reviwed-by:) from
emails. You can give the comamnd a rangeish to add the tags from
the same email to multiple already applied patches.
The regexp used to pick up tags is purposefully quite broad.
From: Ville Syrjälä
Add the "add-link" command so that you can add the Link: tag to
patches that failed to apply directly.
Signed-off-by: Ville Syrjälä
---
dim | 39 +++
1 file changed, 39
== Series Details ==
Series: series starting with [1/7] drm/i915: Move residency calculation into
intel_pm.c
URL : https://patchwork.freedesktop.org/series/21217/
State : warning
== Summary ==
Series 21217v1 Series without cover letter
On Mon, Mar 13, 2017 at 05:10:28PM +0100, Maarten Lankhorst wrote:
> As a proof of concept, first try to convert intel_tv, which is a rarely
> used connector. It has 5 properties, tv format and 4 margins.
>
> I'm less certain about the state behavior itself, should we pass a size
> parameter to
From: Ville Syrjälä
We currently hold a vblank referenced while trying to evade the
vblank, but we drop it as soon as we've done that. After all the
planes have been committed we are quite likely to grab a new vblank
reference for delivering the flip event. This
== Series Details ==
Series: drm/i915: Replace irq_seqno_barrier on hws write with a clflush (rev2)
URL : https://patchwork.freedesktop.org/series/21210/
State : failure
== Summary ==
Series 21210v2 drm/i915: Replace irq_seqno_barrier on hws write with a clflush
From: Ville Syrjälä
VLV is a sloth and so doing anything extra between the first and last
get_vblank() is likely to increase the chance of failing the test.
Let's move the make_busy() stuff to happen before the first get_vblank()
to reduce the amount of work done
On 3/13/2017 3:17 PM, Chris Wilson wrote:
On Mon, Mar 13, 2017 at 10:28:34AM +0530, Kamble, Sagar A wrote:
On 3/12/2017 6:29 PM, Chris Wilson wrote:
On Sat, Mar 11, 2017 at 04:17:34AM -, Patchwork wrote:
== Series Details ==
Series: series starting with [1/3] drm/i915/guc: Release GuC
On 3/14/2017 10:18 AM, Chris Wilson wrote:
We take the runtime pm wakelock during i915_handle_error() to ensure
that all paths that reach the error capture keep the device awake during
the hw reads. However, we need to extend that from the reset handler to
include the earlier capture routines.
We take the runtime pm wakelock during i915_handle_error() to ensure
that all paths that reach the error capture keep the device awake during
the hw reads. However, we need to extend that from the reset handler to
include the earlier capture routines.
Reported-by: Antonio Argenziano
On Tue, Mar 14, 2017 at 10:05:08AM -0700, Michel Thierry wrote:
> On 3/14/2017 2:09 AM, Chris Wilson wrote:
> >On Mon, Mar 13, 2017 at 05:26:06PM -0700, Michel Thierry wrote:
> >>At least in bxt (with decoupled mmio), it prevents a warning while
> >>capturing the reg state:
> >>
> >> [ ] WARNING:
On 3/14/2017 2:09 AM, Chris Wilson wrote:
On Mon, Mar 13, 2017 at 05:26:06PM -0700, Michel Thierry wrote:
At least in bxt (with decoupled mmio), it prevents a warning while
capturing the reg state:
[ ] WARNING: assert_rpm_wakelock_held.part.4+0x1e/0x20 [i915]
[ ] RPM wakelock ref not held
Hi Daniel,
I investigated the CI failures because of this patch. These
were essentially the failures which were always there but hidden because
they used be DEBUG_KMS messages for link failures so never got caught by CI.
But now my patch actually throws DRM_ERROR if the link training fails at RBR
On 13/03/17 17:26, Michel Thierry wrote:
At least in bxt (with decoupled mmio), it prevents a warning while
capturing the reg state:
[ ] WARNING: assert_rpm_wakelock_held.part.4+0x1e/0x20 [i915]
[ ] RPM wakelock ref not held during HW access
[ ] Call Trace:
[ ] dump_stack+0x63/0x87
On 14/03/2017 09:34, Chris Wilson wrote:
It turns out that we may want to restore the original
engine->submit_request (and engine->schedule) callbacks from more than
just the guc <-> execlists transition. Move this to a vfunc so we can
have a common interface.
Signed-off-by: Chris Wilson
== Series Details ==
Series: series starting with [v3,1/2] drm/i915: Move engine->submit_request
selection to a vfunc
URL : https://patchwork.freedesktop.org/series/21203/
State : success
== Summary ==
Series 21203v1 Series without cover letter
kasan is very good at detecting use-after-free. However, our requests
are allocated from a rcu-typesafe slab that is important for performance
but makes kasan less effective. When enabling kasan we are intentionally
looking for memory errors, disable the use of our caches to improve the
likelihood
This is what I currently use for cherry-picking fixes. Completely
non-interactive and stateless.
Signed-off-by: Jani Nikula
---
dim | 93 +
1 file changed, 61 insertions(+), 32 deletions(-)
diff --git
Signed-off-by: Jani Nikula
---
dim | 37 ++---
1 file changed, 30 insertions(+), 7 deletions(-)
diff --git a/dim b/dim
index 6d3b9734b348..621f60471697 100755
--- a/dim
+++ b/dim
@@ -660,22 +660,44 @@ function dim_apply_next_fixes
On Tue, Mar 14, 2017 at 10:12:59AM +0100, Daniel Vetter wrote:
> On Mon, Mar 13, 2017 at 12:42:54PM -0400, Sean Paul wrote:
> > On Wed, Mar 08, 2017 at 03:12:35PM +0100, Daniel Vetter wrote:
> > > +/**
> > > + * struct drm_prime_file_private - per-file tracking for PRIME
> > > + *
> > > + * This
We need to ensure that we always serialize updates to the bottom-half
using the breadcrumbs.irq_lock so that we don't race with a concurrent
interrupt handler. This is most important just prior to leaving the
waiter (when the intel_wait will be overwritten), so make sure we are
not the current
On Tue, Mar 14, 2017 at 03:28:14PM +0100, Arkadiusz Hiler wrote:
> `guc_firmware_path` and `huc_firmware_path` module parameters are added.
>
> Using the parameter disables version checks and loads desired firmware
> instead of the default one.
>
> v2: make params unsafe && notice about disabled
From: Ville Syrjälä
All platforms that lack double buffered watermarks will need to
handle the legacy cursor updates in the same way. So let's extract the
logic to determine the plane visibility into a small helper. For
simplicity we'll make the function DTRT for
From: Ville Syrjälä
Use intel_wm_plane_visible() to determine cursor visibility for SKL+
also. Previously SKL+ would check the actual visibility which now
conflicts with the assumptions in intel_legacy_cursor_update().
We also change SKL+ to compute the cursor
Let intel_guc_init_fw() focus on determining and fetching the correct
firmware.
This patch introduces intel_uc_sanitize_options() that is called from
intel_sanitize_options().
Then, if we have GuC, we can call intel_guc_init_fw() conditionally
and we do not have to do the internal checks.
v2:
`guc_firmware_path` and `huc_firmware_path` module parameters are added.
Using the parameter disables version checks and loads desired firmware
instead of the default one.
v2: make params unsafe && notice about disabled fw check (J. Lahtinen)
Cc: Chris Wilson
Cc:
Current version of intel_guc_init_hw() does a lot:
- cares about submission
- loads huc
- implement WA
This change offloads some of the logic to intel_uc_init_hw(), which now
cares about the above.
v2: rename guc_hw_reset and fix typo in define name (M. Wajdeczko)
v3: rename once again
v4:
intel_{h,g}uc_init_fw selects correct firmware and then triggers it's
preparation (fetch + initial parsing).
This change separates out select steps, so those can be called by
the sanitize_options().
Then, during the init_fw(), we prepare the firmware if the firmware was
selected.
Cc: Michal
Currently fw->path values can represent one of three possible states:
1) NULL - device without the uC
2) '\0' - device with the uC but have no firmware
3) else - device with the uC and we have firmware
Second case is used only to WARN at a later stage.
We can WARN right away and merge cases
Used to obtain "dev_priv" from huc struct pointer.
We already have similar thing for guc.
Cc: Michal Wajdeczko
Signed-off-by: Arkadiusz Hiler
Reviewed-by: Michal Wajdeczko
---
drivers/gpu/drm/i915/i915_drv.h |
The file fits better.
Additionally rename it to intel_uc_prepare_fw(), as the function does
more than simple fetch.
`obj` cleanup in the function is also fixed (i.e. removed). In the fail
scenario it was always 'put' but there's no possible flow that
initializes the obj properly and then goes to
GuC historically has two "startup" functions called _init() and _setup()
Then HuC came with it's _init() and _load().
This commit renames intel_guc_setup() and intel_huc_load() to
*uc_init_hw() as they called from the i915_gem_init_hw().
The aim is to be consistent in that entry points called
Instead of calling intel_guc_init() and intel_huc_init() one by one this
patch introduces intel_uc_init_fw() function that calls them both.
Called functions are renamed accordingly.
Trying to have subject_verb_object ordering and more descriptive names,
the intel_huc_init() and intel_guc_init()
---
drivers/gpu/drm/i915/i915_params.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_params.c
b/drivers/gpu/drm/i915/i915_params.c
index b6a7e36..9dcc8a0 100644
--- a/drivers/gpu/drm/i915/i915_params.c
+++ b/drivers/gpu/drm/i915/i915_params.c
Externs are implicit and we generally try to avoid them.
Cc: Michal Wajdeczko
Signed-off-by: Arkadiusz Hiler
Reviewed-by: Joonas Lahtinen
---
drivers/gpu/drm/i915/intel_uc.h | 12 ++--
1 file
Reasoning
=
General GuC/HuC cleanup simplifying logic and moving chunks around as the area
got pretty rusty.
This is the first part of effort to clean it up.
A lot of logic were extracted from intel_guc_load() to other functions - it did
not only handle the actual loading but had WA
On Fri, 2017-03-10 at 12:27 +0200, Ander Conselvan De Oliveira wrote:
> On Fri, 2017-03-10 at 12:18 +0200, Ander Conselvan de Oliveira wrote:
> > Some of the kms_cursor_crc subtests where failing on Geminilake. The
> > root cause was an error on programming the pre-CSC gamma tables, which
> > led
On Fri, 2017-03-10 at 13:18 +, Patchwork wrote:
> == Series Details ==
>
> Series: drm/i915/glk: Improve rounding caused by pre-CSC gamma tables
> URL : https://patchwork.freedesktop.org/series/21049/
> State : success
Pushed. Thanks for reviewing.
Ander
>
> == Summary ==
>
> Series
Hi,
On 14.03.2017 12:48, Eero Tamminen wrote:
On 11.03.2017 03:14, Kenneth Graunke wrote:
On systems without LLC, drm_intel_gem_bo_map_unsynchronized() has
had the surprising behavior of doing a synchronized GTT mapping.
This is obviously not what the user of the API wanted.
Eric left a
On Tue, Mar 14, 2017 at 03:17:26PM +0200, Mika Kuoppala wrote:
> Use intel_rc6_residency to get benefit for increased resolution
> in byt/chv.
>
> Signed-off-by: Mika Kuoppala
I was thinking what's the point, but
Reviewed-by: Chris Wilson
On Tue, Mar 14, 2017 at 03:17:24PM +0200, Mika Kuoppala wrote:
> Change the granularity from milliseconds to microseconds
> when returning rc6 residencies. This is in preparation
> for increased resolution on some platforms.
>
> Signed-off-by: Mika Kuoppala
> ---
>
On Tue, Mar 14, 2017 at 03:17:25PM +0200, Mika Kuoppala wrote:
> The high counter value bit can be used to get 8 bits more
> of range out of the same residency counter registers.
Please do note that it is internally a 40bit register with a 32bit
window (and a similar comment in code).
> Lets
On Mon, Mar 13, 2017 at 03:29:20PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:55PM +0100, Daniel Vetter wrote:
> > With all drivers converted there's only legacy dri1 drivers using it.
> > Not going to touch those, instead just hide it like we've done with
> > other dri1 driver hooks
On Tue, Mar 14, 2017 at 10:21:09PM +0900, Sergey Senozhatsky wrote:
> Hello,
>
> [ 530.698622] ==
> [ 530.698623] WARNING: possible circular locking dependency detected
> [ 530.698626] 4.11.0-rc2-mm1-dbg-00167-gdb8a9941614c-dirty #222 Not
On Tue, Mar 14, 2017 at 01:30:40PM +, Chris Wilson wrote:
> On Tue, Mar 14, 2017 at 03:17:27PM +0200, Mika Kuoppala wrote:
> > Avoid more costly punit access and use the local cpu clock.
> > The time diff between separate processor units is irrelevant in
> > our rc0 residency granularity so we
On Mon, 2017-03-13 at 16:54 +0530, Shashank Sharma wrote:
> Geminilake platform sports a native HDMI 2.0 controller, and is
> capable of driving pixel-clocks upto 594Mhz. HDMI 2.0 spec
> mendates scrambling for these higher clocks, for reduced RF footprint.
>
> This patch checks if the monitor
On Wed, Mar 08, 2017 at 04:45:13PM +0100, Daniel Vetter wrote:
> On Wed, Mar 08, 2017 at 03:07:48PM +, Chris Wilson wrote:
> > On Wed, Mar 08, 2017 at 03:12:45PM +0100, Daniel Vetter wrote:
> > > There's really not a reason afaics that we can't just clean up
> > > everything at the end, in the
Chris Wilson writes:
> On Tue, Mar 14, 2017 at 03:17:29PM +0200, Mika Kuoppala wrote:
>> After ei reset, upclock as default if we don't have a previous
>> timestamp at hand. We might at sometimes waste one interval
>> of more power but also be more agile if we need to
On Tue, Mar 14, 2017 at 03:17:29PM +0200, Mika Kuoppala wrote:
> After ei reset, upclock as default if we don't have a previous
> timestamp at hand. We might at sometimes waste one interval
> of more power but also be more agile if we need to ramp up.
Why? There is nothing special about the first
Manual pointer manipulation is error prone. Let compiler calculate
right offsets for us in case we need to change ads layout.
v2: don't call it object (Chris)
v3: restyle offset assignments (Chris)
Signed-off-by: Michal Wajdeczko
Cc: Oscar Mateo
On Mon, Mar 13, 2017 at 03:35:43PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:57PM +0100, Daniel Vetter wrote:
> > Sadly there's only 1 driver which can use it, everyone else is special
> > for some reason:
> >
> > - gma500 has a horrible runtime PM ioctl wrapper that probably
On Tue, Mar 14, 2017 at 03:17:27PM +0200, Mika Kuoppala wrote:
> Avoid more costly punit access and use the local cpu clock.
> The time diff between separate processor units is irrelevant in
> our rc0 residency granularity so we can ignore it.
>
> Cc: Chris Wilson
> Cc:
On Mon, Mar 13, 2017 at 03:18:05PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:53PM +0100, Daniel Vetter wrote:
> > Again no apparent explanation for the split except hysterical raisins.
> > Merging them also makes it a bit more obviuos what's going on wrt the
> > runtime pm
On Mon, Mar 13, 2017 at 03:11:05PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:50PM +0100, Daniel Vetter wrote:
> > I didn't spot anything that would require ordering here (well not
> > anywhere else either), and I'm trying to unify at least modern drivers
> > on one close hook.
> >
>
On Mon, Mar 13, 2017 at 01:53:28PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:44PM +0100, Daniel Vetter wrote:
> > Well, mostly drm_file.h, and clean up all related things:
> >
> > - I didnt' figure out the difference between preclose and postclose.
> > The existing explanation in
On Mon, Mar 13, 2017 at 01:53:28PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:44PM +0100, Daniel Vetter wrote:
> > Well, mostly drm_file.h, and clean up all related things:
> >
> > - I didnt' figure out the difference between preclose and postclose.
> > The existing explanation in
After ei reset, upclock as default if we don't have a previous
timestamp at hand. We might at sometimes waste one interval
of more power but also be more agile if we need to ramp up.
Signed-off-by: Mika Kuoppala
---
drivers/gpu/drm/i915/i915_irq.c | 2 ++
1 file
Change the granularity from milliseconds to microseconds
when returning rc6 residencies. This is in preparation
for increased resolution on some platforms.
Signed-off-by: Mika Kuoppala
---
drivers/gpu/drm/i915/i915_drv.h | 4 ++--
drivers/gpu/drm/i915/i915_sysfs.c |
Hello,
[ 530.698622] ==
[ 530.698623] WARNING: possible circular locking dependency detected
[ 530.698626] 4.11.0-rc2-mm1-dbg-00167-gdb8a9941614c-dirty #222 Not tainted
[ 530.698627] --
[
On Mon, Mar 13, 2017 at 01:05:27PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:43PM +0100, Daniel Vetter wrote:
> > We might as well dump the drm_file pointer, that's about as useful
> > a cookie as the pid. Noticed while typing docs for drm_file and friends.
> >
> > Since the only
The high counter value bit can be used to get 8 bits more
of range out of the same residency counter registers.
Lets toggle this bit on and off on vlv/chv while reading the
counters to push the wrap from 13 seconds to 54 minutes.
Reported-by: Len Brown
Cc: Chris Wilson
Plan is to make generic residency calculation utility
function for usage outside of sysfs. As a first step
move residency calculation into intel_pm.c
Signed-off-by: Mika Kuoppala
---
drivers/gpu/drm/i915/i915_drv.h | 2 ++
drivers/gpu/drm/i915/i915_sysfs.c | 27
Set byt rc residency counters high level as chv does by
default. We lose some accuracy on byt but we can do the calculation
without extra hw read on both platforms, as now they behave
identically in this respect.
Cc: Chris Wilson
Cc: Ville Syrjälä
Avoid more costly punit access and use the local cpu clock.
The time diff between separate processor units is irrelevant in
our rc0 residency granularity so we can ignore it.
Cc: Chris Wilson
Cc: Ville Syrjälä
Signed-off-by: Mika Kuoppala
Use intel_rc6_residency to get benefit for increased resolution
in byt/chv.
Signed-off-by: Mika Kuoppala
---
drivers/gpu/drm/i915/i915_debugfs.c | 24
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git
On ma, 2017-03-13 at 14:15 +0100, Arkadiusz Hiler wrote:
> `guc_firmware_path` and `huc_firmware_path` module parameters are added.
>
> Using the parameter disables version checks and loads desired firmware
> instead of the default one.
>
> v2: make params unsafe && notice about disabled fw
The rcu_barrier() takes the cpu_hotplug mutex which itself is not
reclaim-safe, and so rcu_barrier() is illegal from inside the shrinker.
[ 309.661373] =
[ 309.661376] [ INFO: possible irq lock inversion dependency detected ]
[
On Tue, Mar 14, 2017 at 11:38:59AM +, Chris Wilson wrote:
> When manually overwriting the HWS, rather than assume irq_seqno_barrier
> does the right thing, we can explicitly flush the cacheline instead.
> This avoids us calling the engine->irq_seqno_barrier() from an illegal
> context:
Drat,
When manually overwriting the HWS, rather than assume irq_seqno_barrier
does the right thing, we can explicitly flush the cacheline instead.
This avoids us calling the engine->irq_seqno_barrier() from an illegal
context:
[ 1472.651797] BUG: scheduling while atomic: migration/0/11/0x0002
[
On Tue 2017-03-14 10:08:23, Thorsten Leemhuis wrote:
> On 06.03.2017 00:01, Pavel Machek wrote:
> >>> mplayer stopped working after a while. Dmesg says:
> >>>
> >>> [ 3000.266533] cdc_ether 2-1.2:1.0 usb0: register 'cdc_ether' at
> > Now I'm pretty sure it is a regression in v4.11-rc0. Any ideas
On Tue, 14 Mar 2017, Damian Dominik Martinez Dreyer wrote:
> I am writing regarding this kernel patch: https://git.kernel.org/pub/sc
> m/linux/kernel/git/stable/stable-queue.git/tree/releases/4.9.9/drm-
> i915-execlists-reset-ring-registers-upon-resume.patch .
That would be
When manually overwriting the HWS, rather than assume irq_seqno_barrier
does the right thing, we can explicitly flush the cacheline instead.
This avoids us calling the engine->irq_seqno_barrier() from an illegal
context:
[ 1472.651797] BUG: scheduling while atomic: migration/0/11/0x0002
[
On Tue, 14 Mar 2017, Jani Nikula wrote:
> Thank you for the backports. I'll likely send a pull request of the
> stuff I have in -fixes now, apply these, and let them simmer until next
> week's pull.
I sent the fixes pull request for -rc3, and pushed Chris' backports
after
Hi,
On 11.03.2017 03:14, Kenneth Graunke wrote:
On systems without LLC, drm_intel_gem_bo_map_unsynchronized() has
had the surprising behavior of doing a synchronized GTT mapping.
This is obviously not what the user of the API wanted.
Eric left a comment indicating a valid concern: if the CPU
On Wed, Mar 08, 2017 at 04:53:58PM +0200, Jani Nikula wrote:
> On Wed, 08 Mar 2017, ville.syrj...@linux.intel.com wrote:
> > From: Ville Syrjälä
> >
> > This reverts commit bb10d4ec3be4b069bfb61c60ca4f708f58f440f1.
> >
> > Since commit c8ebfad7a063 ("drm/i915:
On Mon, Mar 13, 2017 at 04:09:53PM -0700, Manasi Navare wrote:
> On Mon, Mar 13, 2017 at 10:53:23PM +0200, ville.syrj...@linux.intel.com wrote:
> > From: Ville Syrjälä
> >
> > Apparently some DP sinks are a little nuts and cause HPD to drop
> > intermittently
Hi Dave, I'm early this week, here are the drm/i915 fixes for -rc3. The
majority of them are actually Cc: stable, not bugs introduced this
cycle, and almost all of them also have Fixes: annotations.
BR,
Jani.
The following changes since commit 70647f9163aa4fc7090b0d6795d026ebe3897928:
Merge
On 06.03.2017 00:01, Pavel Machek wrote:
>>> mplayer stopped working after a while. Dmesg says:
>>>
>>> [ 3000.266533] cdc_ether 2-1.2:1.0 usb0: register 'cdc_ether' at
> Now I'm pretty sure it is a regression in v4.11-rc0. Any ideas what to
> try? Bisect will be slow and nasty :-(.
@Pavel,
It turns out that we may want to restore the original
engine->submit_request (and engine->schedule) callbacks from more than
just the guc <-> execlists transition. Move this to a vfunc so we can
have a common interface.
Signed-off-by: Chris Wilson
Cc: Tvrtko Ursulin
When we wedge the device, we override engine->submit_request with a nop
to ensure that all in-flight requests are marked in error. However, igt
would like to unwedge the device to test -EIO handling. This requires us
to flush those in-flight requests and restore the original
On Mon, Mar 13, 2017 at 12:42:54PM -0400, Sean Paul wrote:
> On Wed, Mar 08, 2017 at 03:12:35PM +0100, Daniel Vetter wrote:
> > +/**
> > + * struct drm_prime_file_private - per-file tracking for PRIME
> > + *
> > + * This just contains the internal dma_buf and handle caches for
> > each
> > + *
On Mon, Mar 13, 2017 at 09:30:41AM +0100, Peter Zijlstra wrote:
> On Mon, Mar 13, 2017 at 09:15:17AM +0100, Daniel Vetter wrote:
> > On Mon, Mar 13, 2017 at 09:01:57AM +0100, Peter Zijlstra wrote:
> > > On Sun, Mar 12, 2017 at 09:53:40PM +0100, Daniel Vetter wrote:
> > >
> > > > Peter/Ingo,
> > >
On Mon, 13 Mar 2017, Manasi Navare wrote:
> On Mon, Mar 13, 2017 at 11:55:31AM +0200, Jani Nikula wrote:
>> On Sat, 11 Mar 2017, Manasi Navare wrote:
>> > On Fri, Mar 10, 2017 at 03:27:58PM +0200, Jani Nikula wrote:
>> >> The main thing are
On Mon, Mar 13, 2017 at 05:26:06PM -0700, Michel Thierry wrote:
> At least in bxt (with decoupled mmio), it prevents a warning while
> capturing the reg state:
>
> [ ] WARNING: assert_rpm_wakelock_held.part.4+0x1e/0x20 [i915]
> [ ] RPM wakelock ref not held during HW access
> [ ] Call
That's funny. I have a MacBook Pro 12,1 from late 2015. Hibernate
failed for me in 4.9.6 through 4.9.8 (possibly earlier as well, I do
no recall) without the patch. The patch you reference fixed my problem
and apparently many others based on the bug reports:
On Mon, 13 Mar 2017, Chris Wilson wrote:
> On Mon, Mar 13, 2017 at 06:14:56PM +0200, Jani Nikula wrote:
>> 1f7b847d72c3 ("drm/i915: Disable engine->irq_tasklet around resets")
> Done.
>
>> 370a81fb89cb ("drm/i915: Remove unused function intel_ddi_get_link_dpll()")
>
On Mon, Mar 13, 2017 at 12:08:19PM +0100, Maarten Lankhorst wrote:
> Op 13-03-17 om 11:15 schreef Daniel Vetter:
> > On Thu, Mar 09, 2017 at 03:52:04PM +0100, Maarten Lankhorst wrote:
> >> Add a big fat warning in __intel_display_resume that the old state is
> >> invalid, and use the correct state
100 matches
Mail list logo