I had some replies to this mailing list yesterday, but received below
notification:
---
Delivery is delayed to these recipients or groups:
intel-gfx@lists.freedesktop.org
Subject: RE: [RFCv2 PATCH 00/14] gvt: Hacking i915 for GVT context requirement
This message hasn't been delivered yet.
Hi Joe,
Yes, first thing to try is to increase the tries.
Can you please point me to the bug and provide more details like platform,
monitor, cable.
Are you referring to the same issue as Oleksandr reported where a single link
dvi/hdmi cable didn’t work and dual link worked?
Regards,
Sonika
Hi Sonika,
A kernel bug report was opened against Ubuntu [0]. After a kernel
bisect, it was found that reverting the following commit resolved this bug:
commit 237ed86c693d8a8e4db476976aeb30df4deac74b
Author: Sonika Jindal
Date: Tue Sep 15 09:44:20 2015 +0530
On Wed, Feb 24, 2016 at 06:48:35PM +0100, Tore Anderson wrote:
> Hi Ville,
>
> > Well, just to check the details of your particular cable/dongle,
> > maybe you can post the dmesg with drm.debug=0xe with my branch?
> > Or at least the parts that refer to DP dual mode adaptors.
>
> Here:
If power well support is disabled via the i915.disable_power_well module
option we should never enable DC states. Currently we would enable DC
states even in this case during system suspend, where we need to disable
all power wells regardless of the disable_power_well option.
CC: Patrik Jakobsson
During system suspend we need to first disable power wells then
unitialize the display core. In case power well support is disabled we
did this in the wrong order, so fix this up.
Fixes: d314cd43 ("drm/i915: fix handling of the disable_power_well module
option")
CC: sta...@vger.kernel.org
CC:
We can simplify the conditions selecting the target DC state during
runtime by calculating the allowed DC states in advance during driver
loading. This also makes it easier to disable DC states depending on the
i915.disable_power_well module option, added in the next patch.
CC: Patrik Jakobsson
Disabling the DC states when it's already disabled is a valid scenario,
for example during HW state sanitization during driver loading and
resuming or when DC states are disabled via the i915.enable_dc or
disable_power_well option.
CC: Patrik Jakobsson
This patchset simplifies the way how we select the target DC state based
on the platform and module options. It also fixes one ordering problem I
noticed while going through the code, see the first patch.
Imre Deak (4):
drm/i915/skl: Fix power domain suspend sequence
drm/i915/gen9: Sanitize
Hi Ville,
> Well, just to check the details of your particular cable/dongle,
> maybe you can post the dmesg with drm.debug=0xe with my branch?
> Or at least the parts that refer to DP dual mode adaptors.
Here: http://filebin.net/gd91wnltky/dmesg.4.5.0-rc4-g32fa589.txt
I'm assuming the most
On Tue, Feb 23, 2016 at 06:46:26PM +0200, ville.syrj...@linux.intel.com wrote:
> From: Ville Syrjälä
>
> Try to detect the max TMDS clock limit for the DP++ adaptor (if any)
> and take it into account when checking the port clock.
>
> Note that as with the sink
From: Alex Dai
This version of GuC firmware fixes the engine reset issue where golden
context LRC address is treated as page index by mistake. It also fixes
the problem that scheduler stops submiting to one engine when the other
engine work queue is full.
Signed-off-by: Alex
Em Qua, 2016-02-24 às 11:24 +0100, Maarten Lankhorst escreveu:
> Currently we perform our own wait in post_plane_update,
> but the atomic core performs another one in wait_for_vblanks.
> This means that 2 vblanks are done when a fb is changed,
> which is a bit overkill.
>
> Merge them by creating
On 23.02.2016 21:36, Dave Gordon wrote:
On 23/02/16 14:39, Tvrtko Ursulin wrote:
On 23/02/16 14:03, Chris Wilson wrote:
On Tue, Feb 23, 2016 at 01:31:17PM +, Tvrtko Ursulin wrote:
On 23/02/16 13:06, Gabriel Feceoru wrote:
On 23.02.2016 13:05, Tvrtko Ursulin wrote:
Hi,
On
On Tue, Feb 23, 2016 at 08:44:49PM +0100, Tore Anderson wrote:
> Hi Ville,
>
> > "The monitor is connected with a DP+-to-HDMI cable"
> > This and some reading of the DP dual mode spec gave me another idea;
> > The DP->HDMI adaptor may simply be degrading the signal quality too
> > much. According
Minor cleanup, connector and connector_state are always non-NULL here.
Signed-off-by: Maarten Lankhorst
---
drivers/gpu/drm/drm_atomic_helper.c | 15 +--
1 file changed, 5 insertions(+), 10 deletions(-)
diff --git
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
>
> For flexible GEM context creation, we factor out __i915_gem_create_context
> as the core logic of creation a GEM context. After the refactor, it more
> likesa context creation servcie, which is able to create context by
>
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
>
> This patchset is used to discuss and finalize the i915 changes required by
> GVT context. Previously we have discussed about how to hack i915 to meet
> GVT context requirement, and thanks for the idea and comments.
>
> In this
On 24/02/16 13:54, Maarten Lankhorst wrote:
When intel_update_watermarks is called on skylake it inspects
crtc->state, which should show as disabled. This wasn't the case,
and this resulted in a divide-by-zero in
skl_ddb_get_pipe_allocation_limits when intel_update_watermarks was called.
The recent commit [0bdf5a05647a: drm/i915: Add reverse mapping between
port and intel_encoder] introduced a reverse mapping to retrieve
intel_dig_port object from the port number. The code assumed that the
port vs intel_dig_port are 1:1 mapping. But in reality, this was a
too naive assumption.
The current check doesn't handle the case where we don't steal an
encoder, but keep it on the current connector. If we repurpose
disable_conflicting_encoders to do the checking, we just have
to reject the ones that conflict.
Signed-off-by: Maarten Lankhorst
Now that only encoders can be stolen that are part of the state
steal_encoder no longer needs to inspect all connectors,
just those that are part of the atomic state.
Signed-off-by: Maarten Lankhorst
---
drivers/gpu/drm/drm_atomic_helper.c | 11 ++-
1
After trying out various ways to handle encoder stealing better
I came up with a cleaner way.
The first patch cleans up update_output_state and only adds affected connectors.
This is required to determine which connectors are not part of the atomic state,
and allow disabling them if they have
With the addition of crtc_state->connector_mask other connectors from
different crtc's aren't needed any more to determine if a crtc has
connectors, so only call add_affected_connectors on the target crtc.
This allows a cleanup to first remove all current connectors, then
add all set->connectors
On Tue, 23 Feb 2016 20:09:02 +0100,
Martin Kepplinger wrote:
>
> Am 2016-02-23 um 18:14 schrieb Ville Syrjälä:
> > On Tue, Feb 23, 2016 at 05:57:40PM +0100, Takashi Iwai wrote:
> >> On Mon, 22 Feb 2016 22:37:28 +0100,
> >> Martin Kepplinger wrote:
> >>>
> >>> Am 2016-02-22 um 20:10 schrieb
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
>
> Previously the PDPs inside the ring context are updated at:
>
> - When populate a LRC context
> - Before submitting a LRC context (only for 32 bit PPGTT, as the amount
> of used PDPs may change)
>
> This patch postpones the
Hi Kevin:
Now our context switch is covered by context status change notification
handler. In the status change handler we will save render registers. As GVT
context will be a single submission context we will restore the render
registers when the GVT context is scheduled-out by HW.
>
The generic gpio is sequence is parsed from the VBT and the
GPIO table is updated with the North core, South core and
SUS core elements.
v2: Move changes in sideband.c file to new patch(Jani), rebase
v3: Moved the Macro`s to intel_dsi_panel_vbt.c (Jani)
v3 by Jani
- rebase on previous patches
-
Op 24-02-16 om 02:24 schreef Matt Roper:
> On Tue, Feb 23, 2016 at 05:20:13PM -0800, Matt Roper wrote:
>> In addition to calculating final watermarks, let's also pre-calculate a
>> set of intermediate watermark values at atomic check time. These
>> intermediate watermarks are a combination of the
> From: Wang, Zhi A
> Sent: Tuesday, February 23, 2016 9:23 PM
> >> --- a/drivers/gpu/drm/i915/gvt/gvt.c
> >> +++ b/drivers/gpu/drm/i915/gvt/gvt.c
> >> @@ -348,6 +348,10 @@ void *gvt_create_pgt_device(struct drm_i915_private
> *dev_priv)
> >>goto err;
> >>}
> >>
> >> +
Hi Ville,
We will look into this in sometime. Right now team is slightly loaded due to
project milestone.
Last time I looked into this, we dint have this HW to reproduce this issue.
Regards
Shashank
-Original Message-
From: Ville Syrjälä [mailto:ville.syrj...@linux.intel.com]
Sent:
> From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> Sent: Tuesday, February 23, 2016 8:42 PM
>
> Hi,
>
> Code is looking a lot better.
>
> A general question still; why you seem to be preferring multi-stage
> alloc and destroy?
One reason for multi-stage, IMO, is that GVT needs
Op 22-02-16 om 15:33 schreef John Harrison:
> On 18/02/2016 14:49, Chris Wilson wrote:
>> On Thu, Feb 18, 2016 at 02:24:06PM +, john.c.harri...@intel.com wrote:
>>> From: John Harrison
>>>
>>> The fence object used inside the request structure requires a sequence
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
>
> Let the core logic of context creation service creats the GEM context by
> context creation params.
>
> Now it provides following options for context creation:
> - Need to create legacy context for this GEM context?
> - Need to
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
> diff --git a/drivers/gpu/drm/i915/intel_lrc.h
> b/drivers/gpu/drm/i915/intel_lrc.h
> index e6cda3e..528c4fb 100644
> --- a/drivers/gpu/drm/i915/intel_lrc.h
> +++ b/drivers/gpu/drm/i915/intel_lrc.h
> @@ -97,10 +97,18 @@ static
From: Yogesh Mohan Marimuthu
The GPIO configuration and register offsets are different from
baytrail for cherrytrail. Port the gpio programming accordingly
for cherrytrail in this patch.
v2: Removing the duplication of parsing
v3: Moved the macro def to
On Wed, 24 Feb 2016 08:51:32 +0100,
Takashi Iwai wrote:
>
> Since dig_port_map[] is used only for the audio callback, we can
> assign it dynamically just before the callbacks.
>
> Could you try the patch below? (It's totally untested.)
The one below might be safer, it's setting in the mutex
24.02.2016 06:40, Jindal, Sonika написав:
Do you have logs for the failure with the single link hdmi cable and
the register dump which you have given for the working case?
If not, can you please capture the logs and register dump.
Also which platform is this?
If it is live status related issue,
23.02.2016 17:08, Ville Syrjälä написав:
In the meantime I think as a workaround I think you could use
something like video=HDMI-A-1:e on the kernel command line (not sure
I got the connector name right for your system). I think that should
result in the live status check to be skipped, at least
Instead of failing with -EINVAL when conflicting encoders are found,
the legacy set_config will disable other connectors when encoders
conflict.
With the cleanup to update_output_state this is a lot easier to
implement. set_config only adds connectors to the state that are
modified, and because
Hi,
On 10/02/16 18:01, Vinay Belgaumkar wrote:
These tests were initially reviewed/merged under the gem_softpin title.
They use softpinning and userptr mechanism to share buffers between
CPU and GPU.
The userptr part was decoupled from them recently. Adding these tests
under a different name
When intel_update_watermarks is called on skylake it inspects
crtc->state, which should show as disabled. This wasn't the case,
and this resulted in a divide-by-zero in
skl_ddb_get_pipe_allocation_limits when intel_update_watermarks was called.
[ cut here ]
WARNING: CPU:
There's no need to have a separate function to get the crtc
which is stolen, this can already be found when actually
stealing the encoder.
drm_for_each_connector already checks for connection_mutex, so
use that macro now.
Signed-off-by: Maarten Lankhorst
---
> From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> Sent: Tuesday, February 23, 2016 8:54 PM
> > diff --git a/drivers/gpu/drm/i915/gvt/Makefile
> > b/drivers/gpu/drm/i915/gvt/Makefile
> > new file mode 100644
> > index 000..959305f
> > --- /dev/null
> > +++
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
> diff --git a/drivers/gpu/drm/i915/gvt/gvt.c b/drivers/gpu/drm/i915/gvt/gvt.c
> index 52cfa32..2099b7e 100644
> --- a/drivers/gpu/drm/i915/gvt/gvt.c
> +++ b/drivers/gpu/drm/i915/gvt/gvt.c
> @@ -348,6 +348,10 @@ void
Hi:
Had a question about "BIOS GT RC6 option". If I disabled the BIOS GT
RC6 option, does that mean RC6 will be disabled no matter how driver
configure RC_CONTROL registers on core platform?
Looks currently on some SoC platform e.g. BXT, the status of "BIOS GT
RC6 option" will be reflected
> From: Wang, Zhi A
> Sent: Wednesday, February 24, 2016 5:19 PM
>
> Hi Kevin:
> Now our context switch is covered by context status change notification
> handler. In the
> status change handler we will save render registers. As GVT context will be a
> single
> submission context we will
Hey,
Op 18-02-16 om 13:59 schreef Ville Syrjälä:
> On Thu, Feb 18, 2016 at 01:43:11PM +0100, Daniel Vetter wrote:
>> On Thu, Feb 18, 2016 at 12:18:53PM +0100, Maarten Lankhorst wrote:
>>> Op 18-02-16 om 12:07 schreef Daniel Vetter:
On Thu, Feb 18, 2016 at 09:54:43AM +0100, Maarten Lankhorst
Currently we perform our own wait in post_plane_update,
but the atomic core performs another one in wait_for_vblanks.
This means that 2 vblanks are done when a fb is changed,
which is a bit overkill.
Merge them by creating a helper function that takes a crtc mask
for the planes to wait on.
The
From: Chris Wilson
Currently emit-request starts writing to the ring and reserves space
for a workaround to be emitted later whilst submitting the request. It
is easier to read if the caller only allocates sufficient space for
its access (then the reader can quickly
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
> -static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt)
> +static int gen8_ppgtt_init(struct i915_hw_ppgtt *ppgtt, int
> address_space_mode)
address_space_mode -> address_bits?
> {
> int ret;
>
> @@ -1518,7 +1524,7 @@
> From: Wang, Zhi A
> Sent: Thursday, February 18, 2016 7:42 PM
> diff --git a/drivers/gpu/drm/i915/intel_lrc.c
> b/drivers/gpu/drm/i915/intel_lrc.c
> index 1c0366a..2a6d6fe 100644
> --- a/drivers/gpu/drm/i915/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/intel_lrc.c
> @@ -479,6 +479,40 @@ static void
52 matches
Mail list logo