During a hibernate/resume cycle, the whole system is reset, including
the GuC and the doorbell hardware. Then the system is booted up, drivers
are loaded, etc -- the GuC firmware may be loaded and set running at this
point. But then, the booted kernel is replaced by the hibernated image,
and this
On Tue, Jun 07, 2016 at 02:31:07PM +0300, Joonas Lahtinen wrote:
> On pe, 2016-06-03 at 17:36 +0100, Chris Wilson wrote:
> > i915_gem_idle_work_handler(struct work_struct *work)
> > {
> > struct drm_i915_private *dev_priv =
> > - container_of(work, typeof(*dev_priv),
From: Ville Syrjälä
With the output_types bitmask there's no need to loop through
the encoders anymore when checking for HDMI+non-HDMI cloning.
v2: Use output_types bitmask
Signed-off-by: Ville Syrjälä
---
From: Ville Syrjälä
Since we now have the output_types bitmaks in the crtc state, there's no
need to iterate through all the encoders to see if an LVDS or SDVO/HDMI
encoder might be present.
Signed-off-by: Ville Syrjälä
---
From: Ville Syrjälä
has_dsi_encoder was introduced to indicate that the pipe is driving
a DSI encoder. Now that we have the output_types bitmask that can
tell us the same thing, let's just kill has_dsi_encoder.
v2: Rebase, handle BXT DSI transcoder, rewrote commit
From: Ville Syrjälä
INTEL_OUTPUT_DISPLAYPORT hsa been bugging me for a long time. It always
looks out of place besides INTEL_OUTPUT_EDP and INTEL_OUTPUT_DP_MST.
Let's just rename it to INTEL_OUTPUT_DP.
v2: Rebase
Signed-off-by: Ville Syrjälä
From: Ville Syrjälä
Use the new output_types bitmask instead of has_dp_encoder.
To make it less oainlful provide a small helper
(intel_crtc_has_dp_encoder()) to do the bitsy stuff.
v2: Rebase
Signed-off-by: Ville Syrjälä
---
From: Ville Syrjälä
A bunch of places still look for DP encoders manually. Just call
intel_crtc_has_dp_encoder(). Note that many of these places don't
look for EDP or DP_MST, but it's still fine to replace them because
* for audio we don't enable audio on eDP
From: Ville Syrjälä
Now that we have the output_types bitmask in the crtc state, we
can use it to indicate in which mode we want to drive the DDI
encoders. For pre-DDI output_types will instead indicate what
kind of cloning is being done, but as DDI platforms don't
From: Ville Syrjälä
Move the encoder cloning check to happen earlier in the modeset. The
main benefit will be that the debug output from a failed modeset will
be less confusing as output_types can not indicate an invalid
configuration during the later computation
From: Ville Syrjälä
DDI encoders changing encoder->type at random points in time is a
bit of a problem. Earlier I posted a series [1] to split DDI encoders
into separate HDMI and DP encoders, but now that LSPCON is coming
we'll anyway need some way to change the
From: Ville Syrjälä
Now that eDP encoders won't have can_mst==true, we can throw out
the encoder type checks from the MST suspend/resume paths.
Cc: Dave Airlie
Signed-off-by: Ville Syrjälä
---
From: Ville Syrjälä
If we've determined that the encoder is eDP, we shouldn't try to use MST
on it. Or at least the code doesn't seem to expect that since there are
some type==DP checks in the MST code.
Cc: Dave Airlie
Signed-off-by: Ville
From: Ville Syrjälä
With the introduction of the output_types mask, intel_pipe_has_type()
and intel_pipe_will_have_type() are basically the same thing. Replace
them with a new intel_crtc_has_type() (identical to
intel_pipe_will_have_type() actually).
v2: Rebase
From: Ville Syrjälä
Rather than looping through encoders to see which encoder types
are being driven by the pipe, add an output_types bitmask into
the crtc state and populate it prior to compute_config and during
state readout.
v2: Determine output_types before
On Wed, Jun 08, 2016 at 11:04:57AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:36:29PM +0100, Chris Wilson wrote:
> > Ideally, we want to automagically have the GPU respond to the
> > instantaneous load by reclocking itself. However, reclocking occurs
> > relatively slowly, and to the
On Wed, Jun 08, 2016 at 11:14:23AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:36:38PM +0100, Chris Wilson wrote:
> > static inline struct drm_i915_gem_request *
> > +to_request(struct fence *fence)
> > +{
> > + /* We assume that NULL fence/request are interoperable */
> > +
On 08/06/16 09:40, Daniel Vetter wrote:
On Wed, Jun 08, 2016 at 01:57:44PM +0530, Akshu Agrawal wrote:
CHV pipe C hits underrun when we get -ve X values of cursor. To avoid
this we crop the cursor image for by -ve X value and thus use '0' as
least X value.
You're talking about "-ve" here and
On 08/06/16 11:01, Chris Wilson wrote:
On Tue, Jun 07, 2016 at 01:46:53PM +0100, Tvrtko Ursulin wrote:
On 03/06/16 17:08, Chris Wilson wrote:
With only a single callsite for intel_engine_cs->irq_get and ->irq_put,
we can reduce the code size by moving the common preamble into the
caller, and
On 08/06/16 10:48, Chris Wilson wrote:
On Tue, Jun 07, 2016 at 01:04:22PM +0100, Tvrtko Ursulin wrote:
+static int intel_breadcrumbs_signaler(void *arg)
+{
+ struct intel_engine_cs *engine = arg;
+ struct intel_breadcrumbs *b = >breadcrumbs;
+ struct signal *signal;
+
+
On Wed, Jun 08, 2016 at 12:02:01PM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:55:44PM +0100, Chris Wilson wrote:
> > Since we are not concerned with userspace racing itself with set-tiling
> > (the order is indeterminant even if we take a lock), then we can safely
> > read back the
Hi,
On 06/07/2016 05:18 PM, Boris Brezillon wrote:
For all outputs except DSI we have a 1:1 relationship between connectors
and encoders and the driver is relying on the atomic helpers: we can
drop the custom ->best_encoder() and let the core call
drm_atomic_helper_best_encoder() for us.
On Fri, Jun 03, 2016 at 05:55:52PM +0100, Chris Wilson wrote:
> Treat the VMA as the primary struct responsible for tracking bindings
> into the GPU's VM. That is we want to treat the VMA returned after we
> pin an object into the VM as the cookie we hold and eventually release
> when unpinning.
On Wed, Jun 08, 2016 at 11:41:01AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:55:21PM +0100, Chris Wilson wrote:
> > Our GPUs impose certain requirements upon buffers that depend upon how
> > exactly they are used. Typically this is expressed as that they require
> > a larger surface
On Fri, Jun 03, 2016 at 05:55:47PM +0100, Chris Wilson wrote:
> The error state is purposefully racy as we expect it to be called at any
> time and so have avoided any locking whilst capturing the crash dump.
> However, with multi-engine GPUs and multiple CPUs, those races can
> manifest into
On Wed, Jun 08, 2016 at 11:57:21AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:55:37PM +0100, Chris Wilson wrote:
> > Signed-off-by: Chris Wilson
>
> Shouldn't we only do this as a last resort, i.e. in the oom notifier?
> Commit message is a bit sparse on
On Wed, Jun 08, 2016 at 11:59:25AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:55:42PM +0100, Chris Wilson wrote:
> > We only need to take the struct_mutex if the object is pinned to the
> > display engine and so requires checking for clflush. (The race with
> > userspace pinning the
On Fri, Jun 03, 2016 at 05:55:44PM +0100, Chris Wilson wrote:
> Since we are not concerned with userspace racing itself with set-tiling
> (the order is indeterminant even if we take a lock), then we can safely
> read back the single obj->tiling_mode and do the static lookup of
> swizzle mode
On Tue, Jun 07, 2016 at 01:46:53PM +0100, Tvrtko Ursulin wrote:
>
> On 03/06/16 17:08, Chris Wilson wrote:
> >With only a single callsite for intel_engine_cs->irq_get and ->irq_put,
> >we can reduce the code size by moving the common preamble into the
> >caller, and we can also eliminate the
On Fri, Jun 03, 2016 at 05:55:43PM +0100, Chris Wilson wrote:
> We don't need to incur the overhead of checking whether the object is
> pinned prior to changing its madvise. If the object is pinned, the
> madvise will not take effect until it is unpinned and so we cannot free
> the pages being
== Series Details ==
Series: Support for creating/using Stolen memory backed objects (rev16)
URL : https://patchwork.freedesktop.org/series/659/
State : failure
== Summary ==
Applying: drm/i915: Add support for mapping an object page by page
Applying: drm/i915: Introduce
Hi Boris,
Thanks for the patch.
Acked-by: Vincent Abriou
Vincent
On 06/07/2016 01:48 PM, Boris Brezillon wrote:
> All outputs have a 1:1 relationship between connectors and encoders
> and the driver is relying on the atomic helpers: we can drop the custom
>
On Fri, Jun 03, 2016 at 05:55:42PM +0100, Chris Wilson wrote:
> We only need to take the struct_mutex if the object is pinned to the
> display engine and so requires checking for clflush. (The race with
> userspace pinning the object to a framebuffer is irrelevant.)
>
> v2: Use access once for
On 08/06/16 10:35, Chris Wilson wrote:
On Mon, Jun 06, 2016 at 04:34:27PM +0100, Tvrtko Ursulin wrote:
On 03/06/16 17:08, Chris Wilson wrote:
If we flag the seqno as potentially stale upon receiving an interrupt,
we can use that information to reduce the frequency that we apply the
On Fri, Jun 03, 2016 at 05:55:37PM +0100, Chris Wilson wrote:
> Signed-off-by: Chris Wilson
Shouldn't we only do this as a last resort, i.e. in the oom notifier?
Commit message is a bit sparse on the motivation here ;-)
-Daniel
> ---
>
On Tue, Jun 07, 2016 at 01:31:25PM +0100, Tvrtko Ursulin wrote:
> >@@ -418,20 +423,23 @@ int intel_engine_enable_signaling(struct
> >drm_i915_gem_request *request)
> > struct intel_engine_cs *engine = request->engine;
> > struct intel_breadcrumbs *b = >breadcrumbs;
> > struct rb_node
On Fri, Jun 03, 2016 at 05:55:28PM +0100, Chris Wilson wrote:
> We are motivated to avoid using a bitfield for obj->active for a couple
> of reasons. Firstly, we wish to document our lockless read of obj->active
> using READ_ONCE inside i915_gem_busy_ioctl() and that requires an
> integral type
From: Ankitprasad Sharma
This patch adds support for extending the pread/pwrite functionality
for objects not backed by shmem. The access will be made through
gtt interface. This will cover objects backed by stolen memory as well
as other non-shmem backed objects.
From: Ankitprasad Sharma
Propagating correct error codes to userspace by using ERR_PTR and
PTR_ERR macros for stolen memory based object allocation. We generally
return -ENOMEM to the user whenever there is a failure in object
allocation. This patch helps user to
From: Ankitprasad Sharma
When constructing a batchbuffer, it is sometimes crucial to know the
largest hole into which we can fit a fenceable buffer (for example when
handling very large objects on gen2 and gen3). This depends on the
fragmentation of pinned buffers
From: Chris Wilson
If we run out of stolen memory when trying to allocate an object, see if
we can reap enough purgeable objects to free up enough contiguous free
space for the allocation. This is in principle very much like evicting
objects to free up enough contiguous space in the vma when
From: Ankitprasad Sharma
This patch adds support for clearing buffer objects via CPU/GTT. This
is particularly useful for clearing out the non shmem backed objects.
Currently intend to use this only for buffers allocated from stolen
region.
v2: Added kernel doc
From: Ankitprasad Sharma
In pwrite_fast, map an object page by page if obj_ggtt_pin fails. First,
we try a nonblocking pin for the whole object (since that is fastest if
reused), then failing that we try to grab one page in the mappable
aperture. It also allows us
From: Ankitprasad Sharma
The BIOS RapidStartTechnology may corrupt the stolen memory across S3
suspend due to unalarmed hibernation, in which case we will not be able
to preserve the User data stored in the stolen region. Hence this patch
tries to identify
From: Chris Wilson
Ville reminded us that stolen memory is not preserved across
hibernation, and a result of this was that context objects now being
allocated from stolen were being corrupted on S4 and promptly hanging
the GPU on resume.
We want to utilise stolen for
From: Chris Wilson
This utility function is a companion to i915_gem_object_get_page() that
uses the same cached iterator for the scatterlist to perform fast
sequential lookup of the dma address associated with any page within the
object.
Signed-off-by: Chris Wilson
From: Ankitprasad Sharma
Extend the drm_i915_gem_create structure to add support for
creating Stolen memory backed objects. Added a new flag through
which user can specify the preference to allocate the object from
stolen memory, which if set, an attempt will be
From: Chris Wilson
Introduced a new vm specfic callback insert_page() to program a single pte in
ggtt or ppgtt. This allows us to map a single page in to the mappable aperture
space. This can be iterated over to access the whole object by using space as
meagre as page
From: Ankitprasad Sharma
This patch series adds support for creating/using Stolen memory backed
objects.
Despite being a unified memory architecture (UMA) some bits of memory
are more equal than others. In particular we have the thorny issue of
stolen memory,
On Tue, Jun 07, 2016 at 01:04:22PM +0100, Tvrtko Ursulin wrote:
> >+static int intel_breadcrumbs_signaler(void *arg)
> >+{
> >+struct intel_engine_cs *engine = arg;
> >+struct intel_breadcrumbs *b = >breadcrumbs;
> >+struct signal *signal;
> >+
> >+/* Install ourselves with high
On Fri, Jun 03, 2016 at 05:55:25PM +0100, Chris Wilson wrote:
> Since i915_gem_obj_ggtt_pin() is an idiom breaking curry function for
> i915_gem_object_ggtt_pin(), spare us the confustion and remove it.
> Removing it now simplifies later patches to change the i915_vma_pin()
> (and friends)
On Fri, Jun 03, 2016 at 05:55:21PM +0100, Chris Wilson wrote:
> Our GPUs impose certain requirements upon buffers that depend upon how
> exactly they are used. Typically this is expressed as that they require
> a larger surface than would be naively computed by pitch * height.
> Normally such
On Mon, Jun 06, 2016 at 04:34:27PM +0100, Tvrtko Ursulin wrote:
>
> On 03/06/16 17:08, Chris Wilson wrote:
> >If we flag the seqno as potentially stale upon receiving an interrupt,
> >we can use that information to reduce the frequency that we apply the
> >heavyweight coherent seqno read (i.e. if
On Fri, Jun 03, 2016 at 05:36:25PM +0100, Chris Wilson wrote:
> Just to see if anyone is awake this series takes us to the VMA leak fix.
> Just the tip of the iceberg when it comes to VMA fixes...
Read through it. I think it'd be good (although yes, painful) if we could
untangle the ring/engine
On Mon, Jun 06, 2016 at 04:09:12PM +0100, Tvrtko Ursulin wrote:
>
> On 03/06/16 17:08, Chris Wilson wrote:
> > if (INTEL_GEN(dev_priv) >= 5) {
> >-ret = intel_init_pipe_control(engine);
> >+ret = intel_init_pipe_control(engine, 4096);
>
> Could be cool to define this
On Mon, Jun 06, 2016 at 03:55:18PM +0100, Tvrtko Ursulin wrote:
>
> On 03/06/16 17:08, Chris Wilson wrote:
> >diff --git a/drivers/gpu/drm/i915/i915_irq.c
> >b/drivers/gpu/drm/i915/i915_irq.c
> >index 2a736f4a0fe5..4013ad92cdc6 100644
> >--- a/drivers/gpu/drm/i915/i915_irq.c
> >+++
On pe, 2016-06-03 at 12:36 +, Tian, Kevin wrote:
> >
> > From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> > Sent: Friday, May 27, 2016 7:32 PM
> >
> > On pe, 2016-05-27 at 10:05 +, Wang, Zhi A wrote:
> > >
> > > For me I think maybe i915 could save the snapshot for GVT,
On Mon, Jun 06, 2016 at 02:44:41PM +0100, Tvrtko Ursulin wrote:
>
> On 03/06/16 17:36, Chris Wilson wrote:
> > Now that we have disambuigated ring and engine, we can use the clearer
> > and more consistent name for the intel_ringbuffer pointer in the
> > request.
>
> This one needs all the
On Fri, Jun 03, 2016 at 05:36:39PM +0100, Chris Wilson wrote:
> Now that we derive requests from struct fence, swap over to its
> nomenclature for references. It's shorter and more idiomatic across the
> kernel.
>
> s/i915_gem_request_reference/i915_gem_request_get/
>
On Fri, Jun 03, 2016 at 05:36:38PM +0100, Chris Wilson wrote:
> dma-buf provides a generic fence class for interoperation between
> drivers. Internally we use the request structure as a fence, and so with
> only a little bit of interfacing we can rebase those requests on top of
> dma-buf fences.
On Wed, Jun 08, 2016 at 10:42:58AM +0200, Daniel Vetter wrote:
> On Fri, Jun 03, 2016 at 05:08:34PM +0100, Chris Wilson wrote:
> > We can forgo queuing the hangcheck from the start of every request to
> > until we wait upon a request. This reduces the overhead of every
> > request, but may
On Fri, Jun 03, 2016 at 05:36:29PM +0100, Chris Wilson wrote:
> Ideally, we want to automagically have the GPU respond to the
> instantaneous load by reclocking itself. However, reclocking occurs
> relatively slowly, and to the client waiting for a result from the GPU,
> too late. To compensate
Hi,
[auto build test ERROR on drm-intel/for-linux-next]
[also build test ERROR on v4.7-rc2 next-20160608]
[if your patch is applied to the wrong git tree, please drop us a note to help
improve the system]
url:
https://github.com/0day-ci/linux/commits/Akshu-Agrawal/Revert-drm-i915-Workaround
On pe, 2016-06-03 at 15:37 +0100, Chris Wilson wrote:
> We only need to force a switch to the kernel context placeholder during
> eviction. All other uses of i915_gpu_idle() just want to wait until
> existing work on the GPU is idle. Rename i915_gpu_idle() to
> i915_gem_wait_for_idle() to avoid
On pe, 2016-06-03 at 15:37 +0100, Chris Wilson wrote:
> The module init/exit routines are a wrapper around the PCI device
> init/exit, so move them across.
>
> Note that in order to avoid exporting the driver struct, instead of
> manipulating driver.features inside i915_init we instead opt to
On ke, 2016-06-08 at 09:41 +0300, Ander Conselvan De Oliveira wrote:
> On Tue, 2016-06-07 at 21:24 +0300, Imre Deak wrote:
> > So far we configured a static lane latency optimization during driver
> > loading/resuming. The specification changed at one point and now this
> > configuration depends
On Fri, Jun 03, 2016 at 05:08:47PM +0100, Chris Wilson wrote:
> We have testcases to ensure that seqno wraparound works fine, so we can
> forgo forcing everyone to encounter seqno wraparound during early
> uptime. seqno wraparound incurs a full GPU stall so not forcing it
> will eliminate one
Use our ioctl overload to enable kcov capture around every ioctl.
---
lib/igt_aux.c | 84 +++
lib/igt_aux.h | 5
2 files changed, 84 insertions(+), 5 deletions(-)
diff --git a/lib/igt_aux.c b/lib/igt_aux.c
index fb1dac2..71067b3
A small utility for extracting kernel coverage using
/sys/kernel/debug/kcov.
Signed-off-by: Chris Wilson
---
lib/Makefile.sources | 2 +
lib/igt_debugfs.c| 20 ++
lib/igt_debugfs.h| 1 +
lib/igt_kcov.c | 188
On ke, 2016-06-08 at 08:04 +0100, Chris Wilson wrote:
> On Tue, Jun 07, 2016 at 11:18:46AM -0400, Zhi Wang wrote:
> >
> > This patch introduces the support of LRC context single submission.
> > As GVT context may come from different guests, which require different
> > configuration of render
On Fri, Jun 03, 2016 at 05:08:34PM +0100, Chris Wilson wrote:
> We can forgo queuing the hangcheck from the start of every request to
> until we wait upon a request. This reduces the overhead of every
> request, but may increase the latency of detecting a hang. Howeever, if
> nothing every waits
On Wed, Jun 08, 2016 at 01:57:44PM +0530, Akshu Agrawal wrote:
> CHV pipe C hits underrun when we get -ve X values of cursor. To avoid
> this we crop the cursor image for by -ve X value and thus use '0' as
> least X value.
You're talking about "-ve" here and there's absolutely no "-ve" anywhere
On ti, 2016-06-07 at 23:01 +0100, Chris Wilson wrote:
> On Tue, Jun 07, 2016 at 11:18:45AM -0400, Zhi Wang wrote:
> >
> > This patch introduces an approach to track the execlist context status
> > change.
> >
> > GVT-g uses GVT context as the "shadow context". The content inside GVT
> > context
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> Currently the addressing mode bit in context descriptor is statically
> generated from the configuration of system-wide PPGTT usage model.
>
> GVT-g will load the PPGTT shadow page table by itself and probably one
> guest is using a different
On ke, 2016-06-08 at 08:08 +0100, Chris Wilson wrote:
> On Tue, Jun 07, 2016 at 11:18:43AM -0400, Zhi Wang wrote:
> >
> > This patch introduces an option for configuring the ring buffer size
> > of a LRC context after the context creation.
> >
> > Signed-off-by: Zhi Wang
>
== Series Details ==
Series: series starting with [1/2] Revert "drm/i915: Workaround CHV pipe C
cursor fail"
URL : https://patchwork.freedesktop.org/series/8431/
State : failure
== Summary ==
CC [M] drivers/gpu/drm/i915/intel_dp_mst.o
CC [M] drivers/gpu/drm/i915/intel_dp.o
CC [M]
On Fri, Jun 03, 2016 at 11:20:19AM +0100, Chris Wilson wrote:
> On Fri, Jun 03, 2016 at 03:44:09PM +0530, Goel, Akash wrote:
> >
> >
> > On 6/3/2016 12:45 PM, Daniel Vetter wrote:
> > >On Thu, Jun 02, 2016 at 12:21:49PM +0200, Johannes Berg wrote:
> > >>On Thu, 2016-06-02 at 10:16 +, Daniel
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> This patch introduces the very basic framework of GVT-g device model,
> includes basic prototypes, definitions, initialization.
>
> v7:
> - Refine the URL link in Kconfig. (Joonas)
> - Refine the introduction of GVT-g host support in Kconfig.
This reverts commit ef8dd37af85a8f37ca3a29074647511e52c56181.
Will use cropped cursor image for -ve co-ordinates instead of using
swcursor. Advantages of cropped cursor being, it will work for non X11
based platform like Chrome and also will use hwcursor.
Signed-off-by: Akshu Agrawal
CHV pipe C hits underrun when we get -ve X values of cursor. To avoid
this we crop the cursor image for by -ve X value and thus use '0' as
least X value.
Signed-off-by: Akshu Agrawal
---
drivers/gpu/drm/i915/intel_display.c | 113 +++
Sure. Moving code in one patch, "offsetof()" guy will be in another patch
> -Original Message-
> From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> Sent: Wednesday, June 08, 2016 10:55 AM
> To: Wang, Zhi A ; ch...@chris-wilson.co.uk; Lv, Zhiyuan
>
On 07/06/16 21:00, Chris Wilson wrote:
On Tue, Jun 07, 2016 at 02:23:34PM +0100, Tvrtko Ursulin wrote:
On 07/06/16 11:54, Dave Gordon wrote:
On 07/06/16 09:43, Patchwork wrote:
== Series Details ==
Series: series starting with [1/3] drm/i915/guc: fix GuC
loading/submission check
URL :
Good idea. :)
> -Original Message-
> From: Chris Wilson [mailto:ch...@chris-wilson.co.uk]
> Sent: Wednesday, June 08, 2016 10:13 AM
> To: Wang, Zhi A
> Cc: Lv, Zhiyuan ; Tian, Kevin ;
> tvrtko.ursu...@linux.intel.com;
Patch title s/ballon/balloon/.
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> This function needs to be changed to have a proper goto teardown path.
> Destructors/fini functions are only expected to be called after a
> successful initialization, so calling it at random phase in init function
OK. Will do in the next version. :)
> -Original Message-
> From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> Sent: Wednesday, June 08, 2016 11:05 AM
> To: Wang, Zhi A ; ch...@chris-wilson.co.uk; Lv, Zhiyuan
> ; Tian, Kevin
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> v5:
> - Let functions take struct drm_i915_private *. (Tvrtko)
>
> - Fold vGPU related active check into the inner functions. (Kevin)
>
I already reviewed this patch, so you should add Reviewed-by: and Cc:
tags. Also good to add Cc: tag for
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> To get the offset of the members in PVINFO page, offsetof() looks much
> better than the tricky approach in current code.
>
> v7:
>
> - Move "offsetof()" modification into a dedicated patch. (Joonas)
>
> Signed-off-by: Zhi Wang
On ti, 2016-06-07 at 11:18 -0400, Zhi Wang wrote:
> As the PVINFO page definition is used by both GVT-g guest (vGPU) and GVT-g
> host (GVT-g kernel device model), factor it out for better code structure.
>
> v7:
> - Split the "offsetof" modification into a dedicated patch. (Joonas)
>
> v3:
> -
On ti, 2016-06-07 at 15:29 +, Wang, Zhi A wrote:
>
> >
> > -Original Message-
> > From: Joonas Lahtinen [mailto:joonas.lahti...@linux.intel.com]
> > Sent: Friday, June 03, 2016 12:40 PM
> > To: Wang, Zhi A ; intel-gfx@lists.freedesktop.org;
> >
On Tue, Jun 07, 2016 at 11:18:44AM -0400, Zhi Wang wrote:
> Currently the addressing mode bit in context descriptor is statically
> generated from the configuration of system-wide PPGTT usage model.
>
> GVT-g will load the PPGTT shadow page table by itself and probably one
> guest is using a
On Tue, Jun 07, 2016 at 11:18:43AM -0400, Zhi Wang wrote:
> This patch introduces an option for configuring the ring buffer size
> of a LRC context after the context creation.
>
> Signed-off-by: Zhi Wang
> ---
> drivers/gpu/drm/i915/i915_drv.h | 1 +
>
On Tue, Jun 07, 2016 at 11:18:46AM -0400, Zhi Wang wrote:
> This patch introduces the support of LRC context single submission.
> As GVT context may come from different guests, which require different
> configuration of render registers. It can't be combined into a dual ELSP
> submission combo.
>
On Tue, Jun 07, 2016 at 11:18:47AM -0400, Zhi Wang wrote:
> GVT workload scheduler needs special host LRC contexts, the so called
> "shadow LRC context" to submit guest workload to host i915. During the
> guest workload submission, workload scheduler fills the shadow LRC
> context with the content
On Tue, 2016-06-07 at 21:24 +0300, Imre Deak wrote:
> So far we configured a static lane latency optimization during driver
> loading/resuming. The specification changed at one point and now this
> configuration depends on the lane count, so move the configuration
> to modeset time accordingly.
>
101 - 194 of 194 matches
Mail list logo