On Tue, 2019-06-25 at 13:17 +1000, Alastair D'Silva wrote:
> From: Alastair D'Silva
>
> With the wider display format, it can become hard to identify how many
> bytes into the line you are looking at.
>
> The patch adds new flags to hex_dump_to_buffer() and print_hex_dump() to
> print vertical
On Tue, 2019-06-25 at 15:06 +1000, Alastair D'Silva wrote:
> The change actions Jani's suggestion:
> https://lkml.org/lkml/2019/6/20/343
I suggest not changing any of the existing uses of
hex_dump_to_buffer and only use hex_dump_to_buffer_ext
when necessary for your extended use cases.
On Tue, 2019-06-25 at 15:06 +1000, Alastair D'Silva wrote:
> On Mon, 2019-06-24 at 22:01 -0700, Joe Perches wrote:
> > On Tue, 2019-06-25 at 13:17 +1000, Alastair D'Silva wrote:
> > > From: Alastair D'Silva
> > >
> > > In order to support additional features, rename hex_dump_to_buffer
> > > to
>
On 06/24/2019 05:48 PM, Chris Wilson wrote:
> Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
> preemption, but mediated gvt does not fully support semaphores.
>
> Signed-off-by: Chris Wilson
> ---
> drivers/gpu/drm/i915/gt/intel_engine_cs.c | 4 ++--
>
On Tue, 2019-06-25 at 13:17 +1000, Alastair D'Silva wrote:
> From: Alastair D'Silva
>
> Apologies for the large CC list, it's a heads up for those responsible
> for subsystems where a prototype change in generic code causes a change
> in those subsystems.
[]
> The default behaviour of hexdump is
On Tue, 2019-06-25 at 13:17 +1000, Alastair D'Silva wrote:
> From: Alastair D'Silva
>
> In order to support additional features, rename hex_dump_to_buffer to
> hex_dump_to_buffer_ext, and replace the ascii bool parameter with flags.
[]
> diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c
>
== Series Details ==
Series: Display uncore (rev2)
URL : https://patchwork.freedesktop.org/series/61735/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6337_full -> Patchwork_13408_full
Summary
---
**SUCCESS**
No
On Fri, Jun 21, 2019 at 8:41 PM Derek Basehore wrote:
>
> Devicetree systems can set panel orientation via a panel binding, but
> there's no way, as is, to propagate this setting to the connector,
> where the property need to be added.
> To address this, this patch sets orientation, as well as
== Series Details ==
Series: Enable Transcoder Port Sync feature for Tiled displays (rev2)
URL : https://patchwork.freedesktop.org/series/59837/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_6337 -> Patchwork_13409
Summary
== Series Details ==
Series: Display uncore (rev2)
URL : https://patchwork.freedesktop.org/series/61735/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6337 -> Patchwork_13408
Summary
---
**SUCCESS**
No
== Series Details ==
Series: Enable Transcoder Port Sync feature for Tiled displays (rev2)
URL : https://patchwork.freedesktop.org/series/59837/
State : warning
== Summary ==
$ dim checkpatch origin/drm-tip
e1dacdba0522 drm/i915/display: Rename update_crtcs() to commit_modeset_enables()
On Thu, 2019-06-20 at 17:36 +, Souza, Jose wrote:
> On Thu, 2019-06-20 at 11:13 +0300, Jani Nikula wrote:
> > On Wed, 19 Jun 2019, José Roberto de Souza
> > wrote:
> > > diff --git a/drivers/gpu/drm/i915/i915_reg.h
> > > b/drivers/gpu/drm/i915/i915_reg.h
> > > index 4fc8dc083766..31fb4de5081c
On Fri, 2019-06-21 at 09:15 -0700, Matt Roper wrote:
> On Thu, Jun 20, 2019 at 05:45:54PM -0700, José Roberto de Souza
> wrote:
> > Now 180, 172.8 and 192 MHz are supported.
> >
> > 180 and 172.8 MHz CD clocks will only be used when audio is not
> > enabled as state by BSpec and implemented in
>
In the transcoder port sync mode, the slave transcoders mask their vblanks
until master transcoder's vblank so while disabling them, make
sure slaves are disabled first and then the masters.
Cc: Ville Syrjälä
Cc: Maarten Lankhorst
Cc: Matt Roper
Cc: Jani Nikula
Signed-off-by: Manasi Navare
After the state is committed, we readout the HW registers and compare
the HW state with the SW state that we just committed.
For Transcdoer port sync, we add master_transcoder and the
salves bitmask to the crtc_state, hence we need to read those during
the HW state readout to avoid pipe state
In case of tiled displays when the two tiles are sent across two CRTCs
over two separate DP SST connectors, we need a mechanism to synchronize
the two CRTCs and their corresponding transcoders.
So use the master-slave mode where there is one master corresponding
to last horizontal and vertical
As per the display enable sequence, we need to follow the enable sequence
for slaves first with DP_TP_CTL set to Idle and configure the transcoder
port sync register to select the corersponding master, then follow the
enable sequence for master leaving DP_TP_CTL to idle.
At this point the
In case of tiled displays where different tiles are displayed across
different ports, we need to synchronize the transcoders involved.
This patch implements the transcoder port sync feature for
synchronizing one master transcoder with one or more slave
transcoders. This is only enbaled in slave
Create a new hook commit_modeset_disables() consistent with the naming
in drm atomic helpers and similar to the commit_modeset_enables() hook.
This helps better organize the disable sequence in atomic_commit_tail()
and move that to this disable hook.
No functional change
Suggested-by: Daniel
First two patches are cleanup patches with no functional changes to
rename the existing update_crtc() hook to commit_modeset_enables() to better
match the drm_atomic naming conventions.
This also prepares to add new hook for ICL since that adds support for
transcoder port sync feature.
Rest of
This clears the transcoder port sync bits of the TRANS_DDI_FUNC_CTL2
register during crtc_disable().
Cc: Ville Syrjälä
Cc: Maarten Lankhorst
Cc: Matt Roper
Cc: Jani Nikula
Signed-off-by: Manasi Navare
---
drivers/gpu/drm/i915/display/intel_display.c | 44
1 file
This patch has no functional changes. This just renames the update_crtcs()
hooks to commit_modeset_enables() to match the drm_atomic helper naming
conventions.
Suggested-by: Daniel Vetter
Cc: Ville Syrjälä
Cc: Maarten Lankhorst
Cc: Matt Roper
Cc: Jani Nikula
Signed-off-by: Manasi Navare
---
On Mon, 2019-06-24 at 15:39 +0300, Ville Syrjälä wrote:
> On Thu, Jun 20, 2019 at 11:33:27PM +, Souza, Jose wrote:
> > On Wed, 2019-06-19 at 20:47 +0300, Ville Syrjälä wrote:
> > > On Tue, Jun 18, 2019 at 03:50:33PM -0700, José Roberto de Souza
> > > wrote:
> > > > Now 180, 172.8 and 192 MHz
== Series Details ==
Series: Display uncore (rev2)
URL : https://patchwork.freedesktop.org/series/61735/
State : warning
== Summary ==
$ dim sparse origin/drm-tip
Sparse version: v0.5.2
Commit: drm/i915: split out uncore_mmio_debug
+drivers/gpu/drm/i915/intel_uncore.c:1192:1: warning: context
Hi Daniel.
> > drm_file.h:
> > - needs drm_magic_t
> > drm_magic_t is a simple typedef, a simple unsigned int would do the trick
>
> I think we could switch over to unsigned int for this.
OK, will send a v2 with this addressed.
There is some preparatory patches to fix build breakage when the
Hi Derek.
On Fri, Jun 21, 2019 at 08:41:02PM -0700, Derek Basehore wrote:
> This adds a helper function for reading the rotation (panel
> orientation) from the device tree.
>
> Signed-off-by: Derek Basehore
> ---
> drivers/gpu/drm/drm_panel.c | 42 +
>
As an example of usage of the new accessors.
Changes done mechanically with some manual post-processing to use
rmw where appropriate and fix line length and formatting.
Signed-off-by: Daniele Ceraolo Spurio
---
drivers/gpu/drm/i915/display/intel_hdmi.c | 278 ++
1 file
More focused proposal for display/GT uncore split.
A few notes/considerations:
- I've copied the MMIO ranges check in patch 3 from a patch from Ville
that did something similar in regards to splitting GT/DE register
access. I'm seeing some check failures on gen 3 and 4 [1], so I'd
A forcewake-less uncore to be used to decouple GT accesses from display
ones to avoid serializing them when there is no need.
New accessors that implicitly use the new uncore have also been added.
Signed-off-by: Daniele Ceraolo Spurio
Cc: Ville Syrjälä
---
drivers/gpu/drm/i915/i915_drv.c
Multiple uncore structures will share the debug infrastructure, so
move it to a common place and add extra locking around it.
Signed-off-by: Daniele Ceraolo Spurio
---
drivers/gpu/drm/i915/i915_drv.c | 1 +
drivers/gpu/drm/i915/i915_drv.h | 1 +
drivers/gpu/drm/i915/intel_uncore.c |
Since we now have a separate object, it is cleaner to have dedicated
functions working on the object to stop and restart the mmio debug.
Apart from the cosmetic changes, this patch introduces 2 functional
updates:
- All calls to check_for_unclaimed_mmio will now return false when
the debug is
On Mon, 24 Jun 2019 at 06:45, Chris Wilson wrote:
>
> Switch to tracking activity via i915_active on individual nodes, only
> keeping a list of retired objects in the cache, and reaping the cache
> when the engine itself idles.
>
> Signed-off-by: Chris Wilson
> ---
>
== Series Details ==
Series: drm/i915/gem: Clear read/write domains for GPU clear
URL : https://patchwork.freedesktop.org/series/62646/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6334_full -> Patchwork_13407_full
Hi all,
I tried using wake_up_var() today and accidentally noticed that it
didn't imply an smp_mb() and specifically requires it through
wake_up_bit() / waitqueue_active().
Now, wake_up_bit() doesn't imply the barrier because it is assumed to be
used with the atomic bitops API which either
On Wed, Jun 19, 2019 at 07:21:48PM +0100, Chris Wilson wrote:
> Quoting Ville Syrjala (2019-06-19 19:03:11)
> > From: Ville Syrjälä
> >
> > The strings we want to print to the on stack buffers should
> > be no more than
> > 8 * 3 + strlen("(GET_SCALED_HDTV_RESOLUTION_SUPPORT)") + 1 = 61
> >
On 24/06/2019 17:00, Chris Wilson wrote:
Quoting Matthew Auld (2019-06-24 16:50:48)
On 24/06/2019 15:16, Chris Wilson wrote:
Update the domains for the write via the GPU so that we do not
shortcircuit any set-domain wait afterwards.
I'm lost. How do we short-circuit the set-domain wait?
If
Add libatomic to the Fedora docker image so it can link binaries that
use __atomic_* functions. Also explicitly add libatomic1 to Debian
docker images as it is needed in particular on non-x86 architectures
for run-time linkage.
Signed-off-by: Guillaume Tucker
---
Notes:
v2: add libatomic1
Add conditional dependency on GCC's libatomic in order to be able to
use the __atomic_* functions instead of the older __sync_* ones. The
libatomic library is only needed when there aren't any native support
on the current architecture, so a linker test is used for this
purpose. This makes
Replace calls to the older __sync_* functions with the new atomic_*
standard ones to be consistent with other tests and improve
portability across CPU architectures. Add dependency of sw_sync on
libatomic.
Signed-off-by: Guillaume Tucker
Reviewed-by: Simon Ser
---
Notes:
v2: use atomic_*
This fixes builds on some architectures, in particular MIPS which
doesn't have __sync_add_and_fetch_8 and __sync_val_compare_and_swap_8
for 64-bit variable handling.
* replace calls to the older __sync_* functions with the new atomic_*
standard ones
* use the _Atomic type modifier as required
This series replaces calls to the __sync_* functions with the more
recent atomic_* ones defined in stdatomic.h in gem_create and
sw_sync. It also adds dependency on libatomic when required, that is
to say when the CPU architecture doesn't provide native support for
some atomic operations. This
On Thu, Jun 20, 2019 at 11:46:10PM +0200, Maarten Lankhorst wrote:
> Instead of directly referencing drm_crtc_state, convert to
> intel_ctc_state and use the base struct. This is useful when we're
> making the split between uapi and hw state, and also makes the
> code slightly more readable.
>
>
On Thu, Jun 20, 2019 at 11:46:09PM +0200, Maarten Lankhorst wrote:
> Signed-off-by: Maarten Lankhorst
> ---
> .../gpu/drm/i915/display/intel_atomic_plane.c | 39 +++--
> .../gpu/drm/i915/display/intel_atomic_plane.h | 5 +-
> drivers/gpu/drm/i915/display/intel_display.c | 58
Quoting Matthew Auld (2019-06-24 16:50:48)
> On 24/06/2019 15:16, Chris Wilson wrote:
> > Update the domains for the write via the GPU so that we do not
> > shortcircuit any set-domain wait afterwards.
>
> I'm lost. How do we short-circuit the set-domain wait?
If we never change the read_domain,
On Thu, Jun 20, 2019 at 11:46:08PM +0200, Maarten Lankhorst wrote:
> Signed-off-by: Maarten Lankhorst
> ---
> drivers/gpu/drm/i915/display/intel_display.c | 14 ++
> 1 file changed, 6 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c
>
On Thu, Jun 20, 2019 at 11:46:07PM +0200, Maarten Lankhorst wrote:
> Signed-off-by: Maarten Lankhorst
> ---
> drivers/gpu/drm/i915/display/intel_display.c | 154 +--
> 1 file changed, 74 insertions(+), 80 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/display/intel_display.c
On 24/06/2019 15:16, Chris Wilson wrote:
Update the domains for the write via the GPU so that we do not
shortcircuit any set-domain wait afterwards.
I'm lost. How do we short-circuit the set-domain wait?
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110978
Fixes: b2dbf8d982a4
On Thu, Jun 20, 2019 at 11:46:05PM +0200, Maarten Lankhorst wrote:
Commit msg missing from some of these.
Apart from that:
Reviewed-by: Ville Syrjälä
> Signed-off-by: Maarten Lankhorst
> ---
> drivers/gpu/drm/i915/display/intel_display.c | 68 ++--
> 1 file changed, 34
On Thu, Jun 20, 2019 at 11:46:06PM +0200, Maarten Lankhorst wrote:
> Instead of passing along drm_crtc_state and drm_atomic_state, pass
> along more intel_atomic_state and intel_crtc_state. This will
> make the code more readable by not casting between drm state
> and intel state all the time.
>
On Fri, Jun 21, 2019 at 07:28:30PM -0400, Alex Deucher wrote:
> On Thu, Jun 20, 2019 at 10:26 AM Ville Syrjala
> wrote:
> >
> > From: Ville Syrjälä
> >
> > Ilia pointed out some oddball crap in the i915 aspect ratio handling.
> > While looking at that I noticed a bunch of fail in the core as
== Series Details ==
Series: drm/i915/gem: Clear read/write domains for GPU clear
URL : https://patchwork.freedesktop.org/series/62646/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6334 -> Patchwork_13407
Summary
---
== Series Details ==
Series: series starting with [1/3] drm/i915/execlists: Switch to using
STOP_RING instead of a semaphore
URL : https://patchwork.freedesktop.org/series/62641/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_6334 -> Patchwork_13405
On Mon, Jun 24, 2019 at 03:58:00PM +0200, Christian König wrote:
> Am 24.06.19 um 13:23 schrieb Koenig, Christian:
> > Am 21.06.19 um 18:27 schrieb Daniel Vetter:
> >
> > > So I pondered a few ideas while working out:
> > >
> > > 1) We drop this filtering. Importer needs to keep track of all its
On Mon, Jun 24, 2019 at 11:23:34AM +, Koenig, Christian wrote:
> Am 21.06.19 um 18:27 schrieb Daniel Vetter:
> >>> Your scenario here is new, and iirc my suggestion back then was to
> >>> count the number of pending mappings so you don't go around calling
> >>> ->invalidate on mappings that
== Series Details ==
Series: series starting with [1/6] dma-buf: add dynamic DMA-buf handling v11
URL : https://patchwork.freedesktop.org/series/62644/
State : failure
== Summary ==
Applying: dma-buf: add dynamic DMA-buf handling v11
Applying: drm/ttm: remove the backing store if no placement
Update the domains for the write via the GPU so that we do not
shortcircuit any set-domain wait afterwards.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=110978
Fixes: b2dbf8d982a4 ("drm/i915/blt: Remove recursive vma->lock")
Signed-off-by: Chris Wilson
Cc: Matthew Auld
---
This way we can even pipeline imported BO evictions.
v2: Limit this to only cases when the parent object uses a separate
reservation object as well. This fixes another OOM problem.
Signed-off-by: Christian König
---
drivers/gpu/drm/ttm/ttm_bo_util.c | 20 +++-
1 file
The caching of SGT's is actually quite harmful and should probably removed
altogether when all drivers are audited.
Start by providing a separate DMA-buf export implementation in amdgpu.
This is also a prerequisite of unpinned DMA-buf handling.
v2: fix unintended recursion, remove debugging
Pipeline removal of the BOs backing store when no placement is given
during validation.
Signed-off-by: Christian König
---
drivers/gpu/drm/ttm/ttm_bo.c | 12
1 file changed, 12 insertions(+)
diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c
index
Instead of relying on the DRM functions just implement our own import
functions. This prepares support for taking care of unpinned DMA-buf.
v2: enable for all exporters, not just amdgpu, fix invalidation
handling, lock reservation object while setting callback
v3: change to new dma_buf attach
Avoid that we ping/pong the buffers when we stop to pin DMA-buf
exports by using the allowed domains for exported buffers.
Signed-off-by: Christian König
---
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git
On the exporter side we add optional explicit pinning callbacks. If those
callbacks are implemented the framework no longer caches sg tables and the
map/unmap callbacks are always called with the lock of the reservation object
held.
On the importer side we add an optional invalidate callback.
Am 24.06.19 um 13:23 schrieb Koenig, Christian:
Am 21.06.19 um 18:27 schrieb Daniel Vetter:
So I pondered a few ideas while working out:
1) We drop this filtering. Importer needs to keep track of all its
mappings and filter out invalidates that aren't for that specific importer
(either
On Fri, Jun 21, 2019 at 08:41:04PM -0700, Derek Basehore wrote:
> Not every platform needs quirk detection for panel orientation, so
> split the drm_connector_init_panel_orientation_property into two
> functions. One for platforms without the need for quirks, and the
> other for platforms that
As this engine owns the lock around rq->sched.link (for those waiters
submitted to this engine), we can use that link as an element in a local
list. We can thus replace the recursive algorithm with an iterative walk
over the ordered list of waiters.
Signed-off-by: Chris Wilson
---
With everything now known at the point of adding the request to the
context's timeline, we can move the emission of the final breadcrumb to
the request add and pull it out of the CS interrupt service handler.
Signed-off-by: Chris Wilson
---
drivers/gpu/drm/i915/i915_request.c | 6 +++---
1 file
Not all mediated environments yet support HW semaphores, so we should
avoid using one for our preempt-to-busy barrier and instead request that
the CS be paused and not advance on to the next execlist.
There's a natural advantage with the register writes being serialised
with the writes to the
On Thu, Jun 20, 2019 at 11:33:27PM +, Souza, Jose wrote:
> On Wed, 2019-06-19 at 20:47 +0300, Ville Syrjälä wrote:
> > On Tue, Jun 18, 2019 at 03:50:33PM -0700, José Roberto de Souza
> > wrote:
> > > Now 180, 172.8 and 192 MHz are supported.
> > >
> > > 180 and 172.8 MHz CD clocks will only
On Thu, Jun 20, 2019 at 10:25:42PM +0200, Sam Ravnborg wrote:
> Hi Ville.
>
> On Thu, Jun 20, 2019 at 09:50:48PM +0300, Ville Syrjala wrote:
> > From: Ville Syrjälä
> >
> > Decode the mode flags when printing the modeline so that I
> > no longer have to decode the hex number myself.
> You are
Quoting Patchwork (2019-06-24 09:55:07)
> Possible regressions
>
> * igt@gem_ctx_switch@basic-queue-heavy:
> - shard-skl: [PASS][1] -> [INCOMPLETE][2]
>[1]:
> https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_6331/shard-skl4/igt@gem_ctx_swi...@basic-queue-heavy.html
>
== Series Details ==
Series: drm/i915/execlists: Always clear ring_pause if we do not submit (rev2)
URL : https://patchwork.freedesktop.org/series/62540/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6331_full -> Patchwork_13403_full
Am 21.06.19 um 18:27 schrieb Daniel Vetter:
>>> Your scenario here is new, and iirc my suggestion back then was to
>>> count the number of pending mappings so you don't go around calling
>>> ->invalidate on mappings that don't exist.
>> Well the key point is we don't call invalidate on mappings,
On Mon, 24 Jun 2019, Emil Velikov wrote:
> On Mon, 24 Jun 2019 at 09:21, Daniel Vetter wrote:
>
>> > drm_legacy.h
>> > - needs drm_map_type and drm_map_flags. Seems legit.
>>
>> enum in uapi, sweet (never do this, it's not portable across compilers,
>> #defines all the way).
>
> Don't look too
Quoting Chris Wilson (2019-06-24 11:37:06)
> Quoting Patchwork (2019-06-24 11:30:47)
> > == Series Details ==
> >
> > Series: drm/i915/execlists: Disable preemption under GVT (rev5)
> > URL : https://patchwork.freedesktop.org/series/62533/
> > State : failure
> >
> > == Summary ==
> >
> > CI
Quoting Patchwork (2019-06-24 11:30:47)
> == Series Details ==
>
> Series: drm/i915/execlists: Disable preemption under GVT (rev5)
> URL : https://patchwork.freedesktop.org/series/62533/
> State : failure
>
> == Summary ==
>
> CI Bug Log - changes from CI_DRM_6331 -> Patchwork_13404
>
== Series Details ==
Series: drm/i915/execlists: Disable preemption under GVT (rev5)
URL : https://patchwork.freedesktop.org/series/62533/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_6331 -> Patchwork_13404
Summary
Chris Wilson writes:
> In the unlikely case (thank you CI!), we may find ourselves wanting to
> issue a preemption but having no runnable requests left. In this case,
> we set the semaphore before computing the preemption and so must unset
> it before forgetting (or else we leave the machine
Chris Wilson writes:
> Quoting Mika Kuoppala (2019-06-24 10:03:48)
>> Chris Wilson writes:
>>
>> > In the unlikely case (thank you CI!), we may find ourselves wanting to
>> > issue a preemption but having no runnable requests left. In this case,
>> > we set the semaphore before computing the
On Mon, 24 Jun 2019 at 09:21, Daniel Vetter wrote:
> > drm_legacy.h
> > - needs drm_map_type and drm_map_flags. Seems legit.
>
> enum in uapi, sweet (never do this, it's not portable across compilers,
> #defines all the way).
Don't look too closely then
$ git grep enum -- include/uapi/drm/ | wc
== Series Details ==
Series: drm/i915/execlists: Always clear ring_pause if we do not submit (rev2)
URL : https://patchwork.freedesktop.org/series/62540/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6331 -> Patchwork_13403
Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
preemption, but mediated gvt does not fully support semaphores.
Signed-off-by: Chris Wilson
---
drivers/gpu/drm/i915/gt/intel_engine_cs.c | 4 ++--
drivers/gpu/drm/i915/gt/intel_lrc.c | 21 -
Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
preemption, but mediated gvt does not fully support semaphores.
Signed-off-by: Chris Wilson
---
drivers/gpu/drm/i915/gt/intel_engine_cs.c | 4 ++--
drivers/gpu/drm/i915/gt/intel_lrc.c | 21 -
2
== Series Details ==
Series: drm/i915/execlists: Always clear ring_pause if we do not submit (rev2)
URL : https://patchwork.freedesktop.org/series/62540/
State : warning
== Summary ==
$ dim checkpatch origin/drm-tip
975f9c471da0 drm/i915/execlists: Always clear ring_pause if we do not submit
== Series Details ==
Series: drm/i915/execlists: Disable preemption under GVT (rev3)
URL : https://patchwork.freedesktop.org/series/62533/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_6331 -> Patchwork_13402
Summary
In the unlikely case (thank you CI!), we may find ourselves wanting to
issue a preemption but having no runnable requests left. In this case,
we set the semaphore before computing the preemption and so must unset
it before forgetting (or else we leave the machine busywaiting until the
next request
Quoting Mika Kuoppala (2019-06-24 10:03:48)
> Chris Wilson writes:
>
> > In the unlikely case (thank you CI!), we may find ourselves wanting to
> > issue a preemption but having no runnable requests left. In this case,
> > we set the semaphore before computing the preemption and so must unset
>
Quoting Zhenyu Wang (2019-06-24 10:02:31)
> On 2019.06.21 20:13:13 +0100, Chris Wilson wrote:
> > Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
> > preemption, but mediated gvt does not fully support semaphores.
> >
>
> Current looks semaphore is still used from
Quoting Mika Kuoppala (2019-06-24 10:03:48)
> Chris Wilson writes:
>
> > In the unlikely case (thank you CI!), we may find ourselves wanting to
> > issue a preemption but having no runnable requests left. In this case,
> > we set the semaphore before computing the preemption and so must unset
>
On 2019.06.21 20:13:13 +0100, Chris Wilson wrote:
> Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
> preemption, but mediated gvt does not fully support semaphores.
>
Current looks semaphore is still used from emit_fini_breadcrumb which
caused gvt error, gvt may support
Chris Wilson writes:
> In the unlikely case (thank you CI!), we may find ourselves wanting to
> issue a preemption but having no runnable requests left. In this case,
> we set the semaphore before computing the preemption and so must unset
> it before forgetting (or else we leave the machine
Preempt-to-busy uses a GPU semaphore to enforce an idle-barrier across
preemption, but mediated gvt does not fully support semaphores.
Signed-off-by: Chris Wilson
---
drivers/gpu/drm/i915/gt/intel_lrc.c | 21 -
1 file changed, 16 insertions(+), 5 deletions(-)
diff --git
== Series Details ==
Series: series starting with [01/19] drm/i915/execlists: Always clear
ring_pause if we do not submit
URL : https://patchwork.freedesktop.org/series/62612/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_6331_full -> Patchwork_13401_full
On Sat, Jun 22, 2019 at 02:11:54PM +0200, Sam Ravnborg wrote:
> include/drm/* shall have no or at least minimal dependencies
> to include/uapi/drm/*.
> Following two patches do a small effort to drop such dependencies.
>
> After these patches there are two users
> of uapi/drm/drm.h left in
== Series Details ==
Series: series starting with [01/19] drm/i915/execlists: Always clear
ring_pause if we do not submit
URL : https://patchwork.freedesktop.org/series/62612/
State : success
== Summary ==
CI Bug Log - changes from CI_DRM_6331 -> Patchwork_13401
== Series Details ==
Series: series starting with [01/19] drm/i915/execlists: Always clear
ring_pause if we do not submit
URL : https://patchwork.freedesktop.org/series/62612/
State : warning
== Summary ==
$ dim sparse origin/drm-tip
Sparse version: v0.5.2
Commit: drm/i915/execlists: Always
== Series Details ==
Series: series starting with [01/19] drm/i915/execlists: Always clear
ring_pause if we do not submit
URL : https://patchwork.freedesktop.org/series/62612/
State : warning
== Summary ==
$ dim checkpatch origin/drm-tip
64d75313cafc drm/i915/execlists: Always clear
96 matches
Mail list logo
