- 30 Mar, 2017 17 commits
-
-
Chris Wilson authored
We pretty print the name of an engine in several places, mostly for debug, but also in the GPU hang report. Using "ring" in the name is archaic (we call those engines now to differentiate them from the multiple rings of commands we execute on each engine), quite verbose and often tautological. We run out of room in our GPU hang report for instance if we have more than a couple of engines hung simultaneously. Bit the bullet and update the strings to reflect the common internal names. Signed-off-by:
Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by:
Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Reviewed-by:
Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/20170330134820.12273-1-chris@chris-wilson.co.uk
-
Chris Wilson authored
Michał Winiarski pointed out that the debugging infrastructure (such as trace_dma_fence_release) likes to pretty print the timeline name, long after we have freed the timeline. Our timelines currently live as part of the GTT (due to the strict ordering we currently use through each) which belong to the context. We aim to free the context and release its hardware resources as soon as we able to (i.e. when the last fence/request using it has been signaled and retired). As the .get_timeline_name is purely a debug feature, rather than extending the lifetime of the context, or splitting it into many different release phases just to keep the name around, replace the timeline name with a constant after the fence has been signaled. This avoids the potential use-after-free. Reported-by:
Krzysztof Olinski <krzysztof.e.olinski@intel.com> Fixes: 80b204bc ("drm/i915: Enable multiple timelines") Signed-off-by:
Michał Winiarski <michal.winiarski@intel.com>
-
https://github.com/01org/gvt-linuxDaniel Vetter authored
gvt-next-2017-03-30 - Add mdev attribute group for per-vgpu info - Time slice based vGPU scheduling QoS support (Gao Ping) - Initial KBL support for E3 server (Han Xu) - other misc. Link: http://patchwork.freedesktop.org/patch/msgid/20170330100516.dkavi3rtlsmnoepi@zhen-hp.sh.intel.comSigned-off-by:
Daniel Vetter <daniel.vetter@ffwll.ch>
-
Chris Wilson authored
Since commit 1233e2db ("drm/i915: Move object backing storage manipulation to its own locking"), i915_gem_object_put_pages() and specifically the i915_gem_gtt_finish_pages() may be called from outside of the struct_mutex and so we can no longer pass I915_WAIT_LOCKED to i915_gem_wait_for_idle. Fixes: 1233e2db ("drm/i915: Move object backing storage manipulation to its own locking") Signed-off-by:
-
Oscar Mateo authored
The should happen as soon as possible, but always within the logic that depends on it (and not interrupting the top-level driver control flow). Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Signed-off-by:
Oscar Mateo <oscar.mateo@intel.com> Reviewed-by:
-
Jani Nikula authored
Leave more breadcrumbs for debuggers. Reviewed-by:
Bob Paauwe <bob.j.paauwe@intel.com> Signed-off-by:
Jani Nikula <jani.nikula@intel.com> Link: http://patchwork.freedesktop.org/patch/msgid/1490783578-6065-3-git-send-email-jani.nikula@intel.com
-
Jani Nikula authored
Seems more sensible this way, and reduces indent for the more common case. Reviewed-by:
-
Jani Nikula authored
Reduce indent. No functional changes. Reviewed-by:
-
Ping Gao authored
The timeslice usage will determine vGPU whether has chance to schedule or not at every vGPU switch checkpoint. Signed-off-by:
Ping Gao <ping.a.gao@intel.com> Reviewed-by:
Kevin Tian <kevin.tian@intel.com> Signed-off-by:
Zhenyu Wang <zhenyuw@linux.intel.com>
-
Ping Gao authored
vGPU resource is allocated by scheduler. To account for non-allocated free cycles, we create an idle vGPU as the placeholder similar to idle task concept, which is useful to handle some corner cases in scheduling policy. Signed-off-by:
-
Ping Gao authored
This method tries to guarantee precision in second level, with the adjustment conducted in every 100ms. At the end of each vGPU switch calculate the sched time and subtract it from the time slice allocated; the allocated time slice for every 100ms together with remaining timeslice, will be used to decide how much timeslice allocated to this vGPU in the next 100ms slice, with the end goal to guarantee weight ratio in second level. Signed-off-by:
-
Ping Gao authored
The weight defines proportional control of physical GPU resource shared between vGPUs. So far the weight is tied to a specific vGPU type, i.e when creating multiple vGPUs with different types, they will inherit different weights. e.g. The weight of type GVTg_V5_2 is 8, the weight of type GVTg_V5_4 is 4, so vGPU of type GVTg_V5_2 has double vGPU resource of vGPU type GVTg_V5_4. TODO: allow user control the weight setting in the future. Signed-off-by:
-
Ping Gao authored
Factor out the scheduler to a more clear structure, the basic logic is to find out next vGPU first and then schedule it. vGPUs were ordered in a LRU list, scheduler scan from the LRU list head and choose the first vGPU who has pending workload. Signed-off-by:
-
Ping Gao authored
Add some statistic routine to collect the time when vGPU is scheduled in/out and the time of the last ctx submission. Signed-off-by:
-
Ping Gao authored
Currently the scheduler is triggered by delayed_work, which doesn't provide precision at microsecond level. Move to hrtimer instead for more accurate control. Signed-off-by:
-
Tina Zhang authored
The variable info is never NULL, which is checked by the caller. This patch removes the redundant info NULL check logic. Fixes: 695fbc08 ("drm/i915/gvt: replace the gvt_err with gvt_vgpu_err") Signed-off-by:
Tina Zhang <tina.zhang@intel.com> Signed-off-by:
-
Zhenyu Wang authored
From commit d1a513be ("drm/i915/gvt: add resolution definition for vGPU type"), small type has been restricted to small resolution, so not require larger high GM size any more. Change to smaller 384M for more VM creation with vGPU enabled which still perform reasonable workload. Fixes: d1a513be ("drm/i915/gvt: add resolution definition for vGPU type") Signed-off-by:
-
- 29 Mar, 2017 15 commits
-
-
Ben Widawsky authored
They're the same, so use the one which makes more sense. Signed-off-by:
Ben Widawsky <ben@bwidawsk.net> Link: http://patchwork.freedesktop.org/patch/msgid/20170324212950.2206-1-ben@bwidawsk.netReviewed-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Signed-off-by:
-
Chris Wilson authored
This reverts commit 6c943de6 ("drm/i915: Skip execlists_dequeue() early if the list is empty"). The validity of using READ_ONCE there depends upon having a mb to coordinate the assignment of engine->execlist_first inside submit_request() and checking prior to taking the spinlock in execlists_dequeue(). We wrote "the update to TASKLET_SCHED incurs a memory barrier making this cross-cpu checking safe", but failed to notice that this mb was *conditional* on the execlists being ready, i.e. there wasn't the required mb when it was most necessary! We could install an unconditional memory barrier to fixup the READ_ONCE(): diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c index 7dd732cb9f57..1ed164b16d44 100644 --- a/drivers/gpu/drm/i915/intel_lrc.c +++ b/drivers/gpu/drm/i915/intel_lrc.c @@ -616,6 +616,7 @@ static void execlists_submit_request(struct drm_i915_gem_request *request) if (insert_request(&request->priotree, &engine->execlist_queue)) { engine->execlist_first = &request->priotree.node; + smp_wmb(); if (execlists_elsp_ready(engine)) But we have opted to remove the race as it should be rarely effective, and saves us having to explain the necessary memory barriers which we quite clearly failed at. Reported-and-tested-by:
-
Chris Wilson authored
Unlocking is dangerous. In this case we combine an early update to the out-of-queue request, because we know that it will be inserted into the correct FIFO priority-ordered slot when it becomes ready in the future. However, given sufficient enthusiasm, it may become ready as we are continuing to reschedule, and so may gazump the FIFO if we have since dropped its spinlock. The result is that it may be executed too early, before its dependencies. v2: Move all work into the second phase over the topological sort. This removes the shortcut on the out-of-rbtree request to ensure that we only adjust its priority after adjusting all of its dependencies. Fixes: 20311bd3 ("drm/i915/scheduler: Execute requests in order of priorities") Testcase: igt/gem_exec_whisper Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com>
-
Michal Wajdeczko authored
We can't sometimes use these macros in other headers due to include and definition order. As i915_utils.h already contains other helper macros move these macros there. v2: checkpatch cleanup for WARN() macro. Signed-off-by:
Michal Wajdeczko <michal.wajdeczko@intel.com> Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Link: http://patchwork.freedesktop.org/patch/msgid/20170328084513.174200-1-michal.wajdeczko@intel.comSigned-off-by:
-
Tina Zhang authored
intel_shadow_wa_ctx is a field of intel_vgpu_workload. container_of() can be used to refine the relation-ship between intel_shadow_wa_ctx and intel_vgpu_workload. This patch removes the useless dereference. v2. add "drm/i915/gvt" prefix. (Zhenyu) Signed-off-by:
-
Xu Han authored
Turn on KBL WS platform support in gvt-g. More platforms would be enabled, after validate. Signed-off-by:
Xu Han <xu.han@intel.com> Signed-off-by:
-
Xu Han authored
Extend function dispatch logic to support KBL platform. Signed-off-by:
-
Xu Han authored
Add some KBL specially registers to save/restore list. Signed-off-by:
-
Xu Han authored
Update MMIO handle policy to KBL platform. Signed-off-by:
-
Xu Han authored
Add KBL platform definition. Signed-off-by:
-
Zhenyu Wang authored
This adds initial attribute group for mdev to hold vGPU related for each mdev device, currently just vGPU id is shown. v2: rename group name as "intel_vgpu" Signed-off-by:
-
Pei Zhang authored
GVT-g will emulate a fixed DPCD data to VM for DP/eDP panel. Update this data to latest DP1.2 with the maximum lane bandwidth of 5.4G/s to support 4K resolution in VM. V3: modify patch comment V2: add inline comment to describe the dpcd_fix_data. Signed-off-by:
Pei Zhang <pei.zhang@intel.com> Signed-off-by:
-
Weinan Li authored
Initialize the correct vreg for virtual monitor. Set PG0/1/2 distribution and fuse download done in SKL_FUSE_STATUS. Set PLL_ENABLE and PLL_LOCK in LCPLL_CTL. Guest may need to check these registers for display monitor detection on Skylake platforms. Signed-off-by:
Weinan Li <weinan.z.li@intel.com> Signed-off-by:
-
Daniel Vetter authored
Backmerge drm-next one more because Dave fumbled the conflict resolution slightly and I didn't notice it. We need Zhenyu's hotfix before he can assemble the gvt pull ... Signed-off-by:Daniel Vetter <daniel.vetter@intel.com>
-
Zhenyu Wang authored
From commit 73dec95e ("drm/i915: Emit to ringbuffer directly"), copy_gma_to_hva() now returns copied data length instead of 0, so need to change error return check for that. Note: Looks this is caused by backmerge conflict resolving, so 4.11-rc4 is not impacted as commit 73dec95e ("drm/i915: Emit to ringbuffer directly") is not in 4.11. But need to fix this before I can apply 4.12 stuff against drm-intel-next correctly. Fixes: e5c1ff14 ("Backmerge tag 'v4.11-rc4' into drm-next") Cc: Dave Airlie <airlied@redhat.com> Cc: Tina Zhang <tina.zhang@intel.com> Signed-off-by:
Dave Airlie <airlied@redhat.com>
-
- 28 Mar, 2017 8 commits
-
-
Jani Nikula authored
Several major vendor USB-C->HDMI converters, in particular the DA200, fail to recover a 5.4 GHz 1 lane signal if the link N is greater than 0x80000. The link M and N depend on the pixel clock and link clock ratio. With current code link N exceeds 0x80000 only when link clock >= 540000 kHz. Except for the eDP intermediate link clocks, at least the four least significant bits are always zero. Just one bit shift right would be enough to bring even the DP 1.4 810000 kHz link clock under 0x80000 link N. The pixel clock for modes that require a link clock >= 540000 kHz would also have several least significant bits zero. Unless the user provides a mode with an odd pixel clock value, we can reduce the numbers to reach the goal, with no loss in precision. The DP spec even mentions sources making choices that "allow for static and relatively small Mvid and Nvid values", thus reducing the link M/N regardless of the sink in question seems justified. Everything here is based on the work and information gathered by Clint Taylor <clinton.a.taylor@intel.com>. This is just an iteration to reduce the parameters regardless of lane count, link rate, or sink. Reference: http://patchwork.freedesktop.org/patch/msgid/1490225256-11667-1-git-send-email-clinton.a.taylor@intel.com Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=93578Tested-by:
Mads <mads@ab3.no> Tested-by:
PJ <foobar@pjmodos.net> Tested-by:
François Guerraz <kubrick@fgv6.net> Tested-by:
Lev Popov <leo@nabam.net> Tested-by:
Igor Krivenko <igor.s.krivenko@gmail.com> Tested-by:
Clint Taylor <clinton.a.taylor@intel.com> Reviewed-by:
Dhinakaran Pandiyan <dhinakaran.pandiyan@intel.com> Acked-by:
-
Chris Wilson authored
Capturing GPU state requires the device to be awake in order to read registers. Normally, this is taken along the error handler, but for the direct debugfs access, we cannot make assumptions about the current device state and so either need to wake it up, or abort. Fixes: 5a4c6f1b ("drm/i915: The return of i915_gpu_info to debugfs") Testcase: igt/pm_rpm/debugfs-read Signed-off-by:
Mika Kuoppala <mika.kuoppala@intel.com>
-
Imre Deak authored
We don't expect the core runtime PM get helpers to return any error, so add a WARN for this. Also print the return value for all the callsites to help debugging. v2: - Don't call pm_runtime_get_sync() as part of initing locals. (Chris) Signed-off-by:
Imre Deak <imre.deak@intel.com> Reviewed-by:
-
Matthew Auld authored
Don't throw a warning if we are given an invalid property id. While here let's also bring back Robert' original idea of catching unhandled enumeration values at compile time. Fixes: eec688e1 ("drm/i915: Add i915 perf infrastructure") Signed-off-by:
Matthew Auld <matthew.auld@intel.com> Cc: Robert Bragg <robert@sixbynine.org> Reviewed-by:
Robert Bragg <robert@sixbynine.org> Signed-off-by:
-
Jani Nikula authored
The old URL works but gives 301 Moved Permanently. Update. Reviewed-by:
-
Daniel Vetter authored
Backmerge drm-next to get at -rc4, which we need to land the 4.12 gvt patches. Signed-off-by: -
Dave Airlie authored
Linux 4.11-rc4 The i915 GVT team need the rc4 code to base some more code on.
-
Matthew Auld authored
If we were to ever encounter a sample_flags mismatch we need to ensure we destroy the stream when we bail. Fixes: d7965152 ("drm/i915: Enable i915 perf stream for Haswell OA unit") Signed-off-by:
-
