drm/i915: Flip guc_id allocation partition
Subsystems (2)
PATCH 1/1 drm/i915: Flip guc_id allocation partition
Move the multi-lrc guc_id from the lower allocation partition (0 to number of multi-lrc guc_ids) to upper allocation partition (number of single-lrc to max guc_ids).
This will help when a native driver transitions to a PF after driver load time. If the perma-pin guc_ids (kernel contexts) are in a low range it is easy reduce total number of guc_ids as the allocated slrc are in a valid range the mlrc range moves to an unused range. Assuming no mlrc are allocated and few slrc are used the native to PF transition is seamless for the guc_id resource.
v2:
(Michal / Tvrtko)
- Add an explaination to commit message of why this patch is needed
(Michal / Piotr)
- Replace marcos with functions
(Michal)
- Rework logic flow in new_mlrc_guc_id
- Unconditionally call bitmap_free
--- .../gpu/drm/i915/gt/uc/intel_guc_submission.c | 88 +++++++++++++------ 1 file changed, 63 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +63 -25
@@ -138,17 +138,6 @@ guc_create_parallel(struct intel_engine_cs **engines,#define GUC_REQUEST_SIZE 64/* bytes */ -/* -* We reserve 1/16 of the guc_ids for multi-lrc as these need to be contiguous -* per the GuC submission interface. A different allocation algorithm is used -* (bitmap vs. ida) between multi-lrc and single-lrc hence the reason to -* partition the guc_id space. We believe the number of multi-lrc contexts in -* use should be low and 1/16 should be sufficient. Minimum of 32 guc_ids for -* multi-lrc. -*/ -#define NUMBER_MULTI_LRC_GUC_ID (guc) \ - ((guc)->submission_state.num_guc_ids / 16) -/* * Below is a set of functions which control the GuC scheduling state which * require a lock.
@@ -1777,11 +1766,6 @@ int intel_guc_submission_init(struct intel_guc *guc)INIT_WORK (&guc->submission_state.destroyed_worker, destroyed_worker_func); - guc->submission_state.guc_ids_bitmap = -bitmap_zalloc (NUMBER_MULTI_LRC_GUC_ID (guc),GFP_KERNEL ); -if (!guc->submission_state.guc_ids_bitmap) -return -ENOMEM ; -spin_lock_init (&guc->timestamp.lock);INIT_DELAYED_WORK (&guc->timestamp.work, guc_timestamp_ping); guc->timestamp.ping_delay = (POLL_TIME_CLKS / gt->clock_frequency +1 ) *HZ ;
@@ -1864,6 +1848,57 @@ static void guc_submit_request(struct i915_request *rq)spin_unlock_irqrestore (&sched_engine->lock, flags); } +/* +* We reserve 1/16 of the guc_ids for multi-lrc as these need to be contiguous +* per the GuC submission interface. A different allocation algorithm is used +* (bitmap vs. ida) between multi-lrc and single-lrc hence the reason to +* partition the guc_id space. We believe the number of multi-lrc contexts in +* use should be low and 1/16 should be sufficient. +*/ +#define MLRC_GUC_ID_RATIO 16 + +static int number_mlrc_guc_id (struct intel_guc *guc) +{ +return guc->submission_state.num_guc_ids /MLRC_GUC_ID_RATIO ; +} + +static int number_slrc_guc_id (struct intel_guc *guc) +{ +return guc->submission_state.num_guc_ids -number_mlrc_guc_id (guc); +} + +static int mlrc_guc_id_base (struct intel_guc *guc) +{ +return number_slrc_guc_id (guc); +} + +static int new_mlrc_guc_id (struct intel_guc *guc,struct intel_context *ce) +{ +int ret; + +GEM_BUG_ON (!intel_context_is_parent (ce)); +GEM_BUG_ON (!guc->submission_state.guc_ids_bitmap); + + ret =bitmap_find_free_region (guc->submission_state.guc_ids_bitmap, +number_mlrc_guc_id (guc), +order_base_2 (ce->parallel.number_children + +1 )); +if (unlikely (ret <0 )) +return ret; + +return ret +mlrc_guc_id_base (guc); +} + +static int new_slrc_guc_id (struct intel_guc *guc,struct intel_context *ce) +{ +GEM_BUG_ON (intel_context_is_parent (ce)); + +return ida_simple_get (&guc->submission_state.guc_ids, +0 ,number_slrc_guc_id (guc), +GFP_KERNEL | __GFP_RETRY_MAYFAIL | + __GFP_NOWARN); +} +static int new_guc_id (struct intel_guc *guc,struct intel_context *ce) {int ret;
@@ -1871,16 +1906,10 @@ static int new_guc_id(struct intel_guc *guc, struct intel_context *ce)GEM_BUG_ON (intel_context_is_child (ce));if (intel_context_is_parent (ce)) - ret =bitmap_find_free_region (guc->submission_state.guc_ids_bitmap, -NUMBER_MULTI_LRC_GUC_ID (guc), -order_base_2 (ce->parallel.number_children - +1 )); + ret =new_mlrc_guc_id (guc, ce);else - ret =ida_simple_get (&guc->submission_state.guc_ids, -NUMBER_MULTI_LRC_GUC_ID (guc), - guc->submission_state.num_guc_ids, -GFP_KERNEL | __GFP_RETRY_MAYFAIL | - __GFP_NOWARN); + ret =new_slrc_guc_id (guc, ce); +if (unlikely (ret <0 ))return ret;
@@ -1990,6 +2019,15 @@ static int pin_guc_id(struct intel_guc *guc, struct intel_context *ce)GEM_BUG_ON (atomic_read (&ce->guc_id.ref)); +/* Outside spin lock so we can sleep on alloc */ +if (unlikely (intel_context_is_parent (ce) && + !guc->submission_state.guc_ids_bitmap)) { + guc->submission_state.guc_ids_bitmap = +bitmap_zalloc (number_mlrc_guc_id (guc),GFP_KERNEL ); +if (!guc->submission_state.guc_ids_bitmap) +return -ENOMEM ; + } + try_again:spin_lock_irqsave (&guc->submission_state.lock, flags);
Signature · 1 line
2.34.1
Patchwork
CI[Intel-gfx] ✗ Fi.CI.BAT: failure for drm/i915: Flip guc_id allocation partition (rev2)
== Series Details ==
== Series Details ==
Series: drm/i915: Flip guc_id allocation partition (rev2)
URL : https://patchwork.freedesktop.org/series/98751/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_11075 -> Patchwork_21990
====================================================
Summary
-------
**FAILURE**
Serious unknown changes coming with Patchwork_21990 absolutely need to be verified manually.
If you think the reported changes have nothing to do with the changes introduced in Patchwork_21990, please notify your bug team to allow them to document this new failure mode, which will reduce false positives in CI.
External URL: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/index.html
Participating hosts (44 -> 42)
------------------------------
Missing (2): fi-bsw-cyan fi-pnv-d510
Possible new issues
-------------------
Here are the unknown changes that may have been introduced in Patchwork_21990:
### IGT changes ###
#### Possible regressions ####
* igt@i915_selftest@live@guc_multi_lrc:
- bat-dg1-6: [PASS][1] -> [DMESG-FAIL][2]
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/bat-dg1-6/igt@i915_selftest@live@guc_multi_lrc.html
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-dg1-6/igt@i915_selftest@live@guc_multi_lrc.html
Known issues
------------
Here are the changes found in Patchwork_21990 that come from known issues:
### IGT changes ###
#### Issues hit ####
* igt@amdgpu/amd_basic@semaphore:
- fi-bdw-5557u: NOTRUN -> [SKIP][3] ([fdo#109271]) +31 similar issues
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-bdw-5557u/igt@amdgpu/[email protected]
* igt@gem_flink_basic@bad-flink:
- fi-skl-6600u: [PASS][4] -> [FAIL][5] ([i915#4547])
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/fi-skl-6600u/igt@[email protected]
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-skl-6600u/igt@[email protected]
* igt@i915_selftest@live@hangcheck:
- bat-dg1-5: NOTRUN -> [DMESG-FAIL][6] ([i915#4494])
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-dg1-5/igt@i915_selftest@[email protected]
- fi-hsw-4770: [PASS][7] -> [INCOMPLETE][8] ([i915#4785])
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/fi-hsw-4770/igt@i915_selftest@[email protected]
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-hsw-4770/igt@i915_selftest@[email protected]
* igt@kms_chamelium@dp-crc-fast:
- fi-bdw-5557u: NOTRUN -> [SKIP][9] ([fdo#109271] / [fdo#111827]) +8 similar issues
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-bdw-5557u/igt@[email protected]
* igt@kms_frontbuffer_tracking@basic:
- fi-cml-u2: [PASS][10] -> [DMESG-WARN][11] ([i915#4269])
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/fi-cml-u2/igt@[email protected]
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-cml-u2/igt@[email protected]
* igt@runner@aborted:
- fi-skl-6600u: NOTRUN -> [FAIL][12] ([i915#4312])
[12]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-skl-6600u/igt@[email protected]
- fi-hsw-4770: NOTRUN -> [FAIL][13] ([fdo#109271] / [i915#1436] / [i915#4312])
[13]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-hsw-4770/igt@[email protected]
- bat-dg1-6: NOTRUN -> [FAIL][14] ([i915#4214] / [i915#4312])
[14]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-dg1-6/igt@[email protected]
#### Possible fixes ####
* igt@core_auth@basic-auth:
- {bat-jsl-2}: [DMESG-WARN][15] ([i915#4562]) -> [PASS][16]
[15]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/bat-jsl-2/igt@[email protected]
[16]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-jsl-2/igt@[email protected]
* igt@gem_exec_suspend@basic-s3@smem:
- fi-bdw-5557u: [INCOMPLETE][17] ([i915#146]) -> [PASS][18]
[17]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/fi-bdw-5557u/igt@gem_exec_suspend@[email protected]
[18]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-bdw-5557u/igt@gem_exec_suspend@[email protected]
* igt@i915_selftest@live@gt_engines:
- bat-dg1-5: [INCOMPLETE][19] ([i915#4418]) -> [PASS][20]
[19]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/bat-dg1-5/igt@i915_selftest@live@gt_engines.html
[20]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-dg1-5/igt@i915_selftest@live@gt_engines.html
* igt@i915_selftest@live@hangcheck:
- {fi-ehl-2}: [DMESG-FAIL][21] -> [PASS][22]
[21]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/fi-ehl-2/igt@i915_selftest@[email protected]
[22]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/fi-ehl-2/igt@i915_selftest@[email protected]
* igt@i915_selftest@live@hugepages:
- {bat-adlp-6}: [DMESG-WARN][23] ([i915#1888]) -> [PASS][24]
[23]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_11075/bat-adlp-6/igt@i915_selftest@[email protected]
[24]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/bat-adlp-6/igt@i915_selftest@[email protected]
{name}: This element is suppressed. This means it is ignored when computing
the status of the difference (SUCCESS, WARNING, or FAILURE).
[fdo#109271]: https://bugs.freedesktop.org/show_bug.cgi?id=109271
[fdo#111827]: https://bugs.freedesktop.org/show_bug.cgi?id=111827
[i915#1436]: https://gitlab.freedesktop.org/drm/intel/issues/1436
[i915#146]: https://gitlab.freedesktop.org/drm/intel/issues/146
[i915#1888]: https://gitlab.freedesktop.org/drm/intel/issues/1888
[i915#4214]: https://gitlab.freedesktop.org/drm/intel/issues/4214
[i915#4269]: https://gitlab.freedesktop.org/drm/intel/issues/4269
[i915#4312]: https://gitlab.freedesktop.org/drm/intel/issues/4312
[i915#4418]: https://gitlab.freedesktop.org/drm/intel/issues/4418
[i915#4494]: https://gitlab.freedesktop.org/drm/intel/issues/4494
[i915#4547]: https://gitlab.freedesktop.org/drm/intel/issues/4547
[i915#4562]: https://gitlab.freedesktop.org/drm/intel/issues/4562
[i915#4785]: https://gitlab.freedesktop.org/drm/intel/issues/4785
Build changes
-------------
* Linux: CI_DRM_11075 -> Patchwork_21990
CI-20190529: 20190529
CI_DRM_11075: 9b371f79eeb9a3867d2483e4038f22f89a854886 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_6326: ec75f64fcbcf4aac58fbf1bf629e8f59b19db4ce @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
Patchwork_21990: 562c51a5dc5cad4239bd80a1f64a01e1092c1830 @ git://anongit.freedesktop.org/gfx-ci/linux
== Linux commits ==
562c51a5dc5c drm/i915: Flip guc_id allocation partition
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_21990/index.html