drm/i915/mtl: Add support for 32 bit OAG formats in MTL
Subsystems (2)
PATCH 1/1 drm/i915/mtl: Add support for 32 bit OAG formats in MTL
As part of OA support for MTL,
- Enable 32 bit OAG formats for MTL.
- 0x200c is repurposed on MTL. Use a separate mux table to verify oa configs passed by user.
- Similar to ACM, OA/CS timestamp is mismatched on MTL. Add MTL to the WA.
- On MTL, gt->scratch was using stolen lmem. An MI_SRM to stolen lmem is hanging. Add a page in noa_wait BO to save/restore GPR registers for the noa_wait logic.
--- drivers/gpu/drm/i915/gt/intel_gt_types.h | 6 --- drivers/gpu/drm/i915/i915_perf.c | 49 ++++++++++++++++++------ 2 files changed, 38 insertions(+), 17 deletions(-)
drivers/gpu/drm/i915/gt/intel_gt_types.h +0 -6
@@ -296,12 +296,6 @@ enum intel_gt_scratch_field {/* 8 bytes */ INTEL_GT_SCRATCH_FIELD_COHERENTL3_WA = 256, - -/* 6 * 8 bytes */ -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR =2048 , - -/* 4 bytes */ - INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 = 2096, };#endif /* __INTEL_GT_TYPES_H__ */
drivers/gpu/drm/i915/i915_perf.c +38 -11
111 lines not on this page · raw message
Signature · 1 line
2.36.1
Patchwork
CI[Intel-gfx] ✗ Fi.CI.SPARSE: warning for drm/i915/mtl: Add support for 32 bit OAG formats in MTL
== Series Details ==
== Series Details ==
Series: drm/i915/mtl: Add support for 32 bit OAG formats in MTL
URL : https://patchwork.freedesktop.org/series/111411/
State : warning
== Summary ==
Error: dim sparse failed
Sparse version: v0.6.2
Fast mode used, each commit won't be checked separately.
Patchwork
CI[Intel-gfx] ✗ Fi.CI.BAT: failure for drm/i915/mtl: Add support for 32 bit OAG formats in MTL
In reply to Umesh Nerlige Ramappa== Series Details ==
== Series Details ==
Series: drm/i915/mtl: Add support for 32 bit OAG formats in MTL
URL : https://patchwork.freedesktop.org/series/111411/
State : failure
== Summary ==
CI Bug Log - changes from CI_DRM_12440 -> Patchwork_111411v1
====================================================
Summary
-------
**FAILURE**
Serious unknown changes coming with Patchwork_111411v1 absolutely need to be verified manually.
If you think the reported changes have nothing to do with the changes introduced in Patchwork_111411v1, 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_111411v1/index.html
Participating hosts (34 -> 31)
------------------------------
Missing (3): bat-kbl-2 bat-adlp-4 bat-jsl-3
Possible new issues
-------------------
Here are the unknown changes that may have been introduced in Patchwork_111411v1:
### CI changes ###
#### Possible regressions ####
* boot:
- fi-ilk-650: [PASS][1] -> [FAIL][2]
[1]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_12440/fi-ilk-650/boot.html
[2]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/fi-ilk-650/boot.html
### IGT changes ###
#### Suppressed ####
The following results come from untrusted machines, tests, or statuses. They do not affect the overall result.
* igt@gem_exec_suspend@basic-s0@smem:
- {bat-rpls-1}: NOTRUN -> [DMESG-WARN][3]
[3]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/bat-rpls-1/igt@gem_exec_suspend@[email protected]
Known issues
------------
Here are the changes found in Patchwork_111411v1 that come from known issues:
### IGT changes ###
#### Issues hit ####
* igt@core_hotunplug@unbind-rebind:
- fi-apl-guc: [PASS][4] -> [INCOMPLETE][5] ([i915#7073])
[4]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_12440/fi-apl-guc/igt@[email protected]
[5]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/fi-apl-guc/igt@[email protected]
#### Possible fixes ####
* igt@gem_tiled_blits@basic:
- fi-pnv-d510: [SKIP][6] ([fdo#109271]) -> [PASS][7] +1 similar issue
[6]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_12440/fi-pnv-d510/igt@[email protected]
[7]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/fi-pnv-d510/igt@[email protected]
* igt@i915_module_load@reload:
- {bat-rpls-2}: [INCOMPLETE][8] ([i915#6434]) -> [PASS][9]
[8]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_12440/bat-rpls-2/igt@[email protected]
[9]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/bat-rpls-2/igt@[email protected]
* igt@i915_selftest@live@reset:
- {bat-rpls-1}: [DMESG-FAIL][10] ([i915#4983]) -> [PASS][11]
[10]: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_12440/bat-rpls-1/igt@i915_selftest@[email protected]
[11]: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/bat-rpls-1/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#4258]: https://gitlab.freedesktop.org/drm/intel/issues/4258
[i915#4312]: https://gitlab.freedesktop.org/drm/intel/issues/4312
[i915#4613]: https://gitlab.freedesktop.org/drm/intel/issues/4613
[i915#4983]: https://gitlab.freedesktop.org/drm/intel/issues/4983
[i915#6434]: https://gitlab.freedesktop.org/drm/intel/issues/6434
[i915#6559]: https://gitlab.freedesktop.org/drm/intel/issues/6559
[i915#6997]: https://gitlab.freedesktop.org/drm/intel/issues/6997
[i915#7073]: https://gitlab.freedesktop.org/drm/intel/issues/7073
[i915#7077]: https://gitlab.freedesktop.org/drm/intel/issues/7077
[i915#7346]: https://gitlab.freedesktop.org/drm/intel/issues/7346
Build changes
-------------
* Linux: CI_DRM_12440 -> Patchwork_111411v1
CI-20190529: 20190529
CI_DRM_12440: d21d6474a37e5d43075a24668807ea40a7ee9fc1 @ git://anongit.freedesktop.org/gfx-ci/linux
IGT_7073: d021d66e389f4a759dc749b5f74f278ecd2e6cbf @ https://gitlab.freedesktop.org/drm/igt-gpu-tools.git
Patchwork_111411v1: d21d6474a37e5d43075a24668807ea40a7ee9fc1 @ git://anongit.freedesktop.org/gfx-ci/linux
### Linux commits
1127d659634b drm/i915/mtl: Add support for 32 bit OAG formats in MTL
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_111411v1/index.html
Quoted 10 lines · Umesh Nerlige Ramappa
As part of OA support for MTL,
- Enable 32 bit OAG formats for MTL.
-0x200cis repurposed on MTL. Use a separate mux table to verify oa configs passed by user.
- Similar to ACM, OA/CS timestamp is mismatched on MTL. Add MTL to the WA.
- On MTL,gt->scratchwas using stolen lmem. AnMI_SRMto stolen lmem is hanging. Add a page innoa_waitBO to save/restore GPR registers for thenoa_waitlogic.
why are all these changes squashed in a single patch?
Quoted 39 lines
Signed-off-by Umesh Nerlige Ramappa <[email protected]>---
drivers/gpu/drm/i915/gt/intel_gt_types.h| 6 ---drivers/gpu/drm/i915/i915_perf.c| 49 ++++++++++++++++++------2 files changed, 38 insertions(+), 17 deletions(-)drivers/gpu/drm/i915/gt/intel_gt_types.h +0
-6@@ -296,12 +296,6 @@ enum intel_gt_scratch_field {/* 8 bytes */ INTEL_GT_SCRATCH_FIELD_COHERENTL3_WA = 256, - -/* 6 * 8 bytes */ -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR =2048 , - -/* 4 bytes */ - INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 = 2096, };#endif /* __INTEL_GT_TYPES_H__ */ drivers/gpu/drm/i915/i915_perf.c +1
-2@@ -1842,8 +1842,7 @@ static u32 *save_restore_register(struct i915_perf_stream *stream, u32 *cs,for (d =0 ; d < dword_count; d++) { *cs++ = cmd; *cs++ =i915_mmio_reg_offset (reg) +4 * d; - *cs++ =intel_gt_scratch_offset (stream->engine->gt, - offset) +4 * d; + *cs++ =i915_ggtt_offset (stream->noa_wait) + offset +4 * d; *cs++ =0 ; }
drivers/gpu/drm/i915/i915_perf.c static int alloc_noa_wait(struct i915_perf_stream *stream) · line 1881
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1@@ -1876,7 +1875,13 @@ static int alloc_noa_wait(struct i915_perf_stream *stream) MI_PREDICATE_RESULT_2_ENGINE(base) :MI_PREDICATE_RESULT_1 (RENDER_RING_BASE ); - bo =i915_gem_object_create_internal (i915,4096 ); +/* +* gt->scratch was being used to save/restore the GPR registers, but on +* MTL the scratch uses stolen lmem. An MI_SRM to this memory region +* causes an engine hang. Instead allocate an additional page here to
humn.. is it because of the pte being wrong? "stolen lmem" in mtl is still system memory... do we know why we'd need this change?
Lucas De Marchi
Quoted 104 lines
drivers/gpu/drm/i915/i915_perf.c +12
-5+* save/restore GPR registers +*/ + bo =i915_gem_object_create_internal (i915,8192 );if (IS_ERR (bo)) {drm_err (&i915->drm,"Failed to allocate NOA wait batchbuffer\n" );@@ -1910,14 +1915,19 @@ static int alloc_noa_wait(struct i915_perf_stream *stream) goto err_unpin; } + stream->noa_wait = vma; + +#define GPR_SAVE_OFFSET 4096 +#define PREDICATE_SAVE_OFFSET 4160 +/* Save registers. */ for (i =0 ; i <N_CS_GPR ; i++) cs =save_restore_register ( stream, cs, true/* save */ ,CS_GPR (i), -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR +8 * i,2 ); +GPR_SAVE_OFFSET +8 * i,2 ); cs =save_restore_register ( stream, cs, true/* save */ , mi_predicate_result, -INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 ,1 ); +PREDICATE_SAVE_OFFSET ,1 );/* First timestamp snapshot location. */ ts0 = cs;@@ -2033,10 +2043,10 @@ static int alloc_noa_wait(struct i915_perf_stream *stream)for (i =0 ; i <N_CS_GPR ; i++) cs =save_restore_register ( stream, cs, false/* restore */ ,CS_GPR (i), -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR +8 * i,2 ); +GPR_SAVE_OFFSET +8 * i,2 ); cs =save_restore_register ( stream, cs, false/* restore */ , mi_predicate_result, -INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 ,1 ); +PREDICATE_SAVE_OFFSET ,1 );/* And return to the ring. */ *cs++ =MI_BATCH_BUFFER_END ;@@ -2046,7 +2056,6 @@ static int alloc_noa_wait(struct i915_perf_stream *stream)i915_gem_object_flush_map (bo);__i915_gem_object_release_map (bo); - stream->noa_wait = vma; goto out_ww;
err_unpin:drivers/gpu/drm/i915/i915_perf.c +21
-3@@ -3127,8 +3136,11 @@ get_sseu_config(struct intel_sseu *out_sseu,*/ u32 i915_perf_oa_timestamp_frequency (struct drm_i915_private *i915) { -/* Wa_18013179988:dg2 */ -if (IS_DG2 (i915)) { +/* +* Wa_18013179988:dg2 +* Wa_14015846243:mtl +*/ +if (IS_DG2 (i915) ||IS_METEORLAKE (i915)) {intel_wakeref_t wakeref;u32 reg, shift;@@ -4306,6 +4318,17 @@ static const struct i915_range gen12_oa_mux_regs[] = { {} }; +/* +* Ref: 14010536224: +* 0x20cc is repurposed on MTL, so use a separate array for MTL. +*/ +static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table (addr, gen7_oa_b_counters);@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs); }static u32 mask_reg_value (u32 reg,u32 val)@@ -4746,6 +4772,7 @@ static void oa_init_supported_formats(struct i915_perf *perf)break ;case INTEL_DG2 : +case INTEL_METEORLAKE :oa_format_add (perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8);oa_format_add (perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8);break ;Signature · 1 line
2.36.1
Quoted 12 lines · Lucas De Marchi
Quoted 10 lines · Umesh Nerlige Ramappa
As part of OA support for MTL,
- Enable 32 bit OAG formats for MTL.
-0x200cis repurposed on MTL. Use a separate mux table to verify oa configs passed by user.
- Similar to ACM, OA/CS timestamp is mismatched on MTL. Add MTL to the WA.
- On MTL,gt->scratchwas using stolen lmem. AnMI_SRMto stolen lmem is hanging. Add a page innoa_waitBO to save/restore GPR registers for thenoa_waitlogic.
Quoted 1 line
why are all these changes squashed in a single patch?
I don't have a good reason other than we had it this way to enable OA for MTL. I can break them into separate patches in the next revision.
Quoted 43 lines
Quoted 43 lines
Signed-off-by Umesh Nerlige Ramappa <[email protected]>---
drivers/gpu/drm/i915/gt/intel_gt_types.h| 6 ---drivers/gpu/drm/i915/i915_perf.c| 49 ++++++++++++++++++------2 files changed, 38 insertions(+), 17 deletions(-)drivers/gpu/drm/i915/gt/intel_gt_types.h +0
-6@@ -296,12 +296,6 @@ enum intel_gt_scratch_field {/* 8 bytes */ INTEL_GT_SCRATCH_FIELD_COHERENTL3_WA = 256, - -/* 6 * 8 bytes */ -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR =2048 , - -/* 4 bytes */ - INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 = 2096, };#endif /* __INTEL_GT_TYPES_H__ */ drivers/gpu/drm/i915/i915_perf.c +1
-2@@ -1842,8 +1842,7 @@ static u32 *save_restore_register(struct i915_perf_stream *stream, u32 *cs,for (d =0 ; d < dword_count; d++) { *cs++ = cmd; *cs++ =i915_mmio_reg_offset (reg) +4 * d; - *cs++ =intel_gt_scratch_offset (stream->engine->gt, - offset) +4 * d; + *cs++ =i915_ggtt_offset (stream->noa_wait) + offset +4 * d; *cs++ =0 ; }@@ -1876,7 +1875,13 @@ static int alloc_noa_wait(struct i915_perf_stream *stream) MI_PREDICATE_RESULT_2_ENGINE(base) :MI_PREDICATE_RESULT_1 (RENDER_RING_BASE );
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c static int alloc_noa_wait(struct i915_perf_stream *stream) · line 1881
Quoted 5 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1- bo =i915_gem_object_create_internal (i915,4096 ); +/* +* gt->scratch was being used to save/restore the GPR registers, but on +* MTL the scratch uses stolen lmem. An MI_SRM to this memory region +* causes an engine hang. Instead allocate an additional page here to humn.. is it because of the pte being wrong?"stolen lmem" in mtl is still system memory... do we know why we'd need this change?
This memory is accessed by the CS to save/restore some registers and that was causing a hang, so I guess mapping this memory to ggtt did not work at the time. I am not sure if this was fixed later, but using a memory that's owned by OA seems to work fine.
Thanks,
Umesh
Quoted 108 lines
Lucas De Marchi
Quoted 104 lines
drivers/gpu/drm/i915/i915_perf.c +12
-5+* save/restore GPR registers +*/ + bo =i915_gem_object_create_internal (i915,8192 );if (IS_ERR (bo)) {drm_err (&i915->drm,"Failed to allocate NOA wait batchbuffer\n" );@@ -1910,14 +1915,19 @@ static int alloc_noa_wait(struct i915_perf_stream *stream) goto err_unpin; } + stream->noa_wait = vma; + +#define GPR_SAVE_OFFSET 4096 +#define PREDICATE_SAVE_OFFSET 4160 +/* Save registers. */ for (i =0 ; i <N_CS_GPR ; i++) cs =save_restore_register ( stream, cs, true/* save */ ,CS_GPR (i), -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR +8 * i,2 ); +GPR_SAVE_OFFSET +8 * i,2 ); cs =save_restore_register ( stream, cs, true/* save */ , mi_predicate_result, -INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 ,1 ); +PREDICATE_SAVE_OFFSET ,1 );/* First timestamp snapshot location. */ ts0 = cs;@@ -2033,10 +2043,10 @@ static int alloc_noa_wait(struct i915_perf_stream *stream)for (i =0 ; i <N_CS_GPR ; i++) cs =save_restore_register ( stream, cs, false/* restore */ ,CS_GPR (i), -INTEL_GT_SCRATCH_FIELD_PERF_CS_GPR +8 * i,2 ); +GPR_SAVE_OFFSET +8 * i,2 ); cs =save_restore_register ( stream, cs, false/* restore */ , mi_predicate_result, -INTEL_GT_SCRATCH_FIELD_PERF_PREDICATE_RESULT_1 ,1 ); +PREDICATE_SAVE_OFFSET ,1 );/* And return to the ring. */ *cs++ =MI_BATCH_BUFFER_END ;@@ -2046,7 +2056,6 @@ static int alloc_noa_wait(struct i915_perf_stream *stream)i915_gem_object_flush_map (bo);__i915_gem_object_release_map (bo); - stream->noa_wait = vma; goto out_ww;
err_unpin:drivers/gpu/drm/i915/i915_perf.c +21
-3@@ -3127,8 +3136,11 @@ get_sseu_config(struct intel_sseu *out_sseu,*/ u32 i915_perf_oa_timestamp_frequency (struct drm_i915_private *i915) { -/* Wa_18013179988:dg2 */ -if (IS_DG2 (i915)) { +/* +* Wa_18013179988:dg2 +* Wa_14015846243:mtl +*/ +if (IS_DG2 (i915) ||IS_METEORLAKE (i915)) {intel_wakeref_t wakeref;u32 reg, shift;@@ -4306,6 +4318,17 @@ static const struct i915_range gen12_oa_mux_regs[] = { {} }; +/* +* Ref: 14010536224: +* 0x20cc is repurposed on MTL, so use a separate array for MTL. +*/ +static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table (addr, gen7_oa_b_counters);@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs); }static u32 mask_reg_value (u32 reg,u32 val)@@ -4746,6 +4772,7 @@ static void oa_init_supported_formats(struct i915_perf *perf)break ;case INTEL_DG2 : +case INTEL_METEORLAKE :oa_format_add (perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8);oa_format_add (perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8);break ;Signature · 1 line
2.36.1
Quoted 1 line · Umesh Nerlige Ramappa
Hi Umesh,
Overall looks ok, just a couple of questions below. Splitting the patches would be nice and easier to review, but I'm almost done with this one ;-)
drivers/gpu/drm/i915/i915_perf.c static int alloc_noa_wait(struct i915_perf_stream *stream) · line 1884
Quoted 12 lines
drivers/gpu/drm/i915/i915_perf.c +7
-1@@ -1876,7 +1875,13 @@ static int alloc_noa_wait(struct i915_perf_stream *stream) MI_PREDICATE_RESULT_2_ENGINE(base) :MI_PREDICATE_RESULT_1 (RENDER_RING_BASE ); - bo =i915_gem_object_create_internal (i915,4096 ); +/* +* gt->scratch was being used to save/restore the GPR registers, but on +* MTL the scratch uses stolen lmem. An MI_SRM to this memory region +* causes an engine hang. Instead allocate an additional page here to +* save/restore GPR registers +*/ + bo =i915_gem_object_create_internal (i915,8192 );
Do we know how much space was used in stream->noa_wait originally? Anyway allocating an additional page is not an issue so ok to skip the question.
drivers/gpu/drm/i915/i915_perf.c static void oa_init_supported_formats(struct i915_perf *perf) · line 4777
Quoted 7 lines
drivers/gpu/drm/i915/i915_perf.c +1
-0@@ -4746,6 +4772,7 @@ static void oa_init_supported_formats(struct i915_perf *perf)break ;case INTEL_DG2 : +case INTEL_METEORLAKE :oa_format_add (perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8);oa_format_add (perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8);
Do these formats also need to be added?
oa_format_add(perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8); oa_format_add(perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8); oa_format_add(perf, I915_OAR_FORMAT_A36u64_B8_C8); oa_format_add(perf, I915_OA_FORMAT_A38u64_R2u64_B8_C8);
Or these are not considered OAG formats?
Review static void oa_init_supported_formats(struct i915_perf *perf) · line 4778
Quoted 1 line
(quoted diff) +0
-0break;
Thanks.
--
Ashutosh
Quoted 10 lines · Ashutosh
drivers/gpu/drm/i915/i915_perf.c static void oa_init_supported_formats(struct i915_perf *perf) · line 4777
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +1
-0@@ -4746,6 +4772,7 @@ static void oa_init_supported_formats(struct i915_perf *perf)break ;case INTEL_DG2 : +case INTEL_METEORLAKE :oa_format_add (perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8);oa_format_add (perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8);Do these formats also need to be added?
Sorry!
Quoted 2 lines
oa_format_add(perf, I915_OAR_FORMAT_A32u40_A4u32_B8_C8); oa_format_add(perf, I915_OA_FORMAT_A24u40_A14u32_B8_C8);
These two are the same as above.
Quoted 2 lines
oa_format_add(perf, I915_OAR_FORMAT_A36u64_B8_C8); oa_format_add(perf, I915_OA_FORMAT_A38u64_R2u64_B8_C8);
And it seems these are not 32 bit (where 32 bit refers to the header size).
Quoted 1 line
Or these are not considered OAG formats?
So it's ok, no changes needed here.
Quoted 6 lines
Review static void oa_init_supported_formats(struct i915_perf *perf) · line 4778
Quoted 2 lines
(quoted diff) +0
-0break;Thanks.
--
Ashutosh
Quoted 4 lines · Umesh Nerlige Ramappa
+/*
+ * Ref: 14010536224:
+ *0x20ccis repurposed on MTL, so use a separate array for MTL.
Wondering if it was WAIT_FOR_RC6_EXIT (seen in gen12_oa_mux_regs) which moved elsewhere and if that needs to be added to the array below too?
Quoted 12 lines
+ */
drivers/gpu/drm/i915/i915_perf.c +7
-0+static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table(addr, gen7_oa_b_counters);
drivers/gpu/drm/i915/i915_perf.c static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr) · line 4378
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs);
But otherwise this is:
If you decide to split the patches, please add my R-b on all the split patches.
Quoted 8 lines · Ashutosh
Quoted 4 lines · Umesh Nerlige Ramappa
+/*
+ * Ref: 14010536224:
+ *0x20ccis repurposed on MTL, so use a separate array for MTL.Wondering if it was
WAIT_FOR_RC6_EXIT(seen ingen12_oa_mux_regs) which moved elsewhere and if that needs to be added to the array below too?
WAIT_FOR_RC6_EXIT (0x20cc) moved elsewhere so it should be "removed" from mtl oa mux array.
Quoted 24 lines
Quoted 13 lines
+ */
drivers/gpu/drm/i915/i915_perf.c +7
-0+static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table(addr, gen7_oa_b_counters);drivers/gpu/drm/i915/i915_perf.c static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr) · line 4378
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs);But otherwise this is:
Quoted 1 line
Reviewed-by Ashutosh Dixit <[email protected]>
I will break them into separate patches though. If the diff is identical, I will carry over your R-b on the split patches. Please let me know if that's a concern.
Thanks,
Umesh
Quoted 2 lines
If you decide to split the patches, please add my R-b on all the split patches.
Quoted 8 lines · Umesh Nerlige Ramappa
Quoted 6 lines · Ashutosh
Quoted 4 lines · Umesh Nerlige Ramappa
+/*
+ * Ref: 14010536224:
+ *0x20ccis repurposed on MTL, so use a separate array for MTL.
Quoted 5 lines
Quoted 2 lines
Wondering if it was
WAIT_FOR_RC6_EXIT(seen ingen12_oa_mux_regs) which moved elsewhere and if that needs to be added to the array below too?
WAIT_FOR_RC6_EXIT(0x20cc) moved elsewhere so it should be "removed" from mtl oa mux array.
What I was saying was let's say WAIT_FOR_RC6_EXIT moved to 0xc0ffee so now should 0xc0ffee be added to mtl_oa_mux_regs?
Quoted 25 lines
Quoted 25 lines
Quoted 14 lines
+ */
drivers/gpu/drm/i915/i915_perf.c +7
-0+static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table(addr, gen7_oa_b_counters);drivers/gpu/drm/i915/i915_perf.c static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr) · line 4378
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs);But otherwise this is:
Quoted 5 lines
Quoted 1 line
Reviewed-by Ashutosh Dixit <[email protected]>I will break them into separate patches though. If the diff is identical, I will carry over your R-b on the split patches. Please let me know if that's a concern.
No I can quickly review again anyway.
Quoted 1 line
Quoted 1 line
If you decide to split the patches, please add my R-b on all the split patches.
Thanks.
--
Ashutosh
Quoted 9 lines · Ashutosh
Quoted 8 lines · Umesh Nerlige Ramappa
Quoted 6 lines · Ashutosh
Quoted 4 lines · Umesh Nerlige Ramappa
+/*
+ * Ref: 14010536224:
+ *0x20ccis repurposed on MTL, so use a separate array for MTL.
Quoted 8 lines
Quoted 5 lines
Quoted 2 lines
Wondering if it was
WAIT_FOR_RC6_EXIT(seen ingen12_oa_mux_regs) which moved elsewhere and if that needs to be added to the array below too?
WAIT_FOR_RC6_EXIT(0x20cc) moved elsewhere so it should be "removed" from mtl oa mux array.What I was saying was let's say
WAIT_FOR_RC6_EXITmoved to0xc0ffeeso now should0xc0ffeebe added tomtl_oa_mux_regs?
oh, sorry, I misread that.
I looked for WAIT_FOR_RC6_EXIT in the bspec and did not find it defined for MTL, so it's dropped completely. If you could confirm, that would be great.
Quoted 26 lines
Quoted 26 lines
Quoted 26 lines
Quoted 15 lines
+ */
drivers/gpu/drm/i915/i915_perf.c +7
-0+static const struct i915_range mtl_oa_mux_regs[] = { + { .start =0x0d00 , .end =0x0d04 },/* RPM_CONFIG[0-1] */ + { .start =0x0d0c , .end =0x0d2c },/* NOA_CONFIG[0-8] */ + { .start =0x9840 , .end =0x9840 },/* GDT_CHICKEN_BITS */ + { .start =0x9884 , .end =0x9888 },/* NOA_WRITE */ +}; +static bool gen7_is_valid_b_counter_addr (struct i915_perf *perf,u32 addr) {return reg_in_range_table(addr, gen7_oa_b_counters);drivers/gpu/drm/i915/i915_perf.c static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr) · line 4378
Quoted 8 lines
drivers/gpu/drm/i915/i915_perf.c +4
-1@@ -4349,7 +4372,10 @@ static bool xehp_is_valid_b_counter_addr(struct i915_perf *perf, u32 addr)static bool gen12_is_valid_mux_addr (struct i915_perf *perf,u32 addr) { -return reg_in_range_table (addr, gen12_oa_mux_regs); +if (IS_METEORLAKE (perf->i915)) +return reg_in_range_table (addr, mtl_oa_mux_regs); +else +return reg_in_range_table (addr, gen12_oa_mux_regs);But otherwise this is:
Quoted 7 lines
Quoted 5 lines
Quoted 1 line
Reviewed-by Ashutosh Dixit <[email protected]>I will break them into separate patches though. If the diff is identical, I will carry over your R-b on the split patches. Please let me know if that's a concern.
No I can quickly review again anyway.
got it, will not add the R-bs.
Thanks,
Umesh
Quoted 6 lines
Quoted 2 lines
Quoted 2 lines
If you decide to split the patches, please add my R-b on all the split patches.
Thanks.
--
Ashutosh