On Wed, Apr 13, 2022 at 12:27:29PM -0700, Umesh Nerlige Ramappa wrote:
> From: Matthew Brost <matthew.br...@intel.com>
> 
> There is bug in DG2 where if the CCS contexts switches out while the RCS
> is running it can cause memory corruption. To workaround this add an
> atomic to a memory address with a value 1 and semaphore wait to the same
> address for a value of 0. The GuC firmware is responsible for writing 0
> to the memory address when it is safe for the context to switch out.
> 
> v2: Add atomic GPU instructions
> v3: Add w/a number (JohnH)
> v4: Set DW length to 9 atomic GPU instruction with inline data
> v5: Rebase (Umesh)
> v6: Split MI_ATOMIC definition into 2 (Daniele)
> v7: (Daniele)
> - For non-inline MI_ATOMIC, len should be 1
> - Make INLINE flag part of MI_ATOMIC_INLINE definition
> 

The internal version log should be dropped before posting upstream.

Matt

> Signed-off-by: Matthew Brost <matthew.br...@intel.com>
> ---
>  drivers/gpu/drm/i915/gt/gen8_engine_cs.c      | 41 +++++++++++++++++++
>  drivers/gpu/drm/i915/gt/intel_engine_types.h  |  8 ++++
>  drivers/gpu/drm/i915/gt/intel_gpu_commands.h  |  7 ++++
>  drivers/gpu/drm/i915/gt/uc/intel_guc.c        |  4 ++
>  drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h   |  1 +
>  .../gpu/drm/i915/gt/uc/intel_guc_submission.c |  4 ++
>  6 files changed, 65 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c 
> b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
> index 047b5a710149..9529c5455bc3 100644
> --- a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
> +++ b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
> @@ -583,6 +583,43 @@ static u32 *gen12_emit_preempt_busywait(struct 
> i915_request *rq, u32 *cs)
>       return cs;
>  }
>  
> +/* Wa_14014475959:dg2 */
> +#define CCS_SEMAPHORE_PPHWSP_OFFSET  0x540
> +static u32 ccs_semaphore_offset(struct i915_request *rq)
> +{
> +     return i915_ggtt_offset(rq->context->state) +
> +             (LRC_PPHWSP_PN * PAGE_SIZE) + CCS_SEMAPHORE_PPHWSP_OFFSET;
> +}
> +
> +/* Wa_14014475959:dg2 */
> +static u32 *ccs_emit_wa_busywait(struct i915_request *rq, u32 *cs)
> +{
> +     int i;
> +
> +     *cs++ = MI_ATOMIC_INLINE | MI_ATOMIC_GLOBAL_GTT | MI_ATOMIC_CS_STALL |
> +             MI_ATOMIC_MOVE;
> +     *cs++ = ccs_semaphore_offset(rq);
> +     *cs++ = 0;
> +     *cs++ = 1;
> +
> +     /*
> +      * When MI_ATOMIC_INLINE_DATA set this command must be 11 DW + (1 NOP)
> +      * to align. 4 DWs above + 8 filler DWs here.
> +      */
> +     for (i = 0; i < 8; ++i)
> +             *cs++ = 0;
> +
> +     *cs++ = MI_SEMAPHORE_WAIT |
> +             MI_SEMAPHORE_GLOBAL_GTT |
> +             MI_SEMAPHORE_POLL |
> +             MI_SEMAPHORE_SAD_EQ_SDD;
> +     *cs++ = 0;
> +     *cs++ = ccs_semaphore_offset(rq);
> +     *cs++ = 0;
> +
> +     return cs;
> +}
> +
>  static __always_inline u32*
>  gen12_emit_fini_breadcrumb_tail(struct i915_request *rq, u32 *cs)
>  {
> @@ -593,6 +630,10 @@ gen12_emit_fini_breadcrumb_tail(struct i915_request *rq, 
> u32 *cs)
>           !intel_uc_uses_guc_submission(&rq->engine->gt->uc))
>               cs = gen12_emit_preempt_busywait(rq, cs);
>  
> +     /* Wa_14014475959:dg2 */
> +     if (intel_engine_uses_wa_hold_ccs_switchout(rq->engine))
> +             cs = ccs_emit_wa_busywait(rq, cs);
> +
>       rq->tail = intel_ring_offset(rq, cs);
>       assert_ring_tail_valid(rq->ring, rq->tail);
>  
> diff --git a/drivers/gpu/drm/i915/gt/intel_engine_types.h 
> b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> index eac20112709c..298f2cc7a879 100644
> --- a/drivers/gpu/drm/i915/gt/intel_engine_types.h
> +++ b/drivers/gpu/drm/i915/gt/intel_engine_types.h
> @@ -529,6 +529,7 @@ struct intel_engine_cs {
>  #define I915_ENGINE_HAS_RCS_REG_STATE  BIT(9)
>  #define I915_ENGINE_HAS_EU_PRIORITY    BIT(10)
>  #define I915_ENGINE_FIRST_RENDER_COMPUTE BIT(11)
> +#define I915_ENGINE_USES_WA_HOLD_CCS_SWITCHOUT BIT(12)
>       unsigned int flags;
>  
>       /*
> @@ -629,6 +630,13 @@ intel_engine_has_relative_mmio(const struct 
> intel_engine_cs * const engine)
>       return engine->flags & I915_ENGINE_HAS_RELATIVE_MMIO;
>  }
>  
> +/* Wa_14014475959:dg2 */
> +static inline bool
> +intel_engine_uses_wa_hold_ccs_switchout(struct intel_engine_cs *engine)
> +{
> +     return engine->flags & I915_ENGINE_USES_WA_HOLD_CCS_SWITCHOUT;
> +}
> +
>  #define instdone_has_slice(dev_priv___, sseu___, slice___) \
>       ((GRAPHICS_VER(dev_priv___) == 7 ? 1 : ((sseu___)->slice_mask)) & 
> BIT(slice___))
>  
> diff --git a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h 
> b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> index 4243be030bc1..f3fe7d4c3234 100644
> --- a/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> +++ b/drivers/gpu/drm/i915/gt/intel_gpu_commands.h
> @@ -134,6 +134,13 @@
>  #define   MI_MEM_VIRTUAL     (1 << 22) /* 945,g33,965 */
>  #define   MI_USE_GGTT                (1 << 22) /* g4x+ */
>  #define MI_STORE_DWORD_INDEX MI_INSTR(0x21, 1)
> +#define MI_ATOMIC            MI_INSTR(0x2f, 1)
> +#define MI_ATOMIC_INLINE     (MI_INSTR(0x2f, 9) | MI_ATOMIC_INLINE_DATA)
> +#define   MI_ATOMIC_GLOBAL_GTT               (1 << 22)
> +#define   MI_ATOMIC_INLINE_DATA              (1 << 18)
> +#define   MI_ATOMIC_CS_STALL         (1 << 17)
> +#define        MI_ATOMIC_MOVE                (0x4 << 8)
> +
>  /*
>   * Official intel docs are somewhat sloppy concerning MI_LOAD_REGISTER_IMM:
>   * - Always issue a MI_NOOP _before_ the MI_LOAD_REGISTER_IMM - otherwise hw
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.c 
> b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> index 830889349756..228070e31ef0 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> @@ -296,6 +296,10 @@ static u32 guc_ctl_wa_flags(struct intel_guc *guc)
>       if (IS_DG2_GRAPHICS_STEP(gt->i915, G10, STEP_A0, STEP_B0))
>               flags |= GUC_WA_GAM_CREDITS;
>  
> +     /* Wa_14014475959:dg2 */
> +     if (IS_DG2(gt->i915))
> +             flags |= GUC_WA_HOLD_CCS_SWITCHOUT;
> +
>       /*
>        * Wa_14012197797:dg2_g10:a0,dg2_g11:a0
>        * Wa_22011391025:dg2_g10,dg2_g11,dg2_g12
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h 
> b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> index 126e67ea1619..e389a3a041a2 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> @@ -102,6 +102,7 @@
>  #define   GUC_WA_DUAL_QUEUE          BIT(11)
>  #define   GUC_WA_RCS_RESET_BEFORE_RC6        BIT(13)
>  #define   GUC_WA_PRE_PARSER          BIT(14)
> +#define   GUC_WA_HOLD_CCS_SWITCHOUT  BIT(17)
>  #define   GUC_WA_POLLCS                      BIT(18)
>  
>  #define GUC_CTL_FEATURE                      2
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c 
> b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> index 38ba9f783312..1cb88e99b040 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> @@ -3897,6 +3897,10 @@ static void guc_default_vfuncs(struct intel_engine_cs 
> *engine)
>       engine->flags |= I915_ENGINE_HAS_PREEMPTION;
>       engine->flags |= I915_ENGINE_HAS_TIMESLICES;
>  
> +     /* Wa_14014475959:dg2 */
> +     if (IS_DG2(engine->i915) && engine->class == COMPUTE_CLASS)
> +             engine->flags |= I915_ENGINE_USES_WA_HOLD_CCS_SWITCHOUT;
> +
>       /*
>        * TODO: GuC supports timeslicing and semaphores as well, but they're
>        * handled by the firmware so some minor tweaks are required before
> -- 
> 2.35.1
> 

Reply via email to