drm/amdgpu/sdma: simplify sdma instance setup
Subsystems (2)
PATCH 1/2 drm/amdgpu/sdma: simplify sdma instance setup
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)
drivers/gpu/drm/amd/amdgpu/cik_sdma.c +6 -6
@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >>2 ; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ; }/**
@@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }
@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA1_MASK ;
@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {
drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +6 -6
@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr; }
@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); }static void sdma_v2_4_ring_insert_nop (struct amdgpu_ring *ring,uint32_t count)
@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );
@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {
drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +6 -8
@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ; }return wptr;
@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)WRITE_ONCE (*wb, (lower_32_bits (ring->wptr) <<2 )); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); } }
@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );
@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +11 -12
@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)DRM_DEBUG ("wptr/doorbell before shift == 0x%016llx\n" , wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >>2 ; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG ("wptr [%i]high== 0x%08x low==0x%08x\n" , - me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;
@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); } }
@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)u32 ref_and_mask =0 ;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;
@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = {
Signature · 6 lines
2.13.6 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
A message that is not in the archive (sent off-list or not yet received): 20180625174235.29172-1-alexander.deucher-5C7GfCeVMHo@public.gmane.org
PATCH 2/2 drm/amdgpu/vce: simplify vce instance setup
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
--- drivers/gpu/drm/amd/amdgpu/vce_v2_0.c | 10 ++++++---- drivers/gpu/drm/amd/amdgpu/vce_v3_0.c | 20 ++++++++++++-------- drivers/gpu/drm/amd/amdgpu/vce_v4_0.c | 16 +++++++++------- 3 files changed, 27 insertions(+), 19 deletions(-)
drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +6 -4
@@ -56,7 +56,7 @@ static uint64_t vce_v2_0_ring_get_rptr(struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )return RREG32 (mmVCE_RB_RPTR);else return RREG32 (mmVCE_RB_RPTR2);
@@ -73,7 +73,7 @@ static uint64_t vce_v2_0_ring_get_wptr(struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )return RREG32 (mmVCE_RB_WPTR);else return RREG32 (mmVCE_RB_WPTR2);
@@ -90,7 +90,7 @@ static void vce_v2_0_ring_set_wptr(struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )WREG32 (mmVCE_RB_WPTR,lower_32_bits (ring->wptr));else WREG32 (mmVCE_RB_WPTR2,lower_32_bits (ring->wptr));
@@ -627,8 +627,10 @@ static void vce_v2_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->vce.num_rings; i++) +for (i =0 ; i < adev->vce.num_rings; i++) { adev->vce.ring[i].funcs = &vce_v2_0_ring_funcs; + adev->vce.ring[i].me = i; + } }static const struct amdgpu_irq_src_funcs vce_v2_0_irq_funcs = {
drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +12 -8
@@ -86,9 +86,9 @@ static uint64_t vce_v3_0_ring_get_rptr(struct amdgpu_ring *ring)else if (adev->vce.harvest_config ==AMDGPU_VCE_HARVEST_VCE0 )WREG32 (mmGRBM_GFX_INDEX,GET_VCE_INSTANCE (1 )); -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 ) v =RREG32 (mmVCE_RB_RPTR); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 ) v =RREG32 (mmVCE_RB_RPTR2);else v =RREG32 (mmVCE_RB_RPTR3);
@@ -118,9 +118,9 @@ static uint64_t vce_v3_0_ring_get_wptr(struct amdgpu_ring *ring)else if (adev->vce.harvest_config ==AMDGPU_VCE_HARVEST_VCE0 )WREG32 (mmGRBM_GFX_INDEX,GET_VCE_INSTANCE (1 )); -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 ) v =RREG32 (mmVCE_RB_WPTR); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 ) v =RREG32 (mmVCE_RB_WPTR2);else v =RREG32 (mmVCE_RB_WPTR3);
@@ -149,9 +149,9 @@ static void vce_v3_0_ring_set_wptr(struct amdgpu_ring *ring)else if (adev->vce.harvest_config ==AMDGPU_VCE_HARVEST_VCE0 )WREG32 (mmGRBM_GFX_INDEX,GET_VCE_INSTANCE (1 )); -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )WREG32 (mmVCE_RB_WPTR,lower_32_bits (ring->wptr)); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 )WREG32 (mmVCE_RB_WPTR2,lower_32_bits (ring->wptr));else WREG32 (mmVCE_RB_WPTR3,lower_32_bits (ring->wptr));
@@ -942,12 +942,16 @@ static void vce_v3_0_set_ring_funcs(struct amdgpu_device *adev)int i;if (adev->asic_type >=CHIP_STONEY ) { -for (i =0 ; i < adev->vce.num_rings; i++) +for (i =0 ; i < adev->vce.num_rings; i++) { adev->vce.ring[i].funcs = &vce_v3_0_ring_vm_funcs; + adev->vce.ring[i].me = i; + }DRM_INFO ("VCE enabled in VM mode\n" ); }else { -for (i =0 ; i < adev->vce.num_rings; i++) +for (i =0 ; i < adev->vce.num_rings; i++) { adev->vce.ring[i].funcs = &vce_v3_0_ring_phys_funcs; + adev->vce.ring[i].me = i; + }DRM_INFO ("VCE enabled in physical mode\n" ); } }
drivers/gpu/drm/amd/amdgpu/vce_v4_0.c +9 -7
@@ -60,9 +60,9 @@ static uint64_t vce_v4_0_ring_get_rptr(struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_RPTR)); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 )return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_RPTR2));else return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_RPTR3));
@@ -82,9 +82,9 @@ static uint64_t vce_v4_0_ring_get_wptr(struct amdgpu_ring *ring)if (ring->use_doorbell)return adev->wb.wb[ring->wptr_offs]; -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_WPTR)); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 )return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_WPTR2));else return RREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_WPTR3));
@@ -108,10 +108,10 @@ static void vce_v4_0_ring_set_wptr(struct amdgpu_ring *ring)return ; } -if (ring == &adev->vce.ring[0 ]) +if (ring->me ==0 )WREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_WPTR),lower_32_bits (ring->wptr)); -else if (ring == &adev->vce.ring[1 ]) +else if (ring->me ==1 )WREG32 (SOC15_REG_OFFSET (VCE ,0 , mmVCE_RB_WPTR2),lower_32_bits (ring->wptr));else
@@ -1088,8 +1088,10 @@ static void vce_v4_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->vce.num_rings; i++) +for (i =0 ; i < adev->vce.num_rings; i++) { adev->vce.ring[i].funcs = &vce_v4_0_ring_vm_funcs; + adev->vce.ring[i].me = i; + }DRM_INFO ("VCE enabled in VM mode\n" ); }
Signature · 6 lines
2.13.6 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Ping on this series?
Quoted 216 lines · Alex Deucher
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
Signed-off-by Alex Deucher <[email protected]>--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)drivers/gpu/drm/amd/amdgpu/cik_sdma.c +6
-6@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >>2 ; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ; }/** @@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA1_MASK ;@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +6
-6@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr; }@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); }static void sdma_v2_4_ring_insert_nop (struct amdgpu_ring *ring,uint32_t count)@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +6
-8@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ; }return wptr;@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)WRITE_ONCE (*wb, (lower_32_bits (ring->wptr) <<2 )); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); } }@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +11
-13@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)DRM_DEBUG ("wptr/doorbell before shift == 0x%016llx\n" , wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >>2 ; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG ("wptr [%i]high== 0x%08x low==0x%08x\n" , - me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); } }@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)u32 ref_and_mask =0 ;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = { --
Quoted 2 lines
2.13.6
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
A message that is not in the archive (sent off-list or not yet received): CADnq5_M5DnXH9rF9HgKhPi32FS5PdKj-OjbM13P0uKSJ4Y3sHg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org
Leo Liu
Reviewed-byLooks good to me. Both patches are
Quoted 217 lines · Alex Deucher
Ping on this series?
Quoted 214 lines · Alex Deucher
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
Signed-off-by Alex Deucher <[email protected]>--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)drivers/gpu/drm/amd/amdgpu/cik_sdma.c +6
-6@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >>2 ; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ; }/** @@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA1_MASK ;@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +6
-6@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr; }@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); }static void sdma_v2_4_ring_insert_nop (struct amdgpu_ring *ring,uint32_t count)@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +6
-8@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ; }return wptr;@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)WRITE_ONCE (*wb, (lower_32_bits (ring->wptr) <<2 )); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); } }@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +11
-12@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)DRM_DEBUG ("wptr/doorbell before shift == 0x%016llx\n" , wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >>2 ; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG ("wptr [%i]high== 0x%08x low==0x%08x\n" , - me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); } }@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)u32 ref_and_mask =0 ;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }
Quoted 8 lines
Quoted 4 lines
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +0
-1static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = { --2.13.6
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
A message that is not in the archive (sent off-list or not yet received): 4487ebe1-8046-a678-1f79-3280f707ee17-5C7GfCeVMHo@public.gmane.org
Quoted 4 lines · Leo Liu
Looks good to me. Both patches are
Reviewed-by Leo Liu <[email protected]>
Thanks. Can you look at the UVD patch as well?
https://patchwork.freedesktop.org/patch/233507/
Alex
Quoted 274 lines · Alex Deucher
Quoted 271 lines
Ping on this series?
Quoted 266 lines · Alex Deucher
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
Signed-off-by Alex Deucher <[email protected]>--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)drivers/gpu/drm/amd/amdgpu/cik_sdma.c +1
-2@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >> 2; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ;}/**
drivers/gpu/drm/amd/amdgpu/cik_sdma.c +2
-3@@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask = GPU_HDP_FLUSH_DONE__SDMA1_MASK;drivers/gpu/drm/amd/amdgpu/cik_sdma.c +3
-1@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +1
-2@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr;}
drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +1
-2@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], lower_32_bits(ring->wptr) << 2); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 );}static void sdma_v2_4_ring_insert_nop(struct
amdgpu_ring*ring,uint32_tcount)drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +1
-1@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else
ref_and_mask= REG_SET_FIELD(ref_and_mask,GPU_HDP_FLUSH_DONE, SDMA1, 1);drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +3
-1@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +1
-3@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 : 1; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr = RREG32(mmSDMA0_GFX_RB_WPTR +sdma_offsets[ring->me]) >> 2; } return wptr;drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +1
-3@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring) WRITE_ONCE(*wb, (lower_32_bits(ring->wptr) << 2)); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 : 1; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], lower_32_bits(ring->wptr) << 2); + WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits(ring->wptr) << 2); }}
drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +1
-1@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else
ref_and_mask= REG_SET_FIELD(ref_and_mask,GPU_HDP_FLUSH_DONE, SDMA1, 1);drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +3
-1@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +2
-3@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring) DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >> 2; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >> 2; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG("wptr [%i]high== 0x%08x low==0x%08x\n",drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +3
-7- me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 : 1; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2)); - WREG32(sdma_v4_0_get_reg_offset(adev, me,
mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits(ring->wptr << 2));
+ WREG32(sdma_v4_0_get_reg_offset(adev,ring->me,mmSDMA0_GFX_RB_WPTR),lower_32_bits(ring->wptr << 2));
+ WREG32(sdma_v4_0_get_reg_offset(adev,ring->me,mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2)); }}
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +1
-1@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) u32 ref_and_mask = 0;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +3
-1@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }
Quoted 10 lines
Quoted 8 lines
Quoted 4 lines
static const struct
amdgpu_irq_src_funcssdma_v4_0_trap_irq_funcs= {
--
2.13.6_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
Christian König
Reviewed-byAm 25.06.2018 um 19:42 schrieb Alex Deucher:
Quoted 4 lines
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
Signed-off-by Alex Deucher <[email protected]>
Quoted 201 lines
--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)drivers/gpu/drm/amd/amdgpu/cik_sdma.c +6
-6@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >>2 ; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ; }/** @@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA1_MASK ;@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +6
-6@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr; }@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); }static void sdma_v2_4_ring_insert_nop (struct amdgpu_ring *ring,uint32_t count)@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +6
-8@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ; }return wptr;@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)WRITE_ONCE (*wb, (lower_32_bits (ring->wptr) <<2 )); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); } }@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +8
-11@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)DRM_DEBUG ("wptr/doorbell before shift == 0x%016llx\n" , wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >>2 ; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG ("wptr [%i]high== 0x%08x low==0x%08x\n" , - me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); } }@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)u32 ref_and_mask =0 ;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) · line 1657
Quoted 10 lines
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +3
-1int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = {
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
James Zhu
Reviewed-byQuoted 205 lines · Alex Deucher
Set the me instance in early init and use that rather than calculating the instance based on the ring pointer.
Signed-off-by Alex Deucher <[email protected]>--- drivers/gpu/drm/amd/amdgpu/cik_sdma.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c | 12 ++++++------ drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c | 14 ++++++-------- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 23 +++++++++++------------ 4 files changed, 29 insertions(+), 32 deletions(-)drivers/gpu/drm/amd/amdgpu/cik_sdma.c +6
-6@@ -177,9 +177,8 @@ static uint64_t cik_sdma_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t cik_sdma_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) &0x3fffc ) >>2 ; +return (RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) &0x3fffc ) >>2 ; }/** @@ -192,9 +191,8 @@ static uint64_t cik_sdma_ring_get_wptr(struct amdgpu_ring *ring)static void cik_sdma_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -u32 me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me], +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me], (lower_32_bits (ring->wptr) <<2 ) &0x3fffc ); }@@ -248,7 +246,7 @@ static void cik_sdma_ring_emit_hdp_flush(struct amdgpu_ring *ring) SDMA_POLL_REG_MEM_EXTRA_FUNC(3)); /* == */u32 ref_and_mask; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA0_MASK ;else ref_and_mask =GPU_HDP_FLUSH_DONE__SDMA1_MASK ;@@ -1290,8 +1288,10 @@ static void cik_sdma_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &cik_sdma_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs cik_sdma_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c +6
-6@@ -202,8 +202,7 @@ static uint64_t sdma_v2_4_ring_get_rptr(struct amdgpu_ring *ring)static uint64_t sdma_v2_4_ring_get_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; +u32 wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ;return wptr; }@@ -218,9 +217,8 @@ static uint64_t sdma_v2_4_ring_get_wptr(struct amdgpu_ring *ring)static void sdma_v2_4_ring_set_wptr (struct amdgpu_ring *ring) {struct amdgpu_device *adev = ring->adev; -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); }static void sdma_v2_4_ring_insert_nop (struct amdgpu_ring *ring,uint32_t count)@@ -273,7 +271,7 @@ static void sdma_v2_4_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1213,8 +1211,10 @@ static void sdma_v2_4_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v2_4_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v2_4_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c +6
-8@@ -365,9 +365,7 @@ static uint64_t sdma_v3_0_ring_get_wptr(struct amdgpu_ring *ring)/* XXX check if swapping is necessary on BE */ wptr = ring->adev->wb.wb[ring->wptr_offs] >>2 ; }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - - wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me]) >>2 ; + wptr =RREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me]) >>2 ; }return wptr;@@ -394,9 +392,7 @@ static void sdma_v3_0_ring_set_wptr(struct amdgpu_ring *ring)WRITE_ONCE (*wb, (lower_32_bits (ring->wptr) <<2 )); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; - -WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[me],lower_32_bits (ring->wptr) <<2 ); +WREG32 (mmSDMA0_GFX_RB_WPTR + sdma_offsets[ring->me],lower_32_bits (ring->wptr) <<2 ); } }@@ -450,7 +446,7 @@ static void sdma_v3_0_ring_emit_hdp_flush(struct amdgpu_ring *ring) {u32 ref_and_mask =0 ; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA0 ,1 );else ref_and_mask =REG_SET_FIELD (ref_and_mask,GPU_HDP_FLUSH_DONE ,SDMA1 ,1 );@@ -1655,8 +1651,10 @@ static void sdma_v3_0_set_ring_funcs(struct amdgpu_device *adev) {int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v3_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v3_0_trap_irq_funcs = {drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +8
-11@@ -296,13 +296,12 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring)DRM_DEBUG ("wptr/doorbell before shift == 0x%016llx\n" , wptr); }else {u32 lowbit, highbit; -int me = (ring == &adev->sdma.instance[0 ].ring) ?0 :1 ; - lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR)) >>2 ; - highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ; + lowbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR)) >>2 ; + highbit =RREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI)) >>2 ;DRM_DEBUG ("wptr [%i]high== 0x%08x low==0x%08x\n" , - me, highbit, lowbit); + ring->me, highbit, lowbit); wptr = highbit; wptr = wptr <<32 ; wptr |= lowbit;@@ -339,17 +338,15 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) ring->doorbell_index, ring->wptr << 2);WDOORBELL64 (ring->doorbell_index, ring->wptr <<2 ); }else { -int me = (ring == &ring->adev->sdma.instance[0 ].ring) ?0 :1 ; -DRM_DEBUG ("Not using doorbell -- " "mmSDMA%i_GFX_RB_WPTR == 0x%08x " "mmSDMA%i_GFX_RB_WPTR_HI == 0x%08x\n" , - me, + ring->me,lower_32_bits (ring->wptr <<2 ), - me, + ring->me,upper_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); -WREG32 (sdma_v4_0_get_reg_offset (adev, me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR),lower_32_bits (ring->wptr <<2 )); +WREG32 (sdma_v4_0_get_reg_offset (adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),upper_32_bits (ring->wptr <<2 )); } }@@ -430,7 +427,7 @@ static void sdma_v4_0_ring_emit_hdp_flush(struct amdgpu_ring *ring)u32 ref_and_mask =0 ;const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio_funcs->hdp_flush_reg; -if (ring == &ring->adev->sdma.instance[0 ].ring) +if (ring->me ==0 ) ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0;else ref_and_mask = nbio_hf_reg->ref_and_mask_sdma1;@@ -1651,8 +1648,10 @@ static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) {
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c static void sdma_v4_0_set_ring_funcs(struct amdgpu_device *adev) · line 1657
Quoted 10 lines
drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +3
-1int i; -for (i =0 ; i < adev->sdma.num_instances; i++) +for (i =0 ; i < adev->sdma.num_instances; i++) { adev->sdma.instance[i].ring.funcs = &sdma_v4_0_ring_funcs; + adev->sdma.instance[i].ring.me = i; + } }static const struct amdgpu_irq_src_funcs sdma_v4_0_trap_irq_funcs = {
_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx