[PATCH 3/8] drm/amdgpu: Change GC(SDMA) register access from MMIO to RLCG

Peng Ju Zhou PengJu.Zhou at amd.com
Thu Apr 8 05:32:31 UTC 2021


In SRIOV environment, KMD should access GC registers
with RLCG if GC indirect access flag enabled.

Change GC register access from MMIO to RLCG.

Signed-off-by: Peng Ju Zhou <PengJu.Zhou at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c | 78 ++++++++++++++------------
 1 file changed, 42 insertions(+), 36 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
index 920fc6d4a127..f72faa132419 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c
@@ -324,9 +324,9 @@ static uint64_t sdma_v5_0_ring_get_wptr(struct amdgpu_ring *ring)
 		wptr = READ_ONCE(*((u64 *)&adev->wb.wb[ring->wptr_offs]));
 		DRM_DEBUG("wptr/doorbell before shift == 0x%016llx\n", wptr);
 	} else {
-		wptr = RREG32(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI));
+		wptr = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI));
 		wptr = wptr << 32;
-		wptr |= RREG32(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR));
+		wptr |= RREG32_RLC(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR));
 		DRM_DEBUG("wptr before shift [%i] wptr == 0x%016llx\n", ring->me, wptr);
 	}
 
@@ -367,9 +367,9 @@ static void sdma_v5_0_ring_set_wptr(struct amdgpu_ring *ring)
 				lower_32_bits(ring->wptr << 2),
 				ring->me,
 				upper_32_bits(ring->wptr << 2));
-		WREG32(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR),
 			lower_32_bits(ring->wptr << 2));
-		WREG32(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, ring->me, mmSDMA0_GFX_RB_WPTR_HI),
 			upper_32_bits(ring->wptr << 2));
 	}
 }
@@ -530,12 +530,12 @@ static void sdma_v5_0_gfx_stop(struct amdgpu_device *adev)
 		amdgpu_ttm_set_buffer_funcs_status(adev, false);
 
 	for (i = 0; i < adev->sdma.num_instances; i++) {
-		rb_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 0);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
-		ib_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		ib_cntl = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
 		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 0);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
 	}
 }
 
@@ -596,11 +596,11 @@ static void sdma_v5_0_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
 		}
 
 		if (enable && amdgpu_sdma_phase_quantum) {
-			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
+			WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
 			       phase_quantum);
-			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
+			WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
 			       phase_quantum);
-			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
+			WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
 			       phase_quantum);
 		}
 		if (!amdgpu_sriov_vf(adev))
@@ -667,58 +667,63 @@ static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev)
 
 		/* Set ring buffer size in dwords */
 		rb_bufsz = order_base_2(ring->ring_size / 4);
-		rb_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
+		rb_cntl = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz);
 #ifdef __BIG_ENDIAN
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1);
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL,
 					RPTR_WRITEBACK_SWAP_ENABLE, 1);
 #endif
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
 
 		/* Initialize the ring buffer's read and write pointers */
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
 
 		/* setup the wptr shadow polling */
 		wptr_gpu_addr = adev->wb.gpu_addr + (ring->wptr_offs * 4);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
 		       lower_32_bits(wptr_gpu_addr));
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
 		       upper_32_bits(wptr_gpu_addr));
-		wptr_poll_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i,
+		wptr_poll_cntl = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i,
 							 mmSDMA0_GFX_RB_WPTR_POLL_CNTL));
 		wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl,
 					       SDMA0_GFX_RB_WPTR_POLL_CNTL,
 					       F32_POLL_ENABLE, 1);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL),
 		       wptr_poll_cntl);
 
 		/* set the wb address whether it's enabled or not */
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
 		       upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
 		       lower_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFC);
 
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1);
 
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE),
+		       ring->gpu_addr >> 8);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI),
+		       ring->gpu_addr >> 40);
 
 		ring->wptr = 0;
 
 		/* before programing wptr to a less value, need set minor_ptr_update first */
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
 
 		if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */
-			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr) << 2);
-			WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
+			WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR),
+			       lower_32_bits(ring->wptr) << 2);
+			WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI),
+			       upper_32_bits(ring->wptr) << 2);
 		}
 
-		doorbell = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
-		doorbell_offset = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET));
+		doorbell = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
+		doorbell_offset = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i,
+						mmSDMA0_GFX_DOORBELL_OFFSET));
 
 		if (ring->use_doorbell) {
 			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
@@ -727,8 +732,9 @@ static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev)
 		} else {
 			doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
 		}
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET),
+		       doorbell_offset);
 
 		adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell,
 						      ring->doorbell_index, 20);
@@ -737,7 +743,7 @@ static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev)
 			sdma_v5_0_ring_set_wptr(ring);
 
 		/* set minor_ptr_update to 0 after wptr programed */
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
 
 		if (!amdgpu_sriov_vf(adev)) {
 			/* set utc l1 enable flag always to 1 */
@@ -771,15 +777,15 @@ static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev)
 
 		/* enable DMA RB */
 		rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1);
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
 
-		ib_cntl = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
+		ib_cntl = RREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
 		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1);
 #ifdef __BIG_ENDIAN
 		ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1);
 #endif
 		/* enable DMA IBs */
-		WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
+		WREG32_RLC(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
 
 		ring->sched.ready = true;
 
-- 
2.17.1



More information about the amd-gfx mailing list