From 51668b0b1bc23d2b0cedf19fd1cc210e85a823a2 Mon Sep 17 00:00:00 2001 From: Frank Min Date: Wed, 28 Jun 2017 20:02:04 +0800 Subject: [PATCH] drm/amdgpu/sdma4: Enable sdma poll mem addr on vega10 for SRIOV While doing flr on VFs, there is possibility to lost the doorbell writing for sdma, so enable poll mem for sdma, then sdma fw would check the pollmem holding wptr. Signed-off-by: Frank Min Signed-off-by: Xiangliang.Yu Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu.h | 1 + drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 48 ++++++++++++++++++++++++-- 2 files changed, 46 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index d02f248b2028..0fa6438c152c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1209,6 +1209,7 @@ struct amdgpu_sdma_instance { struct amdgpu_ring ring; bool burst_nop; + uint32_t poll_mem_offs; }; struct amdgpu_sdma { diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 591f3e7fb508..5c247082f699 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -35,6 +35,7 @@ #include "vega10/MMHUB/mmhub_1_0_offset.h" #include "vega10/MMHUB/mmhub_1_0_sh_mask.h" #include "vega10/HDP/hdp_4_0_offset.h" +#include "vega10/NBIO/nbio_6_1_offset.h" #include "raven1/SDMA0/sdma0_4_1_default.h" #include "soc15_common.h" @@ -287,6 +288,8 @@ static uint64_t sdma_v4_0_ring_get_wptr(struct amdgpu_ring *ring) */ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) { + int i; + u32 offset; struct amdgpu_device *adev = ring->adev; DRM_DEBUG("Setting write pointer\n"); @@ -303,6 +306,17 @@ static void sdma_v4_0_ring_set_wptr(struct amdgpu_ring *ring) adev->wb.wb[ring->wptr_offs + 1] = upper_32_bits(ring->wptr << 2); DRM_DEBUG("calling WDOORBELL64(0x%08x, 0x%016llx)\n", ring->doorbell_index, ring->wptr << 2); + + if (amdgpu_sriov_vf(adev)) { + for (i = 0; i < adev->sdma.num_instances; i++) { + if (&adev->sdma.instance[i].ring == ring) { + offset = adev->sdma.instance[i].poll_mem_offs; + atomic64_set((atomic64_t *)&adev->wb.wb[offset], + (ring->wptr << 2)); + nbio_v6_1_hdp_flush(adev); + } + } + } WDOORBELL64(ring->doorbell_index, ring->wptr << 2); } else { int me = (ring == &ring->adev->sdma.instance[0].ring) ? 0 : 1; @@ -573,9 +587,9 @@ static void sdma_v4_0_enable(struct amdgpu_device *adev, bool enable) static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev) { struct amdgpu_ring *ring; - u32 rb_cntl, ib_cntl; + u32 rb_cntl, ib_cntl, wptr_poll_addr_lo, wptr_poll_addr_hi, wptr_poll_cntl; u32 rb_bufsz; - u32 wb_offset; + u32 wb_offset, poll_offset; u32 doorbell; u32 doorbell_offset; u32 temp; @@ -687,6 +701,21 @@ static int sdma_v4_0_gfx_resume(struct amdgpu_device *adev) if (adev->mman.buffer_funcs_ring == ring) amdgpu_ttm_set_active_vram_size(adev, adev->mc.real_vram_size); + + if (amdgpu_sriov_vf(adev)) { + poll_offset = adev->sdma.instance[i].poll_mem_offs * 4; + + wptr_poll_cntl = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL)); + wptr_poll_addr_lo = RREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO)); + wptr_poll_addr_lo = REG_SET_FIELD(wptr_poll_addr_lo, SDMA0_GFX_RB_WPTR_POLL_ADDR_LO, ADDR, + lower_32_bits(adev->wb.gpu_addr + poll_offset) >> 2); + wptr_poll_addr_hi = upper_32_bits(adev->wb.gpu_addr + poll_offset); + wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, SDMA0_GFX_RB_WPTR_POLL_CNTL, F32_POLL_ENABLE, 1); + + WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO), wptr_poll_addr_lo); + WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI), wptr_poll_addr_hi); + WREG32(sdma_v4_0_get_reg_offset(i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), wptr_poll_cntl); + } } return 0; @@ -1247,6 +1276,15 @@ static int sdma_v4_0_sw_init(void *handle) (i == 0) ? AMDGPU_SDMA_IRQ_TRAP0 : AMDGPU_SDMA_IRQ_TRAP1); + + if (amdgpu_sriov_vf(adev)) { + r = amdgpu_wb_get_64bit(adev, + &adev->sdma.instance[i].poll_mem_offs); + if (r) { + dev_err(adev->dev, "(%d) failed to allocate SDMA poll mem wb.\n", r); + return r; + } + } if (r) return r; } @@ -1259,9 +1297,13 @@ static int sdma_v4_0_sw_fini(void *handle) struct amdgpu_device *adev = (struct amdgpu_device *)handle; int i; - for (i = 0; i < adev->sdma.num_instances; i++) + for (i = 0; i < adev->sdma.num_instances; i++) { amdgpu_ring_fini(&adev->sdma.instance[i].ring); + if (amdgpu_sriov_vf(adev)) + amdgpu_wb_free_64bit(adev, + adev->sdma.instance[i].poll_mem_offs); + } return 0; } -- 2.20.1