iv/linux
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

drm/amdgpu: Modify register access in sdma_v5_2 to use _SOC15 macros

Browse Source 
In SRIOV environment, KMD should access SDMA registers
through RLCG if GC indirect access flag enabled.

Using _SOC15 read/write macros ensures that they go
through RLC when the flag is enabled.

Signed-off-by: Rohit Khaire <rohit.khaire@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Rohit Khaire 2021-06-04 15:34:14 -04:00 committed by Alex Deucher
parent 3e88cbb0d0
commit cf2a22e408
1 changed files with 35 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

70
drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c
View File

@ -500,12 +500,12 @@ static void sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
ib_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
}
}
@ -563,11 +563,11 @@ static void sdma_v5_2_ctx_switch_enable(struct amdgpu_device *adev, bool enable)
f32_cntl = REG_SET_FIELD(f32_cntl, SDMA0_CNTL,
AUTO_CTXSW_ENABLE, enable ? 1 : 0);
if (enable && amdgpu_sdma_phase_quantum) {
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE0_QUANTUM),
phase_quantum);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE1_QUANTUM),
phase_quantum);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_PHASE2_QUANTUM),
phase_quantum);
}
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL), f32_cntl);
@ -625,62 +625,62 @@ static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev)
ring = &adev->sdma.instance[i].ring;
wb_offset = (ring->rptr_offs * 4);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0);
/* Set ring buffer size in dwords */
rb_bufsz = order_base_2(ring->ring_size / 4);
rb_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL));
rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
/* Initialize the ring buffer's read and write pointers */
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO),
lower_32_bits(wptr_gpu_addr));
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI),
upper_32_bits(wptr_gpu_addr));
wptr_poll_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i,
wptr_poll_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL),
WREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI),
upper_32_bits(adev->wb.gpu_addr + wb_offset) & 0xFFFFFFFF);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO),
WREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), ring->gpu_addr >> 40);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), ring->gpu_addr >> 8);
WREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1);
WREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr) << 2);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr) << 2);
}
doorbell = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
doorbell_offset = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET));
doorbell = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL));
doorbell_offset = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET));
if (ring->use_doorbell) {
doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1);
@ -689,8 +689,8 @@ static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev)
} else {
doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0);
}
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), doorbell_offset);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell);
WREG32_SOC15_IP(GC, sdma_v5_2_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,
@ -700,7 +700,7 @@ static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev)
sdma_v5_2_ring_set_wptr(ring);
/* set minor_ptr_update to 0 after wptr programed */
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0);
/* set utc l1 enable flag always to 1 */
temp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL));
@ -711,19 +711,19 @@ static int sdma_v5_2_gfx_resume(struct amdgpu_device *adev)
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_CNTL), temp);
/* Set up RESP_MODE to non-copy addresses */
temp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL));
temp = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL));
temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3);
temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp);
/* program default cache read and write policy */
temp = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE));
temp = RREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE));
/* clean read policy and write policy bits */
temp &= 0xFF0FFF;
temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) |
(CACHE_WRITE_POLICY_L2__DEFAULT << 14) |
SDMA0_UTCL1_PAGE__LLC_NOALLOC_MASK);
WREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp);
if (!amdgpu_sriov_vf(adev)) {
/* unhalt engine */
@ -734,15 +734,15 @@ static int sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl);
ib_cntl = RREG32(sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL));
ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_2_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_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
WREG32_SOC15_IP(GC, sdma_v5_2_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl);
ring->sched.ready = true;