diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c | 890 |
1 files changed, 651 insertions, 239 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c index ce2b22f7e4e4..aae7328973d1 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -25,12 +25,17 @@ #include "amdgpu_jpeg.h" #include "soc15.h" #include "soc15d.h" +#include "jpeg_v2_0.h" #include "jpeg_v4_0_3.h" +#include "mmsch_v4_0_3.h" #include "vcn/vcn_4_0_3_offset.h" #include "vcn/vcn_4_0_3_sh_mask.h" #include "ivsrcid/vcn/irqsrcs_vcn_4_0.h" +#define NORMALIZE_JPEG_REG_OFFSET(offset) \ + (offset & 0x1FFFF) + enum jpeg_engin_status { UVD_PGFSM_STATUS__UVDJ_PWR_ON = 0, UVD_PGFSM_STATUS__UVDJ_PWR_OFF = 2, @@ -38,9 +43,10 @@ enum jpeg_engin_status { static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev); static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev); -static int jpeg_v4_0_3_set_powergating_state(void *handle, +static int jpeg_v4_0_3_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state); static void jpeg_v4_0_3_set_ras_funcs(struct amdgpu_device *adev); +static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring); static int amdgpu_ih_srcid_jpeg[] = { VCN_4_0__SRCID__JPEG_DECODE, @@ -53,18 +59,67 @@ static int amdgpu_ih_srcid_jpeg[] = { VCN_4_0__SRCID__JPEG7_DECODE }; +static const struct amdgpu_hwip_reg_entry jpeg_reg_list_4_0_3[] = { + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_POWER_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_INT_STAT), + SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_SYS_INT_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC0_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_ADDR_MODE), + SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_GFX10_ADDR_CONFIG), + SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_Y_GFX10_TILING_SURFACE), + SOC15_REG_ENTRY_STR(JPEG, 0, regJPEG_DEC_UV_GFX10_TILING_SURFACE), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_PITCH), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JPEG_UV_PITCH), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC1_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC2_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC3_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC4_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC5_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC6_UVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, regUVD_JRBC7_UVD_JRBC_STATUS), +}; + +static inline bool jpeg_v4_0_3_normalizn_reqd(struct amdgpu_device *adev) +{ + return (adev->jpeg.caps & AMDGPU_JPEG_CAPS(RRMT_ENABLED)) == 0; +} + +static inline int jpeg_v4_0_3_core_reg_offset(u32 pipe) +{ + if (pipe) + return ((0x40 * pipe) - 0xc80); + else + return 0; +} + /** * jpeg_v4_0_3_early_init - set function pointers * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Set ring and irq function pointers */ -static int jpeg_v4_0_3_early_init(void *handle) +static int jpeg_v4_0_3_early_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; - adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS; + adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS_4_0_3; jpeg_v4_0_3_set_dec_ring_funcs(adev); jpeg_v4_0_3_set_irq_funcs(adev); @@ -76,13 +131,13 @@ static int jpeg_v4_0_3_early_init(void *handle) /** * jpeg_v4_0_3_sw_init - sw init for JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Load firmware and sw initialization */ -static int jpeg_v4_0_3_sw_init(void *handle) +static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; struct amdgpu_ring *ring; int i, j, r, jpeg_inst; @@ -94,6 +149,18 @@ static int jpeg_v4_0_3_sw_init(void *handle) return r; } + /* JPEG DJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; + + /* JPEG EJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; + r = amdgpu_jpeg_sw_init(adev); if (r) return r; @@ -109,9 +176,20 @@ static int jpeg_v4_0_3_sw_init(void *handle) ring = &adev->jpeg.inst[i].ring_dec[j]; ring->use_doorbell = true; ring->vm_hub = AMDGPU_MMHUB0(adev->jpeg.inst[i].aid_id); - ring->doorbell_index = - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 1 + j + 9 * jpeg_inst; + if (!amdgpu_sriov_vf(adev)) { + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 1 + j + 9 * jpeg_inst; + } else { + if (j < 4) + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 4 + j + 32 * jpeg_inst; + else + ring->doorbell_index = + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 8 + j + 32 * jpeg_inst; + } sprintf(ring->name, "jpeg_dec_%d.%d", adev->jpeg.inst[i].aid_id, j); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0, AMDGPU_RING_PRIO_DEFAULT, NULL); @@ -121,10 +199,8 @@ static int jpeg_v4_0_3_sw_init(void *handle) adev->jpeg.internal.jpeg_pitch[j] = regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET; adev->jpeg.inst[i].external.jpeg_pitch[j] = - SOC15_REG_OFFSET1( - JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_SCRATCH0, - (j ? (0x40 * j - 0xc80) : 0)); + SOC15_REG_OFFSET1(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_SCRATCH0, + jpeg_v4_0_3_core_reg_offset(j)); } } @@ -136,70 +212,215 @@ static int jpeg_v4_0_3_sw_init(void *handle) } } + r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_4_0_3, ARRAY_SIZE(jpeg_reg_list_4_0_3)); + if (r) + return r; + + adev->jpeg.supported_reset = + amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec); + if (!amdgpu_sriov_vf(adev)) + adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; + r = amdgpu_jpeg_sysfs_reset_mask_init(adev); + return 0; } /** * jpeg_v4_0_3_sw_fini - sw fini for JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * JPEG suspend and free up sw allocation */ -static int jpeg_v4_0_3_sw_fini(void *handle) +static int jpeg_v4_0_3_sw_fini(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int r; r = amdgpu_jpeg_suspend(adev); if (r) return r; + amdgpu_jpeg_sysfs_reset_mask_fini(adev); + r = amdgpu_jpeg_sw_fini(adev); return r; } +static int jpeg_v4_0_3_start_sriov(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring; + uint64_t ctx_addr; + uint32_t param, resp, expected; + uint32_t tmp, timeout; + + struct amdgpu_mm_table *table = &adev->virt.mm_table; + uint32_t *table_loc; + uint32_t table_size; + uint32_t size, size_dw, item_offset; + uint32_t init_status; + int i, j, jpeg_inst; + + struct mmsch_v4_0_cmd_direct_write + direct_wt = { {0} }; + struct mmsch_v4_0_cmd_end end = { {0} }; + struct mmsch_v4_0_3_init_header header; + + direct_wt.cmd_header.command_type = + MMSCH_COMMAND__DIRECT_REG_WRITE; + end.cmd_header.command_type = + MMSCH_COMMAND__END; + + for (i = 0; i < adev->jpeg.num_jpeg_inst; i++) { + jpeg_inst = GET_INST(JPEG, i); + + memset(&header, 0, sizeof(struct mmsch_v4_0_3_init_header)); + header.version = MMSCH_VERSION; + header.total_size = sizeof(struct mmsch_v4_0_3_init_header) >> 2; + + table_loc = (uint32_t *)table->cpu_addr; + table_loc += header.total_size; + + item_offset = header.total_size; + + for (j = 0; j < adev->jpeg.num_jpeg_rings; j++) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + table_size = 0; + + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, lower_32_bits(ring->gpu_addr)); + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, upper_32_bits(ring->gpu_addr)); + tmp = SOC15_REG_OFFSET(JPEG, 0, regUVD_JRBC0_UVD_JRBC_RB_SIZE); + MMSCH_V4_0_INSERT_DIRECT_WT(tmp, ring->ring_size / 4); + + if (j <= 3) { + header.mjpegdec0[j].table_offset = item_offset; + header.mjpegdec0[j].init_status = 0; + header.mjpegdec0[j].table_size = table_size; + } else { + header.mjpegdec1[j - 4].table_offset = item_offset; + header.mjpegdec1[j - 4].init_status = 0; + header.mjpegdec1[j - 4].table_size = table_size; + } + header.total_size += table_size; + item_offset += table_size; + } + + MMSCH_V4_0_INSERT_END(); + + /* send init table to MMSCH */ + size = sizeof(struct mmsch_v4_0_3_init_header); + table_loc = (uint32_t *)table->cpu_addr; + memcpy((void *)table_loc, &header, size); + + ctx_addr = table->gpu_addr; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_LO, lower_32_bits(ctx_addr)); + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_ADDR_HI, upper_32_bits(ctx_addr)); + + tmp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID); + tmp &= ~MMSCH_VF_VMID__VF_CTX_VMID_MASK; + tmp |= (0 << MMSCH_VF_VMID__VF_CTX_VMID__SHIFT); + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_VMID, tmp); + + size = header.total_size; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_CTX_SIZE, size); + + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP, 0); + + param = 0x00000001; + WREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_HOST, param); + tmp = 0; + timeout = 1000; + resp = 0; + expected = MMSCH_VF_MAILBOX_RESP__OK; + init_status = + ((struct mmsch_v4_0_3_init_header *)(table_loc))->mjpegdec0[i].init_status; + while (resp != expected) { + resp = RREG32_SOC15(VCN, jpeg_inst, regMMSCH_VF_MAILBOX_RESP); + + if (resp != 0) + break; + udelay(10); + tmp = tmp + 10; + if (tmp >= timeout) { + DRM_ERROR("failed to init MMSCH. TIME-OUT after %d usec"\ + " waiting for regMMSCH_VF_MAILBOX_RESP "\ + "(expected=0x%08x, readback=0x%08x)\n", + tmp, expected, resp); + return -EBUSY; + } + } + if (resp != expected && resp != MMSCH_VF_MAILBOX_RESP__INCOMPLETE && + init_status != MMSCH_VF_ENGINE_STATUS__PASS) + DRM_ERROR("MMSCH init status is incorrect! readback=0x%08x, header init status for jpeg: %x\n", + resp, init_status); + + } + return 0; +} + /** * jpeg_v4_0_3_hw_init - start and test JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * */ -static int jpeg_v4_0_3_hw_init(void *handle) +static int jpeg_v4_0_3_hw_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; struct amdgpu_ring *ring; int i, j, r, jpeg_inst; - for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { - jpeg_inst = GET_INST(JPEG, i); + if (amdgpu_sriov_vf(adev)) { + r = jpeg_v4_0_3_start_sriov(adev); + if (r) + return r; + + for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { + for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + ring->wptr = 0; + ring->wptr_old = 0; + jpeg_v4_0_3_dec_ring_set_wptr(ring); + ring->sched.ready = true; + } + } + } else { + /* This flag is not set for VF, assumed to be disabled always */ + if (RREG32_SOC15(VCN, GET_INST(VCN, 0), regVCN_RRMT_CNTL) & + 0x100) + adev->jpeg.caps |= AMDGPU_JPEG_CAPS(RRMT_ENABLED); - ring = adev->jpeg.inst[i].ring_dec; + for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { + jpeg_inst = GET_INST(JPEG, i); - if (ring->use_doorbell) - adev->nbio.funcs->vcn_doorbell_range( - adev, ring->use_doorbell, - (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + - 9 * jpeg_inst, - adev->jpeg.inst[i].aid_id); + ring = adev->jpeg.inst[i].ring_dec; - for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { - ring = &adev->jpeg.inst[i].ring_dec[j]; if (ring->use_doorbell) - WREG32_SOC15_OFFSET( - VCN, GET_INST(VCN, i), - regVCN_JPEG_DB_CTRL, - (ring->pipe ? (ring->pipe - 0x15) : 0), - ring->doorbell_index + adev->nbio.funcs->vcn_doorbell_range( + adev, ring->use_doorbell, + (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + + 9 * jpeg_inst, + adev->jpeg.inst[i].aid_id); + + for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { + ring = &adev->jpeg.inst[i].ring_dec[j]; + if (ring->use_doorbell) + WREG32_SOC15_OFFSET( + VCN, GET_INST(VCN, i), + regVCN_JPEG_DB_CTRL, + (ring->pipe ? (ring->pipe - 0x15) : 0), + ring->doorbell_index << VCN_JPEG_DB_CTRL__OFFSET__SHIFT | - VCN_JPEG_DB_CTRL__EN_MASK); - r = amdgpu_ring_test_helper(ring); - if (r) - return r; + VCN_JPEG_DB_CTRL__EN_MASK); + r = amdgpu_ring_test_helper(ring); + if (r) + return r; + } } } - DRM_DEV_INFO(adev->dev, "JPEG decode initialized successfully.\n"); return 0; } @@ -207,19 +428,24 @@ static int jpeg_v4_0_3_hw_init(void *handle) /** * jpeg_v4_0_3_hw_fini - stop the hardware block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Stop the JPEG block, mark ring as not ready any more */ -static int jpeg_v4_0_3_hw_fini(void *handle) +static int jpeg_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int ret = 0; cancel_delayed_work_sync(&adev->jpeg.idle_work); - if (adev->jpeg.cur_state != AMD_PG_STATE_GATE) - ret = jpeg_v4_0_3_set_powergating_state(adev, AMD_PG_STATE_GATE); + if (!amdgpu_sriov_vf(adev)) { + if (adev->jpeg.cur_state != AMD_PG_STATE_GATE) + ret = jpeg_v4_0_3_set_powergating_state(ip_block, AMD_PG_STATE_GATE); + } + + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG) && !amdgpu_sriov_vf(adev)) + amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0); return ret; } @@ -227,20 +453,19 @@ static int jpeg_v4_0_3_hw_fini(void *handle) /** * jpeg_v4_0_3_suspend - suspend JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * HW fini and suspend JPEG block */ -static int jpeg_v4_0_3_suspend(void *handle) +static int jpeg_v4_0_3_suspend(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; int r; - r = jpeg_v4_0_3_hw_fini(adev); + r = jpeg_v4_0_3_hw_fini(ip_block); if (r) return r; - r = amdgpu_jpeg_suspend(adev); + r = amdgpu_jpeg_suspend(ip_block->adev); return r; } @@ -248,20 +473,19 @@ static int jpeg_v4_0_3_suspend(void *handle) /** * jpeg_v4_0_3_resume - resume JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Resume firmware and hw init JPEG block */ -static int jpeg_v4_0_3_resume(void *handle) +static int jpeg_v4_0_3_resume(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; int r; - r = amdgpu_jpeg_resume(adev); + r = amdgpu_jpeg_resume(ip_block->adev); if (r) return r; - r = jpeg_v4_0_3_hw_init(adev); + r = jpeg_v4_0_3_hw_init(ip_block); return r; } @@ -316,6 +540,75 @@ static void jpeg_v4_0_3_enable_clock_gating(struct amdgpu_device *adev, int inst WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CGC_GATE, data); } +static void jpeg_v4_0_3_start_inst(struct amdgpu_device *adev, int inst) +{ + int jpeg_inst = GET_INST(JPEG, inst); + + WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG, + 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT); + SOC15_WAIT_ON_RREG(JPEG, jpeg_inst, regUVD_PGFSM_STATUS, + UVD_PGFSM_STATUS__UVDJ_PWR_ON << + UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT, + UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK); + + /* disable anti hang mechanism */ + WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), + 0, ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK); + + /* JPEG disable CGC */ + jpeg_v4_0_3_disable_clock_gating(adev, inst); + + /* MJPEG global tiling registers */ + WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX8_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX10_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + + /* enable JMI channel */ + WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0, + ~UVD_JMI_CNTL__SOFT_RESET_MASK); +} + +static void jpeg_v4_0_3_start_jrbc(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + int jpeg_inst = GET_INST(JPEG, ring->me); + int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe); + + /* enable System Interrupt for JRBC */ + WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regJPEG_SYS_INT_EN), + JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe, + ~(JPEG_SYS_INT_EN__DJRBC0_MASK << ring->pipe)); + + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_LMI_JRBC_RB_VMID, + reg_offset, 0); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JRBC0_UVD_JRBC_RB_CNTL, + reg_offset, + (0x00000001L | 0x00000002L)); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW, + reg_offset, lower_32_bits(ring->gpu_addr)); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH, + reg_offset, upper_32_bits(ring->gpu_addr)); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JRBC0_UVD_JRBC_RB_RPTR, + reg_offset, 0); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JRBC0_UVD_JRBC_RB_WPTR, + reg_offset, 0); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JRBC0_UVD_JRBC_RB_CNTL, + reg_offset, 0x00000002L); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JRBC0_UVD_JRBC_RB_SIZE, + reg_offset, ring->ring_size / 4); + ring->wptr = RREG32_SOC15_OFFSET(JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_RB_WPTR, + reg_offset); +} + /** * jpeg_v4_0_3_start - start JPEG block * @@ -326,84 +619,36 @@ static void jpeg_v4_0_3_enable_clock_gating(struct amdgpu_device *adev, int inst static int jpeg_v4_0_3_start(struct amdgpu_device *adev) { struct amdgpu_ring *ring; - int i, j, jpeg_inst; + int i, j; for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { - jpeg_inst = GET_INST(JPEG, i); - - WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG, - 1 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT); - SOC15_WAIT_ON_RREG( - JPEG, jpeg_inst, regUVD_PGFSM_STATUS, - UVD_PGFSM_STATUS__UVDJ_PWR_ON - << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT, - UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK); - - /* disable anti hang mechanism */ - WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, - regUVD_JPEG_POWER_STATUS), - 0, ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK); - - /* JPEG disable CGC */ - jpeg_v4_0_3_disable_clock_gating(adev, i); - - /* MJPEG global tiling registers */ - WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX8_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - WREG32_SOC15(JPEG, jpeg_inst, regJPEG_DEC_GFX10_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - - /* enable JMI channel */ - WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), 0, - ~UVD_JMI_CNTL__SOFT_RESET_MASK); - + jpeg_v4_0_3_start_inst(adev, i); for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { - unsigned int reg_offset = (j?(0x40 * j - 0xc80):0); - ring = &adev->jpeg.inst[i].ring_dec[j]; - - /* enable System Interrupt for JRBC */ - WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, - regJPEG_SYS_INT_EN), - JPEG_SYS_INT_EN__DJRBC0_MASK << j, - ~(JPEG_SYS_INT_EN__DJRBC0_MASK << j)); - - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JMI0_UVD_LMI_JRBC_RB_VMID, - reg_offset, 0); - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_RB_CNTL, - reg_offset, - (0x00000001L | 0x00000002L)); - WREG32_SOC15_OFFSET( - JPEG, jpeg_inst, - regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_LOW, - reg_offset, lower_32_bits(ring->gpu_addr)); - WREG32_SOC15_OFFSET( - JPEG, jpeg_inst, - regUVD_JMI0_UVD_LMI_JRBC_RB_64BIT_BAR_HIGH, - reg_offset, upper_32_bits(ring->gpu_addr)); - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_RB_RPTR, - reg_offset, 0); - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_RB_WPTR, - reg_offset, 0); - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_RB_CNTL, - reg_offset, 0x00000002L); - WREG32_SOC15_OFFSET(JPEG, jpeg_inst, - regUVD_JRBC0_UVD_JRBC_RB_SIZE, - reg_offset, ring->ring_size / 4); - ring->wptr = RREG32_SOC15_OFFSET( - JPEG, jpeg_inst, regUVD_JRBC0_UVD_JRBC_RB_WPTR, - reg_offset); + jpeg_v4_0_3_start_jrbc(ring); } } return 0; } +static void jpeg_v4_0_3_stop_inst(struct amdgpu_device *adev, int inst) +{ + int jpeg_inst = GET_INST(JPEG, inst); + /* reset JMI */ + WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), + UVD_JMI_CNTL__SOFT_RESET_MASK, + ~UVD_JMI_CNTL__SOFT_RESET_MASK); + + jpeg_v4_0_3_enable_clock_gating(adev, inst); + + /* enable anti hang mechanism */ + WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JPEG_POWER_STATUS), + UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK, + ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK); + +} + /** * jpeg_v4_0_3_stop - stop JPEG block * @@ -413,31 +658,10 @@ static int jpeg_v4_0_3_start(struct amdgpu_device *adev) */ static int jpeg_v4_0_3_stop(struct amdgpu_device *adev) { - int i, jpeg_inst; + int i; - for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { - jpeg_inst = GET_INST(JPEG, i); - /* reset JMI */ - WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, regUVD_JMI_CNTL), - UVD_JMI_CNTL__SOFT_RESET_MASK, - ~UVD_JMI_CNTL__SOFT_RESET_MASK); - - jpeg_v4_0_3_enable_clock_gating(adev, i); - - /* enable anti hang mechanism */ - WREG32_P(SOC15_REG_OFFSET(JPEG, jpeg_inst, - regUVD_JPEG_POWER_STATUS), - UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK, - ~UVD_JPEG_POWER_STATUS__JPEG_POWER_STATUS_MASK); - - WREG32_SOC15(JPEG, jpeg_inst, regUVD_PGFSM_CONFIG, - 2 << UVD_PGFSM_CONFIG__UVDJ_PWR_CONFIG__SHIFT); - SOC15_WAIT_ON_RREG( - JPEG, jpeg_inst, regUVD_PGFSM_STATUS, - UVD_PGFSM_STATUS__UVDJ_PWR_OFF - << UVD_PGFSM_STATUS__UVDJ_PWR_STATUS__SHIFT, - UVD_PGFSM_STATUS__UVDJ_PWR_STATUS_MASK); - } + for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) + jpeg_v4_0_3_stop_inst(adev, i); return 0; } @@ -453,9 +677,8 @@ static uint64_t jpeg_v4_0_3_dec_ring_get_rptr(struct amdgpu_ring *ring) { struct amdgpu_device *adev = ring->adev; - return RREG32_SOC15_OFFSET( - JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_RPTR, - ring->pipe ? (0x40 * ring->pipe - 0xc80) : 0); + return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_RPTR, + jpeg_v4_0_3_core_reg_offset(ring->pipe)); } /** @@ -471,11 +694,16 @@ static uint64_t jpeg_v4_0_3_dec_ring_get_wptr(struct amdgpu_ring *ring) if (ring->use_doorbell) return adev->wb.wb[ring->wptr_offs]; - else - return RREG32_SOC15_OFFSET( - JPEG, GET_INST(JPEG, ring->me), - regUVD_JRBC0_UVD_JRBC_RB_WPTR, - ring->pipe ? (0x40 * ring->pipe - 0xc80) : 0); + + return RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR, + jpeg_v4_0_3_core_reg_offset(ring->pipe)); +} + +void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring) +{ + /* JPEG engine access for HDP flush doesn't work when RRMT is enabled. + * This is a workaround to avoid any HDP flush through JPEG ring. + */ } /** @@ -493,10 +721,8 @@ static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring) adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr); WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr)); } else { - WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), - regUVD_JRBC0_UVD_JRBC_RB_WPTR, - (ring->pipe ? (0x40 * ring->pipe - 0xc80) : - 0), + WREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, ring->me), regUVD_JRBC0_UVD_JRBC_RB_WPTR, + jpeg_v4_0_3_core_reg_offset(ring->pipe), lower_32_bits(ring->wptr)); } } @@ -508,15 +734,18 @@ static void jpeg_v4_0_3_dec_ring_set_wptr(struct amdgpu_ring *ring) * * Write a start command to the ring. */ -static void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring) +void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring) { - amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x62a04); /* PCTL0_MMHUB_DEEPSLEEP_IB */ + if (!amdgpu_sriov_vf(ring->adev)) { + amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, + 0, 0, PACKETJ_TYPE0)); + amdgpu_ring_write(ring, 0x62a04); /* PCTL0_MMHUB_DEEPSLEEP_IB */ - amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x80004000); + amdgpu_ring_write(ring, + PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, 0, + 0, PACKETJ_TYPE0)); + amdgpu_ring_write(ring, 0x80004000); + } } /** @@ -526,15 +755,18 @@ static void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring) * * Write a end command to the ring. */ -static void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring) +void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring) { - amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x62a04); + if (!amdgpu_sriov_vf(ring->adev)) { + amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, + 0, 0, PACKETJ_TYPE0)); + amdgpu_ring_write(ring, 0x62a04); - amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x00004000); + amdgpu_ring_write(ring, + PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, 0, + 0, PACKETJ_TYPE0)); + amdgpu_ring_write(ring, 0x00004000); + } } /** @@ -547,7 +779,7 @@ static void jpeg_v4_0_3_dec_ring_insert_end(struct amdgpu_ring *ring) * * Write a fence and a trap command to the ring. */ -static void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq, +void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq, unsigned int flags) { WARN_ON(flags & AMDGPU_FENCE_FLAG_64BIT); @@ -576,31 +808,11 @@ static void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE4)); amdgpu_ring_write(ring, 0); - if (ring->adev->jpeg.inst[ring->me].aid_id) { - amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET, - 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x4); - } else { - amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6)); - amdgpu_ring_write(ring, 0); - } - - amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x3fbc); - - if (ring->adev->jpeg.inst[ring->me].aid_id) { - amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET, - 0, PACKETJ_CONDITION_CHECK0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x0); - } else { - amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6)); - amdgpu_ring_write(ring, 0); - } + amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6)); + amdgpu_ring_write(ring, 0); - amdgpu_ring_write(ring, PACKETJ(JRBC_DEC_EXTERNAL_REG_WRITE_ADDR, - 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, 0x1); + amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE6)); + amdgpu_ring_write(ring, 0); amdgpu_ring_write(ring, PACKETJ(0, 0, 0, PACKETJ_TYPE7)); amdgpu_ring_write(ring, 0); @@ -616,7 +828,7 @@ static void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, * * Write ring commands to execute the indirect buffer. */ -static void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring, +void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring, struct amdgpu_job *job, struct amdgpu_ib *ib, uint32_t flags) @@ -625,11 +837,15 @@ static void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring, amdgpu_ring_write(ring, PACKETJ(regUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, (vmid | (vmid << 4))); + + if (ring->funcs->parse_cs) + amdgpu_ring_write(ring, 0); + else + amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8))); amdgpu_ring_write(ring, PACKETJ(regUVD_LMI_JPEG_VMID_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, (vmid | (vmid << 4))); + amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8))); amdgpu_ring_write(ring, PACKETJ(regUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); @@ -667,10 +883,16 @@ static void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring, amdgpu_ring_write(ring, 0x2); } -static void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, +void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, uint32_t val, uint32_t mask) { - uint32_t reg_offset = (reg << 2); + uint32_t reg_offset; + + /* Use normalized offsets if required */ + if (jpeg_v4_0_3_normalizn_reqd(ring->adev)) + reg = NORMALIZE_JPEG_REG_OFFSET(reg); + + reg_offset = (reg << 2); amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); @@ -694,7 +916,7 @@ static void jpeg_v4_0_3_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_ amdgpu_ring_write(ring, mask); } -static void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring, +void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring, unsigned int vmid, uint64_t pd_addr) { struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub]; @@ -709,9 +931,15 @@ static void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring, jpeg_v4_0_3_dec_ring_emit_reg_wait(ring, data0, data1, mask); } -static void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val) +void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val) { - uint32_t reg_offset = (reg << 2); + uint32_t reg_offset; + + /* Use normalized offsets if required */ + if (jpeg_v4_0_3_normalizn_reqd(ring->adev)) + reg = NORMALIZE_JPEG_REG_OFFSET(reg); + + reg_offset = (reg << 2); amdgpu_ring_write(ring, PACKETJ(regUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); @@ -727,7 +955,7 @@ static void jpeg_v4_0_3_dec_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t re amdgpu_ring_write(ring, val); } -static void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count) +void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count) { int i; @@ -739,21 +967,17 @@ static void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count) } } -static bool jpeg_v4_0_3_is_idle(void *handle) +static bool jpeg_v4_0_3_is_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; bool ret = false; int i, j; for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { - unsigned int reg_offset = (j?(0x40 * j - 0xc80):0); - - ret &= ((RREG32_SOC15_OFFSET( - JPEG, GET_INST(JPEG, i), - regUVD_JRBC0_UVD_JRBC_STATUS, - reg_offset) & - UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK) == + ret &= ((RREG32_SOC15_OFFSET(JPEG, GET_INST(JPEG, i), + regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j)) & + UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK) == UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK); } } @@ -761,36 +985,33 @@ static bool jpeg_v4_0_3_is_idle(void *handle) return ret; } -static int jpeg_v4_0_3_wait_for_idle(void *handle) +static int jpeg_v4_0_3_wait_for_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int ret = 0; int i, j; for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { - unsigned int reg_offset = (j?(0x40 * j - 0xc80):0); - - ret &= SOC15_WAIT_ON_RREG_OFFSET( - JPEG, GET_INST(JPEG, i), - regUVD_JRBC0_UVD_JRBC_STATUS, reg_offset, + ret &= (SOC15_WAIT_ON_RREG_OFFSET(JPEG, GET_INST(JPEG, i), + regUVD_JRBC0_UVD_JRBC_STATUS, jpeg_v4_0_3_core_reg_offset(j), UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK, - UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK); + UVD_JRBC0_UVD_JRBC_STATUS__RB_JOB_DONE_MASK)); } } return ret; } -static int jpeg_v4_0_3_set_clockgating_state(void *handle, +static int jpeg_v4_0_3_set_clockgating_state(struct amdgpu_ip_block *ip_block, enum amd_clockgating_state state) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - bool enable = (state == AMD_CG_STATE_GATE) ? true : false; + struct amdgpu_device *adev = ip_block->adev; + bool enable = state == AMD_CG_STATE_GATE; int i; for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { if (enable) { - if (!jpeg_v4_0_3_is_idle(handle)) + if (!jpeg_v4_0_3_is_idle(ip_block)) return -EBUSY; jpeg_v4_0_3_enable_clock_gating(adev, i); } else { @@ -800,12 +1021,17 @@ static int jpeg_v4_0_3_set_clockgating_state(void *handle, return 0; } -static int jpeg_v4_0_3_set_powergating_state(void *handle, +static int jpeg_v4_0_3_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int ret; + if (amdgpu_sriov_vf(adev)) { + adev->jpeg.cur_state = AMD_PG_STATE_UNGATE; + return 0; + } + if (state == adev->jpeg.cur_state) return 0; @@ -828,6 +1054,14 @@ static int jpeg_v4_0_3_set_interrupt_state(struct amdgpu_device *adev, return 0; } +static int jpeg_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + return 0; +} + static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -882,10 +1116,47 @@ static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev, return 0; } +static void jpeg_v4_0_3_core_stall_reset(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + int jpeg_inst = GET_INST(JPEG, ring->me); + int reg_offset = jpeg_v4_0_3_core_reg_offset(ring->pipe); + + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_JMI_CLIENT_STALL, + reg_offset, 0x1F); + SOC15_WAIT_ON_RREG_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_JMI_CLIENT_CLEAN_STATUS, + reg_offset, 0x1F, 0x1F); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_JPEG_LMI_DROP, + reg_offset, 0x1F); + WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 1 << ring->pipe); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_UVD_JMI_CLIENT_STALL, + reg_offset, 0x00); + WREG32_SOC15_OFFSET(JPEG, jpeg_inst, + regUVD_JMI0_JPEG_LMI_DROP, + reg_offset, 0x00); + WREG32_SOC15(JPEG, jpeg_inst, regJPEG_CORE_RST_CTRL, 0x00); +} + +static int jpeg_v4_0_3_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) +{ + if (amdgpu_sriov_vf(ring->adev)) + return -EOPNOTSUPP; + + amdgpu_ring_reset_helper_begin(ring, timedout_fence); + jpeg_v4_0_3_core_stall_reset(ring); + jpeg_v4_0_3_start_jrbc(ring); + return amdgpu_ring_reset_helper_end(ring, timedout_fence); +} + static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = { .name = "jpeg_v4_0_3", .early_init = jpeg_v4_0_3_early_init, - .late_init = NULL, .sw_init = jpeg_v4_0_3_sw_init, .sw_fini = jpeg_v4_0_3_sw_fini, .hw_init = jpeg_v4_0_3_hw_init, @@ -894,12 +1165,10 @@ static const struct amd_ip_funcs jpeg_v4_0_3_ip_funcs = { .resume = jpeg_v4_0_3_resume, .is_idle = jpeg_v4_0_3_is_idle, .wait_for_idle = jpeg_v4_0_3_wait_for_idle, - .check_soft_reset = NULL, - .pre_soft_reset = NULL, - .soft_reset = NULL, - .post_soft_reset = NULL, .set_clockgating_state = jpeg_v4_0_3_set_clockgating_state, .set_powergating_state = jpeg_v4_0_3_set_powergating_state, + .dump_ip_state = amdgpu_jpeg_dump_ip_state, + .print_ip_state = amdgpu_jpeg_print_ip_state, }; static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = { @@ -908,16 +1177,18 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = { .get_rptr = jpeg_v4_0_3_dec_ring_get_rptr, .get_wptr = jpeg_v4_0_3_dec_ring_get_wptr, .set_wptr = jpeg_v4_0_3_dec_ring_set_wptr, + .parse_cs = amdgpu_jpeg_dec_parse_cs, .emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + 8 + /* jpeg_v4_0_3_dec_ring_emit_vm_flush */ - 22 + 22 + /* jpeg_v4_0_3_dec_ring_emit_fence x2 vm fence */ + 18 + 18 + /* jpeg_v4_0_3_dec_ring_emit_fence x2 vm fence */ 8 + 16, .emit_ib_size = 22, /* jpeg_v4_0_3_dec_ring_emit_ib */ .emit_ib = jpeg_v4_0_3_dec_ring_emit_ib, .emit_fence = jpeg_v4_0_3_dec_ring_emit_fence, .emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush, + .emit_hdp_flush = jpeg_v4_0_3_ring_emit_hdp_flush, .test_ring = amdgpu_jpeg_dec_ring_test_ring, .test_ib = amdgpu_jpeg_dec_ring_test_ib, .insert_nop = jpeg_v4_0_3_dec_ring_nop, @@ -929,6 +1200,7 @@ static const struct amdgpu_ring_funcs jpeg_v4_0_3_dec_ring_vm_funcs = { .emit_wreg = jpeg_v4_0_3_dec_ring_emit_wreg, .emit_reg_wait = jpeg_v4_0_3_dec_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, + .reset = jpeg_v4_0_3_ring_reset, }; static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev) @@ -945,7 +1217,6 @@ static void jpeg_v4_0_3_set_dec_ring_funcs(struct amdgpu_device *adev) adev->jpeg.inst[i].aid_id = jpeg_inst / adev->jpeg.num_inst_per_aid; } - DRM_DEV_INFO(adev->dev, "JPEG decode is enabled in VM mode\n"); } static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_irq_funcs = { @@ -953,6 +1224,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_irq_funcs = { .process = jpeg_v4_0_3_process_interrupt, }; +static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_ras_irq_funcs = { + .set = jpeg_v4_0_3_set_ras_interrupt_state, + .process = amdgpu_jpeg_process_poison_irq, +}; + static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev) { int i; @@ -961,6 +1237,9 @@ static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev) adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings; } adev->jpeg.inst->irq.funcs = &jpeg_v4_0_3_irq_funcs; + + adev->jpeg.inst->ras_poison_irq.num_types = 1; + adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_3_ras_irq_funcs; } const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block = { @@ -1057,14 +1336,147 @@ static void jpeg_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev) jpeg_v4_0_3_inst_reset_ras_error_count(adev, i); } +static uint32_t jpeg_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_JPEG_V4_0_3_JPEG0: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF); + break; + case AMDGPU_JPEG_V4_0_3_JPEG1: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool jpeg_v4_0_3_query_ras_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst = 0, sub = 0, poison_stat = 0; + + for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++) + for (sub = 0; sub < AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK; sub++) + poison_stat += + jpeg_v4_0_3_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + static const struct amdgpu_ras_block_hw_ops jpeg_v4_0_3_ras_hw_ops = { .query_ras_error_count = jpeg_v4_0_3_query_ras_error_count, .reset_ras_error_count = jpeg_v4_0_3_reset_ras_error_count, + .query_poison_status = jpeg_v4_0_3_query_ras_poison_status, +}; + +static int jpeg_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + struct aca_bank_info info; + u64 misc0; + int ret; + + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; + + misc0 = bank->regs[ACA_REG_IDX_MISC0]; + switch (type) { + case ACA_SMU_TYPE_UE: + bank->aca_err_type = ACA_ERROR_TYPE_UE; + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, + 1ULL); + break; + case ACA_SMU_TYPE_CE: + bank->aca_err_type = ACA_ERROR_TYPE_CE; + ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, + ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; + } + + return ret; +} + +/* reference to smu driver if header file */ +static int jpeg_v4_0_3_err_codes[] = { + 16, 17, 18, 19, 20, 21, 22, 23, /* JPEG[0-7][S|D] */ + 24, 25, 26, 27, 28, 29, 30, 31 +}; + +static bool jpeg_v4_0_3_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + u32 instlo; + + instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]); + instlo &= GENMASK(31, 1); + + if (instlo != mmSMNAID_AID0_MCA_SMU) + return false; + + if (aca_bank_check_error_codes(handle->adev, bank, + jpeg_v4_0_3_err_codes, + ARRAY_SIZE(jpeg_v4_0_3_err_codes))) + return false; + + return true; +} + +static const struct aca_bank_ops jpeg_v4_0_3_aca_bank_ops = { + .aca_bank_parser = jpeg_v4_0_3_aca_bank_parser, + .aca_bank_is_valid = jpeg_v4_0_3_aca_bank_is_valid, }; +static const struct aca_info jpeg_v4_0_3_aca_info = { + .hwip = ACA_HWIP_TYPE_SMU, + .mask = ACA_ERROR_UE_MASK, + .bank_ops = &jpeg_v4_0_3_aca_bank_ops, +}; + +static int jpeg_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) +{ + int r; + + r = amdgpu_ras_block_late_init(adev, ras_block); + if (r) + return r; + + if (amdgpu_ras_is_supported(adev, ras_block->block) && + adev->jpeg.inst->ras_poison_irq.funcs) { + r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0); + if (r) + goto late_fini; + } + + r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG, + &jpeg_v4_0_3_aca_info, NULL); + if (r) + goto late_fini; + + return 0; + +late_fini: + amdgpu_ras_block_late_fini(adev, ras_block); + + return r; +} + static struct amdgpu_jpeg_ras jpeg_v4_0_3_ras = { .ras_block = { .hw_ops = &jpeg_v4_0_3_ras_hw_ops, + .ras_late_init = jpeg_v4_0_3_ras_late_init, }, }; |