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

drm/amdgpu: add cp queue registers for gfx12 ipdump

Browse Source 
Add gfx12 support of CP queue registers for all queues
to be used by devcoredump.

Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sunil Khatri <sunil.khatri@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Sunil Khatri 2024-06-27 18:39:55 +05:30 committed by Alex Deucher
parent f0c6b79bfc
commit 495e6173a4
1 changed files with 109 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c
View File

@ -130,6 +130,49 @@ static const struct amdgpu_hwip_reg_entry gc_reg_list_12_0[] = {
SOC15_REG_ENTRY_STR(GC, 0, regGRBM_STATUS_SE3)
};
static const struct amdgpu_hwip_reg_entry gc_cp_reg_list_12[] = {
/* compute registers */
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_VMID),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PERSISTENT_STATE),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PIPE_PRIORITY),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUEUE_PRIORITY),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_QUANTUM),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_BASE_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_RPTR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_POLL_ADDR_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_CONTROL),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_BASE_ADDR_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_RPTR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_IB_CONTROL),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_REQUEST),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_BASE_ADDR_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_CONTROL),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_RPTR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_EVENTS),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_LO),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_BASE_ADDR_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_CONTROL),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_OFFSET),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CNTL_STACK_SIZE),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_WG_STATE_OFFSET),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_CTX_SAVE_SIZE),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_GDS_RESOURCE_STATE),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_ERROR),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_EOP_WPTR_MEM),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_LO),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_PQ_WPTR_HI),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_OFFSET),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_CNTL_STACK_DW_CNT),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_SUSPEND_WG_STATE_OFFSET),
SOC15_REG_ENTRY_STR(GC, 0, regCP_HQD_DEQUEUE_STATUS)
};
#define DEFAULT_SH_MEM_CONFIG \
((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \
(SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \
@ -1200,6 +1243,7 @@ static void gfx_v12_0_alloc_ip_dump(struct amdgpu_device *adev)
{
uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
uint32_t *ptr;
uint32_t inst;
ptr = kcalloc(reg_count, sizeof(uint32_t), GFP_KERNEL);
if (ptr == NULL) {
@ -1208,6 +1252,19 @@ static void gfx_v12_0_alloc_ip_dump(struct amdgpu_device *adev)
} else {
adev->gfx.ip_dump_core = ptr;
}
/* Allocate memory for compute queue registers for all the instances */
reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
inst = adev->gfx.mec.num_mec * adev->gfx.mec.num_pipe_per_mec *
adev->gfx.mec.num_queue_per_pipe;
ptr = kcalloc(reg_count * inst, sizeof(uint32_t), GFP_KERNEL);
if (ptr == NULL) {
DRM_ERROR("Failed to allocate memory for Compute Queues IP Dump\n");
adev->gfx.ip_dump_compute_queues = NULL;
} else {
adev->gfx.ip_dump_compute_queues = ptr;
}
}
static int gfx_v12_0_sw_init(void *handle)
@ -1404,6 +1461,7 @@ static int gfx_v12_0_sw_fini(void *handle)
gfx_v12_0_free_microcode(adev);
kfree(adev->gfx.ip_dump_core);
kfree(adev->gfx.ip_dump_compute_queues);
return 0;
}
@ -4759,7 +4817,7 @@ static void gfx_v12_0_emit_mem_sync(struct amdgpu_ring *ring)
static void gfx_v12_ip_print(void *handle, struct drm_printer *p)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
uint32_t i;
uint32_t i, j, k, reg, index = 0;
uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
if (!adev->gfx.ip_dump_core)
@ -4769,12 +4827,36 @@ static void gfx_v12_ip_print(void *handle, struct drm_printer *p)
drm_printf(p, "%-50s \t 0x%08x\n",
gc_reg_list_12_0[i].reg_name,
adev->gfx.ip_dump_core[i]);
/* print compute queue registers for all instances */
if (!adev->gfx.ip_dump_compute_queues)
return;
reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
drm_printf(p, "\nnum_mec: %d num_pipe: %d num_queue: %d\n",
adev->gfx.mec.num_mec,
adev->gfx.mec.num_pipe_per_mec,
adev->gfx.mec.num_queue_per_pipe);
for (i = 0; i < adev->gfx.mec.num_mec; i++) {
for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
drm_printf(p, "\nmec %d, pipe %d, queue %d\n", i, j, k);
for (reg = 0; reg < reg_count; reg++) {
drm_printf(p, "%-50s \t 0x%08x\n",
gc_cp_reg_list_12[reg].reg_name,
adev->gfx.ip_dump_compute_queues[index + reg]);
}
index += reg_count;
}
}
}
}
static void gfx_v12_ip_dump(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
uint32_t i;
uint32_t i, j, k, reg, index = 0;
uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0);
if (!adev->gfx.ip_dump_core)
@ -4784,6 +4866,31 @@ static void gfx_v12_ip_dump(void *handle)
for (i = 0; i < reg_count; i++)
adev->gfx.ip_dump_core[i] = RREG32(SOC15_REG_ENTRY_OFFSET(gc_reg_list_12_0[i]));
amdgpu_gfx_off_ctrl(adev, true);
/* dump compute queue registers for all instances */
if (!adev->gfx.ip_dump_compute_queues)
return;
reg_count = ARRAY_SIZE(gc_cp_reg_list_12);
amdgpu_gfx_off_ctrl(adev, false);
mutex_lock(&adev->srbm_mutex);
for (i = 0; i < adev->gfx.mec.num_mec; i++) {
for (j = 0; j < adev->gfx.mec.num_pipe_per_mec; j++) {
for (k = 0; k < adev->gfx.mec.num_queue_per_pipe; k++) {
/* ME0 is for GFX so start from 1 for CP */
soc24_grbm_select(adev, 1+i, j, k, 0);
for (reg = 0; reg < reg_count; reg++) {
adev->gfx.ip_dump_compute_queues[index + reg] =
RREG32(SOC15_REG_ENTRY_OFFSET(
gc_cp_reg_list_12[reg]));
}
index += reg_count;
}
}
}
soc24_grbm_select(adev, 0, 0, 0, 0);
mutex_unlock(&adev->srbm_mutex);
amdgpu_gfx_off_ctrl(adev, true);
}
static const struct amd_ip_funcs gfx_v12_0_ip_funcs = {