drm/amdkfd: add debug set and clear address watch points operation

Shader read, write and atomic memory operations can be alerted to the
debugger as an address watch exception.

Allow the debugger to pass in a watch point to a particular memory
address per device.

Note that there exists only 4 watch points per devices to date, so have
the KFD keep track of what watch points are allocated or not.

Signed-off-by: Jonathan Kim <jonathan.kim@amd.com>
Reviewed-by: Felix Kuehling <felix.kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

8 files changed, 278 insertions, 3 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
index 774ecfc3451a..efd6a72aab4e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_aldebaran.c
@@ -118,6 +118,55 @@ static uint32_t kgd_aldebaran_set_wave_launch_mode(struct amdgpu_device *adev,
 	return data;
 }
 
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_aldebaran_set_address_watch(
+					struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	return watch_address_cntl;
+}
+
+uint32_t kgd_gfx_aldebaran_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls aldebaran_kfd2kgd = {
 	.program_sh_mem_settings = kgd_gfx_v9_program_sh_mem_settings,
 	.set_pasid_vmid_mapping = kgd_gfx_v9_set_pasid_vmid_mapping,
@@ -141,6 +190,8 @@ const struct kfd2kgd_calls aldebaran_kfd2kgd = {
 	.validate_trap_override_request = kgd_aldebaran_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_aldebaran_set_wave_launch_trap_override,
 	.set_wave_launch_mode = kgd_aldebaran_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_aldebaran_set_address_watch,
+	.clear_address_watch = kgd_gfx_aldebaran_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
 	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
 	.program_trap_handler_settings = kgd_gfx_v9_program_trap_handler_settings,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
index fbdc1b7b1e42..6df215aba4c4 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_arcturus.c
@@ -413,6 +413,8 @@ const struct kfd2kgd_calls arcturus_kfd2kgd = {
 	.validate_trap_override_request = kgd_gfx_v9_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_gfx_v9_set_wave_launch_trap_override,
 	.set_wave_launch_mode = kgd_gfx_v9_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v9_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
 	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
 	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
index a7a6edda557f..8ad7a7779e14 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.c
@@ -880,6 +880,82 @@ uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev,
 	return 0;
 }
 
+#define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H)
+uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	/* Turning off this watch point until we set all the registers */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			0);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	/* Enable the watch point */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	WREG32((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
+
 /* kgd_gfx_v10_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values
  * The values read are:
  *     ib_offload_wait_time     -- Wait Count for Indirect Buffer Offloads.
@@ -969,6 +1045,8 @@ const struct kfd2kgd_calls gfx_v10_kfd2kgd = {
 	.validate_trap_override_request = kgd_gfx_v10_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_gfx_v10_set_wave_launch_trap_override,
 	.set_wave_launch_mode = kgd_gfx_v10_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v10_set_address_watch,
+	.clear_address_watch = kgd_gfx_v10_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v10_get_iq_wait_times,
 	.build_grace_period_packet_info = kgd_gfx_v10_build_grace_period_packet_info,
 	.program_trap_handler_settings = program_trap_handler_settings,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
index 3a6aca2b0eaa..e6b70196071a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10.h
@@ -39,6 +39,14 @@ uint32_t kgd_gfx_v10_set_wave_launch_trap_override(struct amdgpu_device *adev,
 uint32_t kgd_gfx_v10_set_wave_launch_mode(struct amdgpu_device *adev,
 					 uint8_t wave_launch_mode,
 					 uint32_t vmid);
+uint32_t kgd_gfx_v10_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid);
+uint32_t kgd_gfx_v10_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id);
 void kgd_gfx_v10_get_iq_wait_times(struct amdgpu_device *adev, uint32_t *wait_times);
 void kgd_gfx_v10_build_grace_period_packet_info(struct amdgpu_device *adev,
 					       uint32_t wait_times,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
index ed36b433a48b..8c8437a4383f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v10_3.c
@@ -678,6 +678,7 @@ const struct kfd2kgd_calls gfx_v10_3_kfd2kgd = {
 	.disable_debug_trap = kgd_gfx_v10_disable_debug_trap,
 	.validate_trap_override_request = kgd_gfx_v10_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_gfx_v10_set_wave_launch_trap_override,
-	.set_wave_launch_mode = kgd_gfx_v10_set_wave_launch_mode
-
+	.set_wave_launch_mode = kgd_gfx_v10_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v10_set_address_watch,
+	.clear_address_watch = kgd_gfx_v10_clear_address_watch
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
index 9711d5128d09..52efa690a3c2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v11.c
@@ -737,6 +737,54 @@ static uint32_t kgd_gfx_v11_set_wave_launch_mode(struct amdgpu_device *adev,
 	return data;
 }
 
+#define TCP_WATCH_STRIDE (regTCP_WATCH1_ADDR_H - regTCP_WATCH0_ADDR_H)
+static uint32_t kgd_gfx_v11_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 7);
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, regTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	return watch_address_cntl;
+}
+
+uint32_t kgd_gfx_v11_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	return 0;
+}
+
 const struct kfd2kgd_calls gfx_v11_kfd2kgd = {
 	.program_sh_mem_settings = program_sh_mem_settings_v11,
 	.set_pasid_vmid_mapping = set_pasid_vmid_mapping_v11,
@@ -757,5 +805,7 @@ const struct kfd2kgd_calls gfx_v11_kfd2kgd = {
 	.disable_debug_trap = kgd_gfx_v11_disable_debug_trap,
 	.validate_trap_override_request = kgd_gfx_v11_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_gfx_v11_set_wave_launch_trap_override,
-	.set_wave_launch_mode = kgd_gfx_v11_set_wave_launch_mode
+	.set_wave_launch_mode = kgd_gfx_v11_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v11_set_address_watch,
+	.clear_address_watch = kgd_gfx_v11_clear_address_watch
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
index 0acc0c18dfe6..51d93fb13ea3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -816,6 +816,81 @@ uint32_t kgd_gfx_v9_set_wave_launch_mode(struct amdgpu_device *adev,
 	return 0;
 }
 
+#define TCP_WATCH_STRIDE (mmTCP_WATCH1_ADDR_H - mmTCP_WATCH0_ADDR_H)
+uint32_t kgd_gfx_v9_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid)
+{
+	uint32_t watch_address_high;
+	uint32_t watch_address_low;
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	watch_address_low = lower_32_bits(watch_address);
+	watch_address_high = upper_32_bits(watch_address) & 0xffff;
+
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VMID,
+			debug_vmid);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MODE,
+			watch_mode);
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			MASK,
+			watch_address_mask >> 6);
+
+	/* Turning off this watch point until we set all the registers */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			0);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_H) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_high);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_ADDR_L) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_low);
+
+	/* Enable the watch point */
+	watch_address_cntl = REG_SET_FIELD(watch_address_cntl,
+			TCP_WATCH0_CNTL,
+			VALID,
+			1);
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
+uint32_t kgd_gfx_v9_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id)
+{
+	uint32_t watch_address_cntl;
+
+	watch_address_cntl = 0;
+
+	WREG32_RLC((SOC15_REG_OFFSET(GC, 0, mmTCP_WATCH0_CNTL) +
+			(watch_id * TCP_WATCH_STRIDE)),
+			watch_address_cntl);
+
+	return 0;
+}
+
 /* kgd_gfx_v9_get_iq_wait_times: Returns the mmCP_IQ_WAIT_TIME1/2 values
  * The values read are:
  *     ib_offload_wait_time     -- Wait Count for Indirect Buffer Offloads.
@@ -1090,6 +1165,8 @@ const struct kfd2kgd_calls gfx_v9_kfd2kgd = {
 	.validate_trap_override_request = kgd_gfx_v9_validate_trap_override_request,
 	.set_wave_launch_trap_override = kgd_gfx_v9_set_wave_launch_trap_override,
 	.set_wave_launch_mode = kgd_gfx_v9_set_wave_launch_mode,
+	.set_address_watch = kgd_gfx_v9_set_address_watch,
+	.clear_address_watch = kgd_gfx_v9_clear_address_watch,
 	.get_iq_wait_times = kgd_gfx_v9_get_iq_wait_times,
 	.build_grace_period_packet_info = kgd_gfx_v9_build_grace_period_packet_info,
 	.get_cu_occupancy = kgd_gfx_v9_get_cu_occupancy,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
index 18f4970ac8e4..5f54bff0db49 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.h
@@ -84,6 +84,14 @@ uint32_t kgd_gfx_v9_set_wave_launch_trap_override(struct amdgpu_device *adev,
 					     uint32_t trap_mask_request,
 					     uint32_t *trap_mask_prev,
 					     uint32_t kfd_dbg_trap_cntl_prev);
+uint32_t kgd_gfx_v9_set_address_watch(struct amdgpu_device *adev,
+					uint64_t watch_address,
+					uint32_t watch_address_mask,
+					uint32_t watch_id,
+					uint32_t watch_mode,
+					uint32_t debug_vmid);
+uint32_t kgd_gfx_v9_clear_address_watch(struct amdgpu_device *adev,
+					uint32_t watch_id);
 void kgd_gfx_v9_get_iq_wait_times(struct amdgpu_device *adev, uint32_t *wait_times);
 void kgd_gfx_v9_build_grace_period_packet_info(struct amdgpu_device *adev,
 					       uint32_t wait_times,