drm/amd/ras: track debug mode for query mode

The RAS manager uses the rascore debug mode setting to control whether
MCA data is read directly or collected through PMFW MCA polling and
ClearMcaOnRead. Track that debug mode state in ras_mgr so
amdgpu_ras_get_error_query_mode() can classify queries as direct or
firmware-backed.

Limit the PMFW polling wait in recovery to firmware-backed MODE1 fatal
recovery so other reset paths, such as MODE2, do not pay the delay
unnecessarily.

Signed-off-by: Xiang Liu <xiang.liu@amd.com>
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

3 files changed, 24 insertions, 2 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index 297f8b04c4eb..35eb87591740 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -2751,7 +2751,8 @@ static void amdgpu_ras_do_recovery(struct work_struct *work)
 		}
 
 		if (amdgpu_ras_get_error_query_mode(adev, &error_query_mode)) {
-			if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY) {
+			if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY &&
+			    (ras->gpu_reset_flags & AMDGPU_RAS_GPU_RESET_MODE1_RESET)) {
 				/* wait 500ms to ensure pmfw polling mca bank info done */
 				msleep(500);
 			}
@@ -4422,6 +4423,10 @@ bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
 
 	if (amdgpu_sriov_vf(adev)) {
 		*error_query_mode = AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY;
+	} else if (amdgpu_uniras_enabled(adev)) {
+		*error_query_mode = amdgpu_ras_mgr_get_debug_mode(adev) ?
+			AMDGPU_RAS_DIRECT_ERROR_QUERY :
+			AMDGPU_RAS_FIRMWARE_ERROR_QUERY;
 	} else {
 		*error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY;
 	}
diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c
index c6c2e2cdaf18..e039249b41af 100644
--- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c
+++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c
@@ -800,9 +800,24 @@ int amdgpu_ras_mgr_lookup_bad_pages_in_a_row(struct amdgpu_device *adev,
 int amdgpu_ras_mgr_set_debug_mode(struct amdgpu_device *adev, bool enable)
 {
 	struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev);
+	int ret;
 
 	if (!ras_mgr || !ras_mgr->ras_core || !ras_mgr->ras_is_ready)
 		return false;
 
-	return ras_core_set_debug_mode(ras_mgr->ras_core, enable);
+	ret = ras_core_set_debug_mode(ras_mgr->ras_core, enable);
+	if (!ret)
+		ras_mgr->is_debug_mode = enable;
+
+	return ret;
+}
+
+bool amdgpu_ras_mgr_get_debug_mode(struct amdgpu_device *adev)
+{
+	struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev);
+
+	if (!ras_mgr || !ras_mgr->ras_core || !ras_mgr->ras_is_ready)
+		return true;
+
+	return ras_mgr->is_debug_mode;
 }
diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h
index a20bb8fdce87..b81fd88fca71 100644
--- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h
+++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h
@@ -53,6 +53,7 @@ struct amdgpu_ras_mgr {
 	void *virt_ras_cmd;
 	uint64_t last_poison_consumption_seqno;
 	bool ras_is_ready;
+	bool is_debug_mode;
 
 	bool is_paused;
 	struct completion ras_event_done;
@@ -86,4 +87,5 @@ int amdgpu_ras_mgr_resume_after_reset(struct amdgpu_device *adev);
 int amdgpu_ras_mgr_lookup_bad_pages_in_a_row(struct amdgpu_device *adev,
 		uint64_t addr, uint64_t *nps_page_addr, uint32_t max_page_count);
 int amdgpu_ras_mgr_set_debug_mode(struct amdgpu_device *adev, bool enable);
+bool amdgpu_ras_mgr_get_debug_mode(struct amdgpu_device *adev);
 #endif