[PATCH 3/3] media: qcom: iris: Improve diagnostics for system and session failures

[Dipayaan Roy <dipayaan.roy@oss.qualcomm.com>]

Log system and session response timeouts, firmware-reported failures,
and failed recovery attempts so the cause of an error is visible.

Add debug messages for unknown sessions, invalid state transitions,
and successful recovery. Avoid a redundant system init timeout message
after firmware reports an error, and rate-limit Gen2 session response
errors.

Signed-off-by: Dipayaan Roy <dipayaan.roy@oss.qualcomm.com>
---
 drivers/media/platform/qcom/iris/iris_core.c              |  2 ++
 drivers/media/platform/qcom/iris/iris_hfi_gen1_response.c |  8 +++++++-
 drivers/media/platform/qcom/iris/iris_hfi_gen2_response.c | 13 ++++++++++++-
 drivers/media/platform/qcom/iris/iris_probe.c             |  8 +++++++-
 drivers/media/platform/qcom/iris/iris_state.c             |  5 ++++-
 drivers/media/platform/qcom/iris/iris_utils.c             |  2 ++
 6 files changed, 34 insertions(+), 4 deletions(-)

diff --git a/drivers/media/platform/qcom/iris/iris_core.c b/drivers/media/platform/qcom/iris/iris_core.c
index d558475986ff..4224feaf1caf 100644
--- a/drivers/media/platform/qcom/iris/iris_core.c
+++ b/drivers/media/platform/qcom/iris/iris_core.c
@@ -43,6 +43,8 @@ static int iris_wait_for_system_response(struct iris_core *core)
 	ret = wait_for_completion_timeout(&core->core_init_done,
 					  msecs_to_jiffies(HW_RESPONSE_TIMEOUT_VALUE));
 	if (!ret) {
+		if (core->state != IRIS_CORE_ERROR)
+			dev_err(core->dev, "system init response timed out\n");
 		core->state = IRIS_CORE_ERROR;
 		return -ETIMEDOUT;
 	}
diff --git a/drivers/media/platform/qcom/iris/iris_hfi_gen1_response.c b/drivers/media/platform/qcom/iris/iris_hfi_gen1_response.c
index 1ce7f97d4087..5c411e69e5fd 100644
--- a/drivers/media/platform/qcom/iris/iris_hfi_gen1_response.c
+++ b/drivers/media/platform/qcom/iris/iris_hfi_gen1_response.c
@@ -284,6 +284,8 @@ static void iris_hfi_gen1_sys_init_done(struct iris_core *core, void *packet)
 	struct hfi_msg_sys_init_done_pkt *pkt = packet;
 
 	if (pkt->error_type != HFI_ERR_NONE) {
+		dev_err(core->dev, "system init failed (HFI error %#x)\n",
+			pkt->error_type);
 		core->state = IRIS_CORE_ERROR;
 		return;
 	}
@@ -740,8 +742,12 @@ static void iris_hfi_gen1_handle_response(struct iris_core *core, void *response
 			struct hfi_msg_session_hdr_pkt *shdr;
 
 			shdr = (struct hfi_msg_session_hdr_pkt *)hdr;
-			if (shdr->error_type != HFI_ERR_NONE)
+			if (shdr->error_type != HFI_ERR_NONE) {
+				dev_err(dev, "session %#x response %#x failed (HFI error %#x)\n",
+					inst->session_id, hdr->pkt_type,
+					shdr->error_type);
 				iris_inst_change_state(inst, IRIS_INST_ERROR);
+			}
 
 			if (pkt_info->pkt == HFI_MSG_SESSION_FLUSH) {
 				if (!(--inst->flush_responses_pending))
diff --git a/drivers/media/platform/qcom/iris/iris_hfi_gen2_response.c b/drivers/media/platform/qcom/iris/iris_hfi_gen2_response.c
index b82f50458fe1..d5b4b27e0549 100644
--- a/drivers/media/platform/qcom/iris/iris_hfi_gen2_response.c
+++ b/drivers/media/platform/qcom/iris/iris_hfi_gen2_response.c
@@ -296,6 +296,8 @@ static int iris_hfi_gen2_handle_system_init(struct iris_core *core,
 					    struct iris_hfi_packet *pkt)
 {
 	if (!(pkt->flags & HFI_FW_FLAGS_SUCCESS)) {
+		dev_err(core->dev, "system command %#x failed (flags %#x)\n",
+			pkt->type, pkt->flags);
 		core->state = IRIS_CORE_ERROR;
 		return 0;
 	}
@@ -929,8 +931,11 @@ static int iris_hfi_gen2_handle_session_response(struct iris_core *core,
 	};
 
 	inst = iris_get_instance(core, hdr->session_id);
-	if (!inst)
+	if (!inst) {
+		dev_dbg(core->dev, "response for unknown session %#x\n",
+			hdr->session_id);
 		return -EINVAL;
+	}
 
 	mutex_lock(&inst->lock);
 	inst_hfi_gen2 = to_iris_inst_hfi_gen2(inst);
@@ -958,6 +963,12 @@ static int iris_hfi_gen2_handle_session_response(struct iris_core *core,
 			if (packet->type > range[i].begin && packet->type < range[i].end) {
 				dequeue |= (packet->type == HFI_CMD_BUFFER);
 				ret = range[i].handle(inst, packet);
+				if (ret && !(packet->flags &
+					     HFI_FW_FLAGS_SESSION_ERROR))
+					dev_err_ratelimited(core->dev,
+							    "session %#x response %#x failed: %d\n",
+							    inst->session_id,
+							    packet->type, ret);
 				if (ret)
 					iris_inst_change_state(inst, IRIS_INST_ERROR);
 			}
diff --git a/drivers/media/platform/qcom/iris/iris_probe.c b/drivers/media/platform/qcom/iris/iris_probe.c
index dd48fa750013..9cefe090466b 100644
--- a/drivers/media/platform/qcom/iris/iris_probe.c
+++ b/drivers/media/platform/qcom/iris/iris_probe.c
@@ -306,9 +306,15 @@ static void iris_sys_error_handler(struct work_struct *work)
 {
 	struct iris_core *core =
 			container_of(work, struct iris_core, sys_error_handler.work);
+	int ret;
 
 	iris_core_deinit(core);
-	iris_core_init(core);
+	ret = iris_core_init(core);
+	if (ret)
+		dev_err(core->dev, "failed to recover from system error: %d\n",
+			ret);
+	else
+		dev_dbg(core->dev, "recovered from system error\n");
 }
 
 static int iris_probe(struct platform_device *pdev)
diff --git a/drivers/media/platform/qcom/iris/iris_state.c b/drivers/media/platform/qcom/iris/iris_state.c
index 5552725c614e..bcb138ccf08c 100644
--- a/drivers/media/platform/qcom/iris/iris_state.c
+++ b/drivers/media/platform/qcom/iris/iris_state.c
@@ -56,8 +56,11 @@ int iris_inst_change_state(struct iris_inst *inst,
 	if (request_state == IRIS_INST_ERROR)
 		goto change_state;
 
-	if (!iris_allow_inst_state_change(inst, request_state))
+	if (!iris_allow_inst_state_change(inst, request_state)) {
+		dev_dbg(inst->core->dev, "session %#x invalid state change %x -> %x\n",
+			inst->session_id, inst->state, request_state);
 		return -EINVAL;
+	}
 
 change_state:
 	dev_dbg(inst->core->dev, "state changed from %x to %x\n",
diff --git a/drivers/media/platform/qcom/iris/iris_utils.c b/drivers/media/platform/qcom/iris/iris_utils.c
index 1096cc4b01c3..5ca4d8e0836b 100644
--- a/drivers/media/platform/qcom/iris/iris_utils.c
+++ b/drivers/media/platform/qcom/iris/iris_utils.c
@@ -78,6 +78,8 @@ int iris_wait_for_session_response(struct iris_inst *inst, bool is_flush)
 	ret = wait_for_completion_timeout(done, msecs_to_jiffies(HW_RESPONSE_TIMEOUT_VALUE));
 	mutex_lock(&inst->lock);
 	if (!ret) {
+		dev_err(inst->core->dev, "session %#x %s response timed out\n",
+			inst->session_id, is_flush ? "flush" : "command");
 		iris_inst_change_state(inst, IRIS_INST_ERROR);
 		return -ETIMEDOUT;
 	}

-- 
2.43.0