This patchset fixes various runtime pm issues while testing pm runtime
on NGD slimbus controller.
During testing it was found that pm refcount was going negative and sometime
transactions are timeout after suspend resume.
These 4 patches fixes those issues and now NGD enters in supend state
and resumes properly without any data timeouts.
Thanks,
srini
Srinivas Kandagatla (4):
slimbus: messaging: start transaction ids from 1 instead of zero
slimbus: messaging: check for valid transaction id
slimbus: ngd: set correct device for pm
slimbus: ngd: reset dma setup during runtime pm
drivers/slimbus/messaging.c | 7 ++++---
drivers/slimbus/qcom-ngd-ctrl.c | 22 +++++++++++++---------
2 files changed, 17 insertions(+), 12 deletions(-)
--
2.21.0
For some reason we ended up using wrong device in some places for pm_runtime calls.
Fix this so that NGG driver can do runtime pm correctly.
Fixes: 917809e2280b ("slimbus: ngd: Add qcom SLIMBus NGD driver")
Signed-off-by: Srinivas Kandagatla <[email protected]>
---
drivers/slimbus/qcom-ngd-ctrl.c | 17 +++++++++--------
1 file changed, 9 insertions(+), 8 deletions(-)
diff --git a/drivers/slimbus/qcom-ngd-ctrl.c b/drivers/slimbus/qcom-ngd-ctrl.c
index c054e83ab636..f3ee8e036372 100644
--- a/drivers/slimbus/qcom-ngd-ctrl.c
+++ b/drivers/slimbus/qcom-ngd-ctrl.c
@@ -618,7 +618,7 @@ static void qcom_slim_ngd_rx(struct qcom_slim_ngd_ctrl *ctrl, u8 *buf)
(mc == SLIM_USR_MC_GENERIC_ACK &&
mt == SLIM_MSG_MT_SRC_REFERRED_USER)) {
slim_msg_response(&ctrl->ctrl, &buf[4], buf[3], len - 4);
- pm_runtime_mark_last_busy(ctrl->dev);
+ pm_runtime_mark_last_busy(ctrl->ctrl.dev);
}
}
@@ -1257,13 +1257,14 @@ static int qcom_slim_ngd_enable(struct qcom_slim_ngd_ctrl *ctrl, bool enable)
}
/* controller state should be in sync with framework state */
complete(&ctrl->qmi.qmi_comp);
- if (!pm_runtime_enabled(ctrl->dev) ||
- !pm_runtime_suspended(ctrl->dev))
- qcom_slim_ngd_runtime_resume(ctrl->dev);
+ if (!pm_runtime_enabled(ctrl->ctrl.dev) ||
+ !pm_runtime_suspended(ctrl->ctrl.dev))
+ qcom_slim_ngd_runtime_resume(ctrl->ctrl.dev);
else
- pm_runtime_resume(ctrl->dev);
- pm_runtime_mark_last_busy(ctrl->dev);
- pm_runtime_put(ctrl->dev);
+ pm_runtime_resume(ctrl->ctrl.dev);
+
+ pm_runtime_mark_last_busy(ctrl->ctrl.dev);
+ pm_runtime_put(ctrl->ctrl.dev);
ret = slim_register_controller(&ctrl->ctrl);
if (ret) {
@@ -1389,7 +1390,7 @@ static int qcom_slim_ngd_ssr_pdr_notify(struct qcom_slim_ngd_ctrl *ctrl,
/* Make sure the last dma xfer is finished */
mutex_lock(&ctrl->tx_lock);
if (ctrl->state != QCOM_SLIM_NGD_CTRL_DOWN) {
- pm_runtime_get_noresume(ctrl->dev);
+ pm_runtime_get_noresume(ctrl->ctrl.dev);
ctrl->state = QCOM_SLIM_NGD_CTRL_DOWN;
qcom_slim_ngd_down(ctrl);
qcom_slim_ngd_exit_dma(ctrl);
--
2.21.0
As tid is unsigned its hard to figure out if the tid is valid or
invalid. So Start the transaction ids from 1 instead of zero
so that we could differentiate between a valid tid and invalid tids
This is useful in cases where controller would add a tid for controller
specific transfers.
Fixes: d3062a210930 ("slimbus: messaging: add slim_alloc/free_txn_tid()")
Signed-off-by: Srinivas Kandagatla <[email protected]>
---
drivers/slimbus/messaging.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/slimbus/messaging.c b/drivers/slimbus/messaging.c
index f2b5d347d227..6097ddc43a35 100644
--- a/drivers/slimbus/messaging.c
+++ b/drivers/slimbus/messaging.c
@@ -66,7 +66,7 @@ int slim_alloc_txn_tid(struct slim_controller *ctrl, struct slim_msg_txn *txn)
int ret = 0;
spin_lock_irqsave(&ctrl->txn_lock, flags);
- ret = idr_alloc_cyclic(&ctrl->tid_idr, txn, 0,
+ ret = idr_alloc_cyclic(&ctrl->tid_idr, txn, 1,
SLIM_MAX_TIDS, GFP_ATOMIC);
if (ret < 0) {
spin_unlock_irqrestore(&ctrl->txn_lock, flags);
--
2.21.0
During suspend/resume NGD remote instance is power cycled along
with remotely controlled bam dma engine.
So Reset the dma configuration during this suspend resume path
so that we are not dealing with any stale dma setup.
Without this transactions timeout after first suspend resume path.
Fixes: 917809e2280b ("slimbus: ngd: Add qcom SLIMBus NGD driver")
Signed-off-by: Srinivas Kandagatla <[email protected]>
---
drivers/slimbus/qcom-ngd-ctrl.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/drivers/slimbus/qcom-ngd-ctrl.c b/drivers/slimbus/qcom-ngd-ctrl.c
index f3ee8e036372..7040293c2ee8 100644
--- a/drivers/slimbus/qcom-ngd-ctrl.c
+++ b/drivers/slimbus/qcom-ngd-ctrl.c
@@ -1080,7 +1080,8 @@ static void qcom_slim_ngd_setup(struct qcom_slim_ngd_ctrl *ctrl)
{
u32 cfg = readl_relaxed(ctrl->ngd->base);
- if (ctrl->state == QCOM_SLIM_NGD_CTRL_DOWN)
+ if (ctrl->state == QCOM_SLIM_NGD_CTRL_DOWN ||
+ ctrl->state == QCOM_SLIM_NGD_CTRL_ASLEEP)
qcom_slim_ngd_init_dma(ctrl);
/* By default enable message queues */
@@ -1131,6 +1132,7 @@ static int qcom_slim_ngd_power_up(struct qcom_slim_ngd_ctrl *ctrl)
dev_info(ctrl->dev, "Subsys restart: ADSP active framer\n");
return 0;
}
+ qcom_slim_ngd_setup(ctrl);
return 0;
}
@@ -1618,6 +1620,7 @@ static int __maybe_unused qcom_slim_ngd_runtime_suspend(struct device *dev)
struct qcom_slim_ngd_ctrl *ctrl = dev_get_drvdata(dev);
int ret = 0;
+ qcom_slim_ngd_exit_dma(ctrl);
if (!ctrl->qmi.handle)
return 0;
--
2.21.0
In some usecases transaction ids are dynamically allocated inside
the controller driver after sending the messages which have generic
acknowledge responses. So check for this before refcounting pm_runtime.
Without this we would end up imbalancing runtime pm count by
doing pm_runtime_put() in both slim_do_transfer() and slim_msg_response()
for a single pm_runtime_get() in slim_do_transfer()
Fixes: d3062a210930 ("slimbus: messaging: add slim_alloc/free_txn_tid()")
Signed-off-by: Srinivas Kandagatla <[email protected]>
---
drivers/slimbus/messaging.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/drivers/slimbus/messaging.c b/drivers/slimbus/messaging.c
index 6097ddc43a35..e5ae26227bdb 100644
--- a/drivers/slimbus/messaging.c
+++ b/drivers/slimbus/messaging.c
@@ -131,7 +131,8 @@ int slim_do_transfer(struct slim_controller *ctrl, struct slim_msg_txn *txn)
goto slim_xfer_err;
}
}
-
+ /* Initialize tid to invalid value */
+ txn->tid = 0;
need_tid = slim_tid_txn(txn->mt, txn->mc);
if (need_tid) {
@@ -163,7 +164,7 @@ int slim_do_transfer(struct slim_controller *ctrl, struct slim_msg_txn *txn)
txn->mt, txn->mc, txn->la, ret);
slim_xfer_err:
- if (!clk_pause_msg && (!need_tid || ret == -ETIMEDOUT)) {
+ if (!clk_pause_msg && (txn->tid == 0 || ret == -ETIMEDOUT)) {
/*
* remove runtime-pm vote if this was TX only, or
* if there was error during this transaction
--
2.21.0