[PATCH v3 00/10] mailbox: imx: Use threaded handler to avoid kworker in imx's remoteproc

[PATCH v3 00/10] mailbox: imx: Use threaded handler to avoid kworker in imx's remoteproc

[Sebastian Andrzej Siewior]

The imx's remoteproc driver uses a kworker from its mailbox callback to
complete the request. The reason is that the imx mailbox driver invokes
the callback from its interrupt handler and the remoteproc callback (at
least the rpmsg-tty) requires a preemptible context.

This works but is problematic in a PREEMPT_RT environment where the
latency of the invocation is important. By scheduling a kworker the
high task priority from the threaded handler is lost and the kworker
competes for CPU ressources with every SCHED_OTHER task in the system.
This can lead to long delays on a busy system with other RT threads
which are less important than the completion of this request.

Looking over other mailbox driver, like the arm_mhu for instance, they
use a threaded interrupt handler to invoke the callback. This avoids the
kworker detour.

The here suggested change utilises a threaded interrupt to invoke the
callback. The primary handler mask the interrupt source so that the
handler can run without getting interrupted by the interrupt again.
Doing so avoids marking the interrupt IRQF_ONESHOT so that in a
shared-interrupt environment the other interrupt can still fire while
the first is masked.

The first four patches are result of the sashiko review. Does not look
critical.

This change was tested on a im93 board with rpmsg-tty driver.

v2…v3: https://lore.kernel.org/r/20260603-imx_mbox_rproc-v2-0-a0059dc3b69a@linutronix.de
  - Forward the error in imx_mu_generic_tx() to the caller (new patch
    #1)
  - Extend the patch description a bit for for "Start splitting the IRQ
    handler" to briefly explain why callbacks are moved to the threaded
    handler.
  - Drop imx_mu_con_priv::pending. The primary handler wakes its
    threaded handler. Once the handler is woken, the pending flag must
    be set and there is no need to set/ clear it.
  - Avoid the double clk_disable_unprepare() if
    devm_mbox_controller_register() fails.

v1…v2: https://lore.kernel.org/r/20260529-imx_mbox_rproc-v1-0-b8ffc36e11e5@linutronix.de
  - Using correct register to enable RXDB event.
  - Update commit description for the "threaded interrupt", "unmasks the
    interrupt" => "masks the interrupt event".
  - Add a shutdown field so that the interrupt does not unmask the
    interrupt if it has been already disabled because the channel is
    about to be shutdown.  A possible race mentioned by sashiko.
  - Use devm_pm_runtime_enable(). This should avoid a possible race
    sashiko mentioned.
  - Use devm_of_platform_populate().

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
Sebastian Andrzej Siewior (10):
      mailbox: imx: Forward the timeout/ error in imx_mu_generic_tx()
      mailbox: imx: Add a channel shutdown field
      mailbox: imx: Use devm_pm_runtime_enable()
      mailbox: imx: use devm_of_platform_populate()
      mailbox: imx: Use channel index instead of zero in imx_mu_specific_rx()
      mailbox: imx: Start splitting the IRQ handler in primary and threaded handler
      mailbox: imx: Move the RX part of the mailbox into the threaded handler
      mailbox: imx: Move the RXDB part of the mailbox into the threaded handler
      mailbox: imx: Don't force-thread the primary handler
      remoteproc: imx_rproc: Invoke the callback directly

 drivers/mailbox/imx-mailbox.c  | 117 ++++++++++++++++++++++++++++++-----------
 drivers/remoteproc/imx_rproc.c |  33 +-----------
 2 files changed, 88 insertions(+), 62 deletions(-)
---
base-commit: b3f94b2b3f3e51ab880a51fc6510e1dafba654ed
change-id: 20260529-imx_mbox_rproc-7d512f5a6f78

Best regards,
-- 
Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[PATCH v3 01/10] mailbox: imx: Forward the timeout/ error in imx_mu_generic_tx()

[Sebastian Andrzej Siewior]

imx_mu_generic_tx() for the IMX_MU_TYPE_TXDB_V2 type polls on a register
which may timeout and is recognized as an error. This error is siltently
dropped and not dropped to the caller.

Forward the error to the caller.

Fixes: b5ef17917f3a7 ("mailbox: imx: fix TXDB_V2 channel race condition")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 246a9a9e39520..0028073be4a71 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -227,6 +227,7 @@ static int imx_mu_generic_tx(struct imx_mu_priv *priv,
 	u32 val;
 	int ret, count;
 
+	ret = 0;
 	switch (cp->type) {
 	case IMX_MU_TYPE_TX:
 		imx_mu_write(priv, *arg, priv->dcfg->xTR + cp->idx * 4);
@@ -259,7 +260,7 @@ static int imx_mu_generic_tx(struct imx_mu_priv *priv,
 		return -EINVAL;
 	}
 
-	return 0;
+	return ret;
 }
 
 static int imx_mu_generic_rx(struct imx_mu_priv *priv,

-- 
2.53.0

Re: [PATCH v3 01/10] mailbox: imx: Forward the timeout/ error in imx_mu_generic_tx()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

New issues:
- [High] Returning a negative error code (-ETIMEDOUT) from the mailbox send_data callback wedges the TX queue permanently.

Pre-existing issues:
- [High] Using readl_poll_timeout() with a timeout inside an atomic section creates up to 100ms of IRQ-disabled latency and is unsafe.
--

commit 00bd1a49fae92272b2f76397f788a118394fe7ae
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Forward the timeout/ error in imx_mu_generic_tx()

This commit modifies imx_mu_generic_tx() to return the actual error code
(such as a timeout) instead of silently dropping it and returning 0. This
was intended to properly notify the caller of failures during the busy-wait
polling loop for TXDB_V2 channels.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> index 246a9a9e39520..0028073be4a71 100644
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c
> @@ -227,6 +227,7 @@ static int imx_mu_generic_tx(struct imx_mu_priv *priv,
>  	u32 val;
>  	int ret, count;
>  
> +	ret = 0;
>  	switch (cp->type) {

[ ... ]

> @@ -259,7 +260,7 @@ static int imx_mu_generic_tx(struct imx_mu_priv *priv,
>  		return -EINVAL;
>  	}
>  
> -	return 0;
> +	return ret;

[Severity: High]
Does returning a negative error code here stall the TX queue?

The mailbox framework does not propagate errors from the send_data
callback back to the caller. Instead, it treats non-zero return values
as a sign the controller is temporarily busy:

    drivers/mailbox/mailbox.c:msg_submit() {
        ...
        err = chan->mbox->ops->send_data(chan, data);
        if (!err) {
            chan->active_req = data;
            chan->msg_count--;
        }
        ...
    }

Because this now returns -ETIMEDOUT, the framework skips dequeuing the
message (chan->msg_count is not decremented) and does not set
chan->active_req. Because err is non-zero, no retry timer is started.

The failed message remains at the head of the TX queue, blocking all
future messages. Subsequent calls to mbox_send_message() will repeatedly
attempt to send the exact same message.

[Severity: High]
This is a pre-existing issue, but does the busy-wait loop in this function
cause high latency with interrupts disabled?

When handling IMX_MU_TYPE_TXDB_V2, imx_mu_generic_tx() calls
readl_poll_timeout() with a 10ms timeout and retries up to 10 times.
However, the caller holds a spinlock with interrupts disabled:

    drivers/mailbox/mailbox.c:msg_submit() {
        scoped_guard(spinlock_irqsave, &chan->lock) {
            ...
            err = chan->mbox->ops->send_data(chan, data);
            ...
        }
    }

This can result in up to 100ms of busy-waiting with interrupts
disabled, which can trigger watchdogs.

Furthermore, readl_poll_timeout() relies on ktime_get(), which the kernel
documentation explicitly forbids using in atomic contexts if a timeout
is specified.

>  }
>  
>  static int imx_mu_generic_rx(struct imx_mu_priv *priv,

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=1

[PATCH v3 02/10] mailbox: imx: Add a channel shutdown field

[Sebastian Andrzej Siewior]

sashiko complained about possible teardown problem. The scenario

 CPU 0                              CPU 1
  imx_mu_isr()                   imx_mu_shutdown()
                                   imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
    imx_mu_specific_rx()
      imx_mu_xcr_rmw(priv, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0), 0);
                                   free_irq()

The RX event remains enabled because in this short window the RX event
was disabled in ->shutdown() while the interrupt was active and then
enabled again by the ISR while ->shutdown waited in free_irq().

This race requires timing and if happens can be problematic on shared
handlers if the "removed" channel triggers an interrupt. In this case
the irq-core will shutdown the interrupt with the "nobody cared"
message.

Introduce imx_mu_con_priv::shutdown to signal that the channel is
shutting down. This flag is set with the lock held (by
imx_mu_xcr_clr_shut()). The unmask side uses imx_mu_xcr_set_act() which
only enables the event if the channel has not been shutdown and
serialises on the same lock.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 40 ++++++++++++++++++++++++++++++++++++----
 1 file changed, 36 insertions(+), 4 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 0028073be4a71..e261b382d5c90 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -80,6 +80,7 @@ struct imx_mu_con_priv {
 	enum imx_mu_chan_type	type;
 	struct mbox_chan	*chan;
 	struct work_struct 	txdb_work;
+	bool			shutdown;
 };
 
 struct imx_mu_priv {
@@ -219,6 +220,36 @@ static u32 imx_mu_xcr_rmw(struct imx_mu_priv *priv, enum imx_mu_xcr type, u32 se
 	return val;
 }
 
+static void imx_mu_xcr_clr_shut(struct imx_mu_priv *priv, struct imx_mu_con_priv *cp,
+				enum imx_mu_xcr type, u32 clr)
+{
+	unsigned long flags;
+	u32 val;
+
+	spin_lock_irqsave(&priv->xcr_lock, flags);
+	cp->shutdown = true;
+
+	val = imx_mu_read(priv, priv->dcfg->xCR[type]);
+	val &= ~clr;
+	imx_mu_write(priv, val, priv->dcfg->xCR[type]);
+	spin_unlock_irqrestore(&priv->xcr_lock, flags);
+}
+
+static void imx_mu_xcr_set_act(struct imx_mu_priv *priv, struct imx_mu_con_priv *cp,
+			       enum imx_mu_xcr type, u32 set)
+{
+	unsigned long flags;
+	u32 val;
+
+	spin_lock_irqsave(&priv->xcr_lock, flags);
+	if (!cp->shutdown) {
+		val = imx_mu_read(priv, priv->dcfg->xCR[type]);
+		val |= set;
+		imx_mu_write(priv, val, priv->dcfg->xCR[type]);
+	}
+	spin_unlock_irqrestore(&priv->xcr_lock, flags);
+}
+
 static int imx_mu_generic_tx(struct imx_mu_priv *priv,
 			     struct imx_mu_con_priv *cp,
 			     void *data)
@@ -377,7 +408,7 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
 		*data++ = imx_mu_read(priv, priv->dcfg->xRR + (i % num_rr) * 4);
 	}
 
-	imx_mu_xcr_rmw(priv, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0), 0);
+	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0));
 	mbox_chan_received_data(cp->chan, (void *)priv->msg);
 
 	return 0;
@@ -605,6 +636,7 @@ static int imx_mu_startup(struct mbox_chan *chan)
 		return ret;
 	}
 
+	cp->shutdown = false;
 	switch (cp->type) {
 	case IMX_MU_TYPE_RX:
 		imx_mu_xcr_rmw(priv, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx), 0);
@@ -639,13 +671,13 @@ static void imx_mu_shutdown(struct mbox_chan *chan)
 
 	switch (cp->type) {
 	case IMX_MU_TYPE_TX:
-		imx_mu_xcr_rmw(priv, IMX_MU_TCR, 0, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx));
+		imx_mu_xcr_clr_shut(priv, cp, IMX_MU_TCR, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx));
 		break;
 	case IMX_MU_TYPE_RX:
-		imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
+		imx_mu_xcr_clr_shut(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 		break;
 	case IMX_MU_TYPE_RXDB:
-		imx_mu_xcr_rmw(priv, IMX_MU_GIER, 0, IMX_MU_xCR_GIEn(priv->dcfg->type, cp->idx));
+		imx_mu_xcr_clr_shut(priv, cp, IMX_MU_GIER, IMX_MU_xCR_GIEn(priv->dcfg->type, cp->idx));
 		break;
 	case IMX_MU_TYPE_RST:
 		imx_mu_xcr_rmw(priv, IMX_MU_CR, IMX_MU_xCR_RST(priv->dcfg->type), 0);

-- 
2.53.0

Re: [PATCH v3 02/10] mailbox: imx: Add a channel shutdown field

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Pre-existing issues:
- [High] The patch introduces `cp->shutdown` to fix a teardown race in the RX path but fails to apply the same fix to the TX path. If a channel is freed while TX messages are queued, the ISR can unconditionally unmask the TX interrupt after `shutdown()` has masked it, leading to a 'nobody cared' interrupt storm.
- [High] Polling jiffies with local IRQs disabled leads to potential deadlock in `imx_mu_tx_waiting_write()`.
- [Medium] Missing `pm_runtime_put_sync()` on the error path of `imx_mu_startup()` causes a permanent power management reference leak.
--

commit ceab4c0d8020e02fd68ce7762cacc2496963937c
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Add a channel shutdown field

sashiko complained about possible teardown problem. The scenario

 CPU 0                              CPU 1
  imx_mu_isr()                   imx_mu_shutdown()
                                   imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
    imx_mu_specific_rx()
      imx_mu_xcr_rmw(priv, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0), 0);
                                   free_irq()

The RX event remains enabled because in this short window the RX event
was disabled in ->shutdown() while the interrupt was active and then
enabled again by the ISR while ->shutdown waited in free_irq().

This race requires timing and if happens can be problematic on shared
handlers if the "removed" channel triggers an interrupt. In this case
the irq-core will shutdown the interrupt with the "nobody cared"
message.

Introduce imx_mu_con_priv::shutdown to signal that the channel is
shutting down. This flag is set with the lock held (by
imx_mu_xcr_clr_shut()). The unmask side uses imx_mu_xcr_set_act() which
only enables the event if the channel has not been shutdown and
serialises on the same lock.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> index 0028073be4a71..e261b382d5c90 100644
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c

[ ... ]

> @@ -605,6 +636,7 @@ static int imx_mu_startup(struct mbox_chan *chan)
>  		return ret;
>  	}
>  
> +	cp->shutdown = false;

[Severity: Medium]
This is a pre-existing issue and was not introduced by the patch under
review, but does this error path leak the PM runtime reference?

In imx_mu_startup(), pm_runtime_get_sync() is called to increment the PM
usage counter. If request_threaded_irq() fails, the function returns
directly without calling pm_runtime_put_sync():

drivers/mailbox/imx-mailbox.c:imx_mu_startup() {
    ...
    pm_runtime_get_sync(priv->dev);
    ...
    ret = request_threaded_irq(...);
    if (ret) {
        dev_err(...);
        return ret;
    }
    ...
}

The mailbox framework handles this failure via mbox_clean_and_put_channel(),
but does not invoke the shutdown() callback when startup() fails.
Consequently, can the device's PM runtime reference be permanently leaked,
preventing the hardware from entering a suspended state?

[ ... ]

> @@ -639,13 +671,13 @@ static void imx_mu_shutdown(struct mbox_chan *chan)
>  
>  	switch (cp->type) {
>  	case IMX_MU_TYPE_TX:
> -		imx_mu_xcr_rmw(priv, IMX_MU_TCR, 0, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx));
> +		imx_mu_xcr_clr_shut(priv, cp, IMX_MU_TCR, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx));
>  		break;

[Severity: High]
This is a pre-existing issue and was not introduced by the patch under
review, but does this patch leave the TX path vulnerable to a similar
teardown race?

The patch introduces cp->shutdown to fix the teardown race in the RX path,
but if a TX channel is freed while messages are queued, can the ISR
unconditionally unmask the TX interrupt after shutdown() has masked it?

When mbox_free_channel() calls imx_mu_shutdown(), it masks the TX interrupt
via imx_mu_xcr_clr_shut(), sets cp->shutdown = true, and waits in free_irq().

If a TX interrupt concurrently fires, imx_mu_isr() masks it and wakes the
threaded handler imx_mu_isr_th(). This handler calls mbox_chan_txdone() and
eventually triggers msg_submit().

Because the queue isn't purged yet, it dequeues the next message and
synchronously calls the driver's TX callback. However, imx_mu_generic_tx()
and imx_mu_specific_tx() unconditionally re-enable the TX interrupt:

drivers/mailbox/imx-mailbox.c:imx_mu_generic_tx() {
    ...
    case IMX_MU_TYPE_TX:
        imx_mu_write(priv, *arg, priv->dcfg->xTR + cp->idx * 4);
        imx_mu_xcr_rmw(priv, IMX_MU_TCR, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx), 0);
        break;
    ...
}

Once the threaded ISR finishes, free_irq() unblocks, leaving the hardware TX
interrupt enabled but without a registered handler. Will this result in a
"nobody cared" kernel warning?

[ ... ]

[Severity: High]
This is a pre-existing issue and was not introduced by the patch under
review, but does polling jiffies with local IRQs disabled lead to a potential
deadlock here?

The imx_mu_tx_waiting_write() function uses an open-coded timeout loop based
on jiffies:

drivers/mailbox/imx-mailbox.c:imx_mu_tx_waiting_write() {
    ...
    do {
        status = imx_mu_read(priv, priv->dcfg->xSR[IMX_MU_TSR]);
        can_write = status & IMX_MU_xSR_TEn(priv->dcfg->type, idx % 4);
    } while (!can_write && time_is_after_jiffies64(timeout_time));
    ...
}

This function is called from the send_data callback (imx_mu_seco_tx), which
the mailbox core invokes with the channel's spinlock (chan->lock) held. This
spinlock unconditionally disables local interrupts.

Because local interrupts are disabled, the timer tick cannot fire on the
local CPU to update jiffies. On a single-core system, or if the local CPU is
responsible for updating the tick, jiffies will never increment.

If the hardware does not promptly set the status bit, will the loop spin
infinitely and cause a hard lockup?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=2

[PATCH v3 03/10] mailbox: imx: Use devm_pm_runtime_enable()

[Sebastian Andrzej Siewior]

sashiko complained about early usage of the device while probe isn't
completed. This can be mitigated by delaying the pm_runtime_enable()
into the removal path instead doing it early. This ensures that in an
error case the device is removed (and imx_mu_shutdown()) before
pm_runtime_disable() so we don't have to do this manually.

For the order to work, lets move devm_mbox_controller_register() until
after the pm-runtime part. So the reverse order will be mbox-controller
removal followed by disabling pm runtime.

Use devm_pm_runtime_enable(), remove manual pm_runtime_disable()
invocations and move the pm_runtime handling in probe before
devm_mbox_controller_register().

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 24 +++++++++++-------------
 1 file changed, 11 insertions(+), 13 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index e261b382d5c90..516a05b64daa1 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -966,38 +966,36 @@ static int imx_mu_probe(struct platform_device *pdev)
 
 	platform_set_drvdata(pdev, priv);
 
-	ret = devm_mbox_controller_register(dev, &priv->mbox);
-	if (ret)
+	ret = devm_pm_runtime_enable(dev);
+	if (ret < 0)
 		goto disable_clk;
 
-	of_platform_populate(dev->of_node, NULL, NULL, dev);
-
-	pm_runtime_enable(dev);
-
 	ret = pm_runtime_resume_and_get(dev);
 	if (ret < 0)
-		goto disable_runtime_pm;
+		goto disable_clk;
 
 	ret = pm_runtime_put_sync(dev);
 	if (ret < 0)
-		goto disable_runtime_pm;
+		goto disable_clk;
 
 	clk_disable_unprepare(priv->clk);
 
+	ret = devm_mbox_controller_register(dev, &priv->mbox);
+	if (ret)
+		goto err_out;
+
+	of_platform_populate(dev->of_node, NULL, NULL, dev);
+
 	return 0;
 
-disable_runtime_pm:
-	pm_runtime_disable(dev);
 disable_clk:
 	clk_disable_unprepare(priv->clk);
+err_out:
 	return ret;
 }
 
 static void imx_mu_remove(struct platform_device *pdev)
 {
-	struct imx_mu_priv *priv = platform_get_drvdata(pdev);
-
-	pm_runtime_disable(priv->dev);
 }
 
 static const struct imx_mu_dcfg imx_mu_cfg_imx6sx = {

-- 
2.53.0

Re: [PATCH v3 03/10] mailbox: imx: Use devm_pm_runtime_enable()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Pre-existing issues:
- [High] Missing error check on `pm_runtime_get_sync()` in `imx_mu_startup()` allows unclocked hardware access.
- [High] Checking `ret < 0` for `pm_runtime_put_sync()` causes spurious probe failures and resource leaks on concurrent PM access.
- [High] Missing `pm_runtime_put_sync()` when `request_irq()` fails in `imx_mu_startup()`.
--

commit 023389d6153d2d912a21b34dd94fa1a45a63f27a
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Use devm_pm_runtime_enable()

This patch updates the imx mailbox driver to use devm_pm_runtime_enable()
and reorders the probe function to delay mailbox controller registration
until after PM runtime initialization. This ensures proper cleanup order
during device removal or probe failure, resolving issues with early device
usage.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> index e261b382d5c90..516a05b64daa1 100644
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c
[ ... ]
> @@ -966,38 +966,36 @@ static int imx_mu_probe(struct platform_device *pdev)
[ ... ]
>  	ret = pm_runtime_resume_and_get(dev);
>  	if (ret < 0)
> -		goto disable_runtime_pm;
> +		goto disable_clk;
>  
>  	ret = pm_runtime_put_sync(dev);
>  	if (ret < 0)
> -		goto disable_runtime_pm;
> +		goto disable_clk;

[Severity: High]
This isn't a bug introduced by this patch, but should the return value
of pm_runtime_put_sync() be treated as a fatal error here?

pm_runtime_put_sync() can return -EAGAIN or -EBUSY if the device cannot be
suspended immediately, such as when another thread concurrently accesses the
device's PM runtime via sysfs.

If this happens, the valid non-error return value causes the driver to
spuriously abort the probe process. It also leaves the device in RPM_ACTIVE,
which leaks the clock enabled by the earlier pm_runtime_resume_and_get() call.

[ ... ]

While reviewing the PM runtime usage, I also noticed two other pre-existing
issues in imx_mu_startup().

[Severity: High]
This is a pre-existing issue, but does this code leave the clock disabled
if the PM runtime resume fails?

drivers/mailbox/imx-mailbox.c:imx_mu_startup() {
	...
	pm_runtime_get_sync(priv->dev);

	spin_lock_irqsave(&priv->xcr_lock, flags);
	imx_mu_xcr_rmw(priv, 0, IMX_MU_xCR_GIEn(cp->type, cp->idx));
	...
}

If pm_runtime_get_sync() fails (for example, if the clock fails to enable),
the return value is ignored and the device remains suspended. The function
then immediately proceeds to access hardware registers via imx_mu_xcr_rmw().
Can accessing unclocked peripheral registers cause a synchronous external
abort and lead to a system crash?

[Severity: High]
This is a pre-existing issue, but does this error path leak the PM runtime
usage counter?

drivers/mailbox/imx-mailbox.c:imx_mu_startup() {
	...
	pm_runtime_get_sync(priv->dev);
	...
	ret = request_irq(priv->irq[cp->type], imx_mu_isr, irq_flag,
			  cp->irq_desc, chan);
	if (ret) {
		dev_err(priv->dev, "Unable to acquire IRQ %d\n",
			priv->irq[cp->type]);
		return ret;
	}
	...
}

If request_irq() fails, the function returns the error directly without
calling pm_runtime_put_sync(). This would permanently leak the PM runtime
usage counter, preventing the device from ever suspending and keeping the
hardware clock continuously enabled.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=3

[PATCH v3 04/10] mailbox: imx: use devm_of_platform_populate()

[Sebastian Andrzej Siewior]

The driver uses of_platform_populate() but does not remove the added
devices on removal. This can lead to "double devices" on module removal
followed by adding the module again.

Use devm_of_platform_populate() to remove the populated devices once the
parent device is removed.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 516a05b64daa1..a8d4e970fb786 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -984,7 +984,7 @@ static int imx_mu_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_out;
 
-	of_platform_populate(dev->of_node, NULL, NULL, dev);
+	devm_of_platform_populate(dev);
 
 	return 0;
 

-- 
2.53.0

Re: [PATCH v3 04/10] mailbox: imx: use devm_of_platform_populate()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The return value of devm_of_platform_populate() is ignored, leading to silent probe success even if child devices fail to populate, and breaking the intended cleanup on module removal.
--

commit 583b7f6c5d004040c54817c853b20be8858487ce
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: use devm_of_platform_populate()

This commit transitions the driver to use the managed function
devm_of_platform_populate(). This is intended to ensure that populated
child devices are properly removed when the parent device is removed,
fixing an issue where module reloads could lead to "double devices".

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c
> @@ -984,7 +984,7 @@ static int imx_mu_probe(struct platform_device *pdev)
>  	if (ret)
>  		goto err_out;
>  
> -	of_platform_populate(dev->of_node, NULL, NULL, dev);
> +	devm_of_platform_populate(dev);

[Severity: High]
Should this code check the return value of devm_of_platform_populate()?

If the function fails and returns an error code, the devres cleanup action
is not registered, but imx_mu_probe() will still return 0. This could
lead to a situation where child devices silently fail to populate, and any
partially populated devices would not be cleaned up on module removal.

For example:

	ret = devm_of_platform_populate(dev);
	if (ret)
		goto err_out;

>  
>  	return 0;
>

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=4

[PATCH v3 05/10] mailbox: imx: Use channel index instead of zero in imx_mu_specific_rx()

[Sebastian Andrzej Siewior]

imx_mu_specific_rx() masks channel 0 and unmasks it again at the end of
the function. Given that at startup the channel index got unmasked it
should do the right job.

This here either unmasks the actual channel or another one but should
have no impact given that it reverses its doing at the end.

Peng Fan commented here:
| For specific rx channel, whether it is i.MX8 SCU or i.MX ELE, actually there is
| only 1 channel as of now, but it seems better to use cp->idx in case more
| channels in future.

Use the channel index instead of zero.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index a8d4e970fb786..408cd083c64ee 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -381,7 +381,7 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
 
 	data = (u32 *)priv->msg;
 
-	imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, 0));
+	imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 	*data++ = imx_mu_read(priv, priv->dcfg->xRR);
 
 	if (priv->dcfg->type & IMX_MU_V2_S4) {
@@ -408,7 +408,7 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
 		*data++ = imx_mu_read(priv, priv->dcfg->xRR + (i % num_rr) * 4);
 	}
 
-	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0));
+	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 	mbox_chan_received_data(cp->chan, (void *)priv->msg);
 
 	return 0;

-- 
2.53.0

Re: [PATCH v3 05/10] mailbox: imx: Use channel index instead of zero in imx_mu_specific_rx()

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 4 potential issue(s) to consider:

Pre-existing issues:
- [High] Unsafe hardware polling inside atomic context leads to deadlocks.
- [High] Error paths in `imx_mu_specific_rx()` leave the RX interrupt permanently masked.
- [High] Error paths in `imx_mu_seco_rxdb()` skip clearing GIP and leave the RXDB interrupt permanently masked.
- [High] PM runtime usage counter leak in `imx_mu_startup()` error path.
--

commit ab09c59a56c0e6ea31b7f95dd9b506c460b17d31
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Use channel index instead of zero in imx_mu_specific_rx()

imx_mu_specific_rx() masks channel 0 and unmasks it again at the end of
the function. Given that at startup the channel index got unmasked it
should do the right job.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> index a8d4e970fb786..408cd083c64ee 100644
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c
> @@ -381,7 +381,7 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
>  
>  	data = (u32 *)priv->msg;
>  
> -	imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, 0));
> +	imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
>  	*data++ = imx_mu_read(priv, priv->dcfg->xRR);
>  
>  	if (priv->dcfg->type & IMX_MU_V2_S4) {

[Severity: High]
This is a pre-existing issue, but the error paths in imx_mu_specific_rx() skip
the corresponding unmask operation.

If size exceeds max_size / 4 or if readl_poll_timeout() times out:

drivers/mailbox/imx-mailbox.c:imx_mu_specific_rx() {
...
	if (size > max_size / 4) {
		dev_err(priv->dev, "Maximal message size (%u bytes) exceeded on RX; got: %i bytes\n", max_size, size << 2);
		return -EINVAL;
	}
...
	for (i = 1; i < size; i++) {
		ret = readl_poll_timeout(...);
		if (ret) {
			dev_err(priv->dev, "timeout read idx %d\n", i);
			return ret;
		}
...
}

Does returning an error here skip unmasking the RIEn interrupt?

In the threaded IRQ handler imx_mu_isr_th(), the unmask is only performed if
this function returns 0. Could this leave the RX interrupt permanently masked
and cause a denial of service for the channel?

[ ... ]

> @@ -408,7 +408,7 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
>  		*data++ = imx_mu_read(priv, priv->dcfg->xRR + (i % num_rr) * 4);
>  	}
>  
> -	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, 0));
> +	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
>  	mbox_chan_received_data(cp->chan, (void *)priv->msg);
>  
>  	return 0;

[Severity: High]
This isn't a bug introduced by this patch, but are there unsafe hardware polling
loops inside atomic contexts elsewhere in this file?

The imx_mu_send_data() callback executes in atomic context (with chan->lock held
and interrupts disabled), but the underlying TX functions wait using mechanisms
that are unsafe in atomic context.

For example, imx_mu_tx_waiting_write():

drivers/mailbox/imx-mailbox.c:imx_mu_tx_waiting_write() {
...
	} while (!can_write && time_is_after_jiffies64(timeout_time));
...
}

There are also readl_poll_timeout() calls in imx_mu_generic_tx() and
imx_mu_specific_tx():

drivers/mailbox/imx-mailbox.c:imx_mu_specific_tx() {
...
		ret = readl_poll_timeout(priv->base + priv->dcfg->xSR[IMX_MU_TSR], xsr, xsr & IMX_MU_xSR_TEn(priv->dcfg->type, i % num_tr), 0, 5 * USEC_PER_SEC);
...
}

Can polling jiffies or using readl_poll_timeout() inside a spinlock with IRQs
disabled cause hard deadlocks, especially if the local CPU is responsible for
advancing the timer tick? Should readl_poll_timeout_atomic() be used here
instead?


[Severity: High]
This is a pre-existing issue, but do error paths in imx_mu_seco_rxdb() skip
clearing the GIP interrupt status?

drivers/mailbox/imx-mailbox.c:imx_mu_seco_rxdb() {
...
	if (byte_size > sizeof(*msg)) {
...
		err = -EINVAL;
		goto error;
	}
...
	for (i = 1; i < msg->hdr.size; i++) {
		err = imx_mu_rx_waiting_read(priv, arg++, i);
		if (err) {
			dev_err(priv->dev, "timeout read idx %d\n", i);
			goto error;
		}
	}
...
error:
	return err;
}

Because these paths jump directly to the error label, they skip the GSR
register write that clears the GIPn interrupt status.

Furthermore, returning an error skips the GIEn unmasking in imx_mu_isr_th().
Could this leave the RXDB interrupt permanently masked and cause spurious
interrupts if the channel is later closed and reopened?


[Severity: High]
This is a pre-existing issue, but does imx_mu_startup() leak the PM runtime
usage counter on error?

drivers/mailbox/imx-mailbox.c:imx_mu_startup() {
...
	ret = pm_runtime_get_sync(priv->dev);
...
	ret = request_threaded_irq(priv->irq[cp->type],
				   imx_mu_isr,
				   imx_mu_isr_th,
				   IRQF_SHARED,
				   cp->irq_desc, cp);
	if (ret) {
		dev_err(priv->dev, "Unable to acquire IRQ %d\n", priv->irq[cp->type]);
		return ret;
	}
...
}

If request_threaded_irq() fails, the function returns immediately without
calling pm_runtime_put_sync().

Since the mailbox core does not call shutdown() if startup() fails, does this
permanently prevent the device and its power domain from suspending?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=5

[PATCH v3 06/10] mailbox: imx: Start splitting the IRQ handler in primary and threaded handler

[Sebastian Andrzej Siewior]

Split the mailbox irq handling into a primary handler (imx_mu_isr()) and
a threaded handler (imx_mu_isr_th()). The primary handler masks the
interrupt event so the threaded handler can run without raising the
interrupt again.

The goal here is to invoke the mailbox core functions (such as
mbox_chan_received_data(), mbox_chan_txdone()) in preemptible context which is
made possible by using an threaded interrupt handler. This in turn means that
mailbox's client callbacks are invoked in preemptible context, too. This then
allows the mailbox client callback to skip an indirection via a workqueue if
it requries preemptible callback.

As a first step, prepare the logic and move TX handling part.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 27 ++++++++++++++++++++++++---
 1 file changed, 24 insertions(+), 3 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 408cd083c64ee..87acc43cb99c4 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -540,11 +540,31 @@ static void imx_mu_txdb_work(struct work_struct *t)
 	mbox_chan_txdone(cp->chan, 0);
 }
 
+static irqreturn_t imx_mu_isr_th(int irq, void *p)
+{
+	struct mbox_chan *chan = p;
+	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
+	struct imx_mu_con_priv *cp = chan->con_priv;
+
+	switch (cp->type) {
+	case IMX_MU_TYPE_TX:
+		mbox_chan_txdone(chan, 0);
+		break;
+
+	default:
+		dev_warn_ratelimited(priv->dev, "Unhandled channel type %d\n",
+				     cp->type);
+		return IRQ_NONE;
+	}
+	return IRQ_HANDLED;
+}
+
 static irqreturn_t imx_mu_isr(int irq, void *p)
 {
 	struct mbox_chan *chan = p;
 	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
 	struct imx_mu_con_priv *cp = chan->con_priv;
+	irqreturn_t ret = IRQ_HANDLED;
 	u32 val, ctrl;
 
 	switch (cp->type) {
@@ -580,7 +600,7 @@ static irqreturn_t imx_mu_isr(int irq, void *p)
 	if ((val == IMX_MU_xSR_TEn(priv->dcfg->type, cp->idx)) &&
 	    (cp->type == IMX_MU_TYPE_TX)) {
 		imx_mu_xcr_rmw(priv, IMX_MU_TCR, 0, IMX_MU_xCR_TIEn(priv->dcfg->type, cp->idx));
-		mbox_chan_txdone(chan, 0);
+		ret = IRQ_WAKE_THREAD;
 	} else if ((val == IMX_MU_xSR_RFn(priv->dcfg->type, cp->idx)) &&
 		   (cp->type == IMX_MU_TYPE_RX)) {
 		priv->dcfg->rx(priv, cp);
@@ -595,7 +615,7 @@ static irqreturn_t imx_mu_isr(int irq, void *p)
 	if (priv->suspend)
 		pm_system_wakeup();
 
-	return IRQ_HANDLED;
+	return ret;
 }
 
 static int imx_mu_send_data(struct mbox_chan *chan, void *data)
@@ -630,7 +650,8 @@ static int imx_mu_startup(struct mbox_chan *chan)
 	if (!(priv->dcfg->type & IMX_MU_V2_IRQ))
 		irq_flag |= IRQF_SHARED;
 
-	ret = request_irq(priv->irq[cp->type], imx_mu_isr, irq_flag, cp->irq_desc, chan);
+	ret = request_threaded_irq(priv->irq[cp->type], imx_mu_isr, imx_mu_isr_th,
+				   irq_flag, cp->irq_desc, chan);
 	if (ret) {
 		dev_err(priv->dev, "Unable to acquire IRQ %d\n", priv->irq[cp->type]);
 		return ret;

-- 
2.53.0

Re: [PATCH v3 06/10] mailbox: imx: Start splitting the IRQ handler in primary and threaded handler

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] Channel shutdown race condition leads to an unhandled IRQ storm on shared interrupt line.
--

commit 7ed2b361db99c9ca608465a64760164124f3d0d2
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Start splitting the IRQ handler in primary and threaded handler

This commit splits the mailbox IRQ handling into primary and threaded
handlers. The primary handler masks the event, and the threaded handler
runs the mailbox core functions like mbox_chan_txdone() in a preemptible
context.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> index 408cd083c64ee..87acc43cb99c4 100644
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c
> @@ -540,11 +540,31 @@ static void imx_mu_txdb_work(struct work_struct *t)
>  	mbox_chan_txdone(cp->chan, 0);
>  }
>  
> +static irqreturn_t imx_mu_isr_th(int irq, void *p)
> +{
> +	struct mbox_chan *chan = p;
> +	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
> +	struct imx_mu_con_priv *cp = chan->con_priv;
> +
> +	switch (cp->type) {
> +	case IMX_MU_TYPE_TX:
> +		mbox_chan_txdone(chan, 0);

[Severity: High]
This is a pre-existing issue, but does moving this into a threaded handler
expose a race condition during channel shutdown?

When mbox_free_channel() is called, it invokes the driver's imx_mu_shutdown()
callback, which masks the TX interrupt and calls free_irq().

Because free_irq() waits for any executing threaded handlers to finish,
it will block if imx_mu_isr_th() is currently executing mbox_chan_txdone().

This in turn triggers the mailbox core to submit the next previously queued
message via msg_submit() -> chan->mbox->ops->send_data().

The send callback (such as imx_mu_generic_tx() or imx_mu_specific_tx()) then
unconditionally unmasks the TX interrupt by calling imx_mu_xcr_rmw(),
completely bypassing any shutdown state.

Once the threaded handler finishes, free_irq() completes and unregisters
the interrupt action, but the hardware interrupt remains unmasked.

Since IRQF_SHARED is used, could the unhandled interrupt fire continuously,
resulting in a storm of IRQ_NONE that eventually causes the kernel to disable
the shared IRQ line and break IPC globally?

> +		break;
> +
> +	default:
> +		dev_warn_ratelimited(priv->dev, "Unhandled channel type %d\n",
> +				     cp->type);
> +		return IRQ_NONE;
> +	}
> +	return IRQ_HANDLED;
> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=6

[PATCH v3 07/10] mailbox: imx: Move the RX part of the mailbox into the threaded handler

[Sebastian Andrzej Siewior]

Move RX callback handling into the threaded handler. This is similar to
the TX side except that we explicitly mask the source interrupt in the
primary handler and unmask it in the threaded handler again after
success. This was done automatically in the TX part.

The masking/ unmasking can be removed from imx_mu_specific_rx() since it
already happens in the primary/ threaded handler before invoking the
channel specific callback.

Move RX channel handling into threaded handler.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 87acc43cb99c4..1219c35b116e4 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -381,7 +381,6 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
 
 	data = (u32 *)priv->msg;
 
-	imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 	*data++ = imx_mu_read(priv, priv->dcfg->xRR);
 
 	if (priv->dcfg->type & IMX_MU_V2_S4) {
@@ -408,7 +407,6 @@ static int imx_mu_specific_rx(struct imx_mu_priv *priv, struct imx_mu_con_priv *
 		*data++ = imx_mu_read(priv, priv->dcfg->xRR + (i % num_rr) * 4);
 	}
 
-	imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 	mbox_chan_received_data(cp->chan, (void *)priv->msg);
 
 	return 0;
@@ -551,6 +549,11 @@ static irqreturn_t imx_mu_isr_th(int irq, void *p)
 		mbox_chan_txdone(chan, 0);
 		break;
 
+	case IMX_MU_TYPE_RX:
+		if (!priv->dcfg->rx(priv, cp))
+			imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
+		break;
+
 	default:
 		dev_warn_ratelimited(priv->dev, "Unhandled channel type %d\n",
 				     cp->type);
@@ -603,7 +606,8 @@ static irqreturn_t imx_mu_isr(int irq, void *p)
 		ret = IRQ_WAKE_THREAD;
 	} else if ((val == IMX_MU_xSR_RFn(priv->dcfg->type, cp->idx)) &&
 		   (cp->type == IMX_MU_TYPE_RX)) {
-		priv->dcfg->rx(priv, cp);
+		imx_mu_xcr_rmw(priv, IMX_MU_RCR, 0, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
+		ret = IRQ_WAKE_THREAD;
 	} else if ((val == IMX_MU_xSR_GIPn(priv->dcfg->type, cp->idx)) &&
 		   (cp->type == IMX_MU_TYPE_RXDB)) {
 		priv->dcfg->rxdb(priv, cp);

-- 
2.53.0

[PATCH v3 08/10] mailbox: imx: Move the RXDB part of the mailbox into the threaded handler

[Sebastian Andrzej Siewior]

Move RXDB callback handling into the threaded handler. This similar to
the RX side and since the imx_mu_dcfg::rxdb callback can return an error, the
interrupt is only enabled on success.

Move RXDB callback handling into the threaded handler.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 1219c35b116e4..1bd434bdff87a 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -554,6 +554,11 @@ static irqreturn_t imx_mu_isr_th(int irq, void *p)
 			imx_mu_xcr_set_act(priv, cp, IMX_MU_RCR, IMX_MU_xCR_RIEn(priv->dcfg->type, cp->idx));
 		break;
 
+	case IMX_MU_TYPE_RXDB:
+		if (!priv->dcfg->rxdb(priv, cp))
+			imx_mu_xcr_set_act(priv, cp, IMX_MU_GIER, IMX_MU_xCR_GIEn(priv->dcfg->type, cp->idx));
+		break;
+
 	default:
 		dev_warn_ratelimited(priv->dev, "Unhandled channel type %d\n",
 				     cp->type);
@@ -610,7 +615,8 @@ static irqreturn_t imx_mu_isr(int irq, void *p)
 		ret = IRQ_WAKE_THREAD;
 	} else if ((val == IMX_MU_xSR_GIPn(priv->dcfg->type, cp->idx)) &&
 		   (cp->type == IMX_MU_TYPE_RXDB)) {
-		priv->dcfg->rxdb(priv, cp);
+		imx_mu_xcr_rmw(priv, IMX_MU_GIER, 0, IMX_MU_xCR_GIEn(priv->dcfg->type, cp->idx));
+		ret = IRQ_WAKE_THREAD;
 	} else {
 		dev_warn_ratelimited(priv->dev, "Not handled interrupt\n");
 		return IRQ_NONE;

-- 
2.53.0

[PATCH v3 09/10] mailbox: imx: Don't force-thread the primary handler

[Sebastian Andrzej Siewior]

The primary interrupt handler (imx_mu_isr()) no longer invokes any
callbacks it only masks the interrupt source and returns. In a
forced-threaded environment the IRQ-core will force-thread the primary
handler which can be avoided.

The primary handler uses a spinlock_t to protect the RMW operation in
imx_mu_xcr_rmw() - nothing that may introduce long latencies.

The lock can be turned into a raw_spinlock_t and then the primary
handler can run in hardirq context even on PREEMPT_RT skipping one
thread.

Make struct imx_mu_priv::xcr_lock a raw_spinlock_t and skip
force-threading the primrary handler by marking it IRQF_NO_THREAD.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/mailbox/imx-mailbox.c | 19 +++++++------------
 1 file changed, 7 insertions(+), 12 deletions(-)

diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
index 1bd434bdff87a..40bad2b3e4d19 100644
--- a/drivers/mailbox/imx-mailbox.c
+++ b/drivers/mailbox/imx-mailbox.c
@@ -87,7 +87,7 @@ struct imx_mu_priv {
 	struct device		*dev;
 	void __iomem		*base;
 	void			*msg;
-	spinlock_t		xcr_lock; /* control register lock */
+	raw_spinlock_t		xcr_lock; /* control register lock */
 
 	struct mbox_controller	mbox;
 	struct mbox_chan	mbox_chans[IMX_MU_CHANS];
@@ -207,15 +207,14 @@ static int imx_mu_rx_waiting_read(struct imx_mu_priv *priv, u32 *val, u32 idx)
 
 static u32 imx_mu_xcr_rmw(struct imx_mu_priv *priv, enum imx_mu_xcr type, u32 set, u32 clr)
 {
-	unsigned long flags;
 	u32 val;
 
-	spin_lock_irqsave(&priv->xcr_lock, flags);
+	guard(raw_spinlock_irqsave)(&priv->xcr_lock);
+
 	val = imx_mu_read(priv, priv->dcfg->xCR[type]);
 	val &= ~clr;
 	val |= set;
 	imx_mu_write(priv, val, priv->dcfg->xCR[type]);
-	spin_unlock_irqrestore(&priv->xcr_lock, flags);
 
 	return val;
 }
@@ -223,31 +222,27 @@ static u32 imx_mu_xcr_rmw(struct imx_mu_priv *priv, enum imx_mu_xcr type, u32 se
 static void imx_mu_xcr_clr_shut(struct imx_mu_priv *priv, struct imx_mu_con_priv *cp,
 				enum imx_mu_xcr type, u32 clr)
 {
-	unsigned long flags;
 	u32 val;
 
-	spin_lock_irqsave(&priv->xcr_lock, flags);
+	guard(raw_spinlock_irqsave)(&priv->xcr_lock);
 	cp->shutdown = true;
 
 	val = imx_mu_read(priv, priv->dcfg->xCR[type]);
 	val &= ~clr;
 	imx_mu_write(priv, val, priv->dcfg->xCR[type]);
-	spin_unlock_irqrestore(&priv->xcr_lock, flags);
 }
 
 static void imx_mu_xcr_set_act(struct imx_mu_priv *priv, struct imx_mu_con_priv *cp,
 			       enum imx_mu_xcr type, u32 set)
 {
-	unsigned long flags;
 	u32 val;
 
-	spin_lock_irqsave(&priv->xcr_lock, flags);
+	guard(raw_spinlock_irqsave)(&priv->xcr_lock);
 	if (!cp->shutdown) {
 		val = imx_mu_read(priv, priv->dcfg->xCR[type]);
 		val |= set;
 		imx_mu_write(priv, val, priv->dcfg->xCR[type]);
 	}
-	spin_unlock_irqrestore(&priv->xcr_lock, flags);
 }
 
 static int imx_mu_generic_tx(struct imx_mu_priv *priv,
@@ -640,7 +635,7 @@ static int imx_mu_startup(struct mbox_chan *chan)
 {
 	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
 	struct imx_mu_con_priv *cp = chan->con_priv;
-	unsigned long irq_flag = 0;
+	unsigned long irq_flag = IRQF_NO_THREAD;
 	int ret;
 
 	pm_runtime_get_sync(priv->dev);
@@ -988,7 +983,7 @@ static int imx_mu_probe(struct platform_device *pdev)
 		goto disable_clk;
 	}
 
-	spin_lock_init(&priv->xcr_lock);
+	raw_spin_lock_init(&priv->xcr_lock);
 
 	priv->mbox.dev = dev;
 	priv->mbox.ops = &imx_mu_ops;

-- 
2.53.0

Re: [PATCH v3 09/10] mailbox: imx: Don't force-thread the primary handler

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [High] The return value of `pm_runtime_get_sync()` is ignored in `imx_mu_startup()`, risking unclocked register accesses and PM reference leaks.
--

commit 14f99570339a0f5e4729690bf8d39e0fdeb47ac3
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

mailbox: imx: Don't force-thread the primary handler

This commit optimizes the i.MX mailbox driver by converting the
xcr_lock to a raw_spinlock_t and marking the primary interrupt handler
with IRQF_NO_THREAD. This prevents the primary handler from being
force-threaded, allowing it to run in hardirq context even on PREEMPT_RT.

> diff --git a/drivers/mailbox/imx-mailbox.c b/drivers/mailbox/imx-mailbox.c
> --- a/drivers/mailbox/imx-mailbox.c
> +++ b/drivers/mailbox/imx-mailbox.c

[ ... ]

> @@ -640,7 +635,7 @@ static int imx_mu_startup(struct mbox_chan *chan)
>  {
>  	struct imx_mu_priv *priv = to_imx_mu_priv(chan->mbox);
>  	struct imx_mu_con_priv *cp = chan->con_priv;
> -	unsigned long irq_flag = 0;
> +	unsigned long irq_flag = IRQF_NO_THREAD;
>  	int ret;
>  
>  	pm_runtime_get_sync(priv->dev);

[Severity: High]
This isn't a bug introduced by this patch, but should the return value of
pm_runtime_get_sync() be checked here?

If the device fails to resume (e.g., the PM domain fails to power on),
pm_runtime_get_sync() returns a negative error code but leaves the PM
usage counter incremented. Because the return value goes unchecked, the
driver might incorrectly assume the hardware is ready and proceed to call
hardware access functions like imx_mu_xcr_rmw() during startup.

Accessing unclocked or unpowered registers on i.MX devices can cause an
asynchronous external abort (bus fault), leading to a system crash. Could
this be updated to use pm_runtime_resume_and_get() instead, allowing the
driver to safely abort the startup on failure?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=9

[PATCH v3 10/10] remoteproc: imx_rproc: Invoke the callback directly

[Sebastian Andrzej Siewior]

The imx-mailbox driver moved the callback invocation into the threaded
IRQ handler. This means the callback is invoked in preemptible context
and there is no need to schedule the kworker for the
imx_rproc_notified_idr_cb() invocation.

This was tested with the rpmsg-tty driver on imx93.

Remove the workqueue handling and invoke the imx_rproc_notified_idr_cb()
callback directly.

Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 drivers/remoteproc/imx_rproc.c | 33 +--------------------------------
 1 file changed, 1 insertion(+), 32 deletions(-)

diff --git a/drivers/remoteproc/imx_rproc.c b/drivers/remoteproc/imx_rproc.c
index 0dd80e688b0ea..c97bc1c401655 100644
--- a/drivers/remoteproc/imx_rproc.c
+++ b/drivers/remoteproc/imx_rproc.c
@@ -24,7 +24,6 @@
 #include <linux/regmap.h>
 #include <linux/remoteproc.h>
 #include <linux/scmi_imx_protocol.h>
-#include <linux/workqueue.h>
 
 #include "imx_rproc.h"
 #include "remoteproc_internal.h"
@@ -115,8 +114,6 @@ struct imx_rproc {
 	struct mbox_client		cl;
 	struct mbox_chan		*tx_ch;
 	struct mbox_chan		*rx_ch;
-	struct work_struct		rproc_work;
-	struct workqueue_struct		*workqueue;
 	void __iomem			*rsc_table;
 	struct imx_sc_ipc		*ipc_handle;
 	struct notifier_block		rproc_nb;
@@ -835,21 +832,11 @@ static int imx_rproc_notified_idr_cb(int id, void *ptr, void *data)
 	return 0;
 }
 
-static void imx_rproc_vq_work(struct work_struct *work)
-{
-	struct imx_rproc *priv = container_of(work, struct imx_rproc,
-					      rproc_work);
-	struct rproc *rproc = priv->rproc;
-
-	idr_for_each(&rproc->notifyids, imx_rproc_notified_idr_cb, rproc);
-}
-
 static void imx_rproc_rx_callback(struct mbox_client *cl, void *msg)
 {
 	struct rproc *rproc = dev_get_drvdata(cl->dev);
-	struct imx_rproc *priv = rproc->priv;
 
-	queue_work(priv->workqueue, &priv->rproc_work);
+	idr_for_each(&rproc->notifyids, imx_rproc_notified_idr_cb, rproc);
 }
 
 static int imx_rproc_xtr_mbox_init(struct rproc *rproc, bool tx_block)
@@ -1214,13 +1201,6 @@ static int imx_rproc_sys_off_handler(struct sys_off_data *data)
 	return NOTIFY_DONE;
 }
 
-static void imx_rproc_destroy_workqueue(void *data)
-{
-	struct workqueue_struct *workqueue = data;
-
-	destroy_workqueue(workqueue);
-}
-
 static int imx_rproc_probe(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
@@ -1249,17 +1229,6 @@ static int imx_rproc_probe(struct platform_device *pdev)
 		priv->ops = dcfg->ops;
 
 	dev_set_drvdata(dev, rproc);
-	priv->workqueue = create_workqueue(dev_name(dev));
-	if (!priv->workqueue) {
-		dev_err(dev, "cannot create workqueue\n");
-		return -ENOMEM;
-	}
-
-	ret = devm_add_action_or_reset(dev, imx_rproc_destroy_workqueue, priv->workqueue);
-	if (ret)
-		return dev_err_probe(dev, ret, "Failed to add devm destroy workqueue action\n");
-
-	INIT_WORK(&priv->rproc_work, imx_rproc_vq_work);
 
 	ret = imx_rproc_xtr_mbox_init(rproc, true);
 	if (ret)

-- 
2.53.0

Re: [PATCH v3 10/10] remoteproc: imx_rproc: Invoke the callback directly

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

Pre-existing issues:
- [High] Unsafe lockless iteration of `rproc->notifyids` in preemptible context risks Use-After-Free of IDR nodes and `rvring` structures during concurrent device teardown.
- [High] If requesting the 'rx' mailbox channel fails, `priv->tx_ch` is freed but its pointer is not cleared, causing a subsequent start to mistakenly treat initialization as successful.
--

commit 6f600102625eeb04d37188290cefc55632750bc2
Author: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

remoteproc: imx_rproc: Invoke the callback directly

The imx-mailbox driver moved the callback invocation into the threaded
IRQ handler. This means the callback is invoked in preemptible context,
so there is no need to schedule a kworker. This patch removes the workqueue
handling and invokes the callback directly.

> diff --git a/drivers/remoteproc/imx_rproc.c b/drivers/remoteproc/imx_rproc.c
> index 0dd80e688b0ea..c97bc1c401655 100644
> --- a/drivers/remoteproc/imx_rproc.c
> +++ b/drivers/remoteproc/imx_rproc.c

[ ... ]

> @@ -835,21 +832,11 @@ static int imx_rproc_notified_idr_cb(int id, void *ptr, void *data)
>  	return 0;
>  }
>  
> -static void imx_rproc_vq_work(struct work_struct *work)
> -{
> -	struct imx_rproc *priv = container_of(work, struct imx_rproc,
> -					      rproc_work);
> -	struct rproc *rproc = priv->rproc;
> -
> -	idr_for_each(&rproc->notifyids, imx_rproc_notified_idr_cb, rproc);
> -}
> -
>  static void imx_rproc_rx_callback(struct mbox_client *cl, void *msg)
>  {
>  	struct rproc *rproc = dev_get_drvdata(cl->dev);
> -	struct imx_rproc *priv = rproc->priv;
>  
> -	queue_work(priv->workqueue, &priv->rproc_work);
> +	idr_for_each(&rproc->notifyids, imx_rproc_notified_idr_cb, rproc);

[Severity: High]
This is a pre-existing issue, but since the iteration now runs directly
in the preemptible threaded IRQ handler, does the lack of RCU protection
risk a use-after-free?

During rproc_shutdown(), imx_rproc_stop() frees the mailbox channel, but
without disabling the IRQ, the callback can run concurrently with
rproc_resource_cleanup().

If the threaded IRQ handler is preempted here, could the RCU grace period
elapse while rproc_resource_cleanup() unregisters the device and removes
entries from the notifyids IDR?

rproc_vq_interrupt() also performs a lockless idr_find() lookup on the same
notifyids IDR:

drivers/remoteproc/remoteproc_virtio.c:rproc_vq_interrupt() {
    ...
    rvring = idr_find(&rproc->notifyids, notifyid);
    ...
}

Would wrapping the idr_for_each() call with rcu_read_lock() and
rcu_read_unlock() be necessary to safely iterate the IDR nodes and prevent
use-after-free of the rvring structures?

>  }
>  
>  static int imx_rproc_xtr_mbox_init(struct rproc *rproc, bool tx_block)

[Severity: High]
This is a pre-existing issue, but I noticed a potential problem in the error
handling within imx_rproc_xtr_mbox_init().

If requesting the rx mailbox channel fails, the code frees the tx channel but
does not clear the pointer:

drivers/remoteproc/imx_rproc.c:imx_rproc_xtr_mbox_init() {
    ...
    priv->rx_ch = mbox_request_channel_byname(cl, "rx");
    if (IS_ERR(priv->rx_ch)) {
        mbox_free_channel(priv->tx_ch);
        return dev_err_probe(...);
    }
    ...
}

Because priv->tx_ch is not set to NULL, and priv->rx_ch contains an ERR_PTR,
the early exit check at the top of the function evaluates to true on any
subsequent call:

    if (priv->tx_ch && priv->rx_ch)
        return 0;

If a user retries starting the remoteproc via sysfs, this will incorrectly
return 0 (success) with a dangling priv->tx_ch pointer. When the remoteproc
later sends a kick, imx_rproc_kick() will use this dangling pointer, which
can cause a use-after-free crash.

Could we set priv->tx_ch to NULL after freeing it to prevent this sequence?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260617-imx_mbox_rproc-v3-0-77948112defc@linutronix.de?part=10