[PATCH 0/2] bus: mhi: Address MHI system suspend abort

[PATCH 0/2] bus: mhi: Address MHI system suspend abort

[Len Brown]

For Upstream:

[PATCH 1/2] bus: mhi: pci_generic: Prevent MHI from aborting system suspend

For Debug of underlying MHI issue:

[RFC 2/2] bus: mhi: host: Add diagnostic print for -EBUSY in mhi_pm_suspend()

thanks,
-Len

[PATCH v2 1/2] bus: mhi: pci_generic: Prevent MHI from aborting system suspend

[Len Brown]

From: Len Brown <len.brown@intel.com>

mhi sometimes aborts system suspend on the thinkpad T16 gen3:

mhi-pci-generic 0000:08:00.0:
... failed to suspend device: -16
... PM: pci_pm_suspend(): mhi_pci_suspend [mhi_pci_generic] returns -16
... PM: dpm_run_callback(): pci_pm_suspend returns -16
... PM: pci_pm_suspend returned -16 after 15145 usecs
... PM: failed to suspend async: error -16

Prevent MHI from spreading this device-specific failure
to the rest of the system.  I.e. "Why is a modem that I don't use
preventing me from suspending my laptop?"

This change does not comprehend the internal root cause of the
mhi failure, which should be found and fixed.

There is also a risk that ignoring the MHI error and continuing
to put the device into PCI D3 will leave it in an error state.

Fixes: d3800c1dce24 ("bus: mhi: pci_generic: Add support for runtime PM")
Signed-off-by: Len Brown <len.brown@intel.com>
---
 drivers/bus/mhi/host/pci_generic.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/bus/mhi/host/pci_generic.c b/drivers/bus/mhi/host/pci_generic.c
index 7193a384cca4..343d1046a25e 100644
--- a/drivers/bus/mhi/host/pci_generic.c
+++ b/drivers/bus/mhi/host/pci_generic.c
@@ -1727,7 +1727,12 @@ static int __maybe_unused mhi_pci_runtime_resume(struct device *dev)
 static int  __maybe_unused mhi_pci_suspend(struct device *dev)
 {
 	pm_runtime_disable(dev);
-	return mhi_pci_runtime_suspend(dev);
+
+	/* Ignore errors (e.g. -EBUSY); a failed modem suspend must not abort system suspend */
+	mhi_pci_runtime_suspend(dev);
+
+	/* Must return 0 to allow system suspend to continue */
+	return 0;
 }
 
 static int __maybe_unused mhi_pci_resume(struct device *dev)
-- 
2.45.2

[RFC 2/2] bus: mhi: host: Add diagnostic print for -EBUSY in mhi_pm_suspend()

[Len Brown]

From: Len Brown <len.brown@intel.com>

mhi_pm_suspend() sometimes returns -EBUSY, but doesn't tell us why.

Print dev_wake and pending_pkts when it does.

Signed-off-by: Len Brown <len.brown@intel.com>
---
 drivers/bus/mhi/host/pm.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/drivers/bus/mhi/host/pm.c b/drivers/bus/mhi/host/pm.c
index ced83fbb1a51..2eb7344b3b39 100644
--- a/drivers/bus/mhi/host/pm.c
+++ b/drivers/bus/mhi/host/pm.c
@@ -871,8 +871,12 @@ int mhi_pm_suspend(struct mhi_controller *mhi_cntrl)
 
 	/* Return busy if there are any pending resources */
 	if (atomic_read(&mhi_cntrl->dev_wake) ||
-	    atomic_read(&mhi_cntrl->pending_pkts))
+	    atomic_read(&mhi_cntrl->pending_pkts)) {
+		dev_err(dev, "DEBUG: suspend busy (early): dev_wake: %d pending_pkts: %d\n",
+			atomic_read(&mhi_cntrl->dev_wake),
+			atomic_read(&mhi_cntrl->pending_pkts));
 		return -EBUSY;
+	}
 
 	/* Take MHI out of M2 state */
 	read_lock_bh(&mhi_cntrl->pm_lock);
@@ -900,6 +904,9 @@ int mhi_pm_suspend(struct mhi_controller *mhi_cntrl)
 	if (atomic_read(&mhi_cntrl->dev_wake) ||
 	    atomic_read(&mhi_cntrl->pending_pkts)) {
 		write_unlock_irq(&mhi_cntrl->pm_lock);
+		dev_err(dev, "DEBUG: suspend busy (recheck): dev_wake: %d pending_pkts: %d\n",
+			atomic_read(&mhi_cntrl->dev_wake),
+			atomic_read(&mhi_cntrl->pending_pkts));
 		return -EBUSY;
 	}
 
-- 
2.45.2

Re: [PATCH v2 1/2] bus: mhi: pci_generic: Prevent MHI from aborting system suspend

[Loic Poulain]

On Fri, Nov 13, 2026 at 4:19 PM Len Brown <lenb@kernel.org> wrote:
>
> From: Len Brown <len.brown@intel.com>
>
> mhi sometimes aborts system suspend on the thinkpad T16 gen3:
>
> mhi-pci-generic 0000:08:00.0:
> ... failed to suspend device: -16
> ... PM: pci_pm_suspend(): mhi_pci_suspend [mhi_pci_generic] returns -16
> ... PM: dpm_run_callback(): pci_pm_suspend returns -16
> ... PM: pci_pm_suspend returned -16 after 15145 usecs
> ... PM: failed to suspend async: error -16
>
> Prevent MHI from spreading this device-specific failure
> to the rest of the system.  I.e. "Why is a modem that I don't use
> preventing me from suspending my laptop?"
>
> This change does not comprehend the internal root cause of the
> mhi failure, which should be found and fixed.
>
> There is also a risk that ignoring the MHI error and continuing
> to put the device into PCI D3 will leave it in an error state.
>
> Fixes: d3800c1dce24 ("bus: mhi: pci_generic: Add support for runtime PM")
> Signed-off-by: Len Brown <len.brown@intel.com>

Reviewed-by: Loic Poulain <loic.poulain@oss.qualcomm.com>

> ---
>  drivers/bus/mhi/host/pci_generic.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/bus/mhi/host/pci_generic.c b/drivers/bus/mhi/host/pci_generic.c
> index 7193a384cca4..343d1046a25e 100644
> --- a/drivers/bus/mhi/host/pci_generic.c
> +++ b/drivers/bus/mhi/host/pci_generic.c
> @@ -1727,7 +1727,12 @@ static int __maybe_unused mhi_pci_runtime_resume(struct device *dev)
>  static int  __maybe_unused mhi_pci_suspend(struct device *dev)
>  {
>         pm_runtime_disable(dev);
> -       return mhi_pci_runtime_suspend(dev);
> +
> +       /* Ignore errors (e.g. -EBUSY); a failed modem suspend must not abort system suspend */
> +       mhi_pci_runtime_suspend(dev);
> +
> +       /* Must return 0 to allow system suspend to continue */
> +       return 0;
>  }
>
>  static int __maybe_unused mhi_pci_resume(struct device *dev)
> --
> 2.45.2
>