2020-10-30 04:14:18

by Bhaumik Bhatt

Subject: [PATCH v3 12/12] bus: mhi: core: Remove MHI event ring IRQ handlers when powering down

While powering down, the device may or may not acknowledge an MHI
RESET issued by host for a graceful shutdown scenario and end up
sending an incoming data packet after tasklets have been killed.
If a rogue device sends this interrupt for a data transfer event
ring update, it can result in a tasklet getting scheduled while a
clean up is ongoing or has completed and cause access to freed
memory leading to a NULL pointer exception. Remove the interrupt
handlers for MHI event rings early on to avoid this scenario.

Signed-off-by: Bhaumik Bhatt <[email protected]>
---
drivers/bus/mhi/core/pm.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/bus/mhi/core/pm.c b/drivers/bus/mhi/core/pm.c
index ffbf6f5..a671f58 100644
--- a/drivers/bus/mhi/core/pm.c
+++ b/drivers/bus/mhi/core/pm.c
@@ -494,6 +494,7 @@ static void mhi_pm_disable_transition(struct mhi_controller *mhi_cntrl)
for (i = 0; i < mhi_cntrl->total_ev_rings; i++, mhi_event++) {
if (mhi_event->offload_ev)
continue;
+ free_irq(mhi_cntrl->irq[mhi_event->irq], mhi_event);
tasklet_kill(&mhi_event->task);
}

@@ -1164,7 +1165,7 @@ void mhi_power_down(struct mhi_controller *mhi_cntrl, bool graceful)
/* Wait for shutdown to complete */
flush_work(&mhi_cntrl->st_worker);

- mhi_deinit_free_irq(mhi_cntrl);
+ free_irq(mhi_cntrl->irq[0], mhi_cntrl);

if (!mhi_cntrl->pre_init) {
/* Free all allocated resources */
--
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

2020-10-30 14:14:14

by Manivannan Sadhasivam

Subject: Re: [PATCH v3 12/12] bus: mhi: core: Remove MHI event ring IRQ handlers when powering down

On Thu, Oct 29, 2020 at 09:10:57PM -0700, Bhaumik Bhatt wrote:
> While powering down, the device may or may not acknowledge an MHI
> RESET issued by host for a graceful shutdown scenario and end up
> sending an incoming data packet after tasklets have been killed.
> If a rogue device sends this interrupt for a data transfer event
> ring update, it can result in a tasklet getting scheduled while a
> clean up is ongoing or has completed and cause access to freed
> memory leading to a NULL pointer exception. Remove the interrupt
> handlers for MHI event rings early on to avoid this scenario.
>
> Signed-off-by: Bhaumik Bhatt <[email protected]>

Reviewed-by: Manivannan Sadhasivam <[email protected]>

Thanks,
Mani

> ---
> drivers/bus/mhi/core/pm.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/bus/mhi/core/pm.c b/drivers/bus/mhi/core/pm.c
> index ffbf6f5..a671f58 100644
> --- a/drivers/bus/mhi/core/pm.c
> +++ b/drivers/bus/mhi/core/pm.c
> @@ -494,6 +494,7 @@ static void mhi_pm_disable_transition(struct mhi_controller *mhi_cntrl)
> for (i = 0; i < mhi_cntrl->total_ev_rings; i++, mhi_event++) {
> if (mhi_event->offload_ev)
> continue;
> + free_irq(mhi_cntrl->irq[mhi_event->irq], mhi_event);
> tasklet_kill(&mhi_event->task);
> }
>
> @@ -1164,7 +1165,7 @@ void mhi_power_down(struct mhi_controller *mhi_cntrl, bool graceful)
> /* Wait for shutdown to complete */
> flush_work(&mhi_cntrl->st_worker);
>
> - mhi_deinit_free_irq(mhi_cntrl);
> + free_irq(mhi_cntrl->irq[0], mhi_cntrl);
>
> if (!mhi_cntrl->pre_init) {
> /* Free all allocated resources */
> --
> The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> a Linux Foundation Collaborative Project
>