Re: [PATCH v13 7/9] drm/xe/amc: Guard xe_amc_work() with runtime PM

["Tauro, Riana" <riana.tauro@intel.com>]

On 20-10-2026 08:06, Raag Jadav wrote:
> xe_device_declare_wedged() callers are expected to hold a runtime PM
> reference. Fix it.
>
> Fixes: a55b76b8bc2c ("drm/xe/i2c: Fix the interrupt handling")
> Signed-off-by: Raag Jadav <raag.jadav@intel.com>
> ---
> v10: Guard wedge callers with runtime PM (Sashiko)
> v11: Use traditional xe_pm_runtime_get/put() with goto (Sashiko)
> ---
>   drivers/gpu/drm/xe/xe_amc.c | 7 +++++++
>   1 file changed, 7 insertions(+)
>
> diff --git a/drivers/gpu/drm/xe/xe_amc.c b/drivers/gpu/drm/xe/xe_amc.c
> index b7390a4ee3ce..1f85df598e56 100644
> --- a/drivers/gpu/drm/xe/xe_amc.c
> +++ b/drivers/gpu/drm/xe/xe_amc.c
> @@ -15,9 +15,11 @@
>   #include "regs/xe_i2c_regs.h"
>   
>   #include "xe_amc.h"
> +#include "xe_any.h"
>   #include "xe_device.h"
>   #include "xe_i2c.h"
>   #include "xe_mmio.h"
> +#include "xe_pm.h"
>   
>   /**
>    * DOC: Add-In Management Controller (AMC)
> @@ -108,11 +110,14 @@ static void xe_amc_work(struct work_struct *work)
>   {
>   	const struct amc_request *request = &amc_get_alert_reason;
>   	struct xe_amc *amc = from_work(amc, work, work);
> +	struct xe_device *xe = xe_any_to_xe(amc->i2c->drm_dev);
>   	u8 alert_reason = AMC_ALERT_UNKNOWN;
>   	struct amc_response response;
>   	struct i2c_client *client;
>   	int ret;
>   
> +	xe_pm_runtime_get(xe);

Even though Sashiko did complain about guard.
Wouldn't usage of guard maintain consistency and be better here?

With or without guard
Reviewed-by: Riana Tauro <riana.tauro@intel.com>


> +
>   	client = amc->i2c->client[XE_I2C_CLIENT_AMC];
>   	if (IS_ERR_OR_NULL(client))
>   		goto out_reassert_interrupt;
> @@ -167,6 +172,8 @@ static void xe_amc_work(struct work_struct *work)
>   		dev_warn(amc->i2c->drm_dev, "unknown AMC alert: %d\n", alert_reason);
>   		break;
>   	}
> +
> +	xe_pm_runtime_put(xe);
>   }
>   
>   void xe_amc_handle_alert(struct xe_i2c *i2c)