[Intel-xe] [PATCH v2 5/5] drm/xe/pm: Init pcode and restore vram on power lost

[Nilawar, Badal]

On 07-06-2023 16:33, Anshuman Gupta wrote:
> Don't init pcode and restore VRAM objects in vain.
> We can rely on primary GT GUC_STATUS to detect whether
> card has really lost power even when d3cold is allowed by xe.
> Adding d3cold.lost_power flag to avoid pcode init and vram
> restoration.
> Also cleaning up the TODO code comment.
> 
> Cc: Rodrigo Vivi <rodrigo.vivi at intel.com>
> Signed-off-by: Anshuman Gupta <anshuman.gupta at intel.com>
> ---
>   drivers/gpu/drm/xe/xe_device_types.h |  3 +++
>   drivers/gpu/drm/xe/xe_guc.c          | 17 +++++++++++++++++
>   drivers/gpu/drm/xe/xe_guc.h          |  1 +
>   drivers/gpu/drm/xe/xe_pci.c          |  2 --
>   drivers/gpu/drm/xe/xe_pm.c           | 15 ++++++++++++---
>   5 files changed, 33 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_device_types.h b/drivers/gpu/drm/xe/xe_device_types.h
> index bdd093d30788..34bf2ea3b3f9 100644
> --- a/drivers/gpu/drm/xe/xe_device_types.h
> +++ b/drivers/gpu/drm/xe/xe_device_types.h
> @@ -382,6 +382,9 @@ struct xe_device {
>   		/** @allowed: Indicates if d3cold is a valid device state */
>   		bool allowed;
>   
> +		/** @power_lost Indicates if card has really lost power. */
> +		bool power_lost;
> +
>   		/**
>   		 * @vram_threshold is the permissible threshold(in megabytes)
>   		 * for vram save/restore. d3cold will be disallowed,
> diff --git a/drivers/gpu/drm/xe/xe_guc.c b/drivers/gpu/drm/xe/xe_guc.c
> index e51d8fb4a354..706b3bc0452e 100644
> --- a/drivers/gpu/drm/xe/xe_guc.c
> +++ b/drivers/gpu/drm/xe/xe_guc.c
> @@ -848,3 +848,20 @@ void xe_guc_print_info(struct xe_guc *guc, struct drm_printer *p)
>   	xe_guc_ct_print(&guc->ct, p, false);
>   	xe_guc_submit_print(guc, p);
>   }
> +
> +/* Srtictly to called from RPM resume handler */
> +bool xe_guc_has_lost_power(struct xe_guc *guc)
> +{
> +	struct xe_gt *gt = guc_to_gt(guc);
> +	u32 status;
> +	int err;
> +
> +	err = xe_force_wake_get(gt_to_fw(gt), XE_FW_GT);
> +	if (err)
> +		return false;
> +
> +	status = xe_mmio_read32(gt, GUC_STATUS);
> +	xe_force_wake_put(gt_to_fw(gt), XE_FW_GT);
Should error check be added here as well?
Regards,
Badal
> +
> +	return  status & GS_MIA_IN_RESET;
> +}
> diff --git a/drivers/gpu/drm/xe/xe_guc.h b/drivers/gpu/drm/xe/xe_guc.h
> index 74a74051f354..3e628bbbe69e 100644
> --- a/drivers/gpu/drm/xe/xe_guc.h
> +++ b/drivers/gpu/drm/xe/xe_guc.h
> @@ -35,6 +35,7 @@ void xe_guc_reset_wait(struct xe_guc *guc);
>   void xe_guc_stop_prepare(struct xe_guc *guc);
>   int xe_guc_stop(struct xe_guc *guc);
>   int xe_guc_start(struct xe_guc *guc);
> +bool xe_guc_has_lost_power(struct xe_guc *guc);
>   
>   static inline u16 xe_engine_class_to_guc_class(enum xe_engine_class class)
>   {
> diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
> index 6cf66ce25039..b5cd3b69c98e 100644
> --- a/drivers/gpu/drm/xe/xe_pci.c
> +++ b/drivers/gpu/drm/xe/xe_pci.c
> @@ -829,8 +829,6 @@ static int xe_pci_runtime_idle(struct device *dev)
>   		 * but maybe include some other conditions. So, before
>   		 * we can re-enable the D3cold, we need to:
>   		 * 1. rewrite the VRAM save / restore to avoid buffer object locks
> -		 * 2. at resume, detect if we really lost power and avoid memory
> -		 *    restoration if we were only up to d3cold
>   		 */
>   		xe->d3cold.allowed = false;
>   	}
> diff --git a/drivers/gpu/drm/xe/xe_pm.c b/drivers/gpu/drm/xe/xe_pm.c
> index 9872a800b0b3..e37ec8aca609 100644
> --- a/drivers/gpu/drm/xe/xe_pm.c
> +++ b/drivers/gpu/drm/xe/xe_pm.c
> @@ -16,6 +16,7 @@
>   #include "xe_display.h"
>   #include "xe_ggtt.h"
>   #include "xe_gt.h"
> +#include "xe_guc.h"
>   #include "xe_irq.h"
>   #include "xe_pcode.h"
>   
> @@ -192,11 +193,19 @@ int xe_pm_runtime_suspend(struct xe_device *xe)
>   
>   int xe_pm_runtime_resume(struct xe_device *xe)
>   {
> -	struct xe_gt *gt;
> +	struct xe_gt *gt, *gt0;
>   	u8 id;
>   	int err;
>   
> -	if (xe->d3cold.allowed) {
> +	/*
> +	 * It can be possible that xe has allowed d3cold but other pcie devices
> +	 * in gfx card soc would have blocked d3cold, therefore card has not
> +	 * really lost power. Detecting primary Gt power is sufficient.
> +	 */
> +	gt0 = xe_device_get_gt(xe, 0);
> +	xe->d3cold.power_lost = xe_guc_has_lost_power(&gt0->uc.guc);
> +
> +	if (xe->d3cold.allowed && xe->d3cold.power_lost) {
>   		for_each_gt(gt, xe, id) {
>   			err = xe_pcode_init(gt);
>   			if (err)
> @@ -217,7 +226,7 @@ int xe_pm_runtime_resume(struct xe_device *xe)
>   	for_each_gt(gt, xe, id)
>   		xe_gt_resume(gt);
>   
> -	if (xe->d3cold.allowed) {
> +	if (xe->d3cold.allowed && xe->d3cold.power_lost) {
>   		err = xe_bo_restore_user(xe);
>   		if (err)
>   			return err;