[PATCH] drm/panthor: flush FW AS caches in slow reset path

[Liviu Dudau]

On Fri, Aug 16, 2024 at 07:52:49PM +0100, Adrián Larumbe wrote:
> In the off-chance that waiting for the firmware to signal its booted status
> timed out in the fast reset path, one must flush the cache lines for the
> entire FW VM address space before reloading the regions, otherwise stale
> values eventually lead to a scheduler job timeout.
> 
> Signed-off-by: Adrián Larumbe <adrian.larumbe at collabora.com>

Looks good to me!

Acked-by: Liviu Dudau <liviu.dudau at arm.com>

Best regards,
Liviu

> ---
>  drivers/gpu/drm/panthor/panthor_fw.c  |  8 +++++++-
>  drivers/gpu/drm/panthor/panthor_mmu.c | 19 ++++++++++++++++---
>  drivers/gpu/drm/panthor/panthor_mmu.h |  1 +
>  3 files changed, 24 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/gpu/drm/panthor/panthor_fw.c b/drivers/gpu/drm/panthor/panthor_fw.c
> index 857f3f11258a..ef232c0c2049 100644
> --- a/drivers/gpu/drm/panthor/panthor_fw.c
> +++ b/drivers/gpu/drm/panthor/panthor_fw.c
> @@ -1089,6 +1089,12 @@ int panthor_fw_post_reset(struct panthor_device *ptdev)
>  		panthor_fw_stop(ptdev);
>  		ptdev->fw->fast_reset = false;
>  		drm_err(&ptdev->base, "FW fast reset failed, trying a slow reset");
> +
> +		ret = panthor_vm_flush_all(ptdev->fw->vm);
> +		if (ret) {
> +			drm_err(&ptdev->base, "FW slow reset failed (couldn't flush FW's AS l2cache)");
> +			return ret;
> +		}
>  	}
>  
>  	/* Reload all sections, including RO ones. We're not supposed
> @@ -1099,7 +1105,7 @@ int panthor_fw_post_reset(struct panthor_device *ptdev)
>  
>  	ret = panthor_fw_start(ptdev);
>  	if (ret) {
> -		drm_err(&ptdev->base, "FW slow reset failed");
> +		drm_err(&ptdev->base, "FW slow reset failed (couldn't start the FW )");
>  		return ret;
>  	}
>  
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.c b/drivers/gpu/drm/panthor/panthor_mmu.c
> index d47972806d50..a77ee5ce691d 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.c
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.c
> @@ -874,14 +874,27 @@ static int panthor_vm_flush_range(struct panthor_vm *vm, u64 iova, u64 size)
>  	if (!drm_dev_enter(&ptdev->base, &cookie))
>  		return 0;
>  
> -	/* Flush the PTs only if we're already awake */
> -	if (pm_runtime_active(ptdev->base.dev))
> -		ret = mmu_hw_do_operation(vm, iova, size, AS_COMMAND_FLUSH_PT);
> +	/*
> +	 * If we made it this far, that means the device is awake, because
> +	 * upon device suspension, all active VMs are given an AS id of -1
> +	 */
> +	ret = mmu_hw_do_operation(vm, iova, size, AS_COMMAND_FLUSH_PT);
>  
>  	drm_dev_exit(cookie);
>  	return ret;
>  }
>  
> +/**
> + * panthor_vm_flush_all() - Flush L2 caches for the entirety of a VM's AS
> + * @vm: VM whose cache to flush
> + *
> + * Return: 0 on success, a negative error code if flush failed.
> + */
> +int panthor_vm_flush_all(struct panthor_vm *vm)
> +{
> +	return panthor_vm_flush_range(vm, vm->base.mm_start, vm->base.mm_range);
> +}
> +
>  static int panthor_vm_unmap_pages(struct panthor_vm *vm, u64 iova, u64 size)
>  {
>  	struct panthor_device *ptdev = vm->ptdev;
> diff --git a/drivers/gpu/drm/panthor/panthor_mmu.h b/drivers/gpu/drm/panthor/panthor_mmu.h
> index f3c1ed19f973..6788771071e3 100644
> --- a/drivers/gpu/drm/panthor/panthor_mmu.h
> +++ b/drivers/gpu/drm/panthor/panthor_mmu.h
> @@ -31,6 +31,7 @@ panthor_vm_get_bo_for_va(struct panthor_vm *vm, u64 va, u64 *bo_offset);
>  int panthor_vm_active(struct panthor_vm *vm);
>  void panthor_vm_idle(struct panthor_vm *vm);
>  int panthor_vm_as(struct panthor_vm *vm);
> +int panthor_vm_flush_all(struct panthor_vm *vm);
>  
>  struct panthor_heap_pool *
>  panthor_vm_get_heap_pool(struct panthor_vm *vm, bool create);
> -- 
> 2.46.0
> 

-- 
====================
| I would like to |
| fix the world,  |
| but they're not |
| giving me the   |
 \ source code!  /
  ---------------
    ¯\_(ツ)_/¯