[Intel-xe] [PATCH 1/2] Revert "drm/xe: Enable d3cold support for dgfx"

[Matthew Auld]

On 26/04/2023 14:12, Rodrigo Vivi wrote:
> This reverts commit 7c1c3d3898c60f3e96f40737f0d8ddd4b8934aaf.
> 
> Unfortunately we are not ready for this step yet.
> 
> We need to rethink our strategy on D3cold:
> 
> 1. rewrite the VRAM save / restore to avoid buffer object locks
> 2. block D3cold if we have a big amount of device memory in use in order
>     to reduce the latency.
> 3. at resume, detect if we really lost power and avoid memory restoration
>     if we were only up to d3cold
> 
> Reference: https://gitlab.freedesktop.org/drm/xe/kernel/-/issues/256
> Cc: Matthew Auld <matthew.auld at intel.com>
> Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
> Cc: Riana Tauro <riana.tauro at intel.com>
> Signed-off-by: Rodrigo Vivi <rodrigo.vivi at intel.com>

fwiw for the series,
Reviewed-by: Matthew Auld <matthew.auld at intel.com>

> ---
>   drivers/gpu/drm/xe/xe_pci.c | 9 +++++++--
>   1 file changed, 7 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c
> index 14be0259a222..9324aed8a821 100644
> --- a/drivers/gpu/drm/xe/xe_pci.c
> +++ b/drivers/gpu/drm/xe/xe_pci.c
> @@ -762,8 +762,13 @@ static int xe_pci_runtime_idle(struct device *dev)
>   	struct pci_dev *pdev = to_pci_dev(dev);
>   	struct xe_device *xe = pdev_to_xe_device(pdev);
>   
> -	if (IS_DGFX(xe) && !xe_device_mem_access_ongoing(xe))
> -		xe->d3cold_allowed = true;
> +	/*
> +	 * FIXME: d3cold should be allowed (true) if
> +	 * (IS_DGFX(xe) && !xe_device_mem_access_ongoing(xe))
> +	 * however the change to the buddy allocator broke the
> +	 * xe_bo_restore_kernel when the pci device is disabled
> +	 */
> +	 xe->d3cold_allowed = false;
>   
>   	return 0;
>   }