Re: [RFC v1 1/4] dma-mapping: introduce the DMA_ATTR_MAY_SLEEP attribute

[Petr Tesarik <petrtesarik@huaweicloud.com>]

On 3/28/2023 5:57 AM, Christoph Hellwig wrote:
> On Mon, Mar 20, 2023 at 01:28:13PM +0100, Petr Tesarik wrote:
>> From: Petr Tesarik <petr.tesarik.ext@huawei.com>
>>
>> Introduce a DMA attribute to tell the DMA-mapping subsystem that
>> the operation is allowed to sleep.
>>
>> This patch merely adds the flag, which is not used for anything at
>> the moment. It should be used by users who can sleep (e.g. dma-buf
>> ioctls) to allow page reclaim and/or allocations from CMA.
> 
> So what drivers would call this?  As-is it doesn't have any users in
> the series.

Yes, I removed one patch from the RFC series to reduce the Cc list while
I wasn't sure if the proposal would be considered at all.

The full series in my local tree added it to the implementation of
DRM_IOCTL_PRIME_FD_TO_HANDLE:

diff --git a/drivers/gpu/drm/drm_prime.c b/drivers/gpu/drm/drm_prime.c
index f924b8b4ab6b..f32e12445570 100644
--- a/drivers/gpu/drm/drm_prime.c
+++ b/drivers/gpu/drm/drm_prime.c
@@ -635,7 +635,7 @@ struct sg_table *drm_gem_map_dma_buf(struct
dma_buf_attachment *attach,
 		return sgt;

 	ret = dma_map_sgtable(attach->dev, sgt, dir,
-			      DMA_ATTR_SKIP_CPU_SYNC);
+			      DMA_ATTR_SKIP_CPU_SYNC | DMA_ATTR_MAY_SLEEP);
 	if (ret) {
 		sg_free_table(sgt);
 		kfree(sgt);

I also noticed a similar place in udmabuf, but since I don't have a use
case ATM, I haven't added the flag there (yet).

Petr T