Re: [RFC PATCH 3/5] drm/ttm: Use drm_memcpy_from_wc for TTM bo moves

[Thomas Hellström <thomas.hellstrom@xxxxxxxxxxxxxxx>]

On 5/21/21 10:10 AM, Christian König wrote:
> Am 20.05.21 um 17:09 schrieb Thomas Hellström:
> > Use fast wc memcpy for reading out of wc memory for TTM bo moves.
> >
> > Cc: Dave Airlie <airlied@xxxxxxxxx>
> > Cc: Christian König <christian.koenig@xxxxxxx>
> > Cc: Daniel Vetter <daniel.vetter@xxxxxxxx>
> > Signed-off-by: Thomas Hellström <thomas.hellstrom@xxxxxxxxxxxxxxx>
>
> Oh, yes I really wanted to have that in TTM for quite some time.
We should use it for swap copy from WC as well IMO. A todo-task for 
somebody.
> But I'm wondering if we shouldn't fix the memremap stuff first.

Using memremap all over is a fairly big change probably with lots of 
opinions involved all over the place.
What I can do for now is to add a dma_buf_map interface to the memcpy 
itself, to move the aliasing out of TTM to the arch specific code that 
knows what it's doing?

/Thomas


> Christian.
>
> > ---
> >   drivers/gpu/drm/ttm/ttm_bo_util.c | 18 +++++++++++++++++-
> >   1 file changed, 17 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/gpu/drm/ttm/ttm_bo_util.c 
> > b/drivers/gpu/drm/ttm/ttm_bo_util.c
> > index bad9b16e96ba..919ee03f7eb3 100644
> > --- a/drivers/gpu/drm/ttm/ttm_bo_util.c
> > +++ b/drivers/gpu/drm/ttm/ttm_bo_util.c
> > @@ -31,6 +31,7 @@
> >     #include <drm/ttm/ttm_bo_driver.h>
> >   #include <drm/ttm/ttm_placement.h>
> > +#include <drm/drm_memcpy.h>
> >   #include <drm/drm_vma_manager.h>
> >   #include <linux/dma-buf-map.h>
> >   #include <linux/io.h>
> > @@ -185,6 +186,7 @@ void ttm_move_memcpy(struct ttm_buffer_object *bo,
> >       struct ttm_resource *old_mem = &bo->mem;
> >       struct ttm_resource_manager *old_man = ttm_manager_type(bdev, 
> > old_mem->mem_type);
> >       struct dma_buf_map old_map, new_map;
> > +    bool wc_memcpy;
> >       pgoff_t i;
> >         /* Single TTM move. NOP */
> > @@ -208,11 +210,25 @@ void ttm_move_memcpy(struct ttm_buffer_object *bo,
> >           return;
> >       }
> >   +    wc_memcpy = ((!old_man->use_tt || bo->ttm->caching != 
> > ttm_cached) &&
> > +             drm_has_memcpy_from_wc());
> > +
> > +    /*
> > +     * We use some nasty aliasing for drm_memcpy_from_wc, but assuming
> > +     * that we can move to memremapping in the not too distant future,
> > +     * reduce the fragility for now with a build assert.
> > +     */
> > +    BUILD_BUG_ON(offsetof(typeof(old_map), vaddr) !=
> > +             offsetof(typeof(old_map), vaddr_iomem));
> > +
> >       for (i = 0; i < new_mem->num_pages; ++i) {
> >           new_iter->ops->kmap_local(new_iter, &new_map, i);
> >           old_iter->ops->kmap_local(old_iter, &old_map, i);
> >   -        if (!old_map.is_iomem && !new_map.is_iomem) {
> > +        if (wc_memcpy) {
> > +            drm_memcpy_from_wc(new_map.vaddr, old_map.vaddr,
> > +                       PAGE_SIZE);
> > +        } else if (!old_map.is_iomem && !new_map.is_iomem) {
> >               memcpy(new_map.vaddr, old_map.vaddr, PAGE_SIZE);
> >           } else if (!old_map.is_iomem) {
> >               dma_buf_map_memcpy_to(&new_map, old_map.vaddr,