On Tue, Sep 1, 2020 at 12:14 PM Robin Murphy <robin.murphy@xxxxxxx> wrote:
>
> On 2020-08-26 07:32, Marek Szyprowski wrote:
> > The Documentation/DMA-API-HOWTO.txt states that the dma_map_sg() function
> > returns the number of the created entries in the DMA address space.
> > However the subsequent calls to the dma_sync_sg_for_{device,cpu}() and
> > dma_unmap_sg must be called with the original number of the entries
> > passed to the dma_map_sg().
> >
> > struct sg_table is a common structure used for describing a non-contiguous
> > memory buffer, used commonly in the DRM and graphics subsystems. It
> > consists of a scatterlist with memory pages and DMA addresses (sgl entry),
> > as well as the number of scatterlist entries: CPU pages (orig_nents entry)
> > and DMA mapped pages (nents entry).
> >
> > It turned out that it was a common mistake to misuse nents and orig_nents
> > entries, calling DMA-mapping functions with a wrong number of entries or
> > ignoring the number of mapped entries returned by the dma_map_sg()
> > function.
> >
> > To avoid such issues, lets use a common dma-mapping wrappers operating
> > directly on the struct sg_table objects and use scatterlist page
> > iterators where possible. This, almost always, hides references to the
> > nents and orig_nents entries, making the code robust, easier to follow
> > and copy/paste safe.
> >
> > Signed-off-by: Marek Szyprowski <m.szyprowski@xxxxxxxxxxx>
> > Acked-by: Rob Clark <robdclark@xxxxxxxxx>
> > ---
> > drivers/gpu/drm/msm/msm_gem.c | 13 +++++--------
> > drivers/gpu/drm/msm/msm_gpummu.c | 14 ++++++--------
> > drivers/gpu/drm/msm/msm_iommu.c | 2 +-
> > 3 files changed, 12 insertions(+), 17 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/msm/msm_gem.c b/drivers/gpu/drm/msm/msm_gem.c
> > index b2f49152b4d4..8c7ae812b813 100644
> > --- a/drivers/gpu/drm/msm/msm_gem.c
> > +++ b/drivers/gpu/drm/msm/msm_gem.c
> > @@ -53,11 +53,10 @@ static void sync_for_device(struct msm_gem_object *msm_obj)
> > struct device *dev = msm_obj->base.dev->dev;
> >
> > if (get_dma_ops(dev) && IS_ENABLED(CONFIG_ARM64)) {
> > - dma_sync_sg_for_device(dev, msm_obj->sgt->sgl,
> > - msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
> > + dma_sync_sgtable_for_device(dev, msm_obj->sgt,
> > + DMA_BIDIRECTIONAL);
> > } else {
> > - dma_map_sg(dev, msm_obj->sgt->sgl,
> > - msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
> > + dma_map_sgtable(dev, msm_obj->sgt, DMA_BIDIRECTIONAL, 0);
> > }
> > }
> >
> > @@ -66,11 +65,9 @@ static void sync_for_cpu(struct msm_gem_object *msm_obj)
> > struct device *dev = msm_obj->base.dev->dev;
> >
> > if (get_dma_ops(dev) && IS_ENABLED(CONFIG_ARM64)) {
> > - dma_sync_sg_for_cpu(dev, msm_obj->sgt->sgl,
> > - msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
> > + dma_sync_sgtable_for_cpu(dev, msm_obj->sgt, DMA_BIDIRECTIONAL);
> > } else {
> > - dma_unmap_sg(dev, msm_obj->sgt->sgl,
> > - msm_obj->sgt->nents, DMA_BIDIRECTIONAL);
> > + dma_unmap_sgtable(dev, msm_obj->sgt, DMA_BIDIRECTIONAL, 0);
> > }
> > }
> >
> > diff --git a/drivers/gpu/drm/msm/msm_gpummu.c b/drivers/gpu/drm/msm/msm_gpummu.c
> > index 310a31b05faa..319f06c28235 100644
> > --- a/drivers/gpu/drm/msm/msm_gpummu.c
> > +++ b/drivers/gpu/drm/msm/msm_gpummu.c
> > @@ -30,21 +30,19 @@ static int msm_gpummu_map(struct msm_mmu *mmu, uint64_t iova,
> > {
> > struct msm_gpummu *gpummu = to_msm_gpummu(mmu);
> > unsigned idx = (iova - GPUMMU_VA_START) / GPUMMU_PAGE_SIZE;
> > - struct scatterlist *sg;
> > + struct sg_dma_page_iter dma_iter;
> > unsigned prot_bits = 0;
> > - unsigned i, j;
> >
> > if (prot & IOMMU_WRITE)
> > prot_bits |= 1;
> > if (prot & IOMMU_READ)
> > prot_bits |= 2;
> >
> > - for_each_sg(sgt->sgl, sg, sgt->nents, i) {
> > - dma_addr_t addr = sg->dma_address;
> > - for (j = 0; j < sg->length / GPUMMU_PAGE_SIZE; j++, idx++) {
> > - gpummu->table[idx] = addr | prot_bits;
> > - addr += GPUMMU_PAGE_SIZE;
> > - }
> > + for_each_sgtable_dma_page(sgt, &dma_iter, 0) {
> > + dma_addr_t addr = sg_page_iter_dma_address(&dma_iter);
> > +
> > + BUILD_BUG_ON(GPUMMU_PAGE_SIZE != PAGE_SIZE);
> > + gpummu->table[idx++] = addr | prot_bits;
>
> Given that the BUILD_BUG_ON might prevent valid arm64 configs from
> building, how about a simple tweak like:
>
> for (i = 0; i < PAGE_SIZE; i += GPUMMU_PAGE_SIZE)
> gpummu->table[idx++] = i + addr | prot_bits;
> ?
>
> Or alternatively perhaps some more aggressive #ifdefs or makefile tweaks
> to prevent the GPUMMU code building for arm64 at all if it's only
> relevant to 32-bit platforms (which I believe might be the case).
yes, the gpummu path is only used on older armv7 snapdragon and imx5
platforms.. I suppose maybe the easy thing would be to add a stub for
msm_gpummu_new() and msm_gpummu_params() and then only build it on
arm64?
BR,
-R
> Robin.
>
> > }
> >
> > /* we can improve by deferring flush for multiple map() */
> > diff --git a/drivers/gpu/drm/msm/msm_iommu.c b/drivers/gpu/drm/msm/msm_iommu.c
> > index 3a381a9674c9..6c31e65834c6 100644
> > --- a/drivers/gpu/drm/msm/msm_iommu.c
> > +++ b/drivers/gpu/drm/msm/msm_iommu.c
> > @@ -36,7 +36,7 @@ static int msm_iommu_map(struct msm_mmu *mmu, uint64_t iova,
> > struct msm_iommu *iommu = to_msm_iommu(mmu);
> > size_t ret;
> >
> > - ret = iommu_map_sg(iommu->domain, iova, sgt->sgl, sgt->nents, prot);
> > + ret = iommu_map_sgtable(iommu->domain, iova, sgt, prot);
> > WARN_ON(!ret);
> >
> > return (ret == len) ? 0 : -EINVAL;
> >
