From mboxrd@z Thu Jan 1 00:00:00 1970 From: =?UTF-8?q?=C3=98rjan=20Eide?= Subject: [PATCH 2/2] ARM: DMA: Use vm_pgoff for IOMMU mmap Date: Thu, 16 Apr 2015 16:41:52 +0200 Message-ID: <1429195312-25898-2-git-send-email-orjan.eide@arm.com> References: <1429195312-25898-1-git-send-email-orjan.eide@arm.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: In-Reply-To: <1429195312-25898-1-git-send-email-orjan.eide@arm.com> Sender: linux-kernel-owner@vger.kernel.org To: linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, dri-devel@lists.freedesktop.org, linux-rockchip@lists.infradead.org Cc: djkurtz@chromium.org List-Id: linux-rockchip.vger.kernel.org arm_iommu_mmap_attrs() ignores the page offset in vma->vm_pgoff that wa= s specified in the mmap call. Any user of dma_mmap_* will get an unexpected result when the device uses the IOMMU DMA ops. Some DRM driver that use dma_mmap_* seems to depend on it ignoring vma->vm_pgoff. Both the Samsung Exynos DRM driver before v3.18, and th= e current Rockchip DRM driver, use the page offset to pass GEM cookies that are used to look up the GEM buffers and then call dma_mmap_attrs() with the cookie value still in vma->vm_pgoff. rockchip_gem_mmap_buf() in drivers/gpu/drm/rockchip/rockchip_drm_gem.c still does this. It should be fixed before fixing arm_iommu_mmap_attrs(). Signed-off-by: =C3=98rjan Eide --- arch/arm/mm/dma-mapping.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index c274476..07d571b 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -1364,6 +1364,9 @@ static int arm_iommu_mmap_attrs(struct device *de= v, struct vm_area_struct *vma, { unsigned long uaddr =3D vma->vm_start; unsigned long usize =3D vma->vm_end - vma->vm_start; + unsigned long nr_vma_pages =3D usize >> PAGE_SHIFT; + unsigned long nr_pages =3D PAGE_ALIGN(size) >> PAGE_SHIFT; + unsigned long off =3D vma->vm_pgoff; struct page **pages =3D __iommu_get_pages(cpu_addr, attrs); =20 vma->vm_page_prot =3D __get_dma_pgprot(attrs, vma->vm_page_prot); @@ -1371,6 +1374,11 @@ static int arm_iommu_mmap_attrs(struct device *d= ev, struct vm_area_struct *vma, if (!pages) return -ENXIO; =20 + if (off >=3D nr_pages || nr_vma_pages > (nr_pages - off)) + return -ENXIO; + + pages +=3D off; + do { int ret =3D vm_insert_page(vma, uaddr, *pages++); if (ret) { --=20 1.9.1 From mboxrd@z Thu Jan 1 00:00:00 1970 From: orjan.eide@arm.com (=?UTF-8?q?=C3=98rjan=20Eide?=) Date: Thu, 16 Apr 2015 16:41:52 +0200 Subject: [PATCH 2/2] ARM: DMA: Use vm_pgoff for IOMMU mmap In-Reply-To: <1429195312-25898-1-git-send-email-orjan.eide@arm.com> References: <1429195312-25898-1-git-send-email-orjan.eide@arm.com> Message-ID: <1429195312-25898-2-git-send-email-orjan.eide@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org arm_iommu_mmap_attrs() ignores the page offset in vma->vm_pgoff that was specified in the mmap call. Any user of dma_mmap_* will get an unexpected result when the device uses the IOMMU DMA ops. Some DRM driver that use dma_mmap_* seems to depend on it ignoring vma->vm_pgoff. Both the Samsung Exynos DRM driver before v3.18, and the current Rockchip DRM driver, use the page offset to pass GEM cookies that are used to look up the GEM buffers and then call dma_mmap_attrs() with the cookie value still in vma->vm_pgoff. rockchip_gem_mmap_buf() in drivers/gpu/drm/rockchip/rockchip_drm_gem.c still does this. It should be fixed before fixing arm_iommu_mmap_attrs(). Signed-off-by: ?rjan Eide --- arch/arm/mm/dma-mapping.c | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index c274476..07d571b 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -1364,6 +1364,9 @@ static int arm_iommu_mmap_attrs(struct device *dev, struct vm_area_struct *vma, { unsigned long uaddr = vma->vm_start; unsigned long usize = vma->vm_end - vma->vm_start; + unsigned long nr_vma_pages = usize >> PAGE_SHIFT; + unsigned long nr_pages = PAGE_ALIGN(size) >> PAGE_SHIFT; + unsigned long off = vma->vm_pgoff; struct page **pages = __iommu_get_pages(cpu_addr, attrs); vma->vm_page_prot = __get_dma_pgprot(attrs, vma->vm_page_prot); @@ -1371,6 +1374,11 @@ static int arm_iommu_mmap_attrs(struct device *dev, struct vm_area_struct *vma, if (!pages) return -ENXIO; + if (off >= nr_pages || nr_vma_pages > (nr_pages - off)) + return -ENXIO; + + pages += off; + do { int ret = vm_insert_page(vma, uaddr, *pages++); if (ret) { -- 1.9.1