From mboxrd@z Thu Jan 1 00:00:00 1970 From: Ingo Molnar Date: Mon, 17 Nov 2008 08:15:26 +0000 Subject: Re: [PATCH] swiotlb: use coherent_dma_mask in alloc_coherent Message-Id: <20081117081526.GA24603@elte.hu> List-Id: References: <20081117162445C.fujita.tomonori@lab.ntt.co.jp> In-Reply-To: <20081117162445C.fujita.tomonori@lab.ntt.co.jp> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: FUJITA Tomonori Cc: tony.luck@intel.com, linux-kernel@vger.kernel.org, linux-ia64@vger.kernel.org, Joerg Roedel , Andrew Morton * FUJITA Tomonori wrote: > This patch fixes swiotlb to use dev->coherent_dma_mask in > alloc_coherent. Currently, swiotlb uses dev->dma_mask in > alloc_coherent but alloc_coherent is supposed to use > coherent_dma_mask. It could break drivers that uses smaller > coherent_dma_mask than dma_mask (though the current code works for > the majority that use the same mask for coherent_dma_mask and > dma_mask). > > Signed-off-by: FUJITA Tomonori > --- > lib/swiotlb.c | 10 +++++++--- > 1 files changed, 7 insertions(+), 3 deletions(-) Applied it with the changelog below to tip/core/urgent, thanks! I also flagged it for v2.6.28 inclusion. This bug was caused by the removal of the GFP_DMA hack in swiotlb_alloc_coherent() in this cycle. I havent seen it actually reported anywhere - have you perhaps?Or have you found this via code review? Do we know roughly the range of devices/systems where there's a real address range that cannot be DMA-ed to coherently, and an estimation about how frequently they would be affected by this bug? Ingo ---------------> >From 1e74f3000b86969de421ca0da08f42e7d21cbd99 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Mon, 17 Nov 2008 16:24:34 +0900 Subject: [PATCH] swiotlb: use coherent_dma_mask in alloc_coherent Impact: fix DMA buffer allocation coherency bug in certain configs This patch fixes swiotlb to use dev->coherent_dma_mask in swiotlb_alloc_coherent(). coherent_dma_mask is a subset of dma_mask (equal to it most of the time), enumerating the address range that a given device is able to DMA to/from in a cache-coherent way. But currently, swiotlb uses dev->dma_mask in alloc_coherent() implicitly via address_needs_mapping(), but alloc_coherent is really supposed to use coherent_dma_mask. This bug could break drivers that uses smaller coherent_dma_mask than dma_mask (though the current code works for the majority that use the same mask for coherent_dma_mask and dma_mask). Signed-off-by: FUJITA Tomonori Cc: tony.luck@intel.com Signed-off-by: Ingo Molnar --- lib/swiotlb.c | 10 +++++++--- 1 files changed, 7 insertions(+), 3 deletions(-) diff --git a/lib/swiotlb.c b/lib/swiotlb.c index 78330c3..5f6c629 100644 --- a/lib/swiotlb.c +++ b/lib/swiotlb.c @@ -467,9 +467,13 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, dma_addr_t dev_addr; void *ret; int order = get_order(size); + u64 dma_mask = DMA_32BIT_MASK; + + if (hwdev && hwdev->coherent_dma_mask) + dma_mask = hwdev->coherent_dma_mask; ret = (void *)__get_free_pages(flags, order); - if (ret && address_needs_mapping(hwdev, virt_to_bus(ret), size)) { + if (ret && !is_buffer_dma_capable(dma_mask, virt_to_bus(ret), size)) { /* * The allocated memory isn't reachable by the device. * Fall back on swiotlb_map_single(). @@ -493,9 +497,9 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size, dev_addr = virt_to_bus(ret); /* Confirm address can be DMA'd by device */ - if (address_needs_mapping(hwdev, dev_addr, size)) { + if (!is_buffer_dma_capable(dma_mask, dev_addr, size)) { printk("hwdev DMA mask = 0x%016Lx, dev_addr = 0x%016Lx\n", - (unsigned long long)*hwdev->dma_mask, + (unsigned long long)dma_mask, (unsigned long long)dev_addr); /* DMA_TO_DEVICE to avoid memcpy in unmap_single */