]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
dma-buf: fix integer overflow in fill_sg_entry() for buffers >= 8GiB
authorAlex Mastro <amastro@fb.com>
Wed, 26 Nov 2025 01:11:18 +0000 (17:11 -0800)
committerAlex Williamson <alex@shazbot.org>
Fri, 28 Nov 2025 17:06:25 +0000 (10:06 -0700)
fill_sg_entry() splits large DMA buffers into multiple scatter-gather
entries, each holding up to UINT_MAX bytes. When calculating the DMA
address for entries beyond the second one, the expression (i * UINT_MAX)
causes integer overflow due to 32-bit arithmetic.

This manifests when the input arg length >= 8 GiB results in looping for
i >= 2.

Fix by casting i to dma_addr_t before multiplication.

Fixes: 3aa31a8bb11e ("dma-buf: provide phys_vec to scatter-gather mapping routine")
Signed-off-by: Alex Mastro <amastro@fb.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Leon Romanovsky <leon@kernel.org>
Link: https://lore.kernel.org/r/20251125-dma-buf-overflow-v1-1-b70ea1e6c4ba@fb.com
Signed-off-by: Alex Williamson <alex@shazbot.org>
drivers/dma-buf/dma-buf-mapping.c

index b4819811a64a0afdd610277ec0893fbb1248926d..b7352e609fbdfa87be5acf44230c7c30456f0538 100644 (file)
@@ -24,7 +24,7 @@ static struct scatterlist *fill_sg_entry(struct scatterlist *sgl, size_t length,
                 * does not require the CPU list for mapping or unmapping.
                 */
                sg_set_page(sgl, NULL, 0, 0);
-               sg_dma_address(sgl) = addr + i * UINT_MAX;
+               sg_dma_address(sgl) = addr + (dma_addr_t)i * UINT_MAX;
                sg_dma_len(sgl) = len;
                sgl = sg_next(sgl);
        }