tree: https://github.com/AsahiLinux/linux dart/4kpgsize head: 7958b0de7ba6467e2f97cd8628661777d14780d8 commit: 38661a4229b4dc63dc0cae6e1401c3465538f1a1 [47/51] iommu/dma: iommu_dma_map_sg: support granule > PAGE_SIZE config: x86_64-randconfig-c001-20210813 (attached as .config) compiler: clang version 14.0.0 (https://github.com/llvm/llvm-project 62df4df41c939205b2dc0a2a3bfb75b8c1ed74fa) reproduce (this is a W=1 build): wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # https://github.com/AsahiLinux/linux/commit/38661a4229b4dc63dc0cae6e1401c3... git remote add asahilinux https://github.com/AsahiLinux/linux git fetch --no-tags asahilinux dart/4kpgsize git checkout 38661a4229b4dc63dc0cae6e1401c3465538f1a1 # save the attached .config to linux build tree COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64 If you fix the issue, kindly add following tag as appropriate Reported-by: kernel test robot <lkp(a)intel.com&gt; All warnings (new ones prefixed by >>): drivers/iommu/dma-iommu.c:918:18: error: implicit declaration of function 'phys_to_page' [-Werror,-Wimplicit-function-declaration] sg_set_page(s, phys_to_page(sg_phys(s) + s_iova_off), s_length, ^ drivers/iommu/dma-iommu.c:918:18: note: did you mean 'pmd_to_page'? include/linux/mm.h:2293:21: note: 'pmd_to_page' declared here static struct page *pmd_to_page(pmd_t *pmd) ^ sg_set_page(s, phys_to_page(sg_phys(s) + s_iova_off), s_length, ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/scatterlist.h:110:69: note: passing argument to parameter 'page' here static inline void sg_set_page(struct scatterlist *sg, struct page *page, ^ drivers/iommu/dma-iommu.c:965:9: error: implicit declaration of function 'phys_to_page' [-Werror,-Wimplicit-function-declaration] phys_to_page(sg_phys(s) + sg_dma_address(s)), ^ drivers/iommu/dma-iommu.c:965:9: warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'struct page *' [-Wint-conversion] phys_to_page(sg_phys(s) + sg_dma_address(s)), ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/scatterlist.h:110:69: note: passing argument to parameter 'page' here static inline void sg_set_page(struct scatterlist *sg, struct page *page, ^ drivers/iommu/dma-iommu.c:1051:18: error: implicit declaration of function 'phys_to_page' [-Werror,-Wimplicit-function-declaration] sg_set_page(s, phys_to_page(s_phys - s_iova_off), ^ drivers/iommu/dma-iommu.c:1051:18: warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'struct page *' [-Wint-conversion] sg_set_page(s, phys_to_page(s_phys - s_iova_off), ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ include/linux/scatterlist.h:110:69: note: passing argument to parameter 'page' here static inline void sg_set_page(struct scatterlist *sg, struct page *page, ^ 3 warnings and 3 errors generated. vim +918 drivers/iommu/dma-iommu.c 896 897 /* 898 * Prepare a successfully-mapped scatterlist to give back to the caller. 899 * 900 * At this point the segments are already laid out by iommu_dma_map_sg() to 901 * avoid individually crossing any boundaries, so we merely need to check a 902 * segment's start address to avoid concatenating across one. 903 */ 904 static int __finalise_sg(struct device *dev, struct scatterlist *sg, int nents, 905 dma_addr_t dma_addr) 906 { 907 struct scatterlist *s, *cur = sg; 908 unsigned long seg_mask = dma_get_seg_boundary(dev); 909 unsigned int cur_len = 0, max_len = dma_get_max_seg_size(dev); 910 int i, count = 0; 911 912 for_each_sg(sg, s, nents, i) { 913 /* Restore this segment's original unaligned fields first */ 914 unsigned int s_iova_off = sg_dma_address(s); 915 unsigned int s_length = sg_dma_len(s); 916 unsigned int s_iova_len = s->length; 917 919 s_iova_off & ~PAGE_MASK); 920 sg_dma_address(s) = DMA_MAPPING_ERROR; 921 sg_dma_len(s) = 0; 922 923 /* 924 * Now fill in the real DMA data. If... 925 * - there is a valid output segment to append to 926 * - and this segment starts on an IOVA page boundary 927 * - but doesn't fall at a segment boundary 928 * - and wouldn't make the resulting output segment too long 929 */ 930 if (cur_len && !s_iova_off && (dma_addr & seg_mask) && 931 (max_len - cur_len >= s_length)) { 932 /* ...then concatenate it with the previous one */ 933 cur_len += s_length; 934 } else { 935 /* Otherwise start the next output segment */ 936 if (i > 0) 937 cur = sg_next(cur); 938 cur_len = s_length; 939 count++; 940 941 sg_dma_address(cur) = dma_addr + s_iova_off; 942 } 943 944 sg_dma_len(cur) = cur_len; 945 dma_addr += s_iova_len; 946 947 if (s_length + s_iova_off < s_iova_len) 948 cur_len = 0; 949 } 950 return count; 951 } 952 --- 0-DAY CI Kernel Test Service, Intel Corporation https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org