Merge branch 'akpm-current/current'
[ I do a git show 2db9f815409f and it doesn't show that hugetlb.c was
modified but it the git log the merge commit is listed. I don't know git well enough to explain it. But I think this was a merge issue. - dan ]
Hello Stephen Rothwell,
The patch 2db9f815409f: "Merge branch 'akpm-current/current'" from Aug 10, 2021, leads to the following
Smatch static checker warning:
mm/hugetlb.c:2172 alloc_buddy_huge_page_with_mpol() error: uninitialized symbol 'page'.
mm/hugetlb.c
2151 static
2152 struct page *alloc_buddy_huge_page_with_mpol(struct hstate *h,
2153 struct vm_area_struct *vma, unsigned long addr)
2154 {
2155 struct page *page;This should be "struct page *page = NULL;"
2156 struct mempolicy *mpol;
2157 gfp_t gfp_mask = htlb_alloc_mask(h);
2158 int nid;
2159 nodemask_t *nodemask;
2160
2161 nid = huge_node(vma, addr, gfp_mask, &mpol, &nodemask);
2162 if (mpol_is_preferred_many(mpol)) {
2163 gfp_t gfp = gfp_mask | __GFP_NOWARN;
2164
2165 gfp &= ~(__GFP_DIRECT_RECLAIM | __GFP_NOFAIL);
2166 page = alloc_surplus_huge_page(h, gfp, nid, nodemask, false);
2167
2168 /* Fallback to all nodes if page==NULL */
2169 nodemask = NULL;
2170 }
2171
--> 2172 if (!page)
2173 page = alloc_surplus_huge_page(h, gfp_mask, nid, nodemask, false);
2174 mpol_cond_put(mpol);
2175 return page;
2176 }regards,
dan carpenter
Hi Dan,
Quoted 14 lines · Dan Carpenter
[ I do a git show
2db9f815409fand it doesn't show thathugetlb.cwas
modified but it the git log the merge commit is listed. I don't know git well enough to explain it. But I think this was a merge issue. - dan ]Hello Stephen Rothwell,
The patch
2db9f815409f: "Merge branch 'akpm-current/current'" from Aug 10, 2021, leads to the following
Smatch static checker warning:mm/hugetlb.c:2172 alloc_buddy_huge_page_with_mpol() error: uninitialized symbol 'page'.
Quoted 8 lines
mm/hugetlb.c2151 static 2152 struct page *alloc_buddy_huge_page_with_mpol(struct hstate *h, 2153 struct vm_area_struct *vma, unsigned long addr) 2154 { 2155 struct page *page;This should be "struct page *page = NULL;"
Thanks for the report! This was also reported by Nathan Chancellor,
who gave a quick fix: https://lore.kernel.org/lkml/[email protected]/
and Andrew has taken it to the -mm tree.
Sorry for the inconvenience.
- Feng