From: Yu Zhao <yuzhao@xxxxxxxxxx> Subject: mm/memory.c: replace is_zero_pfn with is_huge_zero_pmd for thp For hugely mapped thp, we use is_huge_zero_pmd() to check if it's zero page or not. We do fill ptes with my_zero_pfn() when we split zero thp pmd, but this is not what we have in vm_normal_page_pmd() -- pmd_trans_huge_lock() makes sure of it. This is a trivial fix for /proc/pid/numa_maps, and AFAIK nobody complains about it. Gerald Schaefer asked: : Maybe the description could also mention the symptom of this bug? : I would assume that it affects anon/dirty accounting in gather_pte_stats(), : for huge mappings, if zero page mappings are not correctly recognized. I came across this while I was looking at the code, so I'm not aware of any symptom. Link: http://lkml.kernel.org/r/20191108192629.201556-1-yuzhao@xxxxxxxxxx Signed-off-by: Yu Zhao <yuzhao@xxxxxxxxxx> Acked-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx> Cc: Ralph Campbell <rcampbell@xxxxxxxxxx> Cc: Will Deacon <will@xxxxxxxxxx> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx> Cc: "Aneesh Kumar K . V" <aneesh.kumar@xxxxxxxxxxxxx> Cc: Dave Airlie <airlied@xxxxxxxxxx> Cc: Thomas Hellstrom <thellstrom@xxxxxxxxxx> Cc: Souptick Joarder <jrdr.linux@xxxxxxxxx> Cc: Gerald Schaefer <gerald.schaefer@xxxxxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- mm/memory.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) --- a/mm/memory.c~mm-replace-is_zero_pfn-with-is_huge_zero_pmd-for-thp +++ a/mm/memory.c @@ -672,7 +672,7 @@ struct page *vm_normal_page_pmd(struct v if (pmd_devmap(pmd)) return NULL; - if (is_zero_pfn(pfn)) + if (is_huge_zero_pmd(pmd)) return NULL; if (unlikely(pfn > highest_memmap_pfn)) return NULL; _
