On Mon, 1 Feb 2016 16:26:09 +0300 "Kirill A. Shutemov" <[email protected]> wrote:
> Maybe I miss some point, but I don't see a reason why we try to queue > pages from non migratable VMAs. > > The only case when we can queue pages from such VMA is MPOL_MF_STRICT > plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU, > but gfp mask is not sutable for migaration (see mapping_gfp_mask() check > in vma_migratable()). That's looks like a bug to me. > > Let's filter out non-migratable vma at start of queue_pages_test_walk() > and go to queue_pages_pte_range() only if MPOL_MF_MOVE or > MPOL_MF_MOVE_ALL flag is set. Conflicts with http://ozlabs.org/~akpm/mmots/broken-out/mm-mempolicy-skip-vm_hugetlb-and-vm_mixedmap-vma-for-lazy-mbind.patch. I resolved it thusly, please review: --- a/mm/mempolicy.c~mempolicy-do-not-try-to-queue-pages-from-vma_migratable +++ a/mm/mempolicy.c @@ -548,8 +548,7 @@ retry: goto retry; } - if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) - migrate_page_add(page, qp->pagelist, flags); + migrate_page_add(page, qp->pagelist, flags); } pte_unmap_unlock(pte - 1, ptl); cond_resched(); @@ -625,7 +624,7 @@ static int queue_pages_test_walk(unsigne unsigned long endvma = vma->vm_end; unsigned long flags = qp->flags; - if (vma->vm_flags & VM_PFNMAP) + if (!vma_migratable(vma)) return 1; if (endvma > end) @@ -644,17 +643,15 @@ static int queue_pages_test_walk(unsigne if (flags & MPOL_MF_LAZY) { /* Similar to task_numa_work, skip inaccessible VMAs */ - if (vma_migratable(vma) && !is_vm_hugetlb_page(vma) && + if (!is_vm_hugetlb_page(vma) && (vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)) && !(vma->vm_flags & VM_MIXEDMAP)) change_prot_numa(vma, start, endvma); return 1; } - if ((flags & MPOL_MF_STRICT) || - ((flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) && - vma_migratable(vma))) - /* queue pages from current vma */ + /* queue pages from current vma */ + if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) return 0; return 1; } _
