Re: [PATCH v4 1/4] hugetlb: Pass vma into huge_pte_alloc() and huge_pmd_share()

From: Peter Xu
Date: Thu Mar 04 2021 - 11:49:39 EST


On Fri, Mar 05, 2021 at 12:31:00AM +0800, kernel test robot wrote:
> Hi Peter,
>
> Thank you for the patch! Yet something to improve:
>
> [auto build test ERROR on arm64/for-next/core]
> [also build test ERROR on linux/master linus/master v5.12-rc1 next-20210304]
> [cannot apply to hnaz-linux-mm/master]
> [If your patch is applied to the wrong git tree, kindly drop us a note.
> And when submitting patch, we suggest to use '--base' as documented in
> https://git-scm.com/docs/git-format-patch]
>
> url: https://github.com/0day-ci/linux/commits/Peter-Xu/hugetlb-Disable-huge-pmd-unshare-for-uffd-wp/20210219-071334
> base: https://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-next/core
> config: mips-randconfig-m031-20210304 (attached as .config)
> compiler: mipsel-linux-gcc (GCC) 9.3.0
> 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/0day-ci/linux/commit/7ede06d6f63e59db4b9dee54f78eeac0c9ca17e4
> git remote add linux-review https://github.com/0day-ci/linux
> git fetch --no-tags linux-review Peter-Xu/hugetlb-Disable-huge-pmd-unshare-for-uffd-wp/20210219-071334
> git checkout 7ede06d6f63e59db4b9dee54f78eeac0c9ca17e4
> # save the attached .config to linux build tree
> COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-9.3.0 make.cross ARCH=mips
>
> If you fix the issue, kindly add following tag as appropriate
> Reported-by: kernel test robot <lkp@xxxxxxxxx>
>
> All errors (new ones prefixed by >>):
>
> >> mm/hugetlb.c:5376:8: error: conflicting types for 'huge_pmd_share'
> 5376 | pte_t *huge_pmd_share(struct mm_struct *mm, struct vm_area_struct vma,
> | ^~~~~~~~~~~~~~
> In file included from mm/hugetlb.c:39:
> include/linux/hugetlb.h:155:8: note: previous declaration of 'huge_pmd_share' was here
> 155 | pte_t *huge_pmd_share(struct mm_struct *mm, struct vm_area_struct *vma,
> | ^~~~~~~~~~~~~~
>
>
> vim +/huge_pmd_share +5376 mm/hugetlb.c
>
> 5343
> 5344 /*
> 5345 * unmap huge page backed by shared pte.
> 5346 *
> 5347 * Hugetlb pte page is ref counted at the time of mapping. If pte is shared
> 5348 * indicated by page_count > 1, unmap is achieved by clearing pud and
> 5349 * decrementing the ref count. If count == 1, the pte page is not shared.
> 5350 *
> 5351 * Called with page table lock held and i_mmap_rwsem held in write mode.
> 5352 *
> 5353 * returns: 1 successfully unmapped a shared pte page
> 5354 * 0 the underlying pte page is not shared, or it is the last user
> 5355 */
> 5356 int huge_pmd_unshare(struct mm_struct *mm, struct vm_area_struct *vma,
> 5357 unsigned long *addr, pte_t *ptep)
> 5358 {
> 5359 pgd_t *pgd = pgd_offset(mm, *addr);
> 5360 p4d_t *p4d = p4d_offset(pgd, *addr);
> 5361 pud_t *pud = pud_offset(p4d, *addr);
> 5362
> 5363 i_mmap_assert_write_locked(vma->vm_file->f_mapping);
> 5364 BUG_ON(page_count(virt_to_page(ptep)) == 0);
> 5365 if (page_count(virt_to_page(ptep)) == 1)
> 5366 return 0;
> 5367
> 5368 pud_clear(pud);
> 5369 put_page(virt_to_page(ptep));
> 5370 mm_dec_nr_pmds(mm);
> 5371 *addr = ALIGN(*addr, HPAGE_SIZE * PTRS_PER_PTE) - HPAGE_SIZE;
> 5372 return 1;
> 5373 }
> 5374 #define want_pmd_share() (1)
> 5375 #else /* !CONFIG_ARCH_WANT_HUGE_PMD_SHARE */
> > 5376 pte_t *huge_pmd_share(struct mm_struct *mm, struct vm_area_struct vma,
> 5377 unsigned long addr, pud_t *pud)
> 5378 {
> 5379 return NULL;
> 5380 }
> 5381

Sorry for this! I think we need to squash below into this patch for
!CONFIG_ARCH_WANT_HUGE_PMD_SHARE:

-----8<-----
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index fc62932c31cb..94ac419f88cd 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -5415,7 +5415,7 @@ int huge_pmd_unshare(struct mm_struct *mm, struct vm_area_struct *vma,
}

#else /* !CONFIG_ARCH_WANT_HUGE_PMD_SHARE */
-pte_t *huge_pmd_share(struct mm_struct *mm, struct vm_area_struct vma,
+pte_t *huge_pmd_share(struct mm_struct *mm, struct vm_area_struct *vma,
unsigned long addr, pud_t *pud)
{
return NULL;
-----8<-----

Andrew, please kindly let me know when a repost is needed.

Thanks,

--
Peter Xu