On 2024/8/28 17:28, SeongJae Park wrote:
Hello,
On Tue, 27 Aug 2024 19:13:21 -0700 Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
The patch titled
Subject: mm: shmem: fix the gfp flag for large folio allocation
has been added to the -mm mm-unstable branch. Its filename is
mm-shmem-support-large-folio-allocation-for-shmem_replace_folio-fix.patch
This patch will shortly appear at
https://git.kernel.org/pub/scm/linux/kernel/git/akpm/25-new.git/tree/patches/mm-shmem-support-large-folio-allocation-for-shmem_replace_folio-fix.patch
This patch will later appear in the mm-unstable branch at
git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
Before you just go and hit "reply", please:
a) Consider who else should be cc'ed
b) Prefer to cc a suitable mailing list as well
c) Ideally: find the original patch on the mailing list and do a
reply-to-all to that, adding suitable additional cc's
*** Remember to use Documentation/process/submit-checklist.rst when testing your code ***
The -mm tree is included into linux-next via the mm-everything
branch at git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm
and is updated there every 2-3 working days
------------------------------------------------------
From: Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>
Subject: mm: shmem: fix the gfp flag for large folio allocation
Date: Tue, 27 Aug 2024 11:06:34 +0800
In shmem_replace_folio(), it may be necessary to allocate a large folio,
so we should update the gfp flags to ensure it is suitable for allocating
the large folio.
Link: https://lkml.kernel.org/r/5b1e9c5a-7f61-4d97-a8d7-41767ca04c77@xxxxxxxxxxxxxxxxx
Signed-off-by: Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>
Cc: Barry Song <baohua@xxxxxxxxxx>
Cc: Chris Li <chrisl@xxxxxxxxxx>
Cc: Daniel Gomez <da.gomez@xxxxxxxxxxx>
Cc: David Hildenbrand <david@xxxxxxxxxx>
Cc: "Huang, Ying" <ying.huang@xxxxxxxxx>
Cc: Hugh Dickins <hughd@xxxxxxxxxx>
Cc: Kefeng Wang <wangkefeng.wang@xxxxxxxxxx>
Cc: Lance Yang <ioworker0@xxxxxxxxx>
Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx>
Cc: Pankaj Raghav <p.raghav@xxxxxxxxxxx>
Cc: Ryan Roberts <ryan.roberts@xxxxxxx>
Cc: Yang Shi <shy828301@xxxxxxxxx>
Cc: Zi Yan <ziy@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
mm/shmem.c | 18 +++++++++++++-----
1 file changed, 13 insertions(+), 5 deletions(-)
--- a/mm/shmem.c~mm-shmem-support-large-folio-allocation-for-shmem_replace_folio-fix
+++ a/mm/shmem.c
@@ -155,7 +155,7 @@ static unsigned long shmem_default_max_i
static int shmem_swapin_folio(struct inode *inode, pgoff_t index,
struct folio **foliop, enum sgp_type sgp, gfp_t gfp,
- struct mm_struct *fault_mm, vm_fault_t *fault_type);
+ struct vm_area_struct *vma, vm_fault_t *fault_type);
static inline struct shmem_sb_info *SHMEM_SB(struct super_block *sb)
{
@@ -1883,7 +1883,8 @@ static bool shmem_should_replace_folio(s
}
static int shmem_replace_folio(struct folio **foliop, gfp_t gfp,
- struct shmem_inode_info *info, pgoff_t index)
+ struct shmem_inode_info *info, pgoff_t index,
+ struct vm_area_struct *vma)
{
struct folio *new, *old = *foliop;
swp_entry_t entry = old->swap;
@@ -1898,6 +1899,12 @@ static int shmem_replace_folio(struct fo
* limit chance of success by further cpuset and node constraints.
*/
gfp &= ~GFP_CONSTRAINT_MASK;
+ if (nr_pages > 1) {
+ gfp_t huge_gfp = vma_thp_gfp_mask(vma);
+
+ gfp = limit_gfp_mask(huge_gfp, gfp);
+ }
+
I just found the above change makes kernel build without
CONFIG_TRANSPARENT_HUGEPAGE fails like below.
ERROR:root:/usr/bin/ld: mm/shmem.o: in function `shmem_swapin_folio':
shmem.c:(.text+0x12d9): undefined reference to `vma_thp_gfp_mask'
collect2: error: ld returned 1 exit status
Specifically, I reproduce the problem with kunit[1].
Maybe the above if statement need to take care of the case, e.g., wrap with
#ifdef CONFIG_TRANSPARENT_HUGEPAGE?
[1] https://github.com/damonitor/damon-tests/blob/master/corr/tests/kunit.sh
Thanks for reporting. Yes, the vma_thp_gfp_mask() should be wrapped with
the CONFIG_TRANSPARENT_HUGEPAGE macro.
Andrew, please help to squash the following fix. Sorry for troubles.
diff --git a/mm/shmem.c b/mm/shmem.c
index 2b0209d6ac9c..553b99cb265e 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1910,11 +1910,13 @@ static int shmem_replace_folio(struct folio
**foliop, gfp_t gfp,
* limit chance of success by further cpuset and node constraints.
*/
gfp &= ~GFP_CONSTRAINT_MASK;
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
if (nr_pages > 1) {
gfp_t huge_gfp = vma_thp_gfp_mask(vma);
gfp = limit_gfp_mask(huge_gfp, gfp);
}
+#endif
new = shmem_alloc_folio(gfp, folio_order(old), info, index);
if (!new)
