Hi All,
With the mTHP shmem stat names cleaned up [2], we can now introduce real "file_"
stats for file-backed memory. This series does that. The stats are useful to
give visibility into how file-backed memory is being allocated. I'm planning to
build upon this with controls to restrict the folio sizes that can be allocatd
for pagecache (subject to test results that demonstrate the value).
---
This applies on top of today's mm-unstable (650b6752c8a3). All mm selftests have
been run; no regressions were observed.
Changes since v1 [1]
====================
- Added patch 2 to tidy up shmem controls; now exposed for order-1, and not
exposed for any unsupported high orders.
- Simplified "stats" subdirectory management with sysfs_merge_group().
- Added R-b/A-b to patch 1; thanks to David, Barry, Baolin, Lance
[1] https://lore.kernel.org/linux-mm/20240711072929.3590000-1-ryan.roberts@xxxxxxx/
[2] https://lore.kernel.org/linux-mm/20240710095503.3193901-1-ryan.roberts@xxxxxxx/
Thanks,
Ryan
Ryan Roberts (3):
mm: Cleanup count_mthp_stat() definition
mm: Tidy up shmem mTHP controls and stats
mm: mTHP stats for pagecache folio allocations
Documentation/admin-guide/mm/transhuge.rst | 13 +++
include/linux/huge_mm.h | 73 +++++++------
include/linux/pagemap.h | 16 ++-
mm/filemap.c | 6 +-
mm/huge_memory.c | 117 +++++++++++++++------
mm/memory.c | 2 -
mm/shmem.c | 6 --
7 files changed, 156 insertions(+), 77 deletions(-)
--
2.43.0
