mm/swap: inline folio_set_swap_entry() and folio_swap_entry()

Let's simply work on the folio directly and remove the helpers.

Link: https://lkml.kernel.org/r/20230821160849.531668-4-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Reviewed-by: Chris Li <chrisl@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/swap: use dedicated entry for swap in folio

Let's stop working on the private field and use an explicit swap field.
We have to move the swp_entry_t typedef.

Link: https://lkml.kernel.org/r/20230821160849.531668-3-david@redhat.com
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Chris Li <chrisl@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/swap: stop using page->private on tail pages for THP_SWAP

Patch series "mm/swap: stop using page->private on tail pages for THP_SWAP
+ cleanups".

This series stops using page->private on tail pages for THP_SWAP, replaces
folio->private by folio->swap for swapcache folios, and starts using
"new_folio" for tail pages that we are splitting to remove the usage of
page->private for swapcache handling completely.

This patch (of 4):

Let's stop using page->private on tail pages, making it possible to just
unconditionally reuse that field in the tail pages of large folios.

The remaining usage of the private field for THP_SWAP is in the THP
splitting code (mm/huge_memory.c), that we'll handle separately later.

Update the THP_SWAP documentation and sanity checks in mm_types.h and
__split_huge_page_tail().

[david@redhat.com: stop using page->private on tail pages for THP_SWAP]
Link: https://lkml.kernel.org/r/6f0a82a3-6948-20d9-580b-be1dbf415701@redhat.com
Link: https://lkml.kernel.org/r/20230821160849.531668-1-david@redhat.com
Link: https://lkml.kernel.org/r/20230821160849.531668-2-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Reviewed-by: Yosry Ahmed <yosryahmed@google.com>
Cc: Dan Streetman <ddstreet@ieee.org>
Cc: Hugh Dickins <hughd@google.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Peter Xu <peterx@redhat.com>
Cc: Seth Jennings <sjenning@redhat.com>
Cc: Vitaly Wool <vitaly.wool@konsulko.com>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests/mm: fix WARNING comparing pointer to 0

Remove comparing pointer to 0 to avoid this warning from coccinelle:

./tools/testing/selftests/mm/map_populate.c:80:16-17: WARNING comparing pointer to 0, suggest !E
./tools/testing/selftests/mm/map_populate.c:80:16-17: WARNING comparing pointer to 0

Link: https://lkml.kernel.org/r/20230817160033.90079-1-tuananhlfc@gmail.com
Signed-off-by: Anh Tuan Phan <tuananhlfc@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests: cgroup: fix test_kmem_memcg_deletion kernel mem check

Currently, not all kernel memory usage is being accounted for. This
commit switches to using the kernel entry within memory.stat which
already includes kernel_stack, pagetables, and slab. The kernel entry
also includes vmalloc and other additional kernel memory use cases which
were missing.

Link: https://lkml.kernel.org/r/bvrhe2tpsts2azaroq4ubp2slawmop6orndsswrewuscw3ugvk@kmemmrttsnc7
Signed-off-by: Lucas Karpinski <lkarpins@redhat.com>
Acked-by: Shakeel Butt <shakeelb@google.com>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Shuah Khan <shuah@kernel.org>
Cc: Tejun Heo <tj@kernel.org>
Cc: Zefan Li <lizefan.x@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: userfaultfd: remove stale comment about core dump locking

Since commit 7f3bfab52cab ("mm/gup: take mmap_lock in get_dump_page()"),
which landed in v5.10, core dumping doesn't enter fault handling without
holding the mmap_lock anymore.  Remove the stale parts of the comments,
but leave the behavior as-is - letting core dumping block on userfault
handling would be a bad idea and could lead to deadlocks if the dumping
process was handling its own userfaults.

Link: https://lkml.kernel.org/r/20230815212216.264445-1-jannh@google.com
Signed-off-by: Jann Horn <jannh@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arm64: mm: use ptep_clear() instead of pte_clear() in clear_flush()

In clear_flush(), the original pte may be a present entry, so we should
use ptep_clear() to let page_table_check track the pte clearing operation,
otherwise it may cause false positive in subsequent set_pte_at().

Link: https://lkml.kernel.org/r/20230810093241.1181142-1-qi.zheng@linux.dev
Fixes: 42b2547137f5 ("arm64/mm: enable ARCH_SUPPORTS_PAGE_TABLE_CHECK")
Signed-off-by: Qi Zheng <zhengqi.arch@bytedance.com>
Acked-by: Will Deacon <will@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Pasha Tatashin <pasha.tatashin@soleen.com>
Cc: Qi Zheng <zhengqi.arch@bytedance.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: call update_mmu_cache_range() in more page fault handling paths

Pass the vm_fault to the architecture to help it make smarter decisions
about which PTEs to insert into the TLB.

Link: https://lkml.kernel.org/r/20230802151406.3735276-39-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

filemap: batch PTE mappings

Call set_pte_range() once per contiguous range of the folio instead of
once per page.  This batches the updates to mm counters and the rmap.

With a will-it-scale.page_fault3 like app (change file write fault testing
to read fault testing.  Trying to upstream it to will-it-scale at [1]) got
15% performance gain on a 48C/96T Cascade Lake test box with 96 processes
running against xfs.

Perf data collected before/after the change:
  18.73%--page_add_file_rmap
          |
           --11.60%--__mod_lruvec_page_state
                     |
                     |--7.40%--__mod_memcg_lruvec_state
                     |          |
                     |           --5.58%--cgroup_rstat_updated
                     |
                      --2.53%--__mod_lruvec_state
                                |
                                 --1.48%--__mod_node_page_state

  9.93%--page_add_file_rmap_range
         |
          --2.67%--__mod_lruvec_page_state
                    |
                    |--1.95%--__mod_memcg_lruvec_state
                    |          |
                    |           --1.57%--cgroup_rstat_updated
                    |
                     --0.61%--__mod_lruvec_state
                               |
                                --0.54%--__mod_node_page_state

The running time of __mode_lruvec_page_state() is reduced about 9%.

[1]: https://github.com/antonblanchard/will-it-scale/pull/37

Link: https://lkml.kernel.org/r/20230802151406.3735276-38-willy@infradead.org
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert do_set_pte() to set_pte_range()

set_pte_range() allows to setup page table entries for a specific
range.  It takes advantage of batched rmap update for large folio.
It now takes care of calling update_mmu_cache_range().

Link: https://lkml.kernel.org/r/20230802151406.3735276-37-willy@infradead.org
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

rmap: add folio_add_file_rmap_range()

folio_add_file_rmap_range() allows to add pte mapping to a specific range
of file folio.  Comparing to page_add_file_rmap(), it batched updates
__lruvec_stat for large folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-36-willy@infradead.org
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

filemap: add filemap_map_folio_range()

filemap_map_folio_range() maps partial/full folio.  Comparing to original
filemap_map_pages(), it updates refcount once per folio instead of per
page and gets minor performance improvement for large folio.

With a will-it-scale.page_fault3 like app (change file write fault testing
to read fault testing.  Trying to upstream it to will-it-scale at [1]),
got 2% performance gain on a 48C/96T Cascade Lake test box with 96
processes running against xfs.

[1]: https://github.com/antonblanchard/will-it-scale/pull/37

Link: https://lkml.kernel.org/r/20230802151406.3735276-35-willy@infradead.org
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: use flush_icache_pages() in do_set_pmd()

Push the iteration over each page down to the architectures (many can
flush the entire THP without iteration).

Link: https://lkml.kernel.org/r/20230802151406.3735276-34-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: tidy up set_ptes definition

Now that all architectures are converted, we can remove the PFN_PTE_SHIFT
ifdef and we can define set_pte_at() unconditionally.

Link: https://lkml.kernel.org/r/20230802151406.3735276-33-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: rationalise flush_icache_pages() and flush_icache_page()

Move the default (no-op) implementation of flush_icache_pages() to
<linux/cacheflush.h> from <asm-generic/cacheflush.h>.  Remove the
flush_icache_page() wrapper from each architecture into
<linux/cacheflush.h>.

Link: https://lkml.kernel.org/r/20230802151406.3735276-32-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: remove page_mapping_file()

This function has no more users.

Link: https://lkml.kernel.org/r/20230802151406.3735276-31-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

xtensa: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().

Link: https://lkml.kernel.org/r/20230802151406.3735276-30-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

x86: implement the new page table range API

Add PFN_PTE_SHIFT and a noop update_mmu_cache_range().

Link: https://lkml.kernel.org/r/20230802151406.3735276-29-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

um: implement the new page table range API

Add PFN_PTE_SHIFT and update_mmu_cache_range().

Link: https://lkml.kernel.org/r/20230802151406.3735276-28-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Richard Weinberger <richard@nod.at>
Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

sparc64: implement the new page table range API

Add set_ptes(), update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().  Convert the PG_dcache_dirty flag from being
per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-27-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

sparc32: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().

Link: https://lkml.kernel.org/r/20230802151406.3735276-26-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

sh: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().  Change the PG_dcache_clean flag from being per-page
to per-folio.  Flush the entire folio containing the pages in
flush_icache_pages() for ease of implementation.

Link: https://lkml.kernel.org/r/20230802151406.3735276-25-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

s390: implement the new page table range API

Add set_ptes() and update_mmu_cache_range().

Link: https://lkml.kernel.org/r/20230802151406.3735276-24-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Sven Schnelle <svens@linux.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

riscv: implement the new page table range API

Add set_ptes(), update_mmu_cache_range() and flush_dcache_folio().  Change
the PG_dcache_clean flag from being per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-23-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

powerpc: implement the new page table range API

Add set_ptes(), update_mmu_cache_range() and flush_dcache_folio().  Change
the PG_arch_1 (aka PG_dcache_dirty) flag from being per-page to per-folio.

[willy@infradead.org: re-export flush_dcache_icache_folio()]
Link: https://lkml.kernel.org/r/ZMx1daYwvD9EM7Cv@casper.infradead.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-22-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

parisc: implement the new page table range API

Add set_ptes(), update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().  Change the PG_arch_1 (aka PG_dcache_dirty) flag
from being per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-21-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: "James E.J. Bottomley" <James.Bottomley@HansenPartnership.com>
Cc: Helge Deller <deller@gmx.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

openrisc: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range() and flush_dcache_folio().
Change the PG_arch_1 (aka PG_dcache_dirty) flag from being per-page to
per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-20-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Jonas Bonn <jonas@southpole.se>
Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi>
Cc: Stafford Horne <shorne@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

nios2: implement the new page table range API

Add set_ptes(), update_mmu_cache_range(), flush_icache_pages() and
flush_dcache_folio().  Change the PG_arch_1 (aka PG_dcache_dirty) flag
from being per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-19-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Dinh Nguyen <dinguyen@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mips: implement the new page table range API

Rename _PFN_SHIFT to PFN_PTE_SHIFT.  Convert a few places
to call set_pte() instead of set_pte_at().  Add set_ptes(),
update_mmu_cache_range(), flush_icache_pages() and flush_dcache_folio().
Change the PG_arch_1 (aka PG_dcache_dirty) flag from being per-page
to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-18-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

microblaze: implement the new page table range API

Rename PFN_SHIFT_OFFSET to PTE_PFN_SHIFT.  Change the calling convention
for set_pte() to be the same as other architectures.  Add
update_mmu_cache_range(), flush_icache_pages() and flush_dcache_folio().

[arnd@arndb.de: mark flush_dcache_folio() inline]
Link: https://lkml.kernel.org/r/20230810141947.1236730-9-arnd@kernel.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-17-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Michal Simek <monstr@monstr.eu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

m68k: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_icache_pages() and
flush_dcache_folio().

Link: https://lkml.kernel.org/r/20230802151406.3735276-16-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

loongarch: implement the new page table range API

Add update_mmu_cache_range() and change _PFN_SHIFT to PFN_PTE_SHIFT.  It
would probably be more efficient to implement __update_tlb() by flushing
the entire folio instead of calling __update_tlb() N times, but I'll leave
that for someone who understands the architecture better.

Link: https://lkml.kernel.org/r/20230802151406.3735276-15-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: WANG Xuerui <kernel@xen0n.name>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

ia64: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range() and flush_dcache_folio().
Change the PG_arch_1 (aka PG_dcache_clean) flag from being per-page to
per-folio, which makes arch_dma_mark_clean() and mark_clean() a little
more exciting.

[willy@infradead.org: fix folio_size() handling]
Link: https://lkml.kernel.org/r/ZNPlOCe8F+nrzPxr@casper.infradead.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-14-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

hexagon: implement the new page table range API

Add PFN_PTE_SHIFT and update_mmu_cache_range().

Link: https://lkml.kernel.org/r/20230802151406.3735276-13-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Brian Cain <bcain@quicinc.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

csky: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range() and flush_dcache_folio().
Change the PG_dcache_clean flag from being per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-12-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Guo Ren <guoren@kernel.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arm64: implement the new page table range API

Add set_ptes(), update_mmu_cache_range() and flush_dcache_folio().  Change
the PG_dcache_clean flag from being per-page to per-folio.

Link: https://lkml.kernel.org/r/20230802151406.3735276-11-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arm: implement the new page table range API

Add set_ptes(), update_mmu_cache_range(), flush_dcache_folio() and
flush_icache_pages().  Change the PG_dcache_clear flag from being per-page
to per-folio which makes __dma_page_dev_to_cpu() a bit more exciting.
Also add flush_cache_pages(), even though this isn't used by generic code
(yet?)

[m.szyprowski@samsung.com: fix potential endless loop in __dma_page_dev_to_cpu()]
Link: https://lkml.kernel.org/r/20230809172737.3574190-1-m.szyprowski@samsung.com
[willy@infradead.org: fix folio conversion in __dma_page_dev_to_cpu()]
Link: https://lkml.kernel.org/r/20230823191852.1556561-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-10-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arc: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range(), flush_dcache_folio()
and flush_icache_pages().

Change the PG_dc_clean flag from being per-page to per-folio (which means
it cannot always be set as we don't know that all pages in this folio were
cleaned).  Enhance the internal flush routines to take the number of pages
to flush.

Link: https://lkml.kernel.org/r/20230802151406.3735276-9-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Vineet Gupta <vgupta@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

alpha: implement the new page table range API

Add PFN_PTE_SHIFT, update_mmu_cache_range() and flush_icache_pages().

Link: https://lkml.kernel.org/r/20230802151406.3735276-8-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Cc: Matt Turner <mattst88@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add default definition of set_ptes()

Most architectures can just define set_pte() and PFN_PTE_SHIFT to use this
definition.  It's also a handy spot to document the guarantees provided by
the MM.

Link: https://lkml.kernel.org/r/20230802151406.3735276-7-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Suggested-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: remove ARCH_IMPLEMENTS_FLUSH_DCACHE_FOLIO

Current best practice is to reuse the name of the function as a define to
indicate that the function is implemented by the architecture.

Link: https://lkml.kernel.org/r/20230802151406.3735276-6-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add folio_flush_mapping()

This is the folio equivalent of page_mapping_file(), but rename it to make
it clear that it's very different from page_file_mapping().
Theoretically, there's nothing flush-only about it, but there are no other
users today, and I doubt there will be; it's almost always more useful to
know the swapfile's mapping or the swapcache's mapping.

Link: https://lkml.kernel.org/r/20230802151406.3735276-5-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add generic flush_icache_pages() and documentation

flush_icache_page() is deprecated but not yet removed, so add a range
version of it.  Change the documentation to refer to
update_mmu_cache_range() instead of update_mmu_cache().

Link: https://lkml.kernel.org/r/20230802151406.3735276-4-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert page_table_check_pte_set() to page_table_check_ptes_set()

Tell the page table check how many PTEs & PFNs we want it to check.

Link: https://lkml.kernel.org/r/20230802151406.3735276-3-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Mike Rapoport (IBM) <rppt@kernel.org>
Acked-by: Pasha Tatashin <pasha.tatashin@soleen.com>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

minmax: add in_range() macro

Patch series "New page table range API", v6.

This patchset changes the API used by the MM to set up page table entries.
The four APIs are:

    set_ptes(mm, addr, ptep, pte, nr)
    update_mmu_cache_range(vma, addr, ptep, nr)
    flush_dcache_folio(folio)
    flush_icache_pages(vma, page, nr)

flush_dcache_folio() isn't technically new, but no architecture
implemented it, so I've done that for them.  The old APIs remain around
but are mostly implemented by calling the new interfaces.

The new APIs are based around setting up N page table entries at once.
The N entries belong to the same PMD, the same folio and the same VMA, so
ptep++ is a legitimate operation, and locking is taken care of for you.
Some architectures can do a better job of it than just a loop, but I have
hesitated to make too deep a change to architectures I don't understand
well.

One thing I have changed in every architecture is that PG_arch_1 is now a
per-folio bit instead of a per-page bit when used for dcache clean/dirty
tracking.  This was something that would have to happen eventually, and it
makes sense to do it now rather than iterate over every page involved in a
cache flush and figure out if it needs to happen.

The point of all this is better performance, and Fengwei Yin has measured
improvement on x86.  I suspect you'll see improvement on your architecture
too.  Try the new will-it-scale test mentioned here:
https://lore.kernel.org/linux-mm/20230206140639.538867-5-fengwei.yin@intel.com/
You'll need to run it on an XFS filesystem and have
CONFIG_TRANSPARENT_HUGEPAGE set.

This patchset is the basis for much of the anonymous large folio work
being done by Ryan, so it's received quite a lot of testing over the last
few months.

This patch (of 38):

Determine if a value lies within a range more efficiently (subtraction +
comparison vs two comparisons and an AND).  It also has useful (under some
circumstances) behaviour if the range exceeds the maximum value of the
type.  Convert all the conflicting definitions of in_range() within the
kernel; some can use the generic definition while others need their own
definition.

Link: https://lkml.kernel.org/r/20230802151406.3735276-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20230802151406.3735276-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: memcg: use rstat for non-hierarchical stats

Currently, memcg uses rstat to maintain aggregated hierarchical stats.
Counters are maintained for hierarchical stats at each memcg.  Rstat
tracks which cgroups have updates on which cpus to keep those counters
fresh on the read-side.

Non-hierarchical stats are currently not covered by rstat.  Their per-cpu
counters are summed up on every read, which is expensive.  The original
implementation did the same.  At some point before rstat, non-hierarchical
aggregated counters were introduced by commit a983b5ebee57 ("mm:
memcontrol: fix excessive complexity in memory.stat reporting").  However,
those counters were updated on the performance critical write-side, which
caused regressions, so they were later removed by commit 815744d75152
("mm: memcontrol: don't batch updates of local VM stats and events").  See
[1] for more detailed history.

Kernel versions in between a983b5ebee57 & 815744d75152 (a year and a half)
enjoyed cheap reads of non-hierarchical stats, specifically on cgroup v1.
When moving to more recent kernels, a performance regression for reading
non-hierarchical stats is observed.

Now that we have rstat, we know exactly which percpu counters have updates
for each stat.  We can maintain non-hierarchical counters again, making
reads much more efficient, without affecting the performance critical
write-side.  Hence, add non-hierarchical (i.e local) counters for the
stats, and extend rstat flushing to keep those up-to-date.

A caveat is that we now need a stats flush before reading
local/non-hierarchical stats through {memcg/lruvec}_page_state_local() or
memcg_events_local(), where we previously only needed a flush to read
hierarchical stats.  Most contexts reading non-hierarchical stats are
already doing a flush, add a flush to the only missing context in
count_shadow_nodes().

With this patch, reading memory.stat from 1000 memcgs is 3x faster on a
machine with 256 cpus on cgroup v1:

 # for i in $(seq 1000); do mkdir /sys/fs/cgroup/memory/cg$i; done
 # time cat /sys/fs/cgroup/memory/cg*/memory.stat > /dev/null
 real  0m0.125s
 user  0m0.005s
 sys  0m0.120s

After:
 real  0m0.032s
 user  0m0.005s
 sys  0m0.027s

To make sure there are no regressions on cgroup v2, I ran an artificial
reclaim/refault stress test [2] that creates (NR_CPUS * 2) cgroups,
assigns them limits, runs a worker process in each cgroup that allocates
tmpfs memory equal to quadruple the limit (to invoke reclaim
continuously), and then reads back the entire file (to invoke refaults).
All workers are run in parallel, and zram is used as a swapping backend.
Both reclaim and refault have conditional stats flushing.  I ran this on a
machine with 112 cpus, once on mm-unstable, and once on mm-unstable with
this patch reverted.

(1) A few runs without this patch:

 # time ./stress_reclaim_refault.sh
 real 0m9.949s
 user 0m0.496s
 sys 14m44.974s

 # time ./stress_reclaim_refault.sh
 real 0m10.049s
 user 0m0.486s
 sys 14m55.791s

 # time ./stress_reclaim_refault.sh
 real 0m9.984s
 user 0m0.481s
 sys 14m53.841s

(2) A few runs with this patch:

 # time ./stress_reclaim_refault.sh
 real 0m9.885s
 user 0m0.486s
 sys 14m48.753s

 # time ./stress_reclaim_refault.sh
 real 0m9.903s
 user 0m0.495s
 sys 14m48.339s

 # time ./stress_reclaim_refault.sh
 real 0m9.861s
 user 0m0.507s
 sys 14m49.317s

No regressions are observed with this patch. There is actually a very
slight improvement. If I have to guess, maybe it's because we avoid
the percpu loop in count_shadow_nodes() when calling
lruvec_page_state_local(), but I could not prove this using perf, it's
probably in the noise.

[1] https://lore.kernel.org/lkml/20230725201811.GA1231514@cmpxchg.org/
[2] https://lore.kernel.org/lkml/CAJD7tkb17x=qwoO37uxyYXLEUVp15BQKR+Xfh7Sg9Hx-wTQ_=w@mail.gmail.com/

Link: https://lkml.kernel.org/r/20230803185046.1385770-1-yosryahmed@google.com
Link: https://lkml.kernel.org/r/20230726153223.821757-2-yosryahmed@google.com
Signed-off-by: Yosry Ahmed <yosryahmed@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Roman Gushchin <roman.gushchin@linux.dev>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Muchun Song <muchun.song@linux.dev>
Cc: Shakeel Butt <shakeelb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: handle userfaults under VMA lock

Enable handle_userfault to operate under VMA lock by releasing VMA lock
instead of mmap_lock and retrying.  Note that FAULT_FLAG_RETRY_NOWAIT
should never be used when handling faults under per-VMA lock protection
because that would break the assumption that lock is dropped on retry.

[surenb@google.com: fix a lockdep issue in vma_assert_write_locked]
Link: https://lkml.kernel.org/r/20230712195652.969194-1-surenb@google.com
Link: https://lkml.kernel.org/r/20230630211957.1341547-7-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: handle swap page faults under per-VMA lock

When page fault is handled under per-VMA lock protection, all swap page
faults are retried with mmap_lock because folio_lock_or_retry has to drop
and reacquire mmap_lock if folio could not be immediately locked.  Follow
the same pattern as mmap_lock to drop per-VMA lock when waiting for folio
and retrying once folio is available.

With this obstacle removed, enable do_swap_page to operate under per-VMA
lock protection.  Drivers implementing ops->migrate_to_ram might still
rely on mmap_lock, therefore we have to fall back to mmap_lock in that
particular case.

Note that the only time do_swap_page calls synchronous swap_readpage is
when SWP_SYNCHRONOUS_IO is set, which is only set for
QUEUE_FLAG_SYNCHRONOUS devices: brd, zram and nvdimms (both btt and pmem).
Therefore we don't sleep in this path, and there's no need to drop the
mmap or per-VMA lock.

Link: https://lkml.kernel.org/r/20230630211957.1341547-6-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Tested-by: Alistair Popple <apopple@nvidia.com>
Reviewed-by: Alistair Popple <apopple@nvidia.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: change folio_lock_or_retry to use vm_fault directly

Change folio_lock_or_retry to accept vm_fault struct and return the
vm_fault_t directly.

Link: https://lkml.kernel.org/r/20230630211957.1341547-5-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Suggested-by: Matthew Wilcox <willy@infradead.org>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: drop per-VMA lock when returning VM_FAULT_RETRY or VM_FAULT_COMPLETED

handle_mm_fault returning VM_FAULT_RETRY or VM_FAULT_COMPLETED means
mmap_lock has been released.  However with per-VMA locks behavior is
different and the caller should still release it.  To make the rules
consistent for the caller, drop the per-VMA lock when returning
VM_FAULT_RETRY or VM_FAULT_COMPLETED.  Currently the only path returning
VM_FAULT_RETRY under per-VMA locks is do_swap_page and no path returns
VM_FAULT_COMPLETED for now.

[willy@infradead.org: fix riscv]
Link: https://lkml.kernel.org/r/CAJuCfpE6GWEx1rPBmNpUfoD5o-gNFz9-UFywzCE2PbEGBiVz7g@mail.gmail.com
Link: https://lkml.kernel.org/r/20230630211957.1341547-4-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: Peter Xu <peterx@redhat.com>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add missing VM_FAULT_RESULT_TRACE name for VM_FAULT_COMPLETED

VM_FAULT_RESULT_TRACE should contain an element for every vm_fault_reason
to be used as flag_array inside trace_print_flags_seq().  The element for
VM_FAULT_COMPLETED is missing, add it.

Link: https://lkml.kernel.org/r/20230630211957.1341547-3-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

swap: remove remnants of polling from read_swap_cache_async

Patch series "Per-VMA lock support for swap and userfaults", v7.

When per-VMA locks were introduced in [1] several types of page faults
would still fall back to mmap_lock to keep the patchset simple.  Among
them are swap and userfault pages.  The main reason for skipping those
cases was the fact that mmap_lock could be dropped while handling these
faults and that required additional logic to be implemented.  Implement
the mechanism to allow per-VMA locks to be dropped for these cases.

First, change handle_mm_fault to drop per-VMA locks when returning
VM_FAULT_RETRY or VM_FAULT_COMPLETED to be consistent with the way
mmap_lock is handled.  Then change folio_lock_or_retry to accept vm_fault
and return vm_fault_t which simplifies later patches.  Finally allow swap
and uffd page faults to be handled under per-VMA locks by dropping per-VMA
and retrying, the same way it's done under mmap_lock.  Naturally, once VMA
lock is dropped that VMA should be assumed unstable and can't be used.

This patch (of 6):

Commit [1] introduced IO polling support duding swapin to reduce swap read
latency for block devices that can be polled.  However later commit [2]
removed polling support.  Therefore it seems safe to remove do_poll
parameter in read_swap_cache_async and always call swap_readpage with
synchronous=false waiting for IO completion in folio_lock_or_retry.

[1] commit 23955622ff8d ("swap: add block io poll in swapin path")
[2] commit 9650b453a3d4 ("block: ignore RWF_HIPRI hint for sync dio")

Link: https://lkml.kernel.org/r/20230630211957.1341547-1-surenb@google.com
Link: https://lkml.kernel.org/r/20230630211957.1341547-2-surenb@google.com
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Suggested-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: "Huang, Ying" <ying.huang@intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Cc: Alistair Popple <apopple@nvidia.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christian Brauner <brauner@kernel.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Josef Bacik <josef@toxicpanda.com>
Cc: Laurent Dufour <ldufour@linux.ibm.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Lorenzo Stoakes <lstoakes@gmail.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Michel Lespinasse <michel@lespinasse.org>
Cc: Minchan Kim <minchan@google.com>
Cc: Pavel Tatashin <pasha.tatashin@soleen.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Punit Agrawal <punit.agrawal@bytedance.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Yu Zhao <yuzhao@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: memory-failure: fix potential page refcnt leak in memory_failure()

put_ref_page() is not called to drop extra refcnt when comes from madvise
in the case pfn is valid but pgmap is NULL leading to page refcnt leak.

Link: https://lkml.kernel.org/r/20230701072837.1994253-1-linmiaohe@huawei.com
Fixes: 1e8aaedb182d ("mm,memory_failure: always pin the page in madvise_inject_error")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Acked-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/memory.c: fix mismerge

Fix a build issue.

Link: https://lkml.kernel.org/r/ZNerqcNS4EBJA/2v@casper.infradead.org
Fixes: 4aaa60dad4d1 ("mm: allow per-VMA locks on file-backed VMAs")
Signed-off-by: Matthew Wilcox <willy@infradead.org>
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202308121909.XNYBtqNI-lkp@intel.com/
Cc: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/khugepaged: fix collapse_pte_mapped_thp() versus uffd

Jann Horn demonstrated how userfaultfd ioctl UFFDIO_COPY into a private
shmem mapping can add valid PTEs to page table collapse_pte_mapped_thp()
thought it had emptied: page lock on the huge page is enough to protect
against WP faults (which find the PTE has been cleared), but not enough to
protect against userfaultfd.  "BUG: Bad rss-counter state" followed.

retract_page_tables() protects against this by checking !vma->anon_vma;
but we know that MADV_COLLAPSE needs to be able to work on private shmem
mappings, even those with an anon_vma prepared for another part of the
mapping; and we know that MADV_COLLAPSE needs to work on shared shmem
mappings which are userfaultfd_armed().  Whether it needs to work on
private shmem mappings which are userfaultfd_armed(), I'm not so sure: but
assume that it does.

Just for this case, take the pmd_lock() two steps earlier: not because it
gives any protection against this case itself, but because ptlock nests
inside it, and it's the dropping of ptlock which let the bug in.  In other
cases, continue to minimize the pmd_lock() hold time.

Link: https://lkml.kernel.org/r/4d31abf5-56c0-9f3d-d12f-c9317936691@google.com
Fixes: 1043173eb5eb ("mm/khugepaged: collapse_pte_mapped_thp() with mmap_read_lock()")
Signed-off-by: Hugh Dickins <hughd@google.com>
Reported-by: Jann Horn <jannh@google.com>
Closes: https://lore.kernel.org/linux-mm/CAG48ez0FxiRC4d3VTu_a9h=rg5FW-kYD5Rg5xo_RDBM0LTTqZQ@mail.gmail.com/
Acked-by: Peter Xu <peterx@redhat.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Matthew Wilcox <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

hugetlb: clear flags in tail pages that will be freed individually

hugetlb manually creates and destroys compound pages.  As such it makes
assumptions about struct page layout.  Commit ebc1baf5c9b4 ("mm: free up a
word in the first tail page") breaks hugetlb.  The following will fix the
breakage.

Link: https://lkml.kernel.org/r/20230822231741.GC4509@monkey
Fixes: ebc1baf5c9b4 ("mm: free up a word in the first tail page")
Signed-off-by: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

merge mm-hotfixes-stable into mm-stable to pick up depended-upon changes

shmem: fix smaps BUG sleeping while atomic

smaps_pte_hole_lookup() is calling shmem_partial_swap_usage() with page
table lock held: but shmem_partial_swap_usage() does cond_resched_rcu() if
need_resched(): "BUG: sleeping function called from invalid context".

Since shmem_partial_swap_usage() is designed to count across a range, but
smaps_pte_hole_lookup() only calls it for a single page slot, just break
out of the loop on the last or only page, before checking need_resched().

Link: https://lkml.kernel.org/r/6fe3b3ec-abdf-332f-5c23-6a3b3a3b11a9@google.com
Fixes: 230100321518 ("mm/smaps: simplify shmem handling of pte holes")
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: <stable@vger.kernel.org> [5.16+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests: cachestat: catch failing fsync test on tmpfs

The cachestat kselftest runs a test on a normal file, which is created
temporarily in the current directory.  Among the tests it runs there is a
call to fsync(), which is expected to clean all dirty pages used by the
file.

However the tmpfs filesystem implements fsync() as noop_fsync(), so the
call will not even attempt to clean anything when this test file happens
to live on a tmpfs instance.  This happens in an initramfs, or when the
current directory is in /dev/shm or sometimes /tmp.

To avoid this test failing wrongly, use statfs() to check which filesystem
the test file lives on.  If that is "tmpfs", we skip the fsync() test.

Since the fsync test is only one part of the "normal file" test, we now
execute this twice, skipping the fsync part on the first call.  This way
only the second test, including the fsync part, would be skipped.

Link: https://lkml.kernel.org/r/20230821160534.3414911-3-andre.przywara@arm.com
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests: cachestat: test for cachestat availability

Patch series "selftests: cachestat: fix run on older kernels", v2.

I ran all kernel selftests on some test machine, and stumbled upon
cachestat failing (among others).  These patches fix the run on older
kernels and when the current directory is on a tmpfs instance.

This patch (of 2):

As cachestat is a new syscall, it won't be available on older kernels, for
instance those running on a development machine.  At the moment the test
reports all tests as "not ok" in this case.

Test for the cachestat syscall availability first, before doing further
tests, and bail out early with a TAP SKIP comment.

This also uses the opportunity to add the proper TAP headers, and add one
check for proper error handling (illegal file descriptor).

Link: https://lkml.kernel.org/r/20230821160534.3414911-1-andre.przywara@arm.com
Link: https://lkml.kernel.org/r/20230821160534.3414911-2-andre.przywara@arm.com
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Nhat Pham <nphamcs@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

maple_tree: disable mas_wr_append() when other readers are possible

The current implementation of append may cause duplicate data and/or
incorrect ranges to be returned to a reader during an update.  Although
this has not been reported or seen, disable the append write operation
while the tree is in rcu mode out of an abundance of caution.

During the analysis of the mas_next_slot() the following was
artificially created by separating the writer and reader code:

Writer:                                 reader:
mas_wr_append
    set end pivot
    updates end metata
    Detects write to last slot
    last slot write is to start of slot
    store current contents in slot
    overwrite old end pivot
                                        mas_next_slot():
                                                read end metadata
                                                read old end pivot
                                                return with incorrect range
    store new value

Alternatively:

Writer:                                 reader:
mas_wr_append
    set end pivot
    updates end metata
    Detects write to last slot
    last lost write to end of slot
    store value
                                        mas_next_slot():
                                                read end metadata
                                                read old end pivot
                                                read new end pivot
                                                return with incorrect range
    set old end pivot

There may be other accesses that are not safe since we are now updating
both metadata and pointers, so disabling append if there could be rcu
readers is the safest action.

Link: https://lkml.kernel.org/r/20230819004356.1454718-2-Liam.Howlett@oracle.com
Fixes: 54a611b60590 ("Maple Tree: add new data structure")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

madvise:madvise_free_pte_range(): don't use mapcount() against large folio for sharing check

Commit 98b211d6415f ("madvise: convert madvise_free_pte_range() to use a
folio") replaced the page_mapcount() with folio_mapcount() to check
whether the folio is shared by other mapping.

It's not correct for large folios. folio_mapcount() returns the total
mapcount of large folio which is not suitable to detect whether the folio
is shared.

Use folio_estimated_sharers() which returns a estimated number of shares.
That means it's not 100% correct. It should be OK for madvise case here.

User-visible effects is that the THP is skipped when user call madvise.
But the correct behavior is THP should be split and processed then.

NOTE: this change is a temporary fix to reduce the user-visible effects
before the long term fix from David is ready.

Link: https://lkml.kernel.org/r/20230808020917.2230692-4-fengwei.yin@intel.com
Fixes: 98b211d6415f ("madvise: convert madvise_free_pte_range() to use a folio")
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

madvise:madvise_free_huge_pmd(): don't use mapcount() against large folio for sharing check

Commit fc986a38b670 ("mm: huge_memory: convert madvise_free_huge_pmd to
use a folio") replaced the page_mapcount() with folio_mapcount() to check
whether the folio is shared by other mapping.

It's not correct for large folios. folio_mapcount() returns the total
mapcount of large folio which is not suitable to detect whether the folio
is shared.

Use folio_estimated_sharers() which returns a estimated number of shares.
That means it's not 100% correct. It should be OK for madvise case here.

User-visible effects is that the THP is skipped when user call madvise.
But the correct behavior is THP should be split and processed then.

NOTE: this change is a temporary fix to reduce the user-visible effects
before the long term fix from David is ready.

Link: https://lkml.kernel.org/r/20230808020917.2230692-3-fengwei.yin@intel.com
Fixes: fc986a38b670 ("mm: huge_memory: convert madvise_free_huge_pmd to use a folio")
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Cc: Yang Shi <shy828301@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

madvise:madvise_cold_or_pageout_pte_range(): don't use mapcount() against large folio for sharing check

Patch series "don't use mapcount() to check large folio sharing", v2.

In madvise_cold_or_pageout_pte_range() and madvise_free_pte_range(),
folio_mapcount() is used to check whether the folio is shared.  But it's
not correct as folio_mapcount() returns total mapcount of large folio.

Use folio_estimated_sharers() here as the estimated number is enough.

This patchset will fix the cases:
User space application call madvise() with MADV_FREE, MADV_COLD and
MADV_PAGEOUT for specific address range. There are THP mapped to the
range. Without the patchset, the THP is skipped. With the patch, the
THP will be split and handled accordingly.

David reported the cow self test skip some cases because of MADV_PAGEOUT
skip THP:
https://lore.kernel.org/linux-mm/9e92e42d-488f-47db-ac9d-75b24cd0d037@intel.com/T/#mbf0f2ec7fbe45da47526de1d7036183981691e81
and I confirmed this patchset make it work again.

This patch (of 3):

Commit 07e8c82b5eff ("madvise: convert madvise_cold_or_pageout_pte_range()
to use folios") replaced the page_mapcount() with folio_mapcount() to
check whether the folio is shared by other mapping.

It's not correct for large folio.  folio_mapcount() returns the total
mapcount of large folio which is not suitable to detect whether the folio
is shared.

Use folio_estimated_sharers() which returns a estimated number of shares.
That means it's not 100% correct.  It should be OK for madvise case here.

User-visible effects is that the THP is skipped when user call madvise.
But the correct behavior is THP should be split and processed then.

NOTE: this change is a temporary fix to reduce the user-visible effects
before the long term fix from David is ready.

Link: https://lkml.kernel.org/r/20230808020917.2230692-1-fengwei.yin@intel.com
Link: https://lkml.kernel.org/r/20230808020917.2230692-2-fengwei.yin@intel.com
Fixes: 07e8c82b5eff ("madvise: convert madvise_cold_or_pageout_pte_range() to use folios")
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Vishal Moola (Oracle) <vishal.moola@gmail.com>
Cc: Yang Shi <shy828301@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert split_huge_pages_pid() to use a folio

Replaces five calls to compound_head with one.

Link: https://lkml.kernel.org/r/20230816151201.3655946-14-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add tail private fields to struct folio

Because THP_SWAP uses page->private for each page, we must not use the
space which overlaps that field for anything which would conflict with
that.  We avoid the conflict on 32-bit systems by disallowing THP_SWAP on
32-bit.

Link: https://lkml.kernel.org/r/20230816151201.3655946-13-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: remove folio_test_transhuge()

This function is misleading; people think it means "Is this a THP", when
all it actually does is check whether this is a large folio.  Remove it;
the one remaining user should have been checking to see whether the folio
is PMD sized or not.

Link: https://lkml.kernel.org/r/20230816151201.3655946-12-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: free up a word in the first tail page

Store the folio order in the low byte of the flags word in the first tail
page.  This frees up the word that was being used to store the order and
dtor bytes previously.

Link: https://lkml.kernel.org/r/20230816151201.3655946-11-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: rearrange page flags

Move PG_writeback into bottom byte so that it can use PG_waiters in a
later patch.  Move PG_head into bottom byte as well to match with where
'order' is moving next.  PG_active and PG_workingset move into the second
byte to make room for them.

By putting PG_head in bit 6, we ensure that it is cleared by assigning the
folio order to the bottom byte of the first tail page (since the order
cannot be larger than 63).

Link: https://lkml.kernel.org/r/20230816151201.3655946-10-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: add large_rmappable page flag

Stored in the first tail page's flags, this flag replaces the destructor.
That removes the last of the destructors, so remove all references to
folio_dtor and compound_dtor.

Link: https://lkml.kernel.org/r/20230816151201.3655946-9-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: remove HUGETLB_PAGE_DTOR

We can use a bit in page[1].flags to indicate that this folio belongs to
hugetlb instead of using a value in page[1].dtors.  That lets
folio_test_hugetlb() become an inline function like it should be.  We can
also get rid of NULL_COMPOUND_DTOR.

Link: https://lkml.kernel.org/r/20230816151201.3655946-8-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: remove free_compound_page() and the compound_page_dtors array

The only remaining destructor is free_compound_page().  Inline it into
destroy_large_folio() and remove the array it used to live in.

Link: https://lkml.kernel.org/r/20230816151201.3655946-7-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert prep_transhuge_page() to folio_prep_large_rmappable()

Match folio_undo_large_rmappable(), and move the casting from page to
folio into the callers (which they were largely doing anyway).

Link: https://lkml.kernel.org/r/20230816151201.3655946-6-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert free_transhuge_folio() to folio_undo_large_rmappable()

Indirect calls are expensive, thanks to Spectre.  Test for
TRANSHUGE_PAGE_DTOR and destroy the folio appropriately.  Move the
free_compound_page() call into destroy_large_folio() to simplify later
patches.

Link: https://lkml.kernel.org/r/20230816151201.3655946-5-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: convert free_huge_page() to free_huge_folio()

Pass a folio instead of the head page to save a few instructions.  Update
the documentation, at least in English.

Link: https://lkml.kernel.org/r/20230816151201.3655946-4-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: call free_huge_page() directly

Indirect calls are expensive, thanks to Spectre.  Call free_huge_page()
directly if the folio belongs to hugetlb.

Link: https://lkml.kernel.org/r/20230816151201.3655946-3-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: David Hildenbrand <david@redhat.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

io_uring: stop calling free_compound_page()

Patch series "Remove _folio_dtor and _folio_order", v2.

This patch (of 13):

folio_put() is the standard way to write this, and it's not appreciably
slower.  This is an enabling patch for removing free_compound_page()
entirely.

Link: https://lkml.kernel.org/r/20230816151201.3655946-1-willy@infradead.org
Link: https://lkml.kernel.org/r/20230816151201.3655946-2-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Cc: Sidhartha Kumar <sidhartha.kumar@oracle.com>
Cc: Yanteng Si <siyanteng@loongson.cn>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftest/mm: ksm_functional_tests: Add PROT_NONE test

Let's test whether merging and unmerging in PROT_NONE areas works as
expected.

Pass a page protection to mmap_and_merge_range(), which will trigger
an mprotect() after writing to the pages, but before enabling merging.

Make sure that unsharing works as expected, by performing a ptrace write
(using /proc/self/mem) and by setting MADV_UNMERGEABLE.

Note that this implicitly tests that ptrace writes in an inaccessible
(PROT_NONE) mapping work as expected.

[david@redhat.com: use sizeof(i) in test_prot_none(), per Peter]
Link: https://lkml.kernel.org/r/e9cdb144-70c7-6596-2377-e675635c94e0@redhat.com
Link: https://lkml.kernel.org/r/20230803143208.383663-8-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: liubo <liubo254@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftest/mm: ksm_functional_tests: test in mmap_and_merge_range() if anything got merged

Let's extend mmap_and_merge_range() to test if anything in the current
process was merged. range_maps_duplicates() is too unreliable for that
use case, so instead look at KSM stats.

Trigger a complete unmerge first, to cleanup the stable tree and
stabilize accounting of merged pages.

Note that we're using /proc/self/ksm_merging_pages instead of
/proc/self/ksm_stat, because that one is available in more existing
kernels.

If /proc/self/ksm_merging_pages can't be opened, we can't perform any
checks and simply skip them.

We have to special-case the shared zeropage for now. But the only user
-- test_unmerge_zero_pages() -- performs its own merge checks.

Link: https://lkml.kernel.org/r/20230803143208.383663-7-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: liubo <liubo254@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

pgtable: improve pte_protnone() comment

Especially the "For PROT_NONE VMAs, the PTEs are not marked
_PAGE_PROTNONE" part is wrong: doing an mprotect(PROT_NONE) will end up
marking all PTEs on x86_64 as _PAGE_PROTNONE, making pte_protnone()
indicate "yes".

So let's improve the comment, so it's easier to grasp which semantics
pte_protnone() actually has.

Link: https://lkml.kernel.org/r/20230803143208.383663-6-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: liubo <liubo254@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/gup: don't implicitly set FOLL_HONOR_NUMA_FAULT

Commit 0b9d705297b2 ("mm: numa: Support NUMA hinting page faults from
gup/gup_fast") from 2012 documented as the primary reason why we would want
to handle NUMA hinting faults from GUP:

  KVM secondary MMU page faults will trigger the NUMA hinting page
  faults through gup_fast -> get_user_pages -> follow_page ->
  handle_mm_fault.

That is still the case today, and relevant KVM code has been converted to
manually set FOLL_HONOR_NUMA_FAULT. So let's stop setting
FOLL_HONOR_NUMA_FAULT for all GUP users and cross fingers that not that
many other ones that really require such handling for autonuma remain.

Possible interaction with MMU notifiers:

 Assume a driver obtains a page using get_user_pages() to map it into
 a secondary MMU, and uses the MMU notifier framework to get notified on
 changes.

 Assume get_user_pages() succeeded on a PROT_NONE-mapped page (because
 FOLL_HONOR_NUMA_FAULT is not set) in an accessible VMA and the page is
 mapped into a secondary MMU. Once user space would turn that mapping
 inaccessible using mprotect(PROT_NONE), the actual PTE in the page table
 might not change. If the MMU notifier would be smart and optimize for that
 case "why notify if the PTE didn't change", that could be problematic.

 At least change_pmd_range() with MMU_NOTIFY_PROTECTION_VMA for now does an
 unconditional mmu_notifier_invalidate_range_start() ->
 mmu_notifier_invalidate_range_end() and should be fine.

 Note that even if a PTE in an accessible VMA is pte_protnone(), the
 underlying page might be accessed by a secondary MMU that does not set
 FOLL_HONOR_NUMA_FAULT, and test_young() MMU notifiers would return "true".

Link: https://lkml.kernel.org/r/20230803143208.383663-5-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: liubo <liubo254@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

kvm: explicitly set FOLL_HONOR_NUMA_FAULT in hva_to_pfn_slow()

KVM is *the* case we know that really wants to honor NUMA hinting falls.
As we want to stop setting FOLL_HONOR_NUMA_FAULT implicitly, set
FOLL_HONOR_NUMA_FAULT whenever we might obtain pages on behalf of a VCPU
to map them into a secondary MMU, and add a comment why.

Do that unconditionally in hva_to_pfn_slow() when calling
get_user_pages_unlocked().

kvmppc_book3s_instantiate_page(), hva_to_pfn_fast() and
gfn_to_page_many_atomic() are similarly used to map pages into a
secondary MMU. However, FOLL_WRITE and get_user_page_fast_only() always
implicitly honor NUMA hinting faults -- as documented for
FOLL_HONOR_NUMA_FAULT -- so we can limit this change to a single location
for now.

Don't set it in check_user_page_hwpoison(), where we really only want to
check if the mapped page is HW-poisoned.

We won't set it for other KVM users of get_user_pages()/pin_user_pages()
* arch/powerpc/kvm/book3s_64_mmu_hv.c: not used to map pages into a
  secondary MMU.
* arch/powerpc/kvm/e500_mmu.c: only used on shared TLB pages with userspace
* arch/s390/kvm/*: s390x only supports a single NUMA node either way
* arch/x86/kvm/svm/sev.c: not used to map pages into a secondary MMU.

This is a preparation for making FOLL_HONOR_NUMA_FAULT no longer
implicitly be set by get_user_pages() and friends.

Link: https://lkml.kernel.org/r/20230803143208.383663-4-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: John Hubbard <jhubbard@nvidia.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: liubo <liubo254@huawei.com>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

merge mm-hotfixes-stable into mm-stable to pick up depended-upon changes

pagemap: remove wait_on_page_locked_killable()

There is no users of wait_on_page_locked_killable(), remove it.

Link: https://lkml.kernel.org/r/20230815030609.39313-1-wangkefeng.wang@huawei.com
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

Rename kmemleak_initialized to kmemleak_late_initialized

The old name is confusing because it implies the completion of earlier
kmemleak_init(), the new name update to kmemleak_late_initial represents
the completion of kmemleak_late_init().

No functional changes.

Link: https://lkml.kernel.org/r/20230815144128.3623103-3-xiaolei.wang@windriver.com
Signed-off-by: Xiaolei Wang <xiaolei.wang@windriver.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/kmemleak: use object_cache instead of kmemleak_initialized to check in set_track_prepare()

Patch series "mm/kmemleak: use object_cache instead of
kmemleak_initialized", v3.

Use object_cache instead of kmemleak_initialized to check in
set_track_prepare(), so that memory leaks after kmemleak_init() can be
recorded and Rename kmemleak_initialized to kmemleak_late_initialized

unreferenced object 0xc674ca80 (size 64):
 comm "swapper/0", pid 1, jiffies 4294938337 (age 204.880s)
 hex dump (first 32 bytes):
  80 55 75 c6 80 54 75 c6 00 55 75 c6 80 52 75 c6 .Uu..Tu..Uu..Ru.
  00 53 75 c6 00 00 00 00 00 00 00 00 00 00 00 00 .Su..........

This patch (of 2):

kmemleak_initialized is set in kmemleak_late_init(), which also means that
there is no call trace which object's memory leak is before
kmemleak_late_init(), so use object_cache instead of kmemleak_initialized
to check in set_track_prepare() to avoid no call trace records when there
is a memory leak in the code between kmemleak_init() and
kmemleak_late_init().

unreferenced object 0xc674ca80 (size 64):
 comm "swapper/0", pid 1, jiffies 4294938337 (age 204.880s)
 hex dump (first 32 bytes):
  80 55 75 c6 80 54 75 c6 00 55 75 c6 80 52 75 c6 .Uu..Tu..Uu..Ru.
  00 53 75 c6 00 00 00 00 00 00 00 00 00 00 00 00 .Su..........

Link: https://lkml.kernel.org/r/20230815144128.3623103-1-xiaolei.wang@windriver.com
Link: https://lkml.kernel.org/r/20230815144128.3623103-2-xiaolei.wang@windriver.com
Fixes: 56a61617dd22 ("mm: use stack_depot for recording kmemleak's backtrace")
Signed-off-by: Xiaolei Wang <xiaolei.wang@windriver.com>
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Zhaoyang Huang <zhaoyang.huang@unisoc.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/ksm: add pages scanned metric

ksm currently maintains several statistics, which let you determine how
successful KSM is at sharing pages.  However it does not contain a metric
to determine how much work it does.

This commit adds the pages scanned metric.  This allows the administrator
to determine how many pages have been scanned over a period of time.

Link: https://lkml.kernel.org/r/20230811193655.2518943-1-shr@devkernel.io
Signed-off-by: Stefan Roesch <shr@devkernel.io>
Acked-by: David Hildenbrand <david@redhat.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Rik van Riel <riel@surriel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm: allow fault_dirty_shared_page() to be called under the VMA lock

By making maybe_unlock_mmap_for_io() handle the VMA lock correctly, we
make fault_dirty_shared_page() safe to be called without the mmap lock
held.

Link: https://lkml.kernel.org/r/20230812002033.1002367-1-willy@infradead.org
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reported-by: David Hildenbrand <david@redhat.com>
Tested-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm/secretmem: use a folio in secretmem_fault()

Saves four implicit call to compound_head().

Link: https://lkml.kernel.org/r/20230812062612.3184990-1-zhangpeng362@huawei.com
Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>
Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Nanyong Sun <sunnanyong@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

writeback: remove unused delaration of bdi_async_bio_wq

It seems it was introduced by commit d3f77dfdc718 ("blkcg: implement
REQ_CGROUP_PUNT") unintentionally, but the definition does not exist,
remove it.

Link: https://lkml.kernel.org/r/20230812110128.482650-1-xiujianfeng@huaweicloud.com
Signed-off-by: Xiu Jianfeng <xiujianfeng@huawei.com>
Acked-by: Tejun Heo <tj@kernel.org>
Cc: Stefan Roesch <shr@devkernel.io>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm,thp: fix smaps THPeligible output alignment

Extract from current /proc/self/smaps output:

Swap:                  0 kB
SwapPss:               0 kB
Locked:                0 kB
THPeligible:    0
ProtectionKey:         0

That's not the alignment shown in Documentation/filesystems/proc.rst: it's
an ugly artifact from missing out the %8 other fields are using; but
there's even one selftest which expects it to look that way.  Hoping no
other smaps parsers depend on THPeligible to look so ugly, fix these.

Link: https://lkml.kernel.org/r/cfb81f7a-f448-5bc2-b0e1-8136fcd1dd8c@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm,thp: fix nodeN/meminfo output alignment

Add one more space to FileHugePages and FilePmdMapped, so the output is
aligned with other rows in /sys/devices/system/node/nodeN/meminfo.

Link: https://lkml.kernel.org/r/be861b50-a790-e041-bcb0-2a987dcfd1a@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mm,thp: no space after colon in Mem-Info fields

Patch series "mm,thp: fix sloppy text output".

Three independent trivial patches, fixing sloppy text output which has
annoyed me; but might risk surprising a parser, so any can be dropped.

This patch (of 3):

The SysRq-m or OOM Mem-Info dmesg showed (long lines containing) ...
shmem:NkB shmem_thp: NkB shmem_pmdmapped: NkB anon_thp: NkB ...

Delete the space after the colon after shmem_thp, shmem_pmdmapped,
anon_thp: as the shmem example shows, no other fields have a space after
the colon in this output.

Link: https://lkml.kernel.org/r/dc264fd6-40bb-6510-db36-9340a5f01d94@google.com
Link: https://lkml.kernel.org/r/c1edd7da-5493-c542-6feb-92452b4dab3b@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

riscv: include asm/cacheflush.h in asm/hugetlb.h

PG_dcache_clean is used in asm/hugetlb.h but defined in asm/cacheflush.h:
builds rely on an accident of that being included via linux/mempolicy.h,
but better include it directly (like arch/sh/include/asm/hugetlb.h does).

Link: https://lkml.kernel.org/r/84bd3b96-8dbe-51b1-d7d1-6e4f9d8937d8@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arm64: include asm/cacheflush.h in asm/hugetlb.h

PG_dcache_clean is used in asm/hugetlb.h but defined in asm/cacheflush.h:
builds rely on an accident of that being included via linux/mempolicy.h,
but better include it directly (like arch/sh/include/asm/hugetlb.h does).

Link: https://lkml.kernel.org/r/bd77cc1b-e83b-f276-9e27-c19e7c9119aa@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

arm: include asm/cacheflush.h in asm/hugetlb.h

Patch series "arch: include asm/cacheflush.h in asm/hugetlb.h".

Three architectures are using PG_dcache_clean in their asm/hugetlb.h,
but relying on accident to include the asm/cacheflush.h which defines it.

This patch (of 3):

PG_dcache_clean is used in asm/hugetlb.h but defined in asm/cacheflush.h:
builds rely on an accident of that being included via linux/mempolicy.h,
but better include it directly (like arch/sh/include/asm/hugetlb.h does).

Link: https://lkml.kernel.org/r/6d2acfa4-7f44-d3b4-b0a8-5495c5985e4c@google.com
Link: https://lkml.kernel.org/r/4b055d0-7b2e-72bf-9b9d-8f3f1cd312d0@google.com
Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

selftests: improve vm.memfd_noexec sysctl tests

This adds proper tests for the nesting functionality of vm.memfd_noexec as
well as some minor cleanups to spawn_*_thread().

Link: https://lkml.kernel.org/r/20230814-memfd-vm-noexec-uapi-fixes-v2-5-7ff9e3e10ba6@cyphar.com
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Jeff Xu <jeffxu@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <shuah@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memfd: replace ratcheting feature from vm.memfd_noexec with hierarchy

This sysctl has the very unusual behaviour of not allowing any user (even
CAP_SYS_ADMIN) to reduce the restriction setting, meaning that if you were
to set this sysctl to a more restrictive option in the host pidns you
would need to reboot your machine in order to reset it.

The justification given in [1] is that this is a security feature and thus
it should not be possible to disable.  Aside from the fact that we have
plenty of security-related sysctls that can be disabled after being
enabled (fs.protected_symlinks for instance), the protection provided by
the sysctl is to stop users from being able to create a binary and then
execute it.  A user with CAP_SYS_ADMIN can trivially do this without
memfd_create(2):

  % cat mount-memfd.c
  #include <fcntl.h>
  #include <string.h>
  #include <stdio.h>
  #include <stdlib.h>
  #include <unistd.h>
  #include <linux/mount.h>

  #define SHELLCODE "#!/bin/echo this file was executed from this totally private tmpfs:"

  int main(void)
  {
   int fsfd = fsopen("tmpfs", FSOPEN_CLOEXEC);
   assert(fsfd >= 0);
   assert(!fsconfig(fsfd, FSCONFIG_CMD_CREATE, NULL, NULL, 2));

   int dfd = fsmount(fsfd, FSMOUNT_CLOEXEC, 0);
   assert(dfd >= 0);

   int execfd = openat(dfd, "exe", O_CREAT | O_RDWR | O_CLOEXEC, 0782);
   assert(execfd >= 0);
   assert(write(execfd, SHELLCODE, strlen(SHELLCODE)) == strlen(SHELLCODE));
   assert(!close(execfd));

   char *execpath = NULL;
   char *argv[] = { "bad-exe", NULL }, *envp[] = { NULL };
   execfd = openat(dfd, "exe", O_PATH | O_CLOEXEC);
   assert(execfd >= 0);
   assert(asprintf(&execpath, "/proc/self/fd/%d", execfd) > 0);
   assert(!execve(execpath, argv, envp));
  }
  % ./mount-memfd
  this file was executed from this totally private tmpfs: /proc/self/fd/5
  %

Given that it is possible for CAP_SYS_ADMIN users to create executable
binaries without memfd_create(2) and without touching the host filesystem
(not to mention the many other things a CAP_SYS_ADMIN process would be
able to do that would be equivalent or worse), it seems strange to cause a
fair amount of headache to admins when there doesn't appear to be an
actual security benefit to blocking this.  There appear to be concerns
about confused-deputy-esque attacks[2] but a confused deputy that can
write to arbitrary sysctls is a bigger security issue than executable
memfds.

/* New API */

The primary requirement from the original author appears to be more based
on the need to be able to restrict an entire system in a hierarchical
manner[3], such that child namespaces cannot re-enable executable memfds.

So, implement that behaviour explicitly -- the vm.memfd_noexec scope is
evaluated up the pidns tree to &init_pid_ns and you have the most
restrictive value applied to you.  The new lower limit you can set
vm.memfd_noexec is whatever limit applies to your parent.

Note that a pidns will inherit a copy of the parent pidns's effective
vm.memfd_noexec setting at unshare() time.  This matches the existing
behaviour, and it also ensures that a pidns will never have its
vm.memfd_noexec setting *lowered* behind its back (but it will be raised
if the parent raises theirs).

/* Backwards Compatibility */

As the previous version of the sysctl didn't allow you to lower the
setting at all, there are no backwards compatibility issues with this
aspect of the change.

However it should be noted that now that the setting is completely
hierarchical.  Previously, a cloned pidns would just copy the current
pidns setting, meaning that if the parent's vm.memfd_noexec was changed it
wouldn't propoagate to existing pid namespaces.  Now, the restriction
applies recursively.  This is a uAPI change, however:

 * The sysctl is very new, having been merged in 6.3.
 * Several aspects of the sysctl were broken up until this patchset and
   the other patchset by Jeff Xu last month.

And thus it seems incredibly unlikely that any real users would run into
this issue. In the worst case, if this causes userspace isues we could
make it so that modifying the setting follows the hierarchical rules but
the restriction checking uses the cached copy.

[1]: https://lore.kernel.org/CABi2SkWnAgHK1i6iqSqPMYuNEhtHBkO8jUuCvmG3RmUB5TKHJw@mail.gmail.com/
[2]: https://lore.kernel.org/CALmYWFs_dNCzw_pW1yRAo4bGCPEtykroEQaowNULp7svwMLjOg@mail.gmail.com/
[3]: https://lore.kernel.org/CALmYWFuahdUF7cT4cm7_TGLqPanuHXJ-hVSfZt7vpTnc18DPrw@mail.gmail.com/

Link: https://lkml.kernel.org/r/20230814-memfd-vm-noexec-uapi-fixes-v2-4-7ff9e3e10ba6@cyphar.com
Fixes: 105ff5339f49 ("mm/memfd: add MFD_NOEXEC_SEAL and MFD_EXEC")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Jeff Xu <jeffxu@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memfd: improve userspace warnings for missing exec-related flags

In order to incentivise userspace to switch to passing MFD_EXEC and
MFD_NOEXEC_SEAL, we need to provide a warning on each attempt to call
memfd_create() without the new flags.  pr_warn_once() is not useful
because on most systems the one warning is burned up during the boot
process (on my system, systemd does this within the first second of boot)
and thus userspace will in practice never see the warnings to push them to
switch to the new flags.

The original patchset[1] used pr_warn_ratelimited(), however there were
concerns about the degree of spam in the kernel log[2,3].  The resulting
inability to detect every case was flagged as an issue at the time[4].

While we could come up with an alternative rate-limiting scheme such as
only outputting the message if vm.memfd_noexec has been modified, or only
outputting the message once for a given task, these alternatives have
downsides that don't make sense given how low-stakes a single kernel
warning message is.  Switching to pr_info_ratelimited() instead should be
fine -- it's possible some monitoring tool will be unhappy with a stream
of warning-level messages but there's already plenty of info-level message
spam in dmesg.

[1]: https://lore.kernel.org/20221215001205.51969-4-jeffxu@google.com/
[2]: https://lore.kernel.org/202212161233.85C9783FB@keescook/
[3]: https://lore.kernel.org/Y5yS8wCnuYGLHMj4@x1n/
[4]: https://lore.kernel.org/f185bb42-b29c-977e-312e-3349eea15383@linuxfoundation.org/

Link: https://lkml.kernel.org/r/20230814-memfd-vm-noexec-uapi-fixes-v2-3-7ff9e3e10ba6@cyphar.com
Fixes: 105ff5339f49 ("mm/memfd: add MFD_NOEXEC_SEAL and MFD_EXEC")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

memfd: do not -EACCES old memfd_create() users with vm.memfd_noexec=2

Given the difficulty of auditing all of userspace to figure out whether
every memfd_create() user has switched to passing MFD_EXEC and
MFD_NOEXEC_SEAL flags, it seems far less distruptive to make it possible
for older programs that don't make use of executable memfds to run under
vm.memfd_noexec=2.  Otherwise, a small dependency change can result in
spurious errors.  For programs that don't use executable memfds, passing
MFD_NOEXEC_SEAL is functionally a no-op and thus having the same

In addition, every failure under vm.memfd_noexec=2 needs to print to the
kernel log so that userspace can figure out where the error came from.
The concerns about pr_warn_ratelimited() spam that caused the switch to
pr_warn_once()[1,2] do not apply to the vm.memfd_noexec=2 case.

This is a user-visible API change, but as it allows programs to do
something that would be blocked before, and the sysctl itself was broken
and recently released, it seems unlikely this will cause any issues.

[1]: https://lore.kernel.org/Y5yS8wCnuYGLHMj4@x1n/
[2]: https://lore.kernel.org/202212161233.85C9783FB@keescook/

Link: https://lkml.kernel.org/r/20230814-memfd-vm-noexec-uapi-fixes-v2-2-7ff9e3e10ba6@cyphar.com
Fixes: 105ff5339f49 ("mm/memfd: add MFD_NOEXEC_SEAL and MFD_EXEC")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Cc: Dominique Martinet <asmadeus@codewreck.org>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Daniel Verkamp <dverkamp@chromium.org>
Cc: Jeff Xu <jeffxu@google.com>
Cc: Kees Cook <keescook@chromium.org>
Cc: Shuah Khan <shuah@kernel.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>