This is the follow-up on [1]:
[PATCH v2 0/8] mm: COW fixes part 3: reliable GUP R/W FOLL_GET of
anonymous pages
After we implemented __HAVE_ARCH_PTE_SWP_EXCLUSIVE on most prominent
enterprise architectures, implement __HAVE_ARCH_PTE_SWP_EXCLUSIVE on all
remaining architectures that support swap PTEs.
This makes sure that exclusive anonymous pages will stay exclusive, even
after they were swapped out -- for example, making GUP R/W FOLL_GET of
anonymous pages reliable. Details can be found in [1].
This primarily fixes remaining known O_DIRECT memory corruptions that can
happen on concurrent swapout, whereby we can lose DMA reads to a page
(modifying the user page by writing to it).
To verify, there are two test cases (requiring swap space, obviously):
(1) The O_DIRECT+swapout test case [2] from Andrea. This test case tries
triggering a race condition.
(2) My vmsplice() test case [3] that tries to detect if the exclusive
marker was lost during swapout, not relying on a race condition.
For example, on 32bit x86 (with and without PAE), my test case fails
without these patches:
$ ./test_swp_exclusive
FAIL: page was replaced during COW
But succeeds with these patches:
$ ./test_swp_exclusive
PASS: page was not replaced during COW
Why implement __HAVE_ARCH_PTE_SWP_EXCLUSIVE for all architectures, even
the ones where swap support might be in a questionable state? This is the
first step towards removing "readable_exclusive" migration entries, and
instead using pte_swp_exclusive() also with (readable) migration entries
instead (as suggested by Peter). The only missing piece for that is
supporting pmd_swp_exclusive() on relevant architectures with THP
migration support.
As all relevant architectures now implement __HAVE_ARCH_PTE_SWP_EXCLUSIVE,,
we can drop __HAVE_ARCH_PTE_SWP_EXCLUSIVE in the last patch.
RFC because some of the swap PTE layouts are really tricky and I really
need some feedback related to deciphering these layouts and "using yet
unused PTE bits in swap PTEs". I tried cross-compiling all relevant setups
(phew, I might only miss some power/nohash variants), but only tested on
x86 so far.
CCing arch maintainers only on this cover letter and on the respective
patch(es).
[1] https://lkml.kernel.org/r/20220329164329.208407-1-david@xxxxxxxxxx
[2] https://gitlab.com/aarcange/kernel-testcases-for-v5.11/-/blob/main/page_count_do_wp_page-swap.c
[3] https://gitlab.com/davidhildenbrand/scratchspace/-/blob/main/test_swp_exclusive.c
David Hildenbrand (26):
mm/debug_vm_pgtable: more pte_swp_exclusive() sanity checks
alpha/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
arc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
arm/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
csky/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
hexagon/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
ia64/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
loongarch/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
m68k/mm: remove dummy __swp definitions for nommu
m68k/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
microblaze/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
mips/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
nios2/mm: refactor swap PTE layout
nios2/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
openrisc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
parisc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
powerpc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE on 32bit book3s
powerpc/nohash/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
riscv/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
sh/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
sparc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE on 32bit
sparc/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE on 64bit
um/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
x86/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE also on 32bit
xtensa/mm: support __HAVE_ARCH_PTE_SWP_EXCLUSIVE
mm: remove __HAVE_ARCH_PTE_SWP_EXCLUSIVE
arch/alpha/include/asm/pgtable.h | 40 ++++++++-
arch/arc/include/asm/pgtable-bits-arcv2.h | 26 +++++-
arch/arm/include/asm/pgtable-2level.h | 3 +
arch/arm/include/asm/pgtable-3level.h | 3 +
arch/arm/include/asm/pgtable.h | 34 ++++++--
arch/arm64/include/asm/pgtable.h | 1 -
arch/csky/abiv1/inc/abi/pgtable-bits.h | 13 ++-
arch/csky/abiv2/inc/abi/pgtable-bits.h | 19 ++--
arch/csky/include/asm/pgtable.h | 17 ++++
arch/hexagon/include/asm/pgtable.h | 36 ++++++--
arch/ia64/include/asm/pgtable.h | 31 ++++++-
arch/loongarch/include/asm/pgtable-bits.h | 4 +
arch/loongarch/include/asm/pgtable.h | 38 +++++++-
arch/m68k/include/asm/mcf_pgtable.h | 35 +++++++-
arch/m68k/include/asm/motorola_pgtable.h | 37 +++++++-
arch/m68k/include/asm/pgtable_no.h | 6 --
arch/m68k/include/asm/sun3_pgtable.h | 38 +++++++-
arch/microblaze/include/asm/pgtable.h | 44 +++++++---
arch/mips/include/asm/pgtable-32.h | 86 ++++++++++++++++---
arch/mips/include/asm/pgtable-64.h | 23 ++++-
arch/mips/include/asm/pgtable.h | 35 ++++++++
arch/nios2/include/asm/pgtable-bits.h | 3 +
arch/nios2/include/asm/pgtable.h | 37 ++++++--
arch/openrisc/include/asm/pgtable.h | 40 +++++++--
arch/parisc/include/asm/pgtable.h | 40 ++++++++-
arch/powerpc/include/asm/book3s/32/pgtable.h | 37 ++++++--
arch/powerpc/include/asm/book3s/64/pgtable.h | 1 -
arch/powerpc/include/asm/nohash/32/pgtable.h | 22 +++--
arch/powerpc/include/asm/nohash/32/pte-40x.h | 6 +-
arch/powerpc/include/asm/nohash/32/pte-44x.h | 18 +---
arch/powerpc/include/asm/nohash/32/pte-85xx.h | 4 +-
arch/powerpc/include/asm/nohash/64/pgtable.h | 24 +++++-
arch/powerpc/include/asm/nohash/pgtable.h | 15 ++++
arch/powerpc/include/asm/nohash/pte-e500.h | 1 -
arch/riscv/include/asm/pgtable-bits.h | 3 +
arch/riscv/include/asm/pgtable.h | 28 ++++--
arch/s390/include/asm/pgtable.h | 1 -
arch/sh/include/asm/pgtable_32.h | 53 +++++++++---
arch/sparc/include/asm/pgtable_32.h | 26 +++++-
arch/sparc/include/asm/pgtable_64.h | 37 +++++++-
arch/sparc/include/asm/pgtsrmmu.h | 14 +--
arch/um/include/asm/pgtable.h | 36 +++++++-
arch/x86/include/asm/pgtable-2level.h | 26 ++++--
arch/x86/include/asm/pgtable-3level.h | 26 +++++-
arch/x86/include/asm/pgtable.h | 3 -
arch/xtensa/include/asm/pgtable.h | 31 +++++--
include/linux/pgtable.h | 29 -------
mm/debug_vm_pgtable.c | 25 +++++-
mm/memory.c | 4 -
mm/rmap.c | 11 ---
50 files changed, 943 insertions(+), 227 deletions(-)
--
2.38.1
