)]}'
{
  "commit": "18fbf5151d2c0bfe433c7428eef03cabf5fdb2fa",
  "tree": "6e151850bf46f900bbf9dcc682e0c75d91924865",
  "parents": [
    "5e6ff28676dd92a608eb00eeb8d1319ad34024dc",
    "0685630fdccb62dcb0e3f44525a40578da5f6dc8"
  ],
  "author": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Thu Aug 27 09:17:06 2026 -0700"
  },
  "committer": {
    "name": "Linus Torvalds",
    "email": "torvalds@linux-foundation.org",
    "time": "Thu Aug 27 09:17:06 2026 -0700"
  },
  "message": "Merge tag \u0027mm-stable-2026-08-26-15-22\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm\n\nPull more MM updates from Andrew Morton:\n\n - \"mm/rmap: index MAP_PRIVATE file-backed folios by anonymous pgoff\"\n   (Lorenzo Stoakes)\n\n   Index MAP_PRIVATE file-backed folios by their anonymous page offset\n   to resolve confusion around reverse mapping for zeroed and CoW\u0027d\n   file-backed memory.\n\n   Use this new VMA anonymous page offset tracking to eliminate index\n   conflicts and lay the foundation for scalable CoW performance\n   improvements.\n\n - \"promote mapped executable folios after first usage for MGLRU\"\n   (Baolin Wang)\n\n   Make MGLRU\u0027s protection of mapped executable file folios more\n   reliable. Follow the classical LRU\u0027s logic, promoting mapped\n   executable file folios after their first usage to give executable\n   code a better chance to stay in memory and improve workload\n   performance.\n\n - \"mm: vmscan: fix node reclaim ignoring swappiness parameter\" (Ridong\n   Chen)\n\n   Fix per-node proactive reclaim interface\u0027s ignoring the swappiness\n   parameter when CONFIG_MEMCG is disabled by consolidating\n   sc_swappiness() into a single function that checks\n   proactive_swappiness regardless of kernel configuration.\n\n - \"mm/vmscan: reduce lru_lock contention via vmstat-derived\n   scan-balance cost\" (Usama Arif)\n\n   Reduce lru_lock contention in the reclaim path by deriving\n   scan-balance costs from vmstat counters rather than lock-acquired\n   producer updates.\n\n   Read and decay these cost signals on the reclaim side under a\n   dedicated per-lruvec lock, reducing total LRU lock wait time by over\n   60% without impacting scan throughput.\n\n - \"zram: fix zram issues reported by sashiko\" (Sergey Senozhatsky)\n\n   Fix two low-risk zram bugs which Sashiko spotted in drive-by review.\n\n - \"Honor XA_FLAGS_ACCOUNT in xas_split_alloc() and charge to folio\u0027s\n   memcg\" (Zi Yan)\n\n   Fix xas_split_alloc() by enabling target folio memcg charging during\n   splits and adding the missing __GFP_ACCOUNT flag for proper XArray\n   node memory accounting.\n\n - \"selftests/mm: use pattern matching in .gitignore\" (Pratyush Mallick)\n\n   Replace hardcoded binary names in selftests/mm/.gitignore with a\n   generic pattern-matching rule to automatically ignore generated test\n   files and avoid manual updates when adding new tests.\n\n - \"mm/page_ext: remove pgdat_page_ext_init()\" (Sang-Heon Jeon)\n\n   Make the incompatibility between FLATMEM and NUMA explicit in\n   mm/Kconfig and remove the unused pgdat_page_ext_init() function.\n\n - \"zram: fix zstd error paths and add parameter validation\" (Haoqin\n   Huang)\n\n   Clean up zram compression backends by removing redundant error\n   cleanup, adding parameter and dictionary validation, auto-prefixing\n   algorithm error logs, and resetting parameters prior to\n   reinitialization.\n\n - \"zram: fix stale scan bounds after reinitialization\" (Longlong Xia)\n\n   Prevent out-of-bounds slot accesses during concurrent zram resets by\n   moving table scan bound calculations under dev_lock in\n   writeback_store() and read_block_state().\n\n - \"add anon mTHP collapse test cases\" (Baolin Wang)\n\n   Extend selftests helper functions to support arbitrary page orders\n   and add new test cases and options for mTHP collapse in khugepaged.\n\n - \"selftests/mm: Handle unsupported and transient test conditions\"\n   (Muhammad Usama Anjum)\n\n   Update MM selftests to report a SKIP status instead of a failure when\n   required kernel or filesystem features are unsupported, while adding\n   retry logic for transient page migration errors.\n\n - \"mm/zswap: Fixes and improves the zswap shrink\" (Hao Jia)\n\n   Fix the missing zswap global shrinker when CONFIG_MEMCG is disabled\n   and extend shrink_memcg() to support batch writeback for improved\n   writeback efficiency.\n\n - \"alloc_tag: introduce IOCTL-based filtering for MAP\" (Suren\n   Baghdasaryan)\n\n   Introduce an IOCTL-based binary interface for memory allocation\n   profiling that enables kernel-side filtering before per-CPU counter\n   aggregation.\n\n   This eliminates the text-parsing overhead of /proc/allocinfo and\n   provides up to a 20x speedup by transferring only filtered allocation\n   data to userspace.\n\n - \"better block swap batching and a different take on swap_ops v5\"\n   (Christoph Hellwig)\n\n   Refactor block swap I/O to use swap_iocb for batching instead of\n   single-bio requests and rebase the swap_ops interface, achieving\n   faster swap throughput during kernel builds.\n\n - \"mm: kmemleak: reduce transient false positives by confirming leaks\"\n   (Catalin Marinas)\n\n   Reduce false-positive kmemleak reports by combining two kmemleak\n   enhancements that add a second confirmation scan and a configurable\n   minimum unreferenced scan count module parameter.\n\n - \"mm: kmemleak: default min_unref_scans to 2 for verbose kernels\"\n   (Breno Leitao)\n\n   Auto-scanning kernels can generate false-positive memory leak reports\n   on single scans, so this patch defaults min_unref_scans to 2 when\n   CONFIG_DEBUG_KMEMLEAK_VERBOSE is enabled to require a second\n   confirming scan.\n\n - \"swap_ops updates\" (Christoph Hellwig)\n\n   Batching I/O for synchronous swap devices causes performance\n   regressions and filesystem-based swap suffers from double-indirection\n   overhead. This series resolves both issues by reintroducing per-folio\n   writes for synchronous swap and allowing filesystems to directly\n   export their own swap_ops.\n\n - \"mm/khugepaged: several cleanups\" (Nico Pache)\n\n   khugepaged accumulated redundant state-checking patterns and outdated\n   comments following mTHP integration. Introduce dedicated helpers for\n   PTE validation and event counting while refreshing the internal\n   documentation.\n\n - \"maple_tree: lock checking and clean ups\" (Liam Howlett)\n\n   Syzbot reports incorrectly blame memory management exit paths for\n   locking bugs, maple tree erase operations risk allocation failures\n   without gfp flags and internal documentation lacks clarity.\n\n   Improve lock error detection, update docs, fix race and allocation\n   edge cases and optimize erase allocations using a fallback to\n   GFP_KERNEL | GFP_NOFAIL.\n\n* tag \u0027mm-stable-2026-08-26-15-22\u0027 of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (172 commits)\n  selftests/proc: make proc-maps-race work with READ_IMPLIES_EXEC\n  memcg: move LRU size accounting on reparenting instead of copying it\n  mm/vmscan: fix comment logic in balance_pgdat\n  maple_tree: add helper mas_make_walkable()\n  maple_tree: avoid extra gap calculation\n  maple_tree: fix argument name in header\n  maple_tree: change two GFP flags in tests\n  maple_tree: document erase and allocations better\n  maple_tree: avoid mas_erase() and mtree_erase() failures\n  maple_tree: document that erase may use GFP_KERNEL for allocations\n  maple_tree: catch race in mas_alloc_cyclic()\n  maple_tree: add bulk parent set helper\n  maple_tree: micro optimisation of mas_wr_store_type()\n  maple_tree: optimise mas_wr_node_store() when not in rcu mode\n  maple_tree: use prefetched value in mas_wr_store_type()\n  maple_tree: clarify comments on mas_nomem()\n  maple_tree: drop MAPLE_ALLOC_SLOTS\n  maple_tree: drop dead code from mas_extend_spanning_null()\n  maple_tree: documentation fix\n  maple_tree: add write lock checking with lockdep sequence numbers\n  ...\n",
  "tree_diff": []
}