Linux-stable-mirror February 2023

linux-stable-mirror@lists.linaro.org

353 participants
1253 discussions

[merged mm-hotfixes-stable] mm-multi-gen-lru-fix-crash-during-cgroup-migration.patch removed from -mm tree

by Andrew Morton

The quilt patch titled Subject: mm: multi-gen LRU: fix crash during cgroup migration has been removed from the -mm tree. Its filename was mm-multi-gen-lru-fix-crash-during-cgroup-migration.patch This patch was dropped because it was merged into the mm-hotfixes-stable branch of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm ------------------------------------------------------ From: Yu Zhao <yuzhao(a)google.com> Subject: mm: multi-gen LRU: fix crash during cgroup migration Date: Sun, 15 Jan 2023 20:44:05 -0700 lru_gen_migrate_mm() assumes lru_gen_add_mm() runs prior to itself. This isn't true for the following scenario: CPU 1 CPU 2 clone() cgroup_can_fork() cgroup_procs_write() cgroup_post_fork() task_lock() lru_gen_migrate_mm() task_unlock() task_lock() lru_gen_add_mm() task_unlock() And when the above happens, kernel crashes because of linked list corruption (mm_struct->lru_gen.list). Link: https://lore.kernel.org/r/20230115134651.30028-1-msizanoen@qtmlabs.xyz/ Link: https://lkml.kernel.org/r/20230116034405.2960276-1-yuzhao@google.com Fixes: bd74fdaea146 ("mm: multi-gen LRU: support page table walks") Signed-off-by: Yu Zhao <yuzhao(a)google.com> Reported-by: msizanoen <msizanoen(a)qtmlabs.xyz> Tested-by: msizanoen <msizanoen(a)qtmlabs.xyz> Cc: <stable(a)vger.kernel.org> [6.1+] Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/vmscan.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) --- a/mm/vmscan.c~mm-multi-gen-lru-fix-crash-during-cgroup-migration +++ a/mm/vmscan.c @@ -3323,13 +3323,16 @@ void lru_gen_migrate_mm(struct mm_struct if (mem_cgroup_disabled()) return; + /* migration can happen before addition */ + if (!mm->lru_gen.memcg) + return; + rcu_read_lock(); memcg = mem_cgroup_from_task(task); rcu_read_unlock(); if (memcg == mm->lru_gen.memcg) return; - VM_WARN_ON_ONCE(!mm->lru_gen.memcg); VM_WARN_ON_ONCE(list_empty(&mm->lru_gen.list)); lru_gen_del_mm(mm); _ Patches currently in -mm which might be from yuzhao(a)google.com are

2 years

[merged mm-hotfixes-stable] mm-khugepaged-fix-anon_vma-race.patch removed from -mm tree

by Andrew Morton

The quilt patch titled Subject: mm/khugepaged: fix ->anon_vma race has been removed from the -mm tree. Its filename was mm-khugepaged-fix-anon_vma-race.patch This patch was dropped because it was merged into the mm-hotfixes-stable branch of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm ------------------------------------------------------ From: Jann Horn <jannh(a)google.com> Subject: mm/khugepaged: fix ->anon_vma race Date: Wed, 11 Jan 2023 14:33:51 +0100 If an ->anon_vma is attached to the VMA, collapse_and_free_pmd() requires it to be locked. Page table traversal is allowed under any one of the mmap lock, the anon_vma lock (if the VMA is associated with an anon_vma), and the mapping lock (if the VMA is associated with a mapping); and so to be able to remove page tables, we must hold all three of them. retract_page_tables() bails out if an ->anon_vma is attached, but does this check before holding the mmap lock (as the comment above the check explains). If we racily merged an existing ->anon_vma (shared with a child process) from a neighboring VMA, subsequent rmap traversals on pages belonging to the child will be able to see the page tables that we are concurrently removing while assuming that nothing else can access them. Repeat the ->anon_vma check once we hold the mmap lock to ensure that there really is no concurrent page table access. Hitting this bug causes a lockdep warning in collapse_and_free_pmd(), in the line "lockdep_assert_held_write(&vma->anon_vma->root->rwsem)". It can also lead to use-after-free access. Link: https://lore.kernel.org/linux-mm/CAG48ez3434wZBKFFbdx4M9j6eUwSUVPd4dxhzW_k_… Link: https://lkml.kernel.org/r/20230111133351.807024-1-jannh@google.com Fixes: f3f0e1d2150b ("khugepaged: add support of collapse for tmpfs/shmem pages") Signed-off-by: Jann Horn <jannh(a)google.com> Reported-by: Zach O'Keefe <zokeefe(a)google.com> Acked-by: Kirill A. Shutemov <kirill.shutemov(a)intel.linux.com> Reviewed-by: Yang Shi <shy828301(a)gmail.com> Cc: David Hildenbrand <david(a)redhat.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/khugepaged.c | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) --- a/mm/khugepaged.c~mm-khugepaged-fix-anon_vma-race +++ a/mm/khugepaged.c @@ -1642,7 +1642,7 @@ static int retract_page_tables(struct ad * has higher cost too. It would also probably require locking * the anon_vma. */ - if (vma->anon_vma) { + if (READ_ONCE(vma->anon_vma)) { result = SCAN_PAGE_ANON; goto next; } @@ -1671,6 +1671,18 @@ static int retract_page_tables(struct ad if ((cc->is_khugepaged || is_target) && mmap_write_trylock(mm)) { /* + * Re-check whether we have an ->anon_vma, because + * collapse_and_free_pmd() requires that either no + * ->anon_vma exists or the anon_vma is locked. + * We already checked ->anon_vma above, but that check + * is racy because ->anon_vma can be populated under the + * mmap lock in read mode. + */ + if (vma->anon_vma) { + result = SCAN_PAGE_ANON; + goto unlock_next; + } + /* * When a vma is registered with uffd-wp, we can't * recycle the pmd pgtable because there can be pte * markers installed. Skip it only, so the rest mm/vma _ Patches currently in -mm which might be from jannh(a)google.com are

2 years

[merged mm-hotfixes-stable] maple_tree-fix-mas_empty_area_rev-lower-bound-validation.patch removed from -mm tree

by Andrew Morton

The quilt patch titled Subject: maple_tree: fix mas_empty_area_rev() lower bound validation has been removed from the -mm tree. Its filename was maple_tree-fix-mas_empty_area_rev-lower-bound-validation.patch This patch was dropped because it was merged into the mm-hotfixes-stable branch of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm ------------------------------------------------------ From: Liam Howlett <liam.howlett(a)oracle.com> Subject: maple_tree: fix mas_empty_area_rev() lower bound validation Date: Wed, 11 Jan 2023 20:02:07 +0000 mas_empty_area_rev() was not correctly validating the start of a gap against the lower limit. This could lead to the range starting lower than the requested minimum. Fix the issue by better validating a gap once one is found. This commit also adds tests to the maple tree test suite for this issue and tests the mas_empty_area() function for similar bound checking. Link: https://lkml.kernel.org/r/20230111200136.1851322-1-Liam.Howlett@oracle.com Link: https://bugzilla.kernel.org/show_bug.cgi?id=216911 Fixes: 54a611b60590 ("Maple Tree: add new data structure") Signed-off-by: Liam R. Howlett <Liam.Howlett(a)oracle.com> Reported-by: <amanieu(a)gmail.com> Link: https://lore.kernel.org/linux-mm/0b9f5425-08d4-8013-aa4c-e620c3b10bb2@leemh… Tested-by: Holger Hoffsttte <holger(a)applied-asynchrony.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- lib/maple_tree.c | 17 +++---- lib/test_maple_tree.c | 89 ++++++++++++++++++++++++++++++++++++++++ 2 files changed, 97 insertions(+), 9 deletions(-) --- a/lib/maple_tree.c~maple_tree-fix-mas_empty_area_rev-lower-bound-validation +++ a/lib/maple_tree.c @@ -4887,7 +4887,7 @@ static bool mas_rev_awalk(struct ma_stat unsigned long *pivots, *gaps; void __rcu **slots; unsigned long gap = 0; - unsigned long max, min, index; + unsigned long max, min; unsigned char offset; if (unlikely(mas_is_err(mas))) @@ -4909,8 +4909,7 @@ static bool mas_rev_awalk(struct ma_stat min = mas_safe_min(mas, pivots, --offset); max = mas_safe_pivot(mas, pivots, offset, type); - index = mas->index; - while (index <= max) { + while (mas->index <= max) { gap = 0; if (gaps) gap = gaps[offset]; @@ -4941,10 +4940,8 @@ static bool mas_rev_awalk(struct ma_stat min = mas_safe_min(mas, pivots, offset); } - if (unlikely(index > max)) { - mas_set_err(mas, -EBUSY); - return false; - } + if (unlikely((mas->index > max) || (size - 1 > max - mas->index))) + goto no_space; if (unlikely(ma_is_leaf(type))) { mas->offset = offset; @@ -4961,9 +4958,11 @@ static bool mas_rev_awalk(struct ma_stat return false; ascend: - if (mte_is_root(mas->node)) - mas_set_err(mas, -EBUSY); + if (!mte_is_root(mas->node)) + return false; +no_space: + mas_set_err(mas, -EBUSY); return false; } --- a/lib/test_maple_tree.c~maple_tree-fix-mas_empty_area_rev-lower-bound-validation +++ a/lib/test_maple_tree.c @@ -2517,6 +2517,91 @@ static noinline void check_bnode_min_spa mt_set_non_kernel(0); } +static noinline void check_empty_area_window(struct maple_tree *mt) +{ + unsigned long i, nr_entries = 20; + MA_STATE(mas, mt, 0, 0); + + for (i = 1; i <= nr_entries; i++) + mtree_store_range(mt, i*10, i*10 + 9, + xa_mk_value(i), GFP_KERNEL); + + /* Create another hole besides the one at 0 */ + mtree_store_range(mt, 160, 169, NULL, GFP_KERNEL); + + /* Check lower bounds that don't fit */ + rcu_read_lock(); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 5, 90, 10) != -EBUSY); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 6, 90, 5) != -EBUSY); + + /* Check lower bound that does fit */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 5, 90, 5) != 0); + MT_BUG_ON(mt, mas.index != 5); + MT_BUG_ON(mt, mas.last != 9); + rcu_read_unlock(); + + /* Check one gap that doesn't fit and one that does */ + rcu_read_lock(); + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 5, 217, 9) != 0); + MT_BUG_ON(mt, mas.index != 161); + MT_BUG_ON(mt, mas.last != 169); + + /* Check one gap that does fit above the min */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 100, 218, 3) != 0); + MT_BUG_ON(mt, mas.index != 216); + MT_BUG_ON(mt, mas.last != 218); + + /* Check size that doesn't fit any gap */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 100, 218, 16) != -EBUSY); + + /* + * Check size that doesn't fit the lower end of the window but + * does fit the gap + */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 167, 200, 4) != -EBUSY); + + /* + * Check size that doesn't fit the upper end of the window but + * does fit the gap + */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area_rev(&mas, 100, 162, 4) != -EBUSY); + + /* Check mas_empty_area forward */ + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 0, 100, 9) != 0); + MT_BUG_ON(mt, mas.index != 0); + MT_BUG_ON(mt, mas.last != 8); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 0, 100, 4) != 0); + MT_BUG_ON(mt, mas.index != 0); + MT_BUG_ON(mt, mas.last != 3); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 0, 100, 11) != -EBUSY); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 5, 100, 6) != -EBUSY); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 0, 8, 10) != -EBUSY); + + mas_reset(&mas); + mas_empty_area(&mas, 100, 165, 3); + + mas_reset(&mas); + MT_BUG_ON(mt, mas_empty_area(&mas, 100, 163, 6) != -EBUSY); + rcu_read_unlock(); +} + static DEFINE_MTREE(tree); static int maple_tree_seed(void) { @@ -2765,6 +2850,10 @@ static int maple_tree_seed(void) check_bnode_min_spanning(&tree); mtree_destroy(&tree); + mt_init_flags(&tree, MT_FLAGS_ALLOC_RANGE); + check_empty_area_window(&tree); + mtree_destroy(&tree); + #if defined(BENCH) skip: #endif _ Patches currently in -mm which might be from liam.howlett(a)oracle.com are

2 years

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror February 2023