From d2406291483775ecddaee929231a39c70c08fda2 Mon Sep 17 00:00:00 2001 From: Peng Zhang Date: Fri, 27 Oct 2023 11:38:45 +0800 Subject: fork: use __mt_dup() to duplicate maple tree in dup_mmap() In dup_mmap(), using __mt_dup() to duplicate the old maple tree and then directly replacing the entries of VMAs in the new maple tree can result in better performance. __mt_dup() uses DFS pre-order to duplicate the maple tree, so it is efficient. The average time complexity of __mt_dup() is O(n), where n is the number of VMAs. The proof of the time complexity is provided in the commit log that introduces __mt_dup(). After duplicating the maple tree, each element is traversed and replaced (ignoring the cases of deletion, which are rare). Since it is only a replacement operation for each element, this process is also O(n). Analyzing the exact time complexity of the previous algorithm is challenging because each insertion can involve appending to a node, pushing data to adjacent nodes, or even splitting nodes. The frequency of each action is difficult to calculate. The worst-case scenario for a single insertion is when the tree undergoes splitting at every level. If we consider each insertion as the worst-case scenario, we can determine that the upper bound of the time complexity is O(n*log(n)), although this is a loose upper bound. However, based on the test data, it appears that the actual time complexity is likely to be O(n). As the entire maple tree is duplicated using __mt_dup(), if dup_mmap() fails, there will be a portion of VMAs that have not been duplicated in the maple tree. To handle this, we mark the failure point with XA_ZERO_ENTRY. In exit_mmap(), if this marker is encountered, stop releasing VMAs that have not been duplicated after this point. There is a "spawn" in byte-unixbench[1], which can be used to test the performance of fork(). I modified it slightly to make it work with different number of VMAs. Below are the test results. The first row shows the number of VMAs. The second and third rows show the number of fork() calls per ten seconds, corresponding to next-20231006 and the this patchset, respectively. The test results were obtained with CPU binding to avoid scheduler load balancing that could cause unstable results. There are still some fluctuations in the test results, but at least they are better than the original performance. 21 121 221 421 821 1621 3221 6421 12821 25621 51221 112100 76261 54227 34035 20195 11112 6017 3161 1606 802 393 114558 83067 65008 45824 28751 16072 8922 4747 2436 1233 599 2.19% 8.92% 19.88% 34.64% 42.37% 44.64% 48.28% 50.17% 51.68% 53.74% 52.42% [1] https://github.com/kdlucas/byte-unixbench/tree/master Link: https://lkml.kernel.org/r/20231027033845.90608-11-zhangpeng.00@bytedance.com Signed-off-by: Peng Zhang Suggested-by: Liam R. Howlett Reviewed-by: Liam R. Howlett Cc: Christian Brauner Cc: Jonathan Corbet Cc: Mateusz Guzik Cc: Mathieu Desnoyers Cc: Matthew Wilcox Cc: Michael S. Tsirkin Cc: Mike Christie Cc: Nicholas Piggin Cc: Peter Zijlstra Cc: Suren Baghdasaryan Signed-off-by: Andrew Morton --- kernel/fork.c | 40 +++++++++++++++++++++++++++++----------- 1 file changed, 29 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 10917c3e1f03..93924392a5c3 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -650,7 +650,6 @@ static __latent_entropy int dup_mmap(struct mm_struct *mm, int retval; unsigned long charge = 0; LIST_HEAD(uf); - VMA_ITERATOR(old_vmi, oldmm, 0); VMA_ITERATOR(vmi, mm, 0); uprobe_start_dup_mmap(); @@ -678,16 +677,22 @@ static __latent_entropy int dup_mmap(struct mm_struct *mm, goto out; khugepaged_fork(mm, oldmm); - retval = vma_iter_bulk_alloc(&vmi, oldmm->map_count); - if (retval) + /* Use __mt_dup() to efficiently build an identical maple tree. */ + retval = __mt_dup(&oldmm->mm_mt, &mm->mm_mt, GFP_KERNEL); + if (unlikely(retval)) goto out; mt_clear_in_rcu(vmi.mas.tree); - for_each_vma(old_vmi, mpnt) { + for_each_vma(vmi, mpnt) { struct file *file; vma_start_write(mpnt); if (mpnt->vm_flags & VM_DONTCOPY) { + retval = vma_iter_clear_gfp(&vmi, mpnt->vm_start, + mpnt->vm_end, GFP_KERNEL); + if (retval) + goto loop_out; + vm_stat_account(mm, mpnt->vm_flags, -vma_pages(mpnt)); continue; } @@ -749,9 +754,11 @@ static __latent_entropy int dup_mmap(struct mm_struct *mm, if (is_vm_hugetlb_page(tmp)) hugetlb_dup_vma_private(tmp); - /* Link the vma into the MT */ - if (vma_iter_bulk_store(&vmi, tmp)) - goto fail_nomem_vmi_store; + /* + * Link the vma into the MT. After using __mt_dup(), memory + * allocation is not necessary here, so it cannot fail. + */ + vma_iter_bulk_store(&vmi, tmp); mm->map_count++; if (!(tmp->vm_flags & VM_WIPEONFORK)) @@ -760,15 +767,28 @@ static __latent_entropy int dup_mmap(struct mm_struct *mm, if (tmp->vm_ops && tmp->vm_ops->open) tmp->vm_ops->open(tmp); - if (retval) + if (retval) { + mpnt = vma_next(&vmi); goto loop_out; + } } /* a new mm has just been created */ retval = arch_dup_mmap(oldmm, mm); loop_out: vma_iter_free(&vmi); - if (!retval) + if (!retval) { mt_set_in_rcu(vmi.mas.tree); + } else if (mpnt) { + /* + * The entire maple tree has already been duplicated. If the + * mmap duplication fails, mark the failure point with + * XA_ZERO_ENTRY. In exit_mmap(), if this marker is encountered, + * stop releasing VMAs that have not been duplicated after this + * point. + */ + mas_set_range(&vmi.mas, mpnt->vm_start, mpnt->vm_end - 1); + mas_store(&vmi.mas, XA_ZERO_ENTRY); + } out: mmap_write_unlock(mm); flush_tlb_mm(oldmm); @@ -778,8 +798,6 @@ fail_uprobe_end: uprobe_end_dup_mmap(); return retval; -fail_nomem_vmi_store: - unlink_anon_vmas(tmp); fail_nomem_anon_vma_fork: mpol_put(vma_policy(tmp)); fail_nomem_policy: -- cgit v1.2.3 From 2853b66b601a265306be709b4d86aaff7d92a0fc Mon Sep 17 00:00:00 2001 From: "Matthew Wilcox (Oracle)" Date: Mon, 11 Dec 2023 16:22:09 +0000 Subject: mm: remove some calls to page_add_new_anon_rmap() We already have the folio in these functions, we just need to use it. folio_add_new_anon_rmap() didn't exist at the time they were converted to folios. Link: https://lkml.kernel.org/r/20231211162214.2146080-5-willy@infradead.org Signed-off-by: Matthew Wilcox (Oracle) Reviewed-by: David Hildenbrand Signed-off-by: Andrew Morton --- kernel/events/uprobes.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c index 435aac1d8c27..8b115fc43f04 100644 --- a/kernel/events/uprobes.c +++ b/kernel/events/uprobes.c @@ -181,7 +181,7 @@ static int __replace_page(struct vm_area_struct *vma, unsigned long addr, if (new_page) { folio_get(new_folio); - page_add_new_anon_rmap(new_page, vma, addr); + folio_add_new_anon_rmap(new_folio, vma, addr); folio_add_lru_vma(new_folio, vma); } else /* no new page, just dec_mm_counter for old_page */ -- cgit v1.2.3 From 5cc9695f06b065168f5c893c8e006b6a8a2c9c91 Mon Sep 17 00:00:00 2001 From: David Hildenbrand Date: Wed, 20 Dec 2023 23:44:48 +0100 Subject: kernel/events/uprobes: page_remove_rmap() -> folio_remove_rmap_pte() Let's convert __replace_page(). Link: https://lkml.kernel.org/r/20231220224504.646757-25-david@redhat.com Signed-off-by: David Hildenbrand Cc: Hugh Dickins Cc: Matthew Wilcox (Oracle) Cc: Muchun Song Cc: Muchun Song Cc: Peter Xu Cc: Ryan Roberts Cc: Yin Fengwei Signed-off-by: Andrew Morton --- kernel/events/uprobes.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c index 8b115fc43f04..485bb0389b48 100644 --- a/kernel/events/uprobes.c +++ b/kernel/events/uprobes.c @@ -198,7 +198,7 @@ static int __replace_page(struct vm_area_struct *vma, unsigned long addr, set_pte_at_notify(mm, addr, pvmw.pte, mk_pte(new_page, vma->vm_page_prot)); - page_remove_rmap(old_page, vma, false); + folio_remove_rmap_pte(old_folio, old_page, vma); if (!folio_mapped(old_folio)) folio_free_swap(old_folio); page_vma_mapped_walk_done(&pvmw); -- cgit v1.2.3 From 61dd3f246b3adaabff3241c586f2210ac91b05a4 Mon Sep 17 00:00:00 2001 From: Kinsey Ho Date: Wed, 27 Dec 2023 14:12:02 +0000 Subject: mm/mglru: add CONFIG_LRU_GEN_WALKS_MMU Add CONFIG_LRU_GEN_WALKS_MMU such that if disabled, the code that walks page tables to promote pages into the youngest generation will not be built. Also improves code readability by adding two helper functions get_mm_state() and get_next_mm(). Link: https://lkml.kernel.org/r/20231227141205.2200125-3-kinseyho@google.com Signed-off-by: Kinsey Ho Co-developed-by: Aneesh Kumar K.V Signed-off-by: Aneesh Kumar K.V Tested-by: Donet Tom Acked-by: Yu Zhao Cc: kernel test robot Signed-off-by: Andrew Morton --- kernel/fork.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 93924392a5c3..56cf276432c8 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -2946,7 +2946,7 @@ pid_t kernel_clone(struct kernel_clone_args *args) get_task_struct(p); } - if (IS_ENABLED(CONFIG_LRU_GEN) && !(clone_flags & CLONE_VM)) { + if (IS_ENABLED(CONFIG_LRU_GEN_WALKS_MMU) && !(clone_flags & CLONE_VM)) { /* lock the task to synchronize with memcg migration */ task_lock(p); lru_gen_add_mm(p->mm); -- cgit v1.2.3 From fd37721803c6e73619108f76ad2e12a9aa5fafaf Mon Sep 17 00:00:00 2001 From: "Kirill A. Shutemov" Date: Thu, 28 Dec 2023 17:47:03 +0300 Subject: mm, treewide: introduce NR_PAGE_ORDERS NR_PAGE_ORDERS defines the number of page orders supported by the page allocator, ranging from 0 to MAX_ORDER, MAX_ORDER + 1 in total. NR_PAGE_ORDERS assists in defining arrays of page orders and allows for more natural iteration over them. [kirill.shutemov@linux.intel.com: fixup for kerneldoc warning] Link: https://lkml.kernel.org/r/20240101111512.7empzyifq7kxtzk3@box Link: https://lkml.kernel.org/r/20231228144704.14033-1-kirill.shutemov@linux.intel.com Signed-off-by: Kirill A. Shutemov Reviewed-by: Zi Yan Cc: Linus Torvalds Signed-off-by: Andrew Morton --- kernel/crash_core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/crash_core.c b/kernel/crash_core.c index d4313b53837e..56cf4ad7abbb 100644 --- a/kernel/crash_core.c +++ b/kernel/crash_core.c @@ -802,7 +802,7 @@ static int __init crash_save_vmcoreinfo_init(void) VMCOREINFO_OFFSET(list_head, prev); VMCOREINFO_OFFSET(vmap_area, va_start); VMCOREINFO_OFFSET(vmap_area, list); - VMCOREINFO_LENGTH(zone.free_area, MAX_ORDER + 1); + VMCOREINFO_LENGTH(zone.free_area, NR_PAGE_ORDERS); log_buf_vmcoreinfo_setup(); VMCOREINFO_LENGTH(free_area.free_list, MIGRATE_TYPES); VMCOREINFO_NUMBER(NR_FREE_PAGES); -- cgit v1.2.3 From 5e0a760b44417f7cadd79de2204d6247109558a0 Mon Sep 17 00:00:00 2001 From: "Kirill A. Shutemov" Date: Thu, 28 Dec 2023 17:47:04 +0300 Subject: mm, treewide: rename MAX_ORDER to MAX_PAGE_ORDER commit 23baf831a32c ("mm, treewide: redefine MAX_ORDER sanely") has changed the definition of MAX_ORDER to be inclusive. This has caused issues with code that was not yet upstream and depended on the previous definition. To draw attention to the altered meaning of the define, rename MAX_ORDER to MAX_PAGE_ORDER. Link: https://lkml.kernel.org/r/20231228144704.14033-2-kirill.shutemov@linux.intel.com Signed-off-by: Kirill A. Shutemov Cc: Linus Torvalds Signed-off-by: Andrew Morton --- kernel/dma/pool.c | 6 +++--- kernel/dma/swiotlb.c | 4 ++-- kernel/events/ring_buffer.c | 10 +++++----- 3 files changed, 10 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/dma/pool.c b/kernel/dma/pool.c index b481c48a31a6..d10613eb0f63 100644 --- a/kernel/dma/pool.c +++ b/kernel/dma/pool.c @@ -84,8 +84,8 @@ static int atomic_pool_expand(struct gen_pool *pool, size_t pool_size, void *addr; int ret = -ENOMEM; - /* Cannot allocate larger than MAX_ORDER */ - order = min(get_order(pool_size), MAX_ORDER); + /* Cannot allocate larger than MAX_PAGE_ORDER */ + order = min(get_order(pool_size), MAX_PAGE_ORDER); do { pool_size = 1 << (PAGE_SHIFT + order); @@ -190,7 +190,7 @@ static int __init dma_atomic_pool_init(void) /* * If coherent_pool was not used on the command line, default the pool - * sizes to 128KB per 1GB of memory, min 128KB, max MAX_ORDER. + * sizes to 128KB per 1GB of memory, min 128KB, max MAX_PAGE_ORDER. */ if (!atomic_pool_size) { unsigned long pages = totalram_pages() / (SZ_1G / SZ_128K); diff --git a/kernel/dma/swiotlb.c b/kernel/dma/swiotlb.c index 33d942615be5..176078bf2215 100644 --- a/kernel/dma/swiotlb.c +++ b/kernel/dma/swiotlb.c @@ -686,8 +686,8 @@ static struct io_tlb_pool *swiotlb_alloc_pool(struct device *dev, size_t pool_size; size_t tlb_size; - if (nslabs > SLABS_PER_PAGE << MAX_ORDER) { - nslabs = SLABS_PER_PAGE << MAX_ORDER; + if (nslabs > SLABS_PER_PAGE << MAX_PAGE_ORDER) { + nslabs = SLABS_PER_PAGE << MAX_PAGE_ORDER; nareas = limit_nareas(nareas, nslabs); } diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c index e8d82c2f07d0..60ed43d1c29e 100644 --- a/kernel/events/ring_buffer.c +++ b/kernel/events/ring_buffer.c @@ -610,8 +610,8 @@ static struct page *rb_alloc_aux_page(int node, int order) { struct page *page; - if (order > MAX_ORDER) - order = MAX_ORDER; + if (order > MAX_PAGE_ORDER) + order = MAX_PAGE_ORDER; do { page = alloc_pages_node(node, PERF_AUX_GFP, order); @@ -702,9 +702,9 @@ int rb_alloc_aux(struct perf_buffer *rb, struct perf_event *event, /* * kcalloc_node() is unable to allocate buffer if the size is larger - * than: PAGE_SIZE << MAX_ORDER; directly bail out in this case. + * than: PAGE_SIZE << MAX_PAGE_ORDER; directly bail out in this case. */ - if (get_order((unsigned long)nr_pages * sizeof(void *)) > MAX_ORDER) + if (get_order((unsigned long)nr_pages * sizeof(void *)) > MAX_PAGE_ORDER) return -ENOMEM; rb->aux_pages = kcalloc_node(nr_pages, sizeof(void *), GFP_KERNEL, node); @@ -821,7 +821,7 @@ struct perf_buffer *rb_alloc(int nr_pages, long watermark, int cpu, int flags) size = sizeof(struct perf_buffer); size += nr_pages * sizeof(void *); - if (order_base_2(size) > PAGE_SHIFT+MAX_ORDER) + if (order_base_2(size) > PAGE_SHIFT+MAX_PAGE_ORDER) goto fail; node = (cpu == -1) ? cpu : cpu_to_node(cpu); -- cgit v1.2.3