Based on kernel version 4.16.1. Page generated on 2018-04-09 11:53 EST.
1 Split page table lock 2 ===================== 3 4 Originally, mm->page_table_lock spinlock protected all page tables of the 5 mm_struct. But this approach leads to poor page fault scalability of 6 multi-threaded applications due high contention on the lock. To improve 7 scalability, split page table lock was introduced. 8 9 With split page table lock we have separate per-table lock to serialize 10 access to the table. At the moment we use split lock for PTE and PMD 11 tables. Access to higher level tables protected by mm->page_table_lock. 12 13 There are helpers to lock/unlock a table and other accessor functions: 14 - pte_offset_map_lock() 15 maps pte and takes PTE table lock, returns pointer to the taken 16 lock; 17 - pte_unmap_unlock() 18 unlocks and unmaps PTE table; 19 - pte_alloc_map_lock() 20 allocates PTE table if needed and take the lock, returns pointer 21 to taken lock or NULL if allocation failed; 22 - pte_lockptr() 23 returns pointer to PTE table lock; 24 - pmd_lock() 25 takes PMD table lock, returns pointer to taken lock; 26 - pmd_lockptr() 27 returns pointer to PMD table lock; 28 29 Split page table lock for PTE tables is enabled compile-time if 30 CONFIG_SPLIT_PTLOCK_CPUS (usually 4) is less or equal to NR_CPUS. 31 If split lock is disabled, all tables guaded by mm->page_table_lock. 32 33 Split page table lock for PMD tables is enabled, if it's enabled for PTE 34 tables and the architecture supports it (see below). 35 36 Hugetlb and split page table lock 37 --------------------------------- 38 39 Hugetlb can support several page sizes. We use split lock only for PMD 40 level, but not for PUD. 41 42 Hugetlb-specific helpers: 43 - huge_pte_lock() 44 takes pmd split lock for PMD_SIZE page, mm->page_table_lock 45 otherwise; 46 - huge_pte_lockptr() 47 returns pointer to table lock; 48 49 Support of split page table lock by an architecture 50 --------------------------------------------------- 51 52 There's no need in special enabling of PTE split page table lock: 53 everything required is done by pgtable_page_ctor() and pgtable_page_dtor(), 54 which must be called on PTE table allocation / freeing. 55 56 Make sure the architecture doesn't use slab allocator for page table 57 allocation: slab uses page->slab_cache for its pages. 58 This field shares storage with page->ptl. 59 60 PMD split lock only makes sense if you have more than two page table 61 levels. 62 63 PMD split lock enabling requires pgtable_pmd_page_ctor() call on PMD table 64 allocation and pgtable_pmd_page_dtor() on freeing. 65 66 Allocation usually happens in pmd_alloc_one(), freeing in pmd_free() and 67 pmd_free_tlb(), but make sure you cover all PMD table allocation / freeing 68 paths: i.e X86_PAE preallocate few PMDs on pgd_alloc(). 69 70 With everything in place you can set CONFIG_ARCH_ENABLE_SPLIT_PMD_PTLOCK. 71 72 NOTE: pgtable_page_ctor() and pgtable_pmd_page_ctor() can fail -- it must 73 be handled properly. 74 75 page->ptl 76 --------- 77 78 page->ptl is used to access split page table lock, where 'page' is struct 79 page of page containing the table. It shares storage with page->private 80 (and few other fields in union). 81 82 To avoid increasing size of struct page and have best performance, we use a 83 trick: 84 - if spinlock_t fits into long, we use page->ptr as spinlock, so we 85 can avoid indirect access and save a cache line. 86 - if size of spinlock_t is bigger then size of long, we use page->ptl as 87 pointer to spinlock_t and allocate it dynamically. This allows to use 88 split lock with enabled DEBUG_SPINLOCK or DEBUG_LOCK_ALLOC, but costs 89 one more cache line for indirect access; 90 91 The spinlock_t allocated in pgtable_page_ctor() for PTE table and in 92 pgtable_pmd_page_ctor() for PMD table. 93 94 Please, never access page->ptl directly -- use appropriate helper.