Hello, thanks to the report. Peter, this isn't immediately clear to me, is it perhaps something related to clang? This is commit 8dec302e87453234fc7ac1cf4d09e4d577a06cf3 From 8dec302e87453234fc7ac1cf4d09e4d577a06cf3 Mon Sep 17 00:00:00 2001 From: Andrea Arcangeli <aarcange(a)redhat.com&gt; Date: Fri, 7 May 2021 11:05:53 -0400 Subject: [PATCH 1/1] mm: gup: pack has_pinned in MMF_HAS_PINNED has_pinned 32bit can be packed in the MMF_HAS_PINNED bit as a noop cleanup. Any atomic_inc/dec to the mm cacheline shared by all threads in pin-fast would reintroduce a loss of SMP scalability to pin-fast, so there's no future potential usefulness to keep an atomic in the mm for this. set_bit(MMF_HAS_PINNED) will be theoretically a bit slower than WRITE_ONCE (atomic_set is equivalent to WRITE_ONCE), but the set_bit (just like atomic_set after this commit) has to be still issued only once per "mm", so the difference between the two will be lost in the noise. will-it-scale "mmap2" shows no change in performance with enterprise config as expected. will-it-scale "pin_fast" retains the > 4000% SMP scalability performance improvement against upstream as expected. This is a noop as far as overall performance and SMP scalability are concerned. Signed-off-by: Andrea Arcangeli <aarcange(a)redhat.com&gt; [peterx: Fix build for task_mmu.c, introduce mm_set_has_pinned_flag, fix comment here and there] Signed-off-by: Peter Xu <peterx(a)redhat.com&gt; Signed-off-by: Andrea Arcangeli <aarcange(a)redhat.com&gt; --- include/linux/mm.h | 2 +- include/linux/mm_types.h | 10 ---------- include/linux/sched/coredump.h | 8 ++++++++ kernel/fork.c | 1 - mm/gup.c | 19 +++++++++++++++---- 5 files changed, 24 insertions(+), 16 deletions(-) diff --git a/include/linux/mm.h b/include/linux/mm.h index 5c297b0a97e1..639b3b27ffb7 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1352,7 +1352,7 @@ static inline bool page_needs_cow_for_dma(struct vm_area_struct *vma, if (!is_cow_mapping(vma->vm_flags)) return false; - if (!atomic_read(&vma->vm_mm->has_pinned)) + if (!test_bit(MMF_HAS_PINNED, &vma->vm_mm->flags)) return false; return page_maybe_dma_pinned(page); diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h index ebfece42843c..7fc7a3320ad9 100644 --- a/include/linux/mm_types.h +++ b/include/linux/mm_types.h @@ -441,16 +441,6 @@ struct mm_struct { */ atomic_t mm_count; - /** - * @has_pinned: Whether this mm has pinned any pages. This can - * be either replaced in the future by @pinned_vm when it - * becomes stable, or grow into a counter on its own. We're - * aggresive on this bit now - even if the pinned pages were - * unpinned later on, we'll still keep this bit set for the - * lifecycle of this mm just for simplicity. - */ - atomic_t has_pinned; - /** * @write_protect_seq: Locked when any thread is write * protecting pages mapped by this mm to enforce a later COW, diff --git a/include/linux/sched/coredump.h b/include/linux/sched/coredump.h index dfd82eab2902..4d9e3a656875 100644 --- a/include/linux/sched/coredump.h +++ b/include/linux/sched/coredump.h @@ -73,6 +73,14 @@ static inline int get_dumpable(struct mm_struct *mm) #define MMF_OOM_VICTIM 25 /* mm is the oom victim */ #define MMF_OOM_REAP_QUEUED 26 /* mm was queued for oom_reaper */ #define MMF_MULTIPROCESS 27 /* mm is shared between processes */ +/* + * MMF_HAS_PINNED: Whether this mm has pinned any pages. This can be either + * replaced in the future by mm.pinned_vm when it becomes stable, or grow into + * a counter on its own. We're aggresive on this bit for now: even if the + * pinned pages were unpinned later on, we'll still keep this bit set for the + * lifecycle of this mm, just for simplicity. + */ +#define MMF_HAS_PINNED 28 /* FOLL_PIN has run, never cleared */ #define MMF_DISABLE_THP_MASK (1 << MMF_DISABLE_THP) #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK |\ diff --git a/kernel/fork.c b/kernel/fork.c index dc06afd725cb..e31541da7dbd 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1029,7 +1029,6 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p, mm_pgtables_bytes_init(mm); mm->map_count = 0; mm->locked_vm = 0; - atomic_set(&mm->has_pinned, 0); atomic64_set(&mm->pinned_vm, 0); memset(&mm->rss_stat, 0, sizeof(mm->rss_stat)); spin_lock_init(&mm->page_table_lock); diff --git a/mm/gup.c b/mm/gup.c index 8ca5e923b54f..d8751840ad0b 100644 --- a/mm/gup.c +++ b/mm/gup.c @@ -1382,6 +1382,17 @@ int fixup_user_fault(struct mm_struct *mm, } EXPORT_SYMBOL_GPL(fixup_user_fault); +/* + * Set the MMF_HAS_PINNED if not set yet; after set it'll be there for the mm's + * lifecycle. Avoid setting the bit unless necessary, or it might cause write + * cache bouncing on large SMP machines for concurrent pinned gups. + */ +static inline void mm_set_has_pinned_flag(unsigned long *mm_flags) +{ + if (!test_bit(MMF_HAS_PINNED, mm_flags)) + set_bit(MMF_HAS_PINNED, mm_flags); +} + /* * Please note that this function, unlike __get_user_pages will not * return 0 for nr_pages > 0 without FOLL_NOWAIT @@ -1404,8 +1415,8 @@ static __always_inline long __get_user_pages_locked(struct mm_struct *mm, BUG_ON(*locked != 1); } - if ((flags & FOLL_PIN) && !atomic_read(&mm->has_pinned)) - atomic_set(&mm->has_pinned, 1); + if (flags & FOLL_PIN) + mm_set_has_pinned_flag(&mm->flags); /* * FOLL_PIN and FOLL_GET are mutually exclusive. Traditional behavior @@ -2741,8 +2752,8 @@ static int internal_get_user_pages_fast(unsigned long start, FOLL_FAST_ONLY))) return -EINVAL; - if ((gup_flags & FOLL_PIN) && !atomic_read(&current->mm->has_pinned)) - atomic_set(&current->mm->has_pinned, 1); + if (gup_flags & FOLL_PIN) + mm_set_has_pinned_flag(&current->mm->flags); if (!(gup_flags & FOLL_FAST_ONLY)) might_lock_read(&current->mm->mmap_lock); On Tue, May 11, 2021 at 06:45:36PM +0800, kernel test robot wrote:

On Tue, May 11, 2021 at 10:23:39AM -0400, Peter Xu wrote: Btw, there's also a space issue with the patch (we need to reindent mm_set_has_pinned_flag so to use tabs rather than spaces), otherwise checkpatch seems to complain. Sorry about both issues! -- Peter Xu