[PATCH v5 03/11] mm/gup: Applies counting method to monitor gup_pgd_range

Leonardo Bras leonardo at linux.ibm.com
Thu Oct 3 11:33:17 AEST 2019


As described, gup_pgd_range is a lockless pagetable walk. So, in order to
monitor against THP split/collapse with the counting method, it's necessary
to bound it with {begin,end}_lockless_pgtbl_walk.

local_irq_{save,restore} is already inside {begin,end}_lockless_pgtbl_walk,
so there is no need to repeat it here.

Signed-off-by: Leonardo Bras <leonardo at linux.ibm.com>
---
 mm/gup.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/mm/gup.c b/mm/gup.c
index 23a9f9c9d377..52e53b4f39d8 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -2319,7 +2319,7 @@ int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
 			  struct page **pages)
 {
 	unsigned long len, end;
-	unsigned long flags;
+	unsigned long irq_mask;
 	int nr = 0;
 
 	start = untagged_addr(start) & PAGE_MASK;
@@ -2345,9 +2345,9 @@ int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
 
 	if (IS_ENABLED(CONFIG_HAVE_FAST_GUP) &&
 	    gup_fast_permitted(start, end)) {
-		local_irq_save(flags);
+		irq_mask = begin_lockless_pgtbl_walk(current->mm);
 		gup_pgd_range(start, end, write ? FOLL_WRITE : 0, pages, &nr);
-		local_irq_restore(flags);
+		end_lockless_pgtbl_walk(current->mm, irq_mask);
 	}
 
 	return nr;
@@ -2414,9 +2414,9 @@ int get_user_pages_fast(unsigned long start, int nr_pages,
 
 	if (IS_ENABLED(CONFIG_HAVE_FAST_GUP) &&
 	    gup_fast_permitted(start, end)) {
-		local_irq_disable();
+		begin_lockless_pgtbl_walk(current->mm);
 		gup_pgd_range(addr, end, gup_flags, pages, &nr);
-		local_irq_enable();
+		end_lockless_pgtbl_walk(current->mm, IRQS_ENABLED);
 		ret = nr;
 	}
 
-- 
2.20.1



More information about the Linuxppc-dev mailing list