Re: Mlocked pages statistics shows bogus value.

[Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>]

Kirill A. Shutemov wrote:
> On Tue, Jan 19, 2016 at 09:46:21PM +0900, Tetsuo Handa wrote:
> > Kirill A. Shutemov wrote:
> > > Oh. Looks like a bug from 2013...
> > > 
> > > Thanks for report.
> > > 
> > > From 6f80a79dc5f65f29899e396942d40f727cd36480 Mon Sep 17 00:00:00 2001
> > > From: "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
> > > Date: Tue, 19 Jan 2016 14:59:19 +0300
> > > Subject: [PATCH] mm: fix mlock accouting
> > > 
> > > Tetsuo Handa reported underflow of NR_MLOCK on munlock.
> > > 
> > > Testcase:
> > > 	#include <stdio.h>
> > > 	#include <stdlib.h>
> > > 	#include <sys/mman.h>
> > > 
> > > 	#define BASE ((void *)0x400000000000)
> > > 	#define SIZE (1UL << 21)
> > > 
> > > 	int main(int argc, char *argv[])
> > > 	{
> > > 		void *addr;
> > > 
> > > 		system("grep Mlocked /proc/meminfo");
> > > 		addr = mmap(BASE, SIZE, PROT_READ | PROT_WRITE,
> > > 				MAP_ANONYMOUS | MAP_PRIVATE | MAP_LOCKED | MAP_FIXED,
> > > 				-1, 0);
> > > 		if (addr == MAP_FAILED)
> > > 			printf("mmap() failed\n"), exit(1);
> > > 		munmap(addr, SIZE);
> > > 		system("grep Mlocked /proc/meminfo");
> > > 		return 0;
> > > 	}
> > > 
> > > It happens on munlock_vma_page() due to unfortunate choice of nr_pages
> > > data type:
> > > 
> > > 	__mod_zone_page_state(zone, NR_MLOCK, -nr_pages);
> > > 
> > > For unsigned int nr_pages, implicitly casted to long in
> > > __mod_zone_page_state(), it becomes something around UINT_MAX.
> > > 
> > > munlock_vma_page() usually called for THP as small pages go though
> > > pagevec.
> > > 
> > > Let's make nr_pages singed int.
> > > 
> > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
> > > Fixes: ff6a6da60b89 ("mm: accelerate munlock() treatment of THP pages")
> > > Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
> > > Cc: Michel Lespinasse <walken@google.com>
> > > ---
> > >  mm/mlock.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/mm/mlock.c b/mm/mlock.c
> > > index e1e2b1207bf2..96f001041928 100644
> > > --- a/mm/mlock.c
> > > +++ b/mm/mlock.c
> > > @@ -175,7 +175,7 @@ static void __munlock_isolation_failed(struct page *page)
> > >   */
> > >  unsigned int munlock_vma_page(struct page *page)
> > >  {
> > > -	unsigned int nr_pages;
> > > +	int nr_pages;
> > >  	struct zone *zone = page_zone(page);
> > >  
> > >  	/* For try_to_munlock() and to serialize with page migration */
> > > -- 
> > >  Kirill A. Shutemov
> > > 

I tested your patch on Linux 4.4 and confirmed that your patch fixed this bug.
Please also send to stable.

Cc: <stable@vger.kernel.org>  [4.4+]

> > 
> > Don't we want to use "long" than "int" for all variables that count number
> > of pages, for recently commit 6cdb18ad98a49f7e9b95d538a0614cde827404b8
> > "mm/vmstat: fix overflow in mod_zone_page_state()" changed to use "long" ?
> 
> Potentially, yes. But here we count number of small pages in the compound
> page. We're far from being able to allocate 8 terabyte pages ;)

That commit says "we have a 9TB system with only one node".
You might encounter such machines in near future. ;-)

> 
> Anyway, it's out-of-scope for this bug fix.
> 
> My "Fixes:" is probably misleading, since we don't have bug visible until
> 6cdb18ad98a4.
> 
> -- 
>  Kirill A. Shutemov
> 

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>