linux-kernel - Re: [Question] Mlocked count will not be decreased

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <93f1b063-6288-d109-117d-d3c1cf152a8e@suse.cz>
Date:   Wed, 24 May 2017 12:32:29 +0200
From:   Vlastimil Babka <vbabka@...e.cz>
To:     Yisheng Xie <xieyisheng1@...wei.com>,
        Kefeng Wang <wangkefeng.wang@...wei.com>, linux-mm@...ck.org,
        linux-kernel@...r.kernel.org, zhongjiang <zhongjiang@...wei.com>,
        Qiuxishi <qiuxishi@...wei.com>
Subject: Re: [Question] Mlocked count will not be decreased

On 05/24/2017 10:32 AM, Yisheng Xie wrote:
> Hi Kefeng，
> Could you please try this patch.
> 
> Thanks
> Yisheng Xie
> -------------
> From a70ae975756e8e97a28d49117ab25684da631689 Mon Sep 17 00:00:00 2001
> From: Yisheng Xie <xieyisheng1@...wei.com>
> Date: Wed, 24 May 2017 16:01:24 +0800
> Subject: [PATCH] mlock: fix mlock count can not decrease in race condition
> 
> Kefeng reported that when run the follow test the mlock count in meminfo
> cannot be decreased:
>  [1] testcase
>  linux:~ # cat test_mlockal
>  grep Mlocked /proc/meminfo
>   for j in `seq 0 10`
>   do
>  	for i in `seq 4 15`
>  	do
>  		./p_mlockall >> log &
>  	done
>  	sleep 0.2
>  done
>  sleep 5 # wait some time to let mlock decrease
>  grep Mlocked /proc/meminfo
> 
>  linux:~ # cat p_mlockall.c
>  #include <sys/mman.h>
>  #include <stdlib.h>
>  #include <stdio.h>
> 
>  #define SPACE_LEN	4096
> 
>  int main(int argc, char ** argv)
>  {
>  	int ret;
>  	void *adr = malloc(SPACE_LEN);
>  	if (!adr)
>  		return -1;
> 
>  	ret = mlockall(MCL_CURRENT | MCL_FUTURE);
>  	printf("mlcokall ret = %d\n", ret);
> 
>  	ret = munlockall();
>  	printf("munlcokall ret = %d\n", ret);
> 
>  	free(adr);
>  	return 0;
>  }
> 
> When __munlock_pagevec, we ClearPageMlock but isolation_failed in race
> condition, and we do not count these page into delta_munlocked, which cause mlock

Race condition with what? Who else would isolate our pages?

> counter incorrect for we had Clear the PageMlock and cannot count down
> the number in the feture.
> 
> Fix it by count the number of page whoes PageMlock flag is cleared.
> 
> Reported-by: Kefeng Wang <wangkefeng.wang@...wei.com>
> Signed-off-by: Yisheng Xie <xieyisheng1@...wei.com>

Weird, I can reproduce the issue on my desktop's 4.11 distro kernel, but
not in qemu and small kernel build, for some reason. So I couldn't test
the patch yet. But it's true that before 7225522bb429 ("mm: munlock:
batch non-THP page isolation and munlock+putback using pagevec") we
decreased NR_MLOCK for each pages that passed TestClearPageMlocked(),
and that unintentionally changed with my patch. There should be a Fixes:
tag for that.

> ---
>  mm/mlock.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/mlock.c b/mm/mlock.c
> index c483c5c..71ba5cf 100644
> --- a/mm/mlock.c
> +++ b/mm/mlock.c
> @@ -284,7 +284,7 @@ static void __munlock_pagevec(struct pagevec *pvec, struct zone *zone)
>  {
>  	int i;
>  	int nr = pagevec_count(pvec);
> -	int delta_munlocked;
> +	int munlocked = 0;
>  	struct pagevec pvec_putback;
>  	int pgrescued = 0;
> 
> @@ -296,6 +296,7 @@ static void __munlock_pagevec(struct pagevec *pvec, struct zone *zone)
>  		struct page *page = pvec->pages[i];
> 
>  		if (TestClearPageMlocked(page)) {
> +			munlocked --;
>  			/*
>  			 * We already have pin from follow_page_mask()
>  			 * so we can spare the get_page() here.
> @@ -315,8 +316,8 @@ static void __munlock_pagevec(struct pagevec *pvec, struct zone *zone)
>  		pagevec_add(&pvec_putback, pvec->pages[i]);
>  		pvec->pages[i] = NULL;
>  	}
> -	delta_munlocked = -nr + pagevec_count(&pvec_putback);
> -	__mod_zone_page_state(zone, NR_MLOCK, delta_munlocked);
> +	if (munlocked)

You don't have to if () this, it should be very rare that munlocked will
be 0, and the code works fine even if it is.

> +		__mod_zone_page_state(zone, NR_MLOCK, munlocked);
>  	spin_unlock_irq(zone_lru_lock(zone));
> 
>  	/* Now we can release pins of pages that we are not munlocking */
>