[PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

[PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

[Hillf Danton]

From: Hillf Danton <dhillf@gmail.com>
Subject: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

If huge page is enqueued under the protection of hugetlb_lock, then
the operation is atomic and safe.

Cc: Michal Hocko <mhocko@suse.cz>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Hillf Danton <dhillf@gmail.com>
---

--- a/mm/hugetlb.c	Tue Dec 20 21:26:30 2011
+++ b/mm/hugetlb.c	Fri Dec 23 21:16:28 2011
@@ -901,7 +901,6 @@ retry:
 	h->resv_huge_pages += delta;
 	ret = 0;

-	spin_unlock(&hugetlb_lock);
 	/* Free the needed pages to the hugetlb pool */
 	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
 		if ((--needed) < 0)
@@ -915,6 +914,7 @@ retry:
 		VM_BUG_ON(page_count(page));
 		enqueue_huge_page(h, page);
 	}
+	spin_unlock(&hugetlb_lock);

 	/* Free unnecessary surplus pages to the buddy allocator */
 free:

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

[KAMEZAWA Hiroyuki]

On Fri, 23 Dec 2011 21:35:25 +0800
Hillf Danton <dhillf@gmail.com> wrote:

> From: Hillf Danton <dhillf@gmail.com>
> Subject: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page
> 
> If huge page is enqueued under the protection of hugetlb_lock, then
> the operation is atomic and safe.
> 
> Cc: Michal Hocko <mhocko@suse.cz>
> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Hillf Danton <dhillf@gmail.com>

Hmm, at reporting this kind of bug...it's better to show when it's broken.
Maybe this commit a9869b837c098732bad84939015c0eb391b23e41
and it's better to cc stable.

Acked-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>

> ---
> 
> --- a/mm/hugetlb.c	Tue Dec 20 21:26:30 2011
> +++ b/mm/hugetlb.c	Fri Dec 23 21:16:28 2011
> @@ -901,7 +901,6 @@ retry:
>  	h->resv_huge_pages += delta;
>  	ret = 0;
> 
> -	spin_unlock(&hugetlb_lock);
>  	/* Free the needed pages to the hugetlb pool */
>  	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
>  		if ((--needed) < 0)
> @@ -915,6 +914,7 @@ retry:
>  		VM_BUG_ON(page_count(page));
>  		enqueue_huge_page(h, page);
>  	}
> +	spin_unlock(&hugetlb_lock);
> 
>  	/* Free unnecessary surplus pages to the buddy allocator */
>  free:
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

[Michal Hocko]

On Fri 23-12-11 21:35:25, Hillf Danton wrote:
> From: Hillf Danton <dhillf@gmail.com>
> Subject: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page
> 
> If huge page is enqueued under the protection of hugetlb_lock, then
> the operation is atomic and safe.
> 
> Cc: Michal Hocko <mhocko@suse.cz>
> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Hillf Danton <dhillf@gmail.com>

Yes, looks correct even though the changelog could be more verbose.
The code is broken since .37

Reviewed-by: Michal Hocko <mhocko@suse.cz>

Thanks
> ---
> 
> --- a/mm/hugetlb.c	Tue Dec 20 21:26:30 2011
> +++ b/mm/hugetlb.c	Fri Dec 23 21:16:28 2011
> @@ -901,7 +901,6 @@ retry:
>  	h->resv_huge_pages += delta;
>  	ret = 0;
> 
> -	spin_unlock(&hugetlb_lock);
>  	/* Free the needed pages to the hugetlb pool */
>  	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
>  		if ((--needed) < 0)
> @@ -915,6 +914,7 @@ retry:
>  		VM_BUG_ON(page_count(page));
>  		enqueue_huge_page(h, page);
>  	}
> +	spin_unlock(&hugetlb_lock);
> 
>  	/* Free unnecessary surplus pages to the buddy allocator */
>  free:

-- 
Michal Hocko
SUSE Labs
SUSE LINUX s.r.o.
Lihovarska 1060/12
190 00 Praha 9    
Czech Republic

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page

[Andrew Morton]

On Fri, 23 Dec 2011 21:35:25 +0800
Hillf Danton <dhillf@gmail.com> wrote:

> From: Hillf Danton <dhillf@gmail.com>
> Subject: [PATCH] mm: hugetlb: fix non-atomic enqueue of huge page
> 
> If huge page is enqueued under the protection of hugetlb_lock, then
> the operation is atomic and safe.
> 
> Cc: Michal Hocko <mhocko@suse.cz>
> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
> Cc: Andrew Morton <akpm@linux-foundation.org>
> Signed-off-by: Hillf Danton <dhillf@gmail.com>
> ---
> 
> --- a/mm/hugetlb.c	Tue Dec 20 21:26:30 2011
> +++ b/mm/hugetlb.c	Fri Dec 23 21:16:28 2011
> @@ -901,7 +901,6 @@ retry:
>  	h->resv_huge_pages += delta;
>  	ret = 0;
> 
> -	spin_unlock(&hugetlb_lock);
>  	/* Free the needed pages to the hugetlb pool */
>  	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
>  		if ((--needed) < 0)
> @@ -915,6 +914,7 @@ retry:
>  		VM_BUG_ON(page_count(page));
>  		enqueue_huge_page(h, page);
>  	}
> +	spin_unlock(&hugetlb_lock);
> 
>  	/* Free unnecessary surplus pages to the buddy allocator */

btw,

	/* Free the needed pages to the hugetlb pool */
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
		if ((--needed) < 0)
			break;
		list_del(&page->lru);
		/*
		 * This page is now managed by the hugetlb allocator and has
		 * no users -- drop the buddy allocator's reference.
		 */
		put_page_testzero(page);
		VM_BUG_ON(page_count(page));
		enqueue_huge_page(h, page);
	}
	spin_unlock(&hugetlb_lock);


That VM_BUG_ON() largely duplicates the one in put_page_testzero().

(Putting a VM_BUG_ON() in put_page_testzero() was pretty expensive,
too.  I wonder how many people are enabling VM_BUG_ON()?  We should be
sparing in using these things)

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>