Re: [patch 1/4] mm: page_alloc: remove order assumption from__free_pages_bootmem()

[Andrew Morton]

On Tue, 13 Dec 2011 14:58:28 +0100
Johannes Weiner <hannes@xxxxxxxxxxx> wrote:

> Even though bootmem passes an order with the page to be freed,
> __free_pages_bootmem() assumes that 1 << order is always BITS_PER_LONG
> if non-zero.  While this happens to be true, it's not really robust.
> Remove that assumption and use 1 << order instead.
> 
> Signed-off-by: Johannes Weiner <hannes@xxxxxxxxxxx>
> ---
>  mm/page_alloc.c |    7 ++++---
>  1 files changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index 2b8ba3a..4d5e91c 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -703,13 +703,14 @@ void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
>  		set_page_refcounted(page);
>  		__free_page(page);
>  	} else {
> -		int loop;
> +		unsigned int nr_pages = 1 << order;
> +		unsigned int loop;
>  
>  		prefetchw(page);
> -		for (loop = 0; loop < BITS_PER_LONG; loop++) {
> +		for (loop = 0; loop < nr_pages; loop++) {
>  			struct page *p = &page[loop];
>  
> -			if (loop + 1 < BITS_PER_LONG)
> +			if (loop + 1 < nr_pages)
>  				prefetchw(p + 1);
>  			__ClearPageReserved(p);
>  			set_page_count(p, 0);

Tejun has recently secretly snuck the below (rather old) patch into
linux-next's page_alloc.c.  I think I got everything fixed up right -
please check.

commit 53348f27168534561c0c814843bbf181314374f4
Author:     Tejun Heo <tj@xxxxxxxxxx>
AuthorDate: Tue Jul 12 09:58:06 2011 +0200
Commit:     H. Peter Anvin <hpa@xxxxxxxxxxxxxxx>
CommitDate: Wed Jul 13 16:35:56 2011 -0700

    bootmem: Fix __free_pages_bootmem() to use @order properly
    
    a226f6c899 (FRV: Clean up bootmem allocator's page freeing algorithm)
    separated out __free_pages_bootmem() from free_all_bootmem_core().
    __free_pages_bootmem() takes @order argument but it assumes @order is
    either 0 or ilog2(BITS_PER_LONG).  Note that all the current users
    match that assumption and this doesn't cause actual problems.
    
    Fix it by using 1 << order instead of BITS_PER_LONG.
    
    Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
    Link: http://lkml.kernel.org/r/1310457490-3356-3-git-send-email-tj@xxxxxxxxxx
    Cc: David Howells <dhowells@xxxxxxxxxx>
    Signed-off-by: H. Peter Anvin <hpa@xxxxxxxxxxxxxxx>

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 9119faa..b6da6ed 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -705,10 +705,10 @@ void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
 		int loop;
 
 		prefetchw(page);
-		for (loop = 0; loop < BITS_PER_LONG; loop++) {
+		for (loop = 0; loop < (1 << order); loop++) {
 			struct page *p = &page[loop];
 
-			if (loop + 1 < BITS_PER_LONG)
+			if (loop + 1 < (1 << order))
 				prefetchw(p + 1);
 			__ClearPageReserved(p);
 			set_page_count(p, 0);

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/