linux-kernel - Re: [PATCH 1/1] mm/page_owner: fix recursion bug after changing skip entries

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] [day] [month] [year] [list]

Message-ID: <40171c50-4466-6589-3968-024f936e5c7a@suse.cz>
Date:   Wed, 21 Mar 2018 09:25:27 +0100
From:   Vlastimil Babka <vbabka@...e.cz>
To:     Maninder Singh <maninder1.s@...sung.com>,
        akpm@...ux-foundation.org, mhocko@...e.com,
        osalvador@...hadventures.net, gregkh@...uxfoundation.org,
        ayush.m@...sung.com, guptap@...eaurora.org,
        vinmenon@...eaurora.org, gomonovych@...il.com
Cc:     linux-mm@...ck.org, linux-kernel@...r.kernel.org,
        a.sahrawat@...sung.com, pankaj.m@...sung.com,
        Vaneet Narang <v.narang@...sung.com>
Subject: Re: [PATCH 1/1] mm/page_owner: fix recursion bug after changing skip
 entries

On 03/21/2018 05:37 AM, Maninder Singh wrote:
> This patch fixes "5f48f0bd4e368425db4424b9afd1bd251d32367a".
> (mm, page_owner: skip unnecessary stack_trace entries)
> 
> Because if we skip first two entries then logic of checking count
> value as 2 for recursion is broken and code will go in one depth
> recursion.
> 
> so we need to check only one call of _RET_IP(__set_page_owner)
> while checking for recursion.
> 
> Current Backtrace while checking for recursion:-
> 
> (save_stack)             from (__set_page_owner)  // (But recursion returns true here)
> (__set_page_owner)       from (get_page_from_freelist)
> (get_page_from_freelist) from (__alloc_pages_nodemask)
> (__alloc_pages_nodemask) from (depot_save_stack)
> (depot_save_stack)       from (save_stack)       // recursion should return true here
> (save_stack)             from (__set_page_owner)
> (__set_page_owner)       from (get_page_from_freelist)
> (get_page_from_freelist) from (__alloc_pages_nodemask+)
> (__alloc_pages_nodemask) from (depot_save_stack)
> (depot_save_stack)       from (save_stack)
> (save_stack)             from (__set_page_owner)
> (__set_page_owner)       from (get_page_from_freelist)
> 
> Correct Backtrace with fix:
> 
> (save_stack)             from (__set_page_owner) // recursion returned true here
> (__set_page_owner)       from (get_page_from_freelist)
> (get_page_from_freelist) from (__alloc_pages_nodemask+)
> (__alloc_pages_nodemask) from (depot_save_stack)
> (depot_save_stack)       from (save_stack)
> (save_stack)             from (__set_page_owner)
> (__set_page_owner)       from (get_page_from_freelist)
> 
> Signed-off-by: Maninder Singh <maninder1.s@...sung.com>
> Signed-off-by: Vaneet Narang <v.narang@...sung.com>
Fixes: 5f48f0bd4e36 ("mm, page_owner: skip unnecessary stack_trace entries")

Good catch.
Acked-by: Vlastimil Babka <vbabka@...e.cz>

> ---
>  mm/page_owner.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/mm/page_owner.c b/mm/page_owner.c
> index 8592543..46ab1c4 100644
> --- a/mm/page_owner.c
> +++ b/mm/page_owner.c
> @@ -123,13 +123,13 @@ void __reset_page_owner(struct page *page, unsigned int order)
>  static inline bool check_recursive_alloc(struct stack_trace *trace,
>  					unsigned long ip)
>  {
> -	int i, count;
> +	int i;
>  
>  	if (!trace->nr_entries)
>  		return false;
>  
> -	for (i = 0, count = 0; i < trace->nr_entries; i++) {
> -		if (trace->entries[i] == ip && ++count == 2)
> +	for (i = 0; i < trace->nr_entries; i++) {
> +		if (trace->entries[i] == ip)
>  			return true;
>  	}
>  
>