linux-kernel - Re: [PATCH 4/6] lib/scatterlist: Do not leak pages when high-order allocation fails

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]

Date:   Wed, 03 Oct 2018 08:31:40 -0700
From:   Bart Van Assche <bvanassche@....org>
To:     Tvrtko Ursulin <tursulin@...ulin.net>, linux-kernel@...r.kernel.org
Cc:     tvrtko.ursulin@...ux.intel.com,
        Tvrtko Ursulin <tvrtko.ursulin@...el.com>,
        Bart Van Assche <bart.vanassche@....com>,
        Hannes Reinecke <hare@...e.com>,
        Johannes Thumshirn <jthumshirn@...e.de>,
        Jens Axboe <axboe@...nel.dk>
Subject: Re: [PATCH 4/6] lib/scatterlist: Do not leak pages when high-order
 allocation fails

On Wed, 2018-09-26 at 15:16 +0100, Tvrtko Ursulin wrote:
> From: Tvrtko Ursulin <tvrtko.ursulin@...el.com>
> 
> If a higher-order allocation fails, the existing abort and cleanup path
> would consider all segments allocated so far as 0-order page allocations
> and would therefore leak memory.
> 
> Fix this by cleaning up using sgl_free_n_order which allows the correct
> page order to be passed in.
> 
> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@...el.com>
> Cc: Bart Van Assche <bart.vanassche@....com>
> Cc: Hannes Reinecke <hare@...e.com>
> Cc: Johannes Thumshirn <jthumshirn@...e.de>
> Cc: Jens Axboe <axboe@...nel.dk>
> ---
>  lib/scatterlist.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/lib/scatterlist.c b/lib/scatterlist.c
> index 3cc01cd82242..0caed79d7291 100644
> --- a/lib/scatterlist.c
> +++ b/lib/scatterlist.c
> @@ -481,7 +481,7 @@ struct scatterlist *sgl_alloc_order(unsigned long length, unsigned int order,
>  {
>  	struct scatterlist *sgl, *sg;
>  	struct page *page;
> -	unsigned int nent, nalloc;
> +	unsigned int nent, nalloc, i;
>  	u32 elem_len;
>  
>  	nent = round_up(length, PAGE_SIZE << order) >> (PAGE_SHIFT + order);
> @@ -501,17 +501,19 @@ struct scatterlist *sgl_alloc_order(unsigned long length, unsigned int order,
>  
>  	sg_init_table(sgl, nalloc);
>  	sg = sgl;
> +	i = 0;
>  	while (length) {
>  		elem_len = min_t(u64, length, PAGE_SIZE << order);
>  		page = alloc_pages(gfp, order);
>  		if (!page) {
> -			sgl_free(sgl);
> +			sgl_free_n_order(sgl, i, order);
>  			return NULL;
>  		}
>  
>  		sg_set_page(sg, page, elem_len, 0);
>  		length -= elem_len;
>  		sg = sg_next(sg);
> +		i++;
>  	}
>  	WARN_ONCE(length, "length = %ld\n", length);
>  	if (nent_p)

sg_init_table() clears all page pointers and sgl_free_n_order() can handle
elements of which the page pointer is zero. So I think if
sgl_free_n_order(sgl, i, order) would be changed into sgl_free_n_order(sgl,
UINT_MAX, order) that the variable 'i' can be left out.

Bart.