linux-kernel - Re: [PATCH 1/5] fs/ntfs3: Use kmalloc

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: <20210901132448.mgkkyvqszc7lsnwy@kari-VirtualBox>
Date:   Wed, 1 Sep 2021 16:24:48 +0300
From:   Kari Argillander <kari.argillander@...il.com>
To:     Joe Perches <joe@...ches.com>
Cc:     Konstantin Komarov <almaz.alexandrovich@...agon-software.com>,
        ntfs3@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/5] fs/ntfs3: Use kmalloc_array over kmalloc with
 multiply

On Tue, Aug 31, 2021 at 07:40:58PM -0700, Joe Perches wrote:
> On Tue, 2021-08-31 at 21:15 +0300, Kari Argillander wrote:
> > If we do not use kmalloc_array we get checkpatch warning. It is also
> > little safer if something goes wrong with coding.
> []
> > diff --git a/fs/ntfs3/index.c b/fs/ntfs3/index.c
> []
> > @@ -707,7 +707,7 @@ static struct NTFS_DE *hdr_find_e(const struct ntfs_index *indx,
> >  		u16 *ptr;
> >  		int new_slots = ALIGN(2 * nslots, 8);
> >  
> > 
> > -		ptr = kmalloc(sizeof(u16) * new_slots, GFP_NOFS);
> > +		ptr = kmalloc_array(new_slots, sizeof(u16), GFP_NOFS);
> >  		if (ptr)
> >  			memcpy(ptr, offs, sizeof(u16) * max_idx);
> 
> This multiplication could also overflow.
> 
> Maybe use krealloc?

Seems to fit lot better here. But as I was watching this it seems that
we do not even need to resize this array. It is quite costly operation
compared to what entry compare cost.

We just need to compare it and if entry diff > 0 then we start entry
table again from zero without need resize array. It may be that we do
not even need to allocate memory. We can probably survive with stack,
but let's think that later.

This can also speed up search a lot. It is quite odd that we always fill
whole table and will not never check if we are over. Worst case is very
very bad. With this change this will be more like jump search but I
think it will be good in this case because we won't need memory allocation.

Thanks Joe.