linux-kernel - Re: [RFC PATCH] coccinelle: check for integer overflow in binary search

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8f454c56-735c-0620-ead4-b59e328b0283@web.de>
Date:   Thu, 5 Sep 2019 14:32:54 +0200
From:   Markus Elfring <Markus.Elfring@....de>
To:     Denis Efremov <efremov@...ux.com>, cocci@...teme.lip6.fr,
        kernel-janitors@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org,
        Gilles Muller <Gilles.Muller@...6.fr>,
        Julia Lawall <julia.lawall@...6.fr>,
        Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Michal Marek <michal.lkml@...kovi.net>,
        Nicolas Palix <nicolas.palix@...g.fr>
Subject: Re: [RFC PATCH] coccinelle: check for integer overflow in binary
 search

> +identifier l, h, m;

Can expressions make sense for these metavariables?


> +@@
> +(
> + while (\(l < h\|l <= h\|(h - l) > 1\|(l + 1) < h\|l < (h - 1)\)) {
> +  ...
> +(
> +  ((l + h)@p / c)
> +|
> +  ((l + h)@p >> c)
> +)
> +  ...
> + }

* I suggest again to look at further possibilities to reduce undesirable
  code duplication also together with the usage of SmPL disjunctions.

* The condition specification might be easier to read with a few
  additional spaces (or the following variant).

* The SmPL ellipses will probably need further considerations.


+@@
+(
+ while (
+(       l \( < \| <= \) h
+|       (h - l) > 1
+|       (h - 1) > l
+|       (l + 1) < h
+)      )
+ {
+ <+...
+ ((l + h)@p \( / \| >> \) c)
+ ...+>
+ }


> +@...ipt:python depends on report@
> +p << r.p;
> +@@
> +
> +msg="WARNING: custom implementation of bsearch is error-prone. "
> +msg+="Consider using lib/bsearch.c or fix the midpoint calculation "
> +msg+="as 'm = l + (h - l) / 2;' to prevent the arithmetic overflow."
> +coccilib.report.print_report(p[0], msg)

The Linux coding style supports to put a long string literal also into a single line.
Thus I find such a message construction nicer without the extra variable “msg”.

Regards,
Markus