bugtraq - RE: gcc 4.1 bug miscompiles pointer range checks, may place you at risk

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

Message-ID: <11352F9641010A418AD5057945A3A6594CBB26@MTV-EXCHANGE.microfocus.com>
Date: Mon, 17 Apr 2006 17:06:28 -0700
From: "Michael Wojcik" <Michael.Wojcik@...rofocus.com>
To: <bugtraq@...urityfocus.com>
Subject: RE: gcc 4.1 bug miscompiles pointer range checks, may place you at risk

> From: Felix von Leitner [mailto:felix-bugtraq@...e.de] 
> Sent: Monday, 17 April, 2006 16:04
> 
> static inline int range_ptrinbuf(const void* buf,unsigned 
> long len,const void* ptr) {
>   register const char* c=(const char*)buf;      /* no pointer 
> arithmetic on void* */
>   return (c && c+len>c && (const char*)ptr-c<len);
> }
> 
> ...
>   assert(range_ptrinbuf(buf,(unsigned long)-1,buf+1)==0);
> 
> Imagine my surprise when this assertion failed.

As far as the C language is concerned, this isn't a compiler "bug".
You've created an invalid pointer in "c+len" (the result neither points
within the object that includes the location c points to, nor one past
it), which invokes Undefined Behavior (ISO 9899:1990 6.5.6 #8).  And
you've compared two pointers which do not point within the same object
("c+len>c"), which also invokes Undefined Behavior (6.5.8 #5).

The behavior of gcc 4.1 in this case might be infelicitous, but it is
not, properly speaking, a bug.

-- 
Michael Wojcik
Principal Software Systems Developer, Micro Focus