Subject: Re: Performance of various memcpy()'s
To: TAMURA Kent <firstname.lastname@example.org>
From: Bang Jun-Young <email@example.com>
Date: 10/29/2002 14:27:17
On Tue, Oct 29, 2002 at 12:47:49PM +0900, TAMURA Kent wrote:
> Please imagine that src (or dst) is 0x12345678 and length (%ecx)
> is 0xffffffff. Copying over the end of the address space is
> always meaningless in NetBSD, and may crash the program
Length shouldn't be larger than the maximum user address, or you're
very likely to get a seg fault. That's a much more common case than
pointer wraparound. IMHO, pointer wraparound check is worthless in
> or may help exploits.
There's no known way to exploit system by memcpy. If there is, why
not _DIAGNOSTIC enabled by default?
Unless there's a better reason to have _DIAGNOSTIC there, I'm going to
get rid of it.
Bang Jun-Young <firstname.lastname@example.org>