有兴趣帮忙看看
http://groups.google.com/group/cdlug_community/browse_thread/thread/a563cfc716601844
On May 17, 10:44 am, Shawn <cit...@gmail.com> wrote:
> *
> hey guys,
> 写exploit测试多一种选择了:-/
>
> 原文:http://opensource.solidot.org/opensource/12/05/17/0030210.shtml
>
> 安全领域的黑客都喜欢使用perl/python这类脚本语言来写针对exploit的测试程序,一个偶然的机会打算试试用GNU
> Guile<http://www.gnu.org/software/guile/>,都是要调用C的接口system()函数,之后发现同样的功能在python
> 2.6下可以正常运行,而在Guile上却不行,之后把这个bug报给了Ragranok的作者(一个信奉GNUist的狂热分子-_-)<http://opensource.solidot.org/opensource/12/04/03/0210254.shtml?tid=44>,几天之后Nalaginrut和Andy
> Wingo的讨论和测试结果出来了,字符串(String)<http://en.wikipedia.org/wiki/String_%28computer_science%29>的
> 实现有两种:1,传统C语言使用的byte string;2,Guile和python 3使用的character
> string。Guile由于使用了character
> string所以导致传递值时大于ascii码时会被直接代替成"?",这主要是由于locale环境造成的,要让byte
> string能运行正常必须使用en_US.ISO-8859-*,完整的示例代码和解决过程在这里<https://github.com/citypw/SCFE/tree/master/security/shellcode/overwri...>
> ,以后黑客们可以使用Guile来写exploit测试了。*
> 有兴趣帮忙看看
> http://groups.google.com/group/cdlug_community/browse_thread/thread/a563cfc716601844
>
如果你只是打算统计一个程序的instruction的数量的话可以用ptrace来实现,量化一个程序的性能精度要考虑的问题太多了,比如你的程序里gcc可能也会加一些只占位流水线但不实际占用指令周期的nop,从硬件上层面和OS层面的量化方法也不一样,如果嫌time精度不够的话就只能从原始paper开始hacking吧。