Re: Did we really need to clear the IF flag at prepare_singlestep() of x86 kprobes?

[Dongdong Deng <libfetion@gmail.com>]

On Wed, Jan 13, 2010 at 12:06 AM, Arjan van de Ven <arjan@infradead.org> wrote:
> On Tue, 12 Jan 2010 19:09:35 +0800
> Dongdong Deng <libfetion@gmail.com> wrote:
>
>> Hi Kprobe experts,
>>
>> I have a doubt about the handling "X86_EFLAGS_IF" at
>> prepare_singlestep(), Could you give me some suggestions?
>
>
> iirc it was a security thing; we used to have some exploits
> due to the linux-abi entry points which caused a mess, and this
> was put there as defensive programming.

Hi Arjan,

Thanks for your explain. :)

Do you means that the user will modify the IF? for example: through
"p->pre_handler(p, regs)" .

But I couldn't image the affect that if user modify the IF flag, could
you give me a detail info about security thing?


BTW:

Before linux 2.5, the debug tarp was initalized as trap gate:

linux-2.4.37/arch/i386/kernel/traps.c:966:	set_trap_gate(1,&debug);

I know kprobes have a long history,  Is it possible that the interrupt
flag of kprobes was introduced at that time?


Thanks,
Dongdong




>
> I could totally misremember this as well of course.
>
>
> --
> Arjan van de Ven        Intel Open Source Technology Centre
> For development, discussion and tips for power savings,
> visit http://www.lesswatts.org
>