Re: Re-send [PATCH] : Validating address in runtime

[Wenji Huang <wenji dot huang at oracle dot com>]

Ananth N Mavinakayanahalli wrote:
> On Mon, Jun 23, 2008 at 10:54:08AM +0800, Wenji Huang wrote:
> > Hi,
>
> Hi Wenji,
>
> > Regarding PR6646, seems no good way to solve it. I just think maybe double 
> > checking the address in runtime to make things safer.
> >
> > This patch adds validating address vs name mapping in runtime for several 
> > symbols which are used in initializing process.	So kernel oops could be 
> > early avoided in case of incorrect vmlinux or wrong address.
>
> Agreed that the oops will be avoided, but this will just shift the
> problem from 'runtime' to compile time. It still won't help SystemTap
> working.

The checking process happens in runtime. sprint_symbol is from 
kallsyms.c, to get the name according to address from /proc/kallsyms in 
runtime kernel.

Yeah, stap will exit if the checking fails.

> In some ways, its better to get an oops just to be able to see if the
> failure modes are the same or not.
>
> Ananth

Regards,
Wenji