Re: [PATCH 1/3] signal(i386): alternative signal stack wraparound occurs

[Ingo Molnar <mingo@elte.hu>]

* Shi Weihua <shiwh@cn.fujitsu.com> wrote:

> > When it's the handler function itself or its callees that cause the 
> > overflow, rather than the signal handler frame setup alone crossing 
> > the boundary, this still won't help.  But I don't see any way to 
> > distinguish that from the valid longjmp case.
> 
> Thank you for your detailed explanation and patch. I tested your 
> patch, unfortunately it can not stop all kinds of overflow.
[...]
> So, the patch I posted is still needed 

thanks, i've picked up your fix for x86.git, for 2.6.25 merging.

> Surely, adding a variable to sched.h is not a good idea. 
> Could you tell me a better place to store the previous esp?

i think sched.h is ok - it has a sas_ss_sp field already. Alternatively, 
if we only want this in x86, we could put it into the thread_struct - 
but i think eventually other architectures would want to use this too, 
right?

	Ingo