1 |
On Tue, 01 Aug 2006 19:57:36 +0200 |
2 |
pageexec@××××××××.hu wrote: |
3 |
|
4 |
> On 31 Jul 2006 at 17:41, Ned Ludd wrote: |
5 |
> > I think the major problem we are facing here is how to cleanly |
6 |
> > upgrade from 3.x to 4.x. symbol names have changed. And using the |
7 |
> > stub/aliases method Peter used in uClibc svn allows the __guard to |
8 |
> > be overwritten. |
9 |
> |
10 |
> what do you mean by that? i thought the guard was in writable memory |
11 |
> anyway... |
12 |
> |
13 |
> > Flags are missing etc. |
14 |
> |
15 |
> what flags? |
16 |
|
17 |
I think solar's referring to -fno-stack-protector-all. Not too worried |
18 |
about that myself, I think we can hide the change in semantics in |
19 |
flag-o-matic.eclass - actual ebuild use of -fno-stack-protector-all is |
20 |
rare. |
21 |
|
22 |
> > Upstream also destroyed the value of the handler. |
23 |
> |
24 |
> in what sense? i thought that 4.x also had a reporting function... |
25 |
|
26 |
It's that gcc-4.x doesn't pass the caller information (function name |
27 |
and line number) to _stack_chk_fail. |
28 |
|
29 |
I've been thinking that a PaX-style register, stack and perhaps map dump |
30 |
might be a good idea for development environments at least. |
31 |
|
32 |
-- |
33 |
Kevin F. Quinn |