Re: PIE support

[Paul Jakma]

On Wed, 30 Nov 2005, Ralf Wildenhues wrote:

> Because: when one day 100 compilers support this, they may use 10 
> different #defines for this notion.  Or not, who knows.

Sure. Then you should possibly also define __PIC__ for completeness / 
symmetry? (relying on gcc to provide that at present right?)

> Yes, but most packages using libtool won't ever need to compile a 
> PIC object without libtool's help.  Unlike PIE, when you do 
> CFLAGS=-fpie.  Which in turn is what we're trying to settle on, 
> currently.  (Well, I am at least ;)

;)

> Sure.  As long as you stick with GCC and GNU ld, there won't be any 
> issues.  If libtool were to specialize on those two only, it'd be 
> fair to say there is little point in having a thing such as libtool 
> at all.

:)

> Well, the more interesting question would probably be the GNU ld version
> you used, and on what hardware.

x86. GNU ld 2.9.1 and also 2.15. Thing is, I'm not sure gcc actually 
used GNU ld. If I do:

LD=gld gcc ....

The binary works.

If I manually compile to object code with gcc and then link by hand 
with gld, it segfaults:

$ gcc -c -fpie -pie -Os test.c -o test.o -lc -lnsl -lsocket
$ gld -fpie  -lc -lnsl -lsocket -o test test.o

It segfaults if I use Solaris ld by hand too, so I must be doing 
something wrong.

In no case, where I let gcc handle linking, do I get anything that is 
obviously a relocatable executable though. (variables in main are at 
same location on each invocation).

> I could not find such discussion (I searched before reporting the bug
> upstream).  Could you point me to it?  Thanks.

I can't I'm afraid. It came up when talking to the user who 
desperately wanted an easier way to build our project's daemons as 
PIE, I can't find the URL back though. :(

regards,
--
Paul Jakma      address@hidden  address@hidden  Key ID: 64A2FF6A
Fortune:
I'm not a real movie star -- I've still got the same wife I started out
with twenty-eight years ago.
                -- Will Rogers