h.m.brand at xs4all
May 20, 2008, 8:32 AM
Post #3 of 3
(100 views)
Permalink
|
|
Re: Change 33843: [perl #54120] [PATCH] [metaconfig] Need more -fstack-protector
[In reply to]
|
|
On Tue, 20 May 2008 11:21:14 -0400 (EDT), Andy Dougherty
<doughera[at]lafayette.edu> wrote:
> On Tue, 20 May 2008, Nicholas Clark via RT wrote:
>
> [patch to fiddle with -fstack-protector]
>
> > : flags used in final linking phase
> > case "$ldflags" in
> > '') if ./venix; then
> > dflt='-i -z'
> > else
> > dflt=''
> > fi
> > case "$ccflags" in
> > *-posix*) dflt="$dflt -posix" ;;
> > esac
> > # See note above about -fstack-protector
> > case "$ccflags" in
> > *-fstack-protector*) dflt="$dflt -fstack-protector" ;;
> > esac
> > ;;
> > *) dflt="$ldflags";;
> > esac
> >
> >
> > I find that as soon as I use -Aldflags, I don't get -fstack-protector added.
> > Which is somewhat daft, at least as an "end user", as I was using *A* not *D*.
>
> Oops. That's not supposed to happen. Sorry about that!
>
> There are actually two separate things going on here: First, the -A is
> supposed to happen after platform-specific hints are done. By the time
> we've reached this code, platform-specific hints have long been finished,
> so the -A should indeed have been applied.
>
> The troublesome bit of code I quoted above is there to allow a use to tell
> Configure: "use this exact string for ldflags, don't mess with it at
> all." The thinking pre-dates Configure's -A flag. That's not what you
> wanted to do, however.
>
> > I don't understand the implementation of Configure enough to know when -A
> > processing is done, but is it done early enough that ldflags is already set by
> > the time that that code above is reached?
>
> Yes, that's exactly it. And further, we've also lost all memory of
> whether it was a -A or a -D addition.
>
> Clearly, that's a problem. I'll need to think on it a bit to figure out
> how to fix it. I think the short-term fix will be to add
> -fstack-protector no matter what. Something like this (sorry no
> metaconfig patch here, but it should apply with a (large) offset)
Hunk #1 succeeded at 377 (offset -4884 lines).
The regenerated Configure seems to be under just a bit more impact.
Regenerated in #33887
> --- perl-current/Configure 2008-05-17 10:31:35.000000000 -0400
> +++ perl-andy/Configure 2008-05-20 11:11:41.000000000 -0400
> @@ -5261,13 +5261,18 @@
> case "$ccflags" in
> *-posix*) dflt="$dflt -posix" ;;
> esac
> - # See note above about -fstack-protector
> - case "$ccflags" in
> - *-fstack-protector*) dflt="$dflt -fstack-protector" ;;
> - esac
> ;;
> *) dflt="$ldflags";;
> esac
> +# See note above about -fstack-protector
> +case "$ccflags" in
> +*-fstack-protector*)
> + case "$dflt" in
> + *-fstack-protector*) ;; # Don't add it again
> + *) dflt="$dflt -fstack-protector" ;;
> + esac
> + ;;
> +esac
>
> : Try to guess additional flags to pick up local libraries.
> for thislibdir in $libpth; do
>
--
H.Merijn Brand Amsterdam Perl Mongers (http://amsterdam.pm.org/)
using & porting perl 5.6.2, 5.8.x, 5.10.x on HP-UX 10.20, 11.00, 11.11,
& 11.23, SuSE 10.1 & 10.2, AIX 5.2, and Cygwin. http://qa.perl.org
http://mirrors.develooper.com/hpux/ http://www.test-smoke.org
http://www.goldmark.org/jeff/stupid-disclaimers/
|
