4:55 a.m.
Please watch for the symptoms.
If you see them in a pdump XEmacs, it's especially important. If you
see them in an unexec (non-pdump) XEmacs, and build and report for the
pdump XEmacs, you will earn immediate Karma sufficient for promotion
to a higher lifeform at your next reincarnation.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
8:44 a.m.
On Tue, Oct 28, 2003 at 01:55:02PM +0900, Stephen J. Turnbull wrote:
Please watch for the symptoms.
If you see them in a pdump XEmacs, it's especially important. If you
see them in an unexec (non-pdump) XEmacs, and build and report for the
pdump XEmacs, you will earn immediate Karma sufficient for promotion
to a higher lifeform at your next reincarnation.
Hmmm, pdump should break, but it should be fixable. 'cept the code
changed so much I have problems reading it again in the first place.
Mumble.
I'll see about adding the kernel patch when I go to 2.6.0t9 or 10.
OG.
4:58 p.m.
On Tue, 28 Oct 2003 09:44:50 +0100, Olivier Galibert <galibert(a)pobox.com> said:
Hmmm, pdump should break, but it should be fixable. 'cept the
code
changed so much I have problems reading it again in the first place.
Mumble.
I'll see about adding the kernel patch when I go to 2.6.0t9 or 10.
pdump does *not* break.
I'm running a RedHat Rawhide/Fedora userspace from Oct 20, and
2.6.0-test8-mm1-exec-shield-nptl kernel. Running with exec-shield = 2
and exe-shield-randomize=1.
So unless they've borked something in the last week, it's a non-problem for
pdump.
5:07 p.m.
On Tue, 28 Oct 2003 09:44:50 +0100, Olivier Galibert <galibert(a)pobox.com> said:
Hmmm, pdump should break, but it should be fixable. 'cept the code
changed so much I have problems reading it again in the first place.
As a follow-up, building with --pdump=no dies a horrid death:
Finding pointers to doc strings...
Finding pointers to doc strings...done
Dumping under the name xemacs
make[1]: *** [xemacs] Segmentation fault
make[1]: *** Deleting file `xemacs'
make[1]: Leaving directory `/home/valdis/src/xemacs-21.5/src'
I'll do up a proper PROBLEMS diff when I get to the office....
5:19 p.m.
Valdis.Kletnieks(a)vt.edu writes:
On Tue, 28 Oct 2003 09:44:50 +0100, Olivier Galibert
<galibert(a)pobox.com> said:
>
> Hmmm, pdump should break, but it should be fixable. 'cept the code
> changed so much I have problems reading it again in the first place.
As a follow-up, building with --pdump=no dies a horrid death:
Finding pointers to doc strings...
Finding pointers to doc strings...done
Dumping under the name xemacs
make[1]: *** [xemacs] Segmentation fault
make[1]: *** Deleting file `xemacs'
make[1]: Leaving directory `/home/valdis/src/xemacs-21.5/src'
I'll do up a proper PROBLEMS diff when I get to the office....
Perhaps configure could check for the breakage and automatically turn
on pdump? That would cause the least amount of confusion for the end
users.
7:07 p.m.
On Wed, 29 Oct 2003 18:19:29 +0100, Hrvoje Niksic said:
Perhaps configure could check for the breakage and automatically
turn
on pdump? That would cause the least amount of confusion for the end
users.
The only safe way to approach it is to just force pdump if it detects
a Fedora or recent-enough RedHat kernel. And we still need a PROBLEMS
entry to deal with the case of upgrading and having an installed XEmacs break.
Quite frankly, I can't scare up any enthusiasm at all for fixing non-pdump
support on an ExecShield system.
Incidentally, I strongly suspect that sites that have installed the PaX or
grsecurity patches will also have the same problems, for the same reasons.
If ExecShield kills it, so will PaX or grsecurity.
3:34 a.m.
>>>> "Valdis" == Valdis Kletnieks
<Valdis.Kletnieks(a)vt.edu> writes:
Valdis> The only safe way to approach it is to just force pdump if
Valdis> it detects a Fedora or recent-enough RedHat kernel.
I believe --pdump is now default on Linux for exactly this reason.
Valdis> And we still need a PROBLEMS entry to deal with the case
Valdis> of upgrading and having an installed XEmacs break.
According to the post from emacs-devel I forwarded, an installed
XEmacs will _not_ break. Same as with "Hannibal Lector" ld, the
"Texas Chainsaw" toolchain can only affect XEmacs if XEmacs is built
with it.
--
Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Ask not how you can "do" free software business;
ask what your business can "do for" free software.
7694
days inactive
7696
days old
6 comments
4 participants
participants (4)
-
Hrvoje Niksic -
Olivier Galibert -
Stephen J. Turnbull -
Valdis Kl=?iso8859-4?Q?=BA?=tnieks