ASLR Leads to Emacs Crash on Build

[xsgb]

Hi, I'm getting a segfault when building emacs from portacle. Didn't took time to investigate. On Debian unstable, gcc (Debian 6.3.0-13) 6.3.0 20170415

make -C lib all
[...]
Loading /home/xs/brx/portacle/build/emacs/lisp/tooltip.el (source)...
Finding pointers to doc strings...
Finding pointers to doc strings...done
Dumping under the name emacs
Makefile:736: recipe for target 'bootstrap-emacs' failed
make[1]: *** [bootstrap-emacs] Segmentation fault
make[1]: Leaving directory '/home/xs/brx/portacle/build/emacs/src'
Makefile:398: recipe for target 'src' failed
make: *** [src] Error 2

[Shinmera]

What exactly did you do to arrive at that? Did you follow the instructions in the build readme exactly?

[xsgb]

Of course, from build/README.md, with all dependencies installed.
Let me know if I can produce some gcc output by increasing the verbosity.

Edit: there's what's the kernel said and it seems not good at all, however seems not coming from emacs

[10851.403755] temacs[5825]: segfault at 73dfc4 ip 00007ffff2c14e31 sp 00007fffffffcfb8 error 4 in libc-2.24.so[7ffff2aec000+195000]

I'm going to upgrade right now and let you informed if it resolves the problem :)

[xsgb]

Okay, it did exacly the same segfault after upgrading Debian, rebooting, "./build.sh clean" and "./build".

[Shinmera]

I'm not sure that this is a problem related to Portacle per se. It looks much more like a bug in Emacs itself. Unfortunately I'm not familiar with its inner workings, so I'm pretty much as much in the dark about this problem as you are.

If you do want to dive into this, you'll have to google around and see if you can dig up anything.

[xsgb]

I'll take a look tomorrow if I can find the time! It will be the opportunity to test on my work machine (Debian 8). There's effectively something faulty between emacs and the installed libc, I'll check with another emacs tag too then.

[xsgb]

It's a bug related to emacs concerning a strange syscall named "personality" The workaround would be to disable « randomize_va_space » feature by doing:

echo 0 > /proc/sys/kernel/randomize_va_space

(even if it's not successful on each try..., It worked at least once for obscure reasons)
https://lists.gnu.org/archive/html/bug-gnu-emacs/2016-05/msg00707.html

More explanations can be found there:
https://lists.gnu.org/archive/html/bug-gnu-emacs/2016-05/msg00893.html

I'm closing the ticket since it's not related to portacle! Thanks for your work and sharing your passion. See you o/

xs

[Shinmera]

That's interesting. Can you see whether changing the kernel parameter circumvents #18 as well?