Bugzilla Bug 147144

I'm trying to use Xephyr from x11-base/xorg-server-1.0.2-r7, but it shows
segmentation fault just after opening its window:

$ Xephyr :5
_XSERVTransSocketOpenCOTSServer: Unable to open socket for inet6
_XSERVTransOpen: transport open failed for inet6/quantum:5
_XSERVTransMakeAllCOTSServerListeners: failed to open listener for inet6
Extended Input Devices not yet supported. Impelement it at line 625 in kinput.c
    xkb_keycodes             { include "xfree86+aliases(qwerty)" };
    xkb_types                { include "complete" };
    xkb_compatibility        { include "complete" };
    xkb_symbols              { include "pc(pc101)+us" };
    xkb_geometry             { include "pc(pc101)" };
Segmentation fault

Here is gdb backtrace:
#0  0x0000000000000000 in ?? ()
#1  0x00000000007050f7 in FreePictFormat ()
#2  0x000000000044d3ee in FreeCursor ()
#3  0x000000000044d7e0 in FreeCursor ()
#4  0x000000000043b967 in ReplyNotSwappd ()
#5  0x00002b6a24a44136 in __libc_start_main () from /lib/libc.so.6
#6  0x0000000000420209 in ?? ()
#7  0x00007fffff882c98 in ?? ()
#8  0x000000000000001c in ?? ()
#9  0x0000000000000002 in ?? ()
#10 0x00007fffff884032 in ?? ()
#11 0x00007fffff884042 in ?? ()
#12 0x0000000000000000 in ?? ()

Output of emerge --info is in the attached file

Created an attachment (id=96673) [details]
Output from emerge --info

You could try compiling it with different CFLAGS.  I think I've had problems
with -funit-at-a-time and X in the past.

(In reply to comment #2)
> You could try compiling it with different CFLAGS.  I think I've had problems
> with -funit-at-a-time and X in the past.
I've just tried to recompile it with CFLAGS="-march=athlon64 -O2", but I still
have the same crash.

Please unmask Xorg 7.1 from testing (xorg-x11, xorg-server, mesa, mesa-progs
and relevant drivers) and give that a try.  Re-open the bug when you've done
so.

I've already switched to xorg-7.1 ( xorg-server-1.1.1-r1 ) but segmentation
fault still occurs, however backtrace is different:

(gdb) bt
#0  0x0000000000000000 in ?? ()
#1  0x0000000000708b7c in PictureDestroyWindow ()
#2  0x000000000044bcfe in FreeClientPixels ()
#3  0x000000000044c023 in FreeClientPixels ()
#4  0x000000000043ae3a in NotImplemented ()
#5  0x00002b4f0abaa134 in __libc_start_main () from /lib/libc.so.6
#6  0x000000000041f6b9 in ?? ()
#7  0x00007fffa0e10ae8 in ?? ()
#8  0x0000000000000000 in ?? ()

By the way I've switched to gcc 4.1.1, new emerge --info is in the attachment.

Created an attachment (id=98009) [details]
New output from emerge --info

This should move upstream.  Please find/file a bug at
https://bugs.freedesktop.org and post the link here.  Thanks!

(In reply to comment #7)
> This should move upstream.  Please find/file a bug at
> https://bugs.freedesktop.org and post the link here.  Thanks!
I've found how to fix the problem. I'm attaching two patches: one general
(which defines KD_MAX_KEYCODE to 255 according to XDisplayKeycodes
documentation) and one amd64-specific. With this patches XEphyr works stable on
x86_64.
I'll post this patches upstream, but may be in meantime it's good to apply it
in xorg-server ebuild (without it Xephyr will not work on amd64 at all).

Created an attachment (id=98042) [details]
Make KD_MAX_KEYCODE=255 according to XDisplayKeycodes documentation

Created an attachment (id=98043) [details]
Fixes crash of Xephyr on amd64

Where are these posted upstream?  We generally don't add patches without
upstream accepting them first.

(In reply to comment #11)
> Where are these posted upstream?  We generally don't add patches without
> upstream accepting them first.

https://bugs.freedesktop.org/show_bug.cgi?id=8417

Thanks!  Give us a poke once upstream has said they're OK.

*** Bug 173179 has been marked as a duplicate of this bug. ***