bug#51864: 28.0.60; Emacs crashes in a few days under native-compilation

bug#51864: 28.0.60; Emacs crashes in a few days under native-compilation

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

After around the 3 to 4 day mark of running a build with native
compilation, Emacs crashes with a mysterious backtrace consisting of a
NULL pointer dereference at some random spot:

#0  0x00007fdf7d9bb8b2 in raise () at /lib64/libpthread.so.0
#1  0x000000000041db0a in terminate_due_to_signal (sig=sig@entry=11, backtrace_limit=backtrace_limit@entry=40) at emacs.c:437
#2  0x000000000041df7d in handle_fatal_signal (sig=sig@entry=11) at sysdep.c:1762
#3  0x000000000051a6b8 in deliver_thread_signal (sig=sig@entry=11, handler=0x41df72 <handle_fatal_signal>) at sysdep.c:1754
#4  0x000000000051a729 in deliver_fatal_thread_signal (sig=11) at sysdep.c:1774
#5  handle_sigsegv (sig=11, siginfo=<optimized out>, arg=<optimized out>) at sysdep.c:1867
#6  0x00007fdf7d9bba20 in <signal handler called> () at /lib64/libpthread.so.0
#7  face_at_buffer_position (w=w@entry=0xf754d38, pos=pos@entry=263485, endptr=endptr@entry=0x7ffec4579f48, limit=<optimized out>, mouse=<optimized out>, base_face_id=base_face_id@entry=-1, attr_filter=<optimized out>) at xfaces.c:6438
#8  0x0000000000594f06 in font_range (pos=pos@entry=263485, pos_byte=pos_byte@entry=271717, limit=limit@entry=0x7ffec4579fe8, w=w@entry=0xf754d38, face=face@entry=0x0, string=string@entry=0x0, ch=62) at font.c:3909
#9  0x00000000005e43c3 in autocmp_chars (rule=0x5b428bd, charpos=charpos@entry=263485, bytepos=271717, limit=263488, limit@entry=263492, win=win@entry=0xf754d38, face=face@entry=0x0, string=0x0, direction=0x0, ch=62) at composite.c:924
#10 0x00000000005e7d72 in find_automatic_composition (pos=pos@entry=263491, limit=263491, 
    limit@entry=-1, backlim=backlim@entry=-1, start=start@entry=0x7ffec457a118, end=end@entry=0x7ffec457a120, gstring=gstring@entry=0x7ffec457a128, string=0x0) at composite.c:1682
#11 0x00000000005e8a9d in composition_adjust_point (last_pt=last_pt@entry=263491, new_pt=new_pt@entry=263491) at composite.c:1769
#12 0x0000000000512c33 in command_loop_1 () at keyboard.c:1608
#13 0x000000000057c807 in internal_condition_case (bfun=bfun@entry=0x511ba0 <command_loop_1>, handlers=handlers@entry=0x90, hfun=hfun@entry=0x5086d0 <cmd_error>) at eval.c:1453
#14 0x0000000000502a6a in command_loop_2 (handlers=handlers@entry=0x90) at keyboard.c:1133
#15 0x000000000057c761 in internal_catch (tag=tag@entry=0xe7f0, func=func@entry=0x502a50 <command_loop_2>, arg=arg@entry=0x90) at eval.c:1184
#16 0x0000000000502a0f in command_loop () at keyboard.c:1111
#17 0x00000000005082cc in recursive_edit_1 () at keyboard.c:720
#18 0x0000000000508613 in Frecursive_edit () at keyboard.c:803
#19 0x0000000000425a73 in main (argc=2, argv=<optimized out>) at emacs.c:2310

I don't know how to debug these problems with native compilation, and it
always takes a few days for the bug to appear, but I've never seen it in
a build without native compilation.  While builds with native
compilation reliably crash after 3 to 4 days.

Thanks.


In GNU Emacs 28.0.60 (build 1, x86_64-pc-linux-gnu, X toolkit, cairo version 1.17.4, Xaw3d scroll bars)
 of 2021-11-09 built on computer
Repository revision: 307d164c9c053501df8159719b3b502d720a945a
Repository branch: emacs-28
Windowing system distributor 'The X.Org Foundation', version 11.0.12101002
System Description: Fedora 34 (Workstation Edition)

Configured using:
 'configure --with-x-toolkit=lucid --with-native-compilation
 --cache-file=/tmp/ccache'

Configured features:
ACL CAIRO DBUS FREETYPE GIF GLIB GMP GNUTLS GSETTINGS HARFBUZZ JPEG JSON
LCMS2 LIBSELINUX LIBSYSTEMD LIBXML2 MODULES NATIVE_COMP NOTIFY INOTIFY
PDUMPER PNG RSVG SECCOMP SOUND THREADS TIFF TOOLKIT_SCROLL_BARS X11
XAW3D XDBE XIM XPM LUCID ZLIB

bug#51864: 28.0.60; Emacs crashes in a few days under native-compilation

[Eli Zaretskii]

> Date: Mon, 15 Nov 2021 19:41:16 +0800
> From:  Po Lu via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
> 
> 
> After around the 3 to 4 day mark of running a build with native
> compilation, Emacs crashes with a mysterious backtrace consisting of a
> NULL pointer dereference at some random spot:

Thanks, I installed a fix.

bug#51864: 28.0.60; Emacs crashes in a few days under native-compilation

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

> Thanks, I installed a fix.

It's been 4 days since I installed the fix, and it seems to be working
perfectly (3 day, 21 hour uptime).

So I think it's safe to close this bug now.  Thanks for all the hard
work you've been doing on Emacs.

bug#51864: 28.0.60; Emacs crashes in a few days under native-compilation

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: 51864-done@debbugs.gnu.org
> Date: Fri, 19 Nov 2021 20:44:58 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Thanks, I installed a fix.
> 
> It's been 4 days since I installed the fix, and it seems to be working
> perfectly (3 day, 21 hour uptime).
> 
> So I think it's safe to close this bug now.  Thanks for all the hard
> work you've been doing on Emacs.

Thanks for testing, let's hope this is the right fix.  (There's one
more similar place in another function, but I will make the change
there only on master.)