bug#38912: 27.0.60; PDumper meets segmentation fault when evil is loaded

[Eli Zaretskii <eliz@gnu.org>]

Ping!  Any news in debugging this?

> Date: Mon, 6 Jan 2020 09:10:20 -0800
> From: "Daniel Colascione" <dancol@dancol.org>
> Cc: "NiwTinray" <niwtrx@icloud.com>,
>  "Daniel Colascione" <dancol@dancol.org>,
>  38912@debbugs.gnu.org
> 
> > [Please use "Reply All" to reply, so that the bug address is kept on
> > the CC list.]
> >
> >> From: NiwTinray <niwtrx@icloud.com>
> >> Date: Sun, 5 Jan 2020 13:25:07 +0800
> >>
> >> > I cannot reproduce this from "emacs -Q" because 'use-package' is not a
> >> > known function.  Can you show a recipe starting from "emacs -Q",
> >> > please?
> >>
> >> Here. I've attached a minimal script file that helps reproduce this bug.
> >>
> >> (require 'package)
> >> (package-initialize)
> >> (add-to-list 'package-archives
> >>              '("melpa-stable" . "https://stable.melpa.org/packages/") t)
> >> (unless (package-installed-p 'evil)
> >>   (package-refresh-contents)
> >>   (package-install 'evil))
> >> (require 'evil)
> >> (dump-emacs-portable "/tmp/test.pdmp")
> >>
> >> The script downloads the package "evil" from Melpa stable, load the evil
> >> package
> >> and dumps an image to /tmp/test.pdmp.
> >>
> >> > Also, does this happen if you add -Q to the Emacs invocation after
> >> > dumping?  If not, there's more detail missing in your report: the
> >> > customizations in your init files.
> >>
> >>
> >> Sure. Please download this file, and run the command:
> >>
> >> emacs --batch -Q --script evil.el
> >>
> >> To see the bug happen, load the test.pdmp file:
> >>
> >> emacs -Q --dump-file /tmp/test.pdmp
> >>
> >> You should see a segmentation fault:
> >>
> >> [1]    23369 segmentation fault (core dumped)  emacs -Q --dump-file
> >> /tmp/test.pdmp
> >>
> >> I run debugger inside src/.gdbinit using the command:
> >>
> >> gdb -x .gdbinit --args ./emacs --dump-file /tmp/test.pdmp
> >>
> >> And logged backtrace. See my second attachment:
> >>
> >> (base) omnisky :: ~/emacs/src ‹emacs-27*› » gdb -x .gdbinit --args
> >> ./emacs --dump-file /tmp/test.pdmp
> >> GNU gdb (Ubuntu 7.11.1-0ubuntu1~16.5) 7.11.1
> >> Copyright (C) 2016 Free Software Foundation, Inc.
> >> License GPLv3+: GNU GPL version 3 or later
> >> <http://gnu.org/licenses/gpl.html>
> >> This is free software: you are free to change and redistribute it.
> >> There is NO WARRANTY, to the extent permitted by law.  Type "show
> >> copying"
> >> and "show warranty" for details.
> >> This GDB was configured as "x86_64-linux-gnu".
> >> Type "show configuration" for configuration details.
> >> For bug reporting instructions, please see:
> >> <http://www.gnu.org/software/gdb/bugs/>.
> >> Find the GDB manual and other documentation resources online at:
> >> <http://www.gnu.org/software/gdb/documentation/>.
> >> For help, type "help".
> >> Type "apropos word" to search for commands related to "word"...
> >> Reading symbols from ./emacs...done.
> >> warning: File "/home/ntr/emacs/src/.gdbinit" auto-loading has been
> >> declined by your `auto-load safe-path' set to
> >> "$debugdir:$datadir/auto-load".
> >> To enable execution of this file add
> >> 	add-auto-load-safe-path /home/ntr/emacs/src/.gdbinit
> >> line to your configuration file "/home/ntr/.gdbinit".
> >> To completely disable this security protection add
> >> 	set auto-load safe-path /
> >> line to your configuration file "/home/ntr/.gdbinit".
> >> For more information about this security protection see the
> >> "Auto-loading safe path" section in the GDB manual.  E.g., run from the
> >> shell:
> >> 	info "(gdb)Auto-loading safe path"
> >> SIGINT is used by the debugger.
> >> Are you sure you want to change it? (y or n) [answered Y; input not from
> >> terminal]
> >> Environment variable "DISPLAY" not defined.
> >> TERM = xterm-24bits
> >> Breakpoint 1 at 0x411df0: file emacs.c, line 370.
> >> Breakpoint 2 at 0x4bfe60: file xterm.c, line 10130.
> >> (gdb) r
> >> Starting program: /home/ntr/emacs/src/emacs --dump-file /tmp/test.pdmp
> >> /home/ntr/emacs/src/emacs:
> >> /raid_sdc/home/ntr/anaconda3/lib/libtiff.so.5: no version information
> >> available (required by /home/ntr/emacs/src/emacs)
> >> [Thread debugging using libthread_db enabled]
> >> Using host libthread_db library
> >> "/lib/x86_64-linux-gnu/libthread_db.so.1".
> >>
> >> Program received signal SIGSEGV, Segmentation fault.
> >> 0x00000000004f12d0 in Fcurrent_active_maps (olp=olp@entry=XIL(0x30),
> >> position=position@entry=XIL(0)) at keymap.c:1541
> >> 1541	      && NILP (KVAR (current_kboard,
> >> Voverriding_terminal_local_map))
> >> (gdb) xbacktrace
> >> "key-binding" (0xffffd5c8)
> >> "turn-on-undo-tree-mode" (0xffffd758)
> >> "global-undo-tree-mode-enable-in-buffers" (0xffffd948)
> >> "run-hooks" (0xffffd9e8)
> >> "run-mode-hooks" (0xffffdbc0)
> >> "minibuffer-inactive-mode" (0xffffdd40)
> >> (gdb)
> >
> > In my debug build of Emacs 27.0.60 I get an assertion violation while
> > dumping, which probably is already a sign of trouble.
> >
> > Daniel, any ideas?
> 
> I haven't had a chance over the past few days to repro this problem, but I
> hope to do so this weeekend. The messages about the assertion failure
> *during* dumping do seem likely unrelated. The easiest way to debug this
> particular crash is with rr. Run the original test as "rr record emacs
> [args]", then run "rr replay". The latter will dump you in a GDB prompt.
> Type "cont" and run the replay of Emacs until you get to the SIGSEGV. Run
> "watch -l [expression]" to break out of execution whenever that value
> changes, then run "reverse-cont" to run the replay *backwards* until you
> get to the code that changed the variable with the bad value.
> 
>