bug#63495: 28.2; menu crashes on macos

[Alan Third]

On Tue, Jul 11, 2023 at 08:40:46AM +0300, Eshel Yaron wrote:
> Hi,
> 
> Alan Third <alan@idiocy.org> writes:
> 
> > It's interesting that the crash is happening in
> > runMenuAt:forFrame:keymaps: with it trying to access memory at address
> > 0x3. There's nothing obvious in that function that would be doing
> > that, so it would be good if we could get a backtrace from a debugger
> > showing the exact line that's causing the problem and, if possible,
> > which variable has the value 3.
> 
> There's this backtrace that I've posted in Bug#62402, HTH:
> 
> > * thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS 
> > (code=1, address=0x3)
> >   * frame #0: 0x00000001000aebad emacs`AREF(array=0x0000000000000000, 
> > idx=0) at lisp.h:1947:10
> >     frame #1: 0x00000001000af660 
> > emacs`find_and_return_menu_selection(f=0x00000001100dee30, keymaps=true, 
> > client_data=0x000000011089b888) at menu.c:985:11
> >     frame #2: 0x0000000100380f2b emacs`-[EmacsMenu 
> > runMenuAt:forFrame:keymaps:](self=0x00006000017007c0, 
> > _cmd="runMenuAt:forFrame:keymaps:", p=(x = 2, y = 506), 
> > f=0x00000001100dee30, keymaps=true) at nsmenu.m:767:9
> >     frame #3: 0x0000000100381f00 emacs`ns_menu_show(f=0x00000001100dee30, 
> > x=2, y=2, menuflags=1, title=0x0000000000000000, error=0x00007ff7bfefce80) 
> > at nsmenu.m:1067:9
> >     frame #4: 0x00000001000b1203 
> > emacs`x_popup_menu_1(position=0x000000011804dcb3, menu=0x000000011804e003) 
> > at menu.c:1410:17
> >     frame #5: 0x00000001000b15a2 
> > emacs`Fx_popup_menu(position=0x000000011804dcb3, menu=0x000000011804e003) 
> > at menu.c:1474:10
> >     frame #6: 0x0000000100247c58 emacs`eval_sub(form=0x000000011804dd23) at 
> > eval.c:2503:15
> >     ...
> > 
> > It seems that `find_and_return_menu_selection` in menu.c tries to
> > access the global variable `menu_items` before it's initialized.  I'm
> > not sure when or where it should be initialized though :(

No, sorry, I need the output from a debugger. That should give the
exact line where the failure happens.

Is anyone able to recreate this while running in a debugger? It can be
lldb, doesn't have to be gcc.
-- 
Alan Third