bug#74663: Emacs crashes with Segfault while doing basic code editing

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Zan Owsley]

Hi,


I'm getting these crashes periodically while simply doing some code 
editing. Corfu is active and Eglot is communicating with a language 
server, although this has occurred with a few languages and different 
language servers already. I have not been able to reproduce this with my 
init.el inactive, but since very little of my config seems to be in play 
here, I'm hoping this information is still helpful. My config has been 
super stable for quite a while too, although this problem has shown up 
only since moving to Arch linux from Ubuntu. The backtraces always look 
like this:


GDB 'bt full':


(gdb) bt full
#0  0x00005555557d1be5 in Fget ()
No symbol table info available.
#1  0x000055555571ede2 in parse_modifiers ()
No symbol table info available.
#2  0x0000555555736b49 in ?? ()
No symbol table info available.
#3  0x0000555555724b4b in read_char ()
No symbol table info available.
#4  0x00005555557f5613 in ?? ()
No symbol table info available.
#5  0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#6  0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#7  0x00005555557cc741 in Fapply ()
No symbol table info available.
#8  0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#9  0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#10 0x000055555575d913 in Fall_completions ()
No symbol table info available.
#11 0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#12 0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#13 0x00005555557cc741 in Fapply ()
No symbol table info available.
#14 0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#15 0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#16 0x00005555557cc741 in Fapply ()
No symbol table info available.
#17 0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#18 0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#19 0x00005555557cca2b in Fapply ()
No symbol table info available.
#20 0x0000555555819136 in exec_byte_code ()
No symbol table info available.
#21 0x00005555557cbfbd in Ffuncall ()
No symbol table info available.
#22 0x000055555571c5a1 in timer_check ()
No symbol table info available.
#23 0x000055555571cb86 in ?? ()
No symbol table info available.
#24 0x0000555555727bd7 in ?? ()
No symbol table info available.
#25 0x000055555582680b in wait_reading_process_output ()
No symbol table info available.
#26 0x00005555555cf1cc in sit_for ()
No symbol table info available.
#27 0x0000555555725917 in read_char ()
No symbol table info available.
#28 0x000055555572d3c2 in ?? ()
No symbol table info available.
#29 0x0000555555718259 in ?? ()
No symbol table info available.
#30 0x00005555557b96d6 in internal_condition_case ()
No symbol table info available.
#31 0x00005555557176ef in command_loop_2 ()
No symbol table info available.
#32 0x00005555557b9619 in internal_catch ()
No symbol table info available.
#33 0x0000555555717ba8 in ?? ()
No symbol table info available.
#34 0x0000555555717cbc in recursive_edit_1 ()
No symbol table info available.
#35 0x0000555555717eb2 in Frecursive_edit ()
No symbol table info available.
#36 0x00005555555c2e3b in main ()
No symbol table info available.


GDB 'xbacktrace' command is unavailable. (I am using gdb for the first 
time to provide this info, so I'm not knowledgeable at all.)


I hope this information is helpful is some way. If there's anything else 
I can provide, please let me know!


Best Regards,


Zan Owsley

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Eli Zaretskii]

> Date: Mon, 2 Dec 2024 11:35:38 -0700
> From: Zan Owsley <zanowsley@gmail.com>
> 
> I'm getting these crashes periodically while simply doing some code 
> editing. Corfu is active and Eglot is communicating with a language 
> server, although this has occurred with a few languages and different 
> language servers already. I have not been able to reproduce this with my 
> init.el inactive, but since very little of my config seems to be in play 
> here, I'm hoping this information is still helpful. My config has been 
> super stable for quite a while too, although this problem has shown up 
> only since moving to Arch linux from Ubuntu. The backtraces always look 
> like this:
> 
> 
> GDB 'bt full':
> 
> 
> (gdb) bt full
> #0  0x00005555557d1be5 in Fget ()
> No symbol table info available.
> #1  0x000055555571ede2 in parse_modifiers ()
> No symbol table info available.
> #2  0x0000555555736b49 in ?? ()
> No symbol table info available.
> #3  0x0000555555724b4b in read_char ()

Thanks, I believe this is bug#71744, which was already solved for the
upcoming Emacs 30.  You didn't say which Emacs version you are using,
so I'm guessing it's not Emacs 30?  If so, upgrading or installing the
patch in

  https://debbugs.gnu.org/cgi/bugreport.cgi?bug=71744;att=5;msg=44;filename=0005-Don-t-ignore-Wclobbered-in-keyboard.c.patch

should solve this.

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Eli Zaretskii]

> From: Zan Owsley <zanowsley@gmail.com>
> Date: Tue, 3 Dec 2024 07:12:13 -0700
> 
> You'd think I would at least mention the version right? 🙂 Yeah it's the arch repo I'm using. Thanks - I'll check
> out v30!

Did you succeed in checking out Emacs 30?

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Zan Owsley]

Hi Eli,


Yes I did! It was the first time I built Emacs instead of using a 
package manager. I have tried 27, 28, and 29 before, but this version 
really blew me away. I couldn't believe the performance improvement I 
saw right away, and that's before I rebuilt it with native compilation. 
I have no idea how you sped it up like that, but it really feels nice. 
I've been enjoying 30 so much that it sparked another deep dive into my 
config and writing elisp for my own features this past week. So that's 
been fun.


I looked you up after receiving this email as I was wondering who would 
care to follow up with me. Saw that you're the co-maintainer and won the 
2023 FSF award (a belated congrats!) and figured you're just that kind 
of guy who wants to help. Thanks for all your work - it's clear, even to 
me, that Emacs 30 is a special release. Thanks for getting me to try it 
and good luck with the continuing work.



P.S. You might hear more from me in the future. I think I'd like to get 
involved with contributing to Emacs. I'm not elisp expert yet, but I'm 
getting better. :)

Zan

On 12/14/24 2:44 AM, Eli Zaretskii wrote:
>> From: Zan Owsley <zanowsley@gmail.com>
>> Date: Tue, 3 Dec 2024 07:12:13 -0700
>>
>> You'd think I would at least mention the version right? 🙂 Yeah it's the arch repo I'm using. Thanks - I'll check
>> out v30!
> Did you succeed in checking out Emacs 30?

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Corwin Brust]

On Sat, Dec 14, 2024 at 10:04 AM Zan Owsley <zanowsley@gmail.com> wrote:
>
> P.S. You might hear more from me in the future. I think I'd like to get
> involved with contributing to Emacs. I'm not elisp expert yet, but I'm
> getting better. :)
>

Zan,

What a wonderful email - thank you for your kind words toward Eli and
all the hard-work that has gone into Emacs to prepare for the release
of Emacs 30!  I completely agree and it's nice to see these messages
of encouragement for the thoughtful and diligent devs that bring us
Emacs :)

Can I invite you to complete the Copyright Assignment paperwork to
grant the FSF ownership of any code/changes you provide for Emacs?
This is nearly always a requirement to merge your contributions.
Meanwhile, "doing the paperwork" doesn't obligate you to contribute:
it just removes a barrier should you wish to do so.

If you'd like to do that please reply back to say so and, I expect,
Eli (or another Maint) will send over the forms by private email.

Thanks again for sharing your comments; glad you are loving Emacs 30!

bug#74663: Emacs crashes with Segfault while doing basic code editing

[Eli Zaretskii]

> Date: Sat, 14 Dec 2024 09:04:01 -0700
> Cc: 74663@debbugs.gnu.org
> From: Zan Owsley <zanowsley@gmail.com>
> 
> Yes I did! It was the first time I built Emacs instead of using a 
> package manager. I have tried 27, 28, and 29 before, but this version 
> really blew me away. I couldn't believe the performance improvement I 
> saw right away, and that's before I rebuilt it with native compilation. 
> I have no idea how you sped it up like that, but it really feels nice. 
> I've been enjoying 30 so much that it sparked another deep dive into my 
> config and writing elisp for my own features this past week. So that's 
> been fun.

OK, that's good news.  I guess the bug can be closed now?

Regarding the speed-up you see: Emacs 30 is indeed somewhat faster,
but I doubt that it is revolutionary faster.  I'm guessing that the
improvements you see are at least in part due to the different
configuration and build details.

> I looked you up after receiving this email as I was wondering who would 
> care to follow up with me. Saw that you're the co-maintainer and won the 
> 2023 FSF award (a belated congrats!) and figured you're just that kind 
> of guy who wants to help. Thanks for all your work - it's clear, even to 
> me, that Emacs 30 is a special release. Thanks for getting me to try it 
> and good luck with the continuing work.

Thanks, but I'm not the only one to take the credit.  This is a team
work, and we have a great team.