Fish crashes if I type anything after Ctrl-C

[spacecowboy]

fish version 4.0.1

uname: Darwin Jonass-MacBook-Pro.local 24.3.0 Darwin Kernel Version 24.3.0: Thu Jan 2 20:24:24 PST 2025; root:xnu-11215.81.4~3/RELEASE_ARM64_T6030 arm64

TERM: xterm-ghostty

Additional info: using Starship for my prompt.

I often crash fish when I hit Ctrl-C and type anything after killing something like playwright tests.

In addition, after killing playwright with Ctrl-C, and I hit up to see my previous command, instead I see ^[[A printed. If I hit enter, then it runs the previous command but I have no idea what that is before I press enter.

Was unable to capture it with asciicinema. Including screenshots though.

Please let me know if and how I can provide for info.

In the first screenshot I hit Ctrl-C. Then I hit Ctrl-R because I want to run a previous command and I start typing ligh, then I hit enter because I notice the jibberish.

In the second, I hit up right after Ctrl-C.

[faho]

I can't reproduce this. Tested on 4.0.1.

Can you check if something in your configuration triggers it? Is there any other panic message that points to a different line?

reader.rs:4823 is

fish-shell/src/reader.rs

Line 4823 in 67b6aff

let history_pager = zelf.history_pager.as_mut().unwrap();

, which isn't really helpful.

In addition, after killing playwright with Ctrl-C, and I hit up to see my previous command, instead I see ^[[A printed.

That sounds like fish hasn't reacquired the terminal - if it did, it would both unset echo (which is what causes the terminal to show ^[[A) and enable the kitty keyboard protocol in your case (which would show a different key).

Can you reproduce this with something simpler like cat? It's possible playwright is misbehaving (some node tools like to run multiple processes that all acquire the terminal, which can lead to some remaining running).

[krobelus]

In addition, after killing playwright with Ctrl-C, and I hit up to see my previous command, instead I see ^[[A printed. If I hit enter, then it runs the previous command but I have no idea what that is before I press enter.

that's weird, if the child process actually exited that should never happen. I guess ps could tell us.
Note that you can press ctrl-l to force fish to redraw the commandline

[faho]

that's weird, if the child process actually exited that should never happen

#8947 - some node things run child processes, abandon them and have them also fiddle with the terminal.

They sometimes exit after the main process and restore their terminal modes before they do, so there's a fundamental race.

[spacecowboy]

It would not surprise me at all if this is node-specific. I'll see if I'm able to reproduce this with any other command and update here if am able to do so.

[merement]

Observed the same crash

> fish --version
fish, version 4.0.1
> uname -a
Linux desk-main 6.12.21-4-MANJARO #1 SMP PREEMPT_DYNAMIC Tue, 01 Apr 2025 18:45:30 +0000 x86_64 GNU/Linux

This was observed in the following context. I'm experimenting with a microcontroller dev-board monitored by espflash monitor. The microcontroller hangs, espflash freezes and does not respond to Ctrl-C. Pressing the RESET button on the board usually results in fish reporting Broken pipe and getting back to the prompt. Every once in a while, however, fish crashes with the same error message (different PID, of course).

The crash is not easy to reproduce. First, it happened fairly often, but after I realized that this was not one of my reader.rs that reported the error, it happened only once out of a few dozen deliberate attempts to reproduce. I have a screenshot, but it's not really informative.

Forgot to add. This is a relatively fresh install, and the fish configuration is pretty vanilla: the only changes were the "ayu_dark" color scheme and "Informative" prompt set using fish_config. I'm using KDE, with the default konsole configuration.

[krobelus]

On Mon, Apr 28, 2025 at 09:58:12PM -0700, Misha wrote: merement left a comment (fish-shell/fish-shell#11355) Observed the same crash ``` > fish --version fish, version 4.0.1

Ok. The crash itself should be fixed in 4.0.2 (see the milestone label)

[vaibhavshn]

I am still facing this on 4.6.0, which is latest as of now

[krobelus]

@vaibhavshn can you post in a new issue the crash stacktrace and maybe steps to reproduce?