ctrl+c in notcurses-input on linux results in segfault

[dankamongmen]

Looking to see if #2840 could be reproduced on linux, i ran notcurses-input 3.0.13 in kitty. upon pressing ctrl+c, we immediately crashed.

[dankamongmen]

           PID: 1390889 (notcurses-input)
           UID: 1000 (dank)
           GID: 1000 (dank)
        Signal: 11 (SEGV)
     Timestamp: Tue 2025-01-14 13:40:59 EST (2min 18s ago)
  Command Line: notcurses-input
    Executable: /usr/bin/notcurses-input
 Control Group: /user.slice/user-1000.slice/user@1000.service/kitty-1390312-0.scope
          Unit: user@1000.service
     User Unit: kitty-1390312-0.scope
         Slice: user-1000.slice
     Owner UID: 1000 (dank)
       Boot ID: 59b712c9b389474dbd4353ad59c9c7a0
    Machine ID: 61e1fce29000497e8788854e624c7467
      Hostname: schwarzgerat
       Storage: /var/lib/systemd/coredump/core.notcurses-input.1000.59b712c9b389474dbd4353ad59c9c7a0.1390889.1736880059000000.zst (present)
  Size on Disk: 527.5K
       Message: Process 1390889 (notcurses-input) of user 1000 dumped core.
                
                Module libsystemd.so.0 from deb systemd-257.2-1.amd64
                Module libgcc_s.so.1 from deb gcc-14-14.2.0-13.amd64
                Module libstdc++.so.6 from deb gcc-14-14.2.0-13.amd64
                Stack trace of thread 1390913:
                #0  0x00007f5e4110bb8e drop_signals (libnotcurses-core.so.3 + 0x5fb8e)
                #1  0x00007f5e410e41e8 notcurses_stop_minimal (libnotcurses-core.so.3 + 0x381e8)
                #2  0x00007f5e4110b904 fatal_handler (libnotcurses-core.so.3 + 0x5f904)
                #3  0x00007f5e40c49da0 __restore_rt (libc.so.6 + 0x3fda0)
                #4  0x00007f5e40c9dd8c __pthread_kill_implementation (libc.so.6 + 0x93d8c)
                #5  0x00007f5e40c49d02 __GI_raise (libc.so.6 + 0x3fd02)
                #6  0x00007f5e410d0560 kitty_kbd_txt (libnotcurses-core.so.3 + 0x24560)
                #7  0x00007f5e410d0e5d kitty_kbd (libnotcurses-core.so.3 + 0x24e5d)
                #8  0x00007f5e410d1220 process_escape (libnotcurses-core.so.3 + 0x25220)
                #9  0x00007f5e410d27a8 process_escape (libnotcurses-core.so.3 + 0x267a8)
                #10 0x00007f5e410d30f9 process_ibuf (libnotcurses-core.so.3 + 0x270f9)
                #11 0x00007f5e40c9c043 start_thread (libc.so.6 + 0x92043)
                #12 0x00007f5e40d1a778 __clone3 (libc.so.6 + 0x110778)
                
                Stack trace of thread 1390889:
                #0  0x00007f5e40c989de __futex_abstimed_wait_common64 (libc.so.6 + 0x8e9de)
                #1  0x00007f5e40c9b280 __pthread_cond_wait_common (libc.so.6 + 0x91280)
                #2  0x00007f5e410d45ea internal_get (libnotcurses-core.so.3 + 0x285ea)
                #3  0x000055bc0dd165da n/a (n/a + 0x0)
                ELF object binary architecture: AMD x86-64

[dankamongmen]

ick, this might be a regression from the work in 3.0.12 for #2828

[dankamongmen]

ok, got the problem here, we need handle the altstack == NULL case in drop_signals(), from when we come in via the fatal signal handler. but we're still eliciting the segfault.

[dankamongmen]

alright, well now at least we're not leaving the terminal useless. but why are we crashing? it's in shutdown; it comes after the shutdown process starts:

mark_pipe_ready:475:wrote to readiness pipe
internal_get:2728:draining event readiness pipe 0
reset_term_palette:82:restoring palette via xtpopcolors
notcurses_stop_minimal:173:restored terminal, returning 2
[schwarzgerat](130) $

[dankamongmen]

oh, that's not a segfault, that's SIGINT and correct behavior.

[dankamongmen]

and getting the 130 exit code is also correct for SIGINT (128 + 2). so we're good now.

[dankamongmen]

ok so i think we're done here!