[BUG]Panic: "chunk size must be non-zero" when using 'edit' over multiple SSH sessions

[ha7v]

Bug Description

edit panics with a chunk size must be non-zero error. This is reliably reproducible when connected to a Termux instance via two or more concurrent SSH sessions. The issue seems related to framebuffer initialization when multiple pseudo-terminals (TTYs) are active.

Steps to Reproduce

1. Start sshd on Termux.
2. Establish two or more concurrent SSH connections from any client.
3. In one of the sessions, run edit <any-file>.
4. The program panics. (May require a few attempts).

Actual Behavior

The program panics and crashes immediately, returning the shell prompt with a "zsh: abort" message. The following is the full stack backtrace captured with the RUST_BACKTRACE=full environment variable:

Environment

• Server: Termux v0.118.3(1002)
• Client: Windows PowerShell (and likely any other SSH client)
• Connection Protocol: OpenSSH_10.2p1, OpenSSL 3.5.2 5 Aug 2025 (sshd running on Termux)
• Crashing Program: edit v1.2.1

[DHowett]

multiple pseudo-terminals

An application can categorically not have more than one PTY hooked up to standard input. Judging by the spill on the prompt, I somewhat suspect that your multiplexer of choice is returning a corrupt response to one of the queries edit runs on startup. We should be robust, but also, it should not do that.

[ha7v]

Thanks for the reply! I am new here, so I'll give tmux a try.