I sent \x1b[6n but never get a reply — what's wrong?

Three usual causes. (1) The tty is in cooked mode — the reply (\x1b[ ; R) arrives on stdin and the line discipline holds it until you press Enter. Switch to raw or stty -icanon min 0 time N for a non-blocking read with timeout. (2) The read happens *before* the reply lands — the terminal answer can take dozens of milliseconds, so always pair the request with a bounded read (or select() with timeout) and a fallback. (3) You're inside tmux/screen without passthrough — the multiplexer answers on behalf of the inner pane and may report its own size instead of the host terminal's; for the host you need passthrough or a direct query before the multiplexer attaches. Always implement a 200 ms timeout + ioctl(TIOCGWINSZ) fallback so a missing reply doesn't hang the program.

Should I use DSR \x1b[6n or DA \x1b[c to detect terminal capabilities?

Neither directly — they answer different questions. DSR-6n returns the cursor position, useful for measuring screen size (with the \x1b[9999;9999H trick) but says nothing about feature support. DA (\x1b[c) returns a device-attributes string identifying the emulator class (?1;2c, ?64;..., etc.) — a coarse capability hint but not feature-specific. For "does this terminal support synchronized output / focus events / SGR mouse?" you want DECRQM (\x1b[? $p, see the decrqm slug) which asks about a specific DEC private mode and gets a 0/1/2/3/4 answer. Use the three together when fingerprinting: DA for class, DECRQM for per-feature gates, DSR-6n for size when TIOCGWINSZ isn't available.

DSR — Device Status Report (CSI 5n / CSI 6n)

Ask for terminal status (5n) or the current cursor position (6n) — the reverse channel TUIs use to size the terminal.

Byte forms

Every common string-literal form so you can paste-and-search either direction.

\\x1b[\x1b[5n (status request) \x1b[6n (cursor pos request)

\\033[\033[5n / \033[6n

\\e[\e[5n / \e[6n

ESC [ESC [ Ps n

hex1b 5b <Ps> 6e

Description

Device Status Report. The application sends a request; the terminal answers on stdin. \x1b[5n — "are you OK?". Reply: \x1b[0n (ready) or \x1b[3n (malfunction). Used as a liveness check after enabling exotic modes. \x1b[6n (a.k.a. CPR — Cursor Position Report) — "where is the cursor?". Reply: \x1b[<row>;<col>R. The classic trick to discover terminal size without TIOCGWINSZ is to move the cursor to the end of the screen (\x1b[9999;9999H), DSR-6n, parse the reply, then restore. Modern apps prefer ioctl(TIOCGWINSZ) + SIGWINCH for resizes, but DSR-6n remains the only portable answer on systems without that ioctl (over raw serial, certain CI environments). xterm extension: \x1b[?6n returns the cursor position constrained by the origin-mode flag (DECOM).

Spec citation: ECMA-48 §8.3.35 (DSR) / xterm-ctlseqs

Examples

bash

# Discover terminal size without TIOCGWINSZ:\nstty -echo raw min 0 time 5\nprintf '\033[s\033[9999;9999H\033[6n\033[u'\nIFS=';' read -r -d R esc_row col; stty sane\necho "rows=${esc_row#*[} cols=$col"

python

import sys; sys.stdout.write('\x1b[6n')   # reply arrives on stdin in raw mode

fmt.Print("\x1b[6n")

javascript

process.stdout.write('\x1b[6n')

printf("\x1b[6n"); fflush(stdout); /* parse <row>;<col>R from stdin */

Used in

Real-world tools that emit this sequence — anchors the bytes to commands you've already used.

vim, neovim startup probes`:checkhealth` emits `\x1b[6n` to confirm cursor-position reporting works before enabling block-cursor / mouse features
resize(1) (xterm utility)the canonical implementation of "discover terminal size without ioctl" — moves cursor to `\x1b[9999;9999H`, emits DSR-6n, parses the reply, exports `COLUMNS` and `LINES`
stty -a fallback probesraw `printf '\033[6n'` + bounded `read` is the portable fallback for cases where `TIOCGWINSZ` isn't available (CI runners, serial consoles)
powerlevel10k, starship right-promptscomplex multi-line prompts use DSR-6n to reconcile their wcwidth estimate against the terminal's actual cursor advance
asciinema, ttyrec — capture DSR replies in the recorded session so playback is dimension-accurate

Frequently asked

Short answers to the questions developers actually search for this sequence.

I sent \x1b[6n but never get a reply — what's wrong?: Three usual causes. (1) The tty is in cooked mode — the reply (\x1b[<row>;<col>R) arrives on stdin and the line discipline holds it until you press Enter. Switch to raw or stty -icanon min 0 time N for a non-blocking read with timeout. (2) The read happens *before* the reply lands — the terminal answer can take dozens of milliseconds, so always pair the request with a bounded read (or select() with timeout) and a fallback. (3) You're inside tmux/screen without passthrough — the multiplexer answers on behalf of the inner pane and may report its own size instead of the host terminal's; for the host you need passthrough or a direct query before the multiplexer attaches. Always implement a 200 ms timeout + ioctl(TIOCGWINSZ) fallback so a missing reply doesn't hang the program.
Should I use DSR \x1b[6n or DA \x1b[c to detect terminal capabilities?: Neither directly — they answer different questions. DSR-6n returns the cursor position, useful for measuring screen size (with the \x1b[9999;9999H trick) but says nothing about feature support. DA (\x1b[c) returns a device-attributes string identifying the emulator class (?1;2c, ?64;..., etc.) — a coarse capability hint but not feature-specific. For "does this terminal support synchronized output / focus events / SGR mouse?" you want DECRQM (\x1b[?<mode>$p, see the decrqm slug) which asks about a specific DEC private mode and gets a 0/1/2/3/4 answer. Use the three together when fingerprinting: DA for class, DECRQM for per-feature gates, DSR-6n for size when TIOCGWINSZ isn't available.

Common typos / look-alikes

Wrong forms developers commonly paste from Stack Overflow that look right but aren't — with the canonical correction inline.

\x1b[6n from a non-TTY context (stdout redirected to file / pipe)guard with isatty(STDIN_FILENO) && isatty(STDOUT_FILENO) before emitting
DSR is a *query* — the terminal writes the reply (\x1b[<row>;<col>R) back into the program's stdin. From a pipeline (mycli | tee out.log), stdin still points to the user's TTY but the terminal's reply collides with whatever the user types; from a fully-redirected context (mycli < /dev/null > out.log), the reply has nowhere to go and the cursor query hangs. Always gate \x1b[6n on both stdin AND stdout being TTYs; fall back to tput cols / $LINES / $COLUMNS otherwise.
\x1b[6n then read(2) only the first byte (assuming fixed reply length)read until R (0x52); parse ^\x1b\[(\d+);(\d+)R$ — reply length is variable
The reply \x1b[<row>;<col>R has variable length: row / col are decimal digit strings of unknown width (could be 1–4 digits each on a tall terminal). Fixed-length reads either truncate (leaving R and trailing bytes in the buffer for the next read to mistakenly consume) or block waiting for bytes that never come. Read until you see the literal R final byte, then parse with a regex anchored on the \x1b[ prefix. Set a 100-200 ms timeout so terminals that don't support DSR don't hang the program.
treating \x1b[6n as DA — sending it to detect 256-colour supportuse DA \x1b[c for device attributes; DA3 / XTVERSION \x1b[>q for terminal name+version
DSR (\x1b[6n) reports *cursor position*, not terminal capabilities. To probe terminal features use DA1 \x1b[c (returns service-class identification — e.g. \x1b[?64;1;2;6;9;15;18;21;22c listing implemented features), DA2 \x1b[>c (firmware id), DA3 \x1b[=c / XTVERSION \x1b[>q (terminal name + version). The replies have completely different shapes — DSR ends in R, DA1 ends in c. Conflating them returns a single number pair instead of a capability list and a confused detector skips colour support entirely.

Terminal support

xterm: yes
Linux console (fbcon): yes
macOS Terminal.app: yes
iTerm2: yes
Windows Terminal: yes
cmd.exe / ConPTY: yes
kitty: yes
alacritty: yes
WezTerm: yes
Ghostty: yes
GNOME Terminal: yes
Konsole: yes
tmux: yes
GNU screen: yes

xterm	Linux console (fbcon)	macOS Terminal.app	iTerm2	Windows Terminal	cmd.exe / ConPTY	kitty	alacritty	WezTerm	Ghostty	GNOME Terminal	Konsole	tmux	GNU screen
yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes	yes

Related sequences

In the family cookbook

CSI cookbook · 4. Queries — DA, DA2, DSR, DECRQM