Files
Panama/config/dot/quickshell/scripts/panama-brightness
Gabriel Brown 816ea68f9a Read DDC/CI from the AUX bus, not the EDID bus
With I2C access granted, the helper still found nothing: every connected
monitor was probed on the wrong bus.

A connector has two. The `ddc` symlink points at the classic I2C line
that carries EDID on HDMI and DVI. DisplayPort carries DDC/CI over the
AUX channel instead, which appears as a child directory of the
connector. Both exist on a DP connector and both resolve, so the wrong
choice looks entirely reasonable and simply finds no monitor: this
machine's DP-2 has ddc -> i2c-5, where ddcutil reports "No monitor
detected", while its AUX child i2c-9 answers VCP 0x10 immediately.

This is the guess the previous commit said was unverified, and it was
wrong in the way that mattered. Enumerating from sysfs is still right --
it gives the connector name Hyprland uses and skips empty connectors --
but it has to prefer the AUX child and fall back to the symlink.

The fixture now mirrors sysfs properly: /sys/class/drm/<connector> is a
SYMLINK to the real device directory, and `find` does not follow the
path it is given. A fixture built from plain directories passes whether
or not the code resolves the symlink first, which is a test that agrees
with itself rather than with the kernel.

Verified on hardware: the Kuycon P20 reports 100%, accepts 70 and
returns to 100, and the media keys move it through codex's OSD path.
Both bus-selection mistakes are now caught by the contract.

Claude-Session: https://claude.ai/code/session_01BRvzt4H8XXLPVH5MyYdk9L
2026-08-18 09:51:55 -04:00

158 lines
6.5 KiB
Bash
Executable File

#!/usr/bin/env bash
# External monitor brightness over DDC/CI.
#
# A desktop with no backlight class device has no brightness control at all --
# brightnessctl only sees keyboard and NIC LEDs. The panel itself still has a
# brightness setting, reachable over the monitor's DDC/CI channel (VCP feature
# 0x10), which is what the buttons on the bezel drive.
#
# Usage:
# panama-brightness list -> {"displays":[...],"error":""}
# panama-brightness get <bus> -> integer percent
# panama-brightness set <bus> <pct> -> applies, prints nothing
#
# Displays are enumerated from sysfs rather than from `ddcutil detect`. The
# kernel publishes the connector-to-I2C-bus mapping directly, as
# /sys/class/drm/<card>-<connector>/ddc, along with whether anything is plugged
# in. That is better than parsing detect output in three ways: the format is
# stable where detect's brief output is undocumented, the connector name comes
# out exactly as Hyprland and the Displays page already spell it (DP-2), and
# only connectors with a monitor attached get probed -- one bus on this machine
# instead of fourteen, which is the difference between a fast scan and a slow
# one, since each probe of an empty bus waits for a timeout.
#
# No model name is reported. Hyprland already knows the human-readable
# description of every output, so the UI joins on the connector name rather than
# having two sources of truth for what a monitor is called.
set -uo pipefail
readonly VCP_BRIGHTNESS=0x10
# Test seams. The contract needs to exercise enumeration and parsing on a
# machine whose real monitors it must not touch, so both roots this script
# reads are overridable. Nothing sets them in normal use.
readonly DRM_ROOT="${PANAMA_BRIGHTNESS_DRM_ROOT:-/sys/class/drm}"
readonly DEV_ROOT="${PANAMA_BRIGHTNESS_DEV_ROOT:-/dev}"
emit_error() {
printf '{"displays":[],"error":%s}\n' "$(jq -Rn --arg e "$1" '$e')"
exit 0
}
command -v ddcutil >/dev/null 2>&1 || emit_error 'ddcutil is not installed'
# Reading a VCP value needs read/write access to the monitor's I2C bus. The
# udev rule ddcutil ships grants that to the seat user through uaccess, but only
# to devices created after the rule was installed -- so a machine that installed
# ddcutil without rebooting has the rule in place and no access to show for it.
# That is by far the most likely reason for an empty list, and it is fixable in
# one command, so say so rather than reporting "no displays".
has_accessible_bus() {
local dev
for dev in "$DEV_ROOT"/i2c-*; do
[[ -r "$dev" && -w "$dev" ]] && return 0
done
return 1
}
# The I2C bus that carries DDC/CI for one connector.
#
# There are two, and picking the wrong one finds no monitor at all:
#
# DisplayPort carries DDC/CI over the AUX channel. That adapter shows up as a
# child directory of the connector -- /sys/class/drm/card1-DP-2/i2c-9 -- and
# is the one ddcutil talks to.
#
# The connector's `ddc` symlink points at the classic I2C line used for EDID
# on HDMI and DVI. On a DisplayPort connector it still exists and still
# resolves, but nothing answers on it: this machine's DP-2 has ddc -> i2c-5,
# where ddcutil reports "No monitor detected", while i2c-9 answers VCP 0x10
# immediately.
#
# So prefer the AUX child and fall back to the symlink. Note the readlink: the
# entries under /sys/class/drm are symlinks, and `find` does not follow the path
# it is given, so searching the unresolved path silently finds nothing.
bus_for_connector() {
local path="$1" real aux ddc
real="$(readlink -f "$path")"
aux="$(find "$real" -maxdepth 1 -name 'i2c-*' -printf '%f' -quit 2>/dev/null)"
if [[ -n "$aux" ]]; then
printf '%s' "${aux#i2c-}"
return 0
fi
ddc="$(readlink -f "$path/ddc" 2>/dev/null)" || return 1
[[ -n "$ddc" ]] || return 1
ddc="$(basename "$ddc")"
printf '%s' "${ddc#i2c-}"
}
cmd_list() {
has_accessible_bus || emit_error 'no I2C bus is accessible. ddcutil ships a udev rule that grants this, but only to devices created after it was installed. Run: sudo udevadm control --reload-rules && sudo udevadm trigger --subsystem-match=i2c-dev --subsystem-match=drm'
local rows=() connector bus value path
for path in "$DRM_ROOT"/card*-*; do
[[ -e "$path/ddc" ]] || continue
[[ "$(cat "$path/status" 2>/dev/null)" == "connected" ]] || continue
# card1-DP-2 -> DP-2, the name Hyprland uses.
connector="$(basename "$path")"
connector="${connector#card*-}"
bus="$(bus_for_connector "$path")"
[[ "$bus" =~ ^[0-9]+$ ]] || continue
# A monitor that does not implement 0x10 is not an error; it simply
# cannot be controlled, and is left out rather than shown as a slider
# that does nothing.
value="$(cmd_get "$bus")" || continue
[[ -n "$value" ]] || continue
rows+=("$(jq -cn \
--argjson bus "$bus" \
--arg connector "$connector" \
--argjson value "$value" \
'{bus: $bus, connector: $connector, value: $value}')")
done
if [[ ${#rows[@]} -eq 0 ]]; then
emit_error 'no connected monitor reports DDC/CI brightness. Some panels implement it only when "DDC/CI" is enabled in their on-screen menu.'
fi
printf '{"displays":[%s],"error":""}\n' "$(IFS=,; printf '%s' "${rows[*]}")"
}
# Prints the current brightness as a whole percent, or nothing when the display
# cannot report it. `getvcp --brief` is documented as machine readable and
# answers "VCP 10 C <current> <max>"; the max is almost always 100 but is not
# guaranteed to be, so it is read rather than assumed.
cmd_get() {
local bus="$1" out current max
out="$(timeout 10 ddcutil --bus "$bus" getvcp "$VCP_BRIGHTNESS" --brief 2>/dev/null)" || return 1
read -r _ _ _ current max <<<"$out"
[[ "$current" =~ ^[0-9]+$ && "$max" =~ ^[0-9]+$ && "$max" -gt 0 ]] || return 1
printf '%s' "$(( current * 100 / max ))"
}
cmd_set() {
local bus="$1" percent="$2" max out
[[ "$percent" =~ ^[0-9]+$ ]] || return 1
(( percent > 100 )) && percent=100
out="$(timeout 10 ddcutil --bus "$bus" getvcp "$VCP_BRIGHTNESS" --brief 2>/dev/null)" || return 1
read -r _ _ _ _ max <<<"$out"
[[ "$max" =~ ^[0-9]+$ && "$max" -gt 0 ]] || max=100
timeout 10 ddcutil --bus "$bus" setvcp "$VCP_BRIGHTNESS" "$(( percent * max / 100 ))" >/dev/null 2>&1
}
case "${1:-list}" in
list) cmd_list ;;
get) cmd_get "${2:?bus required}" ;;
set) cmd_set "${2:?bus required}" "${3:?percent required}" ;;
*) printf 'usage: panama-brightness [list|get <bus>|set <bus> <percent>]\n' >&2; exit 2 ;;
esac