Kaj SchittecatandClaude Opus 5
820bc21f46
The T-Display P4 ships EIGHT images: panel (AMOLED/LCD) x radio (SX1262/LR2021)
x C6 stack (esp-hosted/ESP-AT). The C6 axis is new and it is the dangerous one,
because the two builds are identical apart from the artifact name they carry,
and flashing the wrong one at an older unit used to boot-loop the board.
scripts/build/p4-release.sh builds all eight and VERIFIES each image against the
OTA name baked into it before accepting it. That is the same string the device
builds its self-update URL from, so a disagreement with the filename is exactly
the failure that bricks people -- the T-Deck Max self-updating to Pro firmware,
and this table has been wrong three times. A mismatch aborts the release.
It earned that check immediately: building the eight by hand produced two
mislabelled images, and every mechanism meant to prevent it was itself broken.
* build.sh fullclean has NEVER worked from a plain shell. It ran `cmake -E
remove_directory` above the esp-idf/export.sh that puts cmake on PATH, so it
printed that it had removed the tree, exited 0, and removed nothing. Now a
plain rm -rf, and it verifies the directory is gone instead of assuming.
This matters beyond tidiness: the hosted and ESP-AT builds SHARE a build dir
(the C6 axis does not change P4_BUILD_TARGET), the backend is read from the
environment at CMake CONFIGURE time, and CMake does not reconfigure because
an env var changed -- so a clean that silently did nothing yields an
incremental build of the OTHER variant, with no configure output to show for
it. That is precisely how the two mislabelled images were produced.
* Deleting CMakeCache.txt alone is not a clean either; it leaves CMakeFiles/
behind and the next configure dies in CMakeDetermineCCompiler.
* Detecting failure by grepping a log for "error:" misses a failed CMake
configure, which prints "Configuring incomplete" and no such line -- and the
previous application.bin is still sitting in the directory to be picked up.
Exit codes now.
gen-flasher-meta.py gains the four ESP-AT entries, labelled "older unit, ESP-AT
C6": someone choosing in the web flasher has no serial log and needs to pick on
something they can check.
scripts/run-host-tests.sh runs test/ as one command. Nothing ever did, which is
why test_touch_prefs_schema had been failing to COMPILE since v64 appended
report_ping and report_done_n without updating its sums -- and a static_assert
that does not compile is not a guard. That one exists to catch the mid-struct
insert that broke v44 and shipped the poisoned beta_57 layout. Repaired, with a
note where the next person appending a field will read it. 17 pass, 0 fail, 2
skipped with their reasons stated rather than quietly dropped.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
162 lines
8.2 KiB
Bash
Executable File
162 lines
8.2 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
# Build every T-Display P4 release image and verify each one is what it claims.
|
|
#
|
|
# The P4 is not a PlatformIO env, so release.sh cannot build it: its bins are
|
|
# produced here and injected into the release tree afterwards. That used to be a
|
|
# handful of ad-hoc commands, which stopped being safe at EIGHT variants:
|
|
#
|
|
# panel AMOLED (RM69A10, default) or LCD (HI8561, WADA_P4_LCD=1)
|
|
# radio SX1262 (default) or LR2021 (WADA_P4_LR2021=1)
|
|
# C6 stack esp-hosted (default) or ESP-AT (WADA_P4_LEGACY_AT=1)
|
|
#
|
|
# The C6 axis is the dangerous one. Current V1 units ship the C6 with esp-hosted
|
|
# firmware; older units still carry the factory ESP-AT. Flash the wrong one and
|
|
# the C6 is unreachable -- which before b52c71f boot-looped the board, and still
|
|
# means no Wi-Fi and no BLE. The two builds are otherwise identical, so NOTHING
|
|
# about the binary tells you which it is except the artifact name it carries.
|
|
#
|
|
# TWO traps this script exists to close, both hit on 2026-10-05:
|
|
#
|
|
# 1. The CMake cache. Each variant has its own build dir, but a dir reused with
|
|
# different env vars keeps the OLD selection: a build asked for as hosted came
|
|
# out as ESP-AT, flag silently ignored. The cache is cleared per variant here.
|
|
# 2. Collecting from the wrong directory. build/tdisplay_p4 is only the AMOLED
|
|
# SX1262 variant; the others live in build/tdisplay_p4_lcd and friends.
|
|
# Copying the first one four times produces four identical "variants" that
|
|
# all look fine until someone flashes one.
|
|
#
|
|
# So every image is verified against the OTA name baked into it before it is
|
|
# accepted. That is the same string the device builds its self-update URL from,
|
|
# so if it disagrees with the filename the artifact is wrong in the way that
|
|
# matters -- a device would fetch the other variant's firmware. The T-Deck Max
|
|
# shipped exactly that bug (a Max self-updating to Pro firmware) and this table
|
|
# has been wrong three times. The check is cheap; a mismatch aborts the release.
|
|
#
|
|
# Usage:
|
|
# scripts/build/p4-release.sh beta_88 [outdir]
|
|
set -euo pipefail
|
|
|
|
TAG="${1:?usage: p4-release.sh <tag> [outdir]}"
|
|
ROOT="$(cd "$(dirname "$0")/../.." && pwd)"
|
|
OUT="${2:-$ROOT/out/p4_$TAG}"
|
|
P4="$ROOT/tdisplay_p4"
|
|
# esptool with ESP32-P4 support: the IDF env's, not whatever is on PATH.
|
|
IDFPY="$ROOT/tanmatsu/esp-idf-tools/python_env/idf5.5_py3.9_env/bin/python"
|
|
|
|
[ -x "$IDFPY" ] || { echo "ABORT: IDF python not found at $IDFPY" >&2; exit 1; }
|
|
mkdir -p "$OUT"
|
|
|
|
# name | build dir | env assignments ("-" = none)
|
|
VARIANTS=(
|
|
"wadamesh-tdisplay-p4|tdisplay_p4|-"
|
|
"wadamesh-tdisplay-p4-lr2021|tdisplay_p4_lr2021|WADA_P4_LR2021=1"
|
|
"wadamesh-tdisplay-p4-lcd|tdisplay_p4_lcd|WADA_P4_LCD=1"
|
|
"wadamesh-tdisplay-p4-lcd-lr2021|tdisplay_p4_lcd_lr2021|WADA_P4_LCD=1 WADA_P4_LR2021=1"
|
|
"wadamesh-tdisplay-p4-at|tdisplay_p4|WADA_P4_LEGACY_AT=1"
|
|
"wadamesh-tdisplay-p4-lr2021-at|tdisplay_p4_lr2021|WADA_P4_LR2021=1 WADA_P4_LEGACY_AT=1"
|
|
"wadamesh-tdisplay-p4-lcd-at|tdisplay_p4_lcd|WADA_P4_LCD=1 WADA_P4_LEGACY_AT=1"
|
|
"wadamesh-tdisplay-p4-lcd-lr2021-at|tdisplay_p4_lcd_lr2021|WADA_P4_LCD=1 WADA_P4_LR2021=1 WADA_P4_LEGACY_AT=1"
|
|
)
|
|
|
|
cd "$P4"
|
|
fail=0
|
|
for v in "${VARIANTS[@]}"; do
|
|
name="${v%%|*}"; rest="${v#*|}"
|
|
dir="${rest%%|*}"; envs="${rest#*|}"
|
|
echo "=== $name (build/$dir)"
|
|
|
|
# Trap 1: never inherit another variant's selection. Two dirs are shared by an
|
|
# esp-hosted and an ESP-AT build (the C6 axis does not change the dir name), so
|
|
# the previous occupant has to be cleared -- but ONLY by a real clean. Deleting
|
|
# CMakeCache.txt alone leaves CMakeFiles/ behind and the next configure dies in
|
|
# CMakeDetermineCCompiler with "No such file or directory". build.sh fullclean
|
|
# removes the tree properly and keeps the patched managed_components.
|
|
# A marker records what this dir last built, so re-running the same variant
|
|
# stays incremental instead of paying for a full rebuild every time.
|
|
# Two of these dirs are shared by an esp-hosted and an ESP-AT build, because the
|
|
# C6 axis does not change the dir name. The backend is read from the
|
|
# ENVIRONMENT at CMake CONFIGURE time (if(DEFINED ENV{WADA_P4_LEGACY_AT}) in
|
|
# main/CMakeLists.txt), and CMake does not reconfigure just because an env var
|
|
# changed -- so an incremental build in a dir that was configured the other way
|
|
# silently produces the OTHER variant, with no configure output to show for it.
|
|
# That is what made the first two runs of this script ship mislabelled images.
|
|
#
|
|
# So: always start these from nothing, and VERIFY the dir is gone rather than
|
|
# assuming the clean worked. The first version suppressed fullclean's output
|
|
# and trusted it; it had not run, and the only evidence was a missing configure
|
|
# line buried in a build log. A full rebuild per variant is the right price for
|
|
# a script that runs once per release.
|
|
if [ -d "build/$dir" ]; then
|
|
if [ "$envs" = "-" ]; then
|
|
env -u WADA_P4_LCD -u WADA_P4_LR2021 -u WADA_P4_LEGACY_AT ./build.sh fullclean >> "/tmp/p4rel_$name.clean.log" 2>&1 || true
|
|
else
|
|
env -u WADA_P4_LCD -u WADA_P4_LR2021 -u WADA_P4_LEGACY_AT $envs ./build.sh fullclean >> "/tmp/p4rel_$name.clean.log" 2>&1 || true
|
|
fi
|
|
fi
|
|
if [ -d "build/$dir" ]; then
|
|
echo " FAIL: build/$dir still present after fullclean; refusing to build on a stale tree"
|
|
echo " (see /tmp/p4rel_$name.clean.log)"
|
|
fail=$((fail+1)); continue
|
|
fi
|
|
|
|
# `env -u` every switch, then set only this variant's: an exported var in the
|
|
# caller's shell would otherwise leak in and silently change the build.
|
|
rc=0
|
|
if [ "$envs" = "-" ]; then
|
|
env -u WADA_P4_LCD -u WADA_P4_LR2021 -u WADA_P4_LEGACY_AT \
|
|
WADA_FW_TAG="$TAG" ./build.sh build > "/tmp/p4rel_$name.log" 2>&1 || rc=$?
|
|
else
|
|
env -u WADA_P4_LCD -u WADA_P4_LR2021 -u WADA_P4_LEGACY_AT \
|
|
$envs WADA_FW_TAG="$TAG" ./build.sh build > "/tmp/p4rel_$name.log" 2>&1 || rc=$?
|
|
fi
|
|
|
|
B="build/$dir"
|
|
app="$B/application.bin" # trap 2: this variant's dir, not the default one
|
|
# Check the EXIT CODE, not the log text. A failed CMake configure prints
|
|
# "Configuring incomplete, errors occurred!" and no "error:" line at all, so a
|
|
# grep for compiler errors waved it through -- and because the previous build's
|
|
# application.bin was still sitting in the dir, the next step happily read that
|
|
# instead. Only the id check downstream caught it.
|
|
if [ "$rc" -ne 0 ]; then
|
|
echo " FAIL: build exited $rc (see /tmp/p4rel_$name.log)"; fail=$((fail+1)); continue
|
|
fi
|
|
if [ ! -f "$app" ]; then
|
|
echo " FAIL: no application.bin (see /tmp/p4rel_$name.log)"; fail=$((fail+1)); continue
|
|
fi
|
|
|
|
# The verification that matters: what does the image say it is?
|
|
# NOTE: no `grep -q` and no `head` in these pipelines. Both exit as soon as
|
|
# they have what they need, which SIGPIPEs `strings`, and under `set -o
|
|
# pipefail` that non-zero status reads as "check failed" on a perfectly good
|
|
# image. Read the whole stream and decide afterwards.
|
|
ids="$(strings -a "$app" | grep -oE 'wadamesh-tdisplay-p4[a-z0-9-]*' | sort -u || true)"
|
|
got="$(printf '%s\n' "$ids" | sed -n '1p')"
|
|
if [ "$got" != "$name" ]; then
|
|
echo " FAIL: image identifies as '$got', expected '$name'"
|
|
echo " A device built from this would self-update to the WRONG variant."
|
|
fail=$((fail+1)); continue
|
|
fi
|
|
# And that it carries this release's tag, not a stale cached one.
|
|
tagged="$(strings -a "$app" | grep -cE "(^|[^0-9a-z])${TAG}([^0-9]|$)" || true)"
|
|
if [ "${tagged:-0}" = "0" ]; then
|
|
echo " FAIL: $TAG is not embedded (stale build dir?)"; fail=$((fail+1)); continue
|
|
fi
|
|
|
|
cp "$app" "$OUT/$name.bin"
|
|
"$IDFPY" -m esptool --chip esp32p4 merge_bin --flash_mode dio --flash_size 16MB --flash_freq 80m \
|
|
-o "$OUT/$name-merged.bin" \
|
|
0x2000 "$B/bootloader/bootloader.bin" \
|
|
0x8000 "$B/partition_table/partition-table.bin" \
|
|
0xf000 "$B/ota_data_initial.bin" \
|
|
0x20000 "$app" > /dev/null
|
|
echo " ok $(stat -f%z "$app") bytes id=$got"
|
|
done
|
|
|
|
echo
|
|
if [ "$fail" -ne 0 ]; then
|
|
echo "ABORT: $fail of ${#VARIANTS[@]} P4 variants failed. Nothing is safe to ship." >&2
|
|
exit 1
|
|
fi
|
|
echo "all ${#VARIANTS[@]} P4 variants built and verified -> $OUT"
|
|
echo "Copy both .bin and -merged.bin for each into releases/<CHANNEL>/$TAG/ AND latest[-beta]/."
|