Merge branch 'esp8266-arduino-toolchain' into esp8266-native-pch

# Conflicts:
#	esphome/components/esp8266/__init__.py

.github/actions/cache-arduino8266/action.yml

@@ -0,0 +1,39 @@
+name: Cache Arduino ESP8266
+description: >
+  Resolve the pinned Arduino core and xtensa toolchain versions and cache the
+  native ESP8266 install (~110 MB framework + toolchain; no ccache store, the
+  seed job saves before any compile runs). Exports
+  ESPHOME_ARDUINO8266_PREFIX to the job so every later step installs into
+  the cached path; the Python venv must already be restored. Mirrors
+  cache-esp-idf: only dev-branch pushes write the shared cache, everything
+  else restores.
+runs:
+  using: composite
+  steps:
+    - name: Resolve the native toolchain cache key
+      # Versions are pinned in code, not a hashable file; resolve them so a
+      # bump changes the cache key. Assignment form so errexit catches a
+      # resolver failure.
+      id: version
+      shell: bash
+      run: |
+        # One owner for the install prefix: exported here and referenced by
+        # the cache steps below via env, so the caller's install and the
+        # cached path cannot diverge.
+        echo "ESPHOME_ARDUINO8266_PREFIX=$HOME/.esphome-arduino8266" >> "$GITHUB_ENV"
+        . venv/bin/activate
+        key=$(python -c 'from esphome.components.esp8266 import RECOMMENDED_ARDUINO_FRAMEWORK_VERSION as f; from esphome.arduino8266.framework import TOOLCHAIN_VERSION as t; print(f"{f}-{t}")')
+        [ -n "$key" ] || exit 1
+        echo "key=$key" >> "$GITHUB_OUTPUT"
+    - name: Cache the native toolchain (write on dev)
+      if: github.ref == 'refs/heads/dev'
+      uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+      with:
+        path: ${{ env.ESPHOME_ARDUINO8266_PREFIX }}
+        key: ${{ runner.os }}-esp8266-native-${{ steps.version.outputs.key }}
+    - name: Restore the native toolchain (off dev)
+      if: github.ref != 'refs/heads/dev'
+      uses: actions/cache/restore@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+      with:
+        path: ${{ env.ESPHOME_ARDUINO8266_PREFIX }}
+        key: ${{ runner.os }}-esp8266-native-${{ steps.version.outputs.key }}

.github/workflows/ci.yml

@@ -102,6 +102,8 @@ jobs:
       device-builder: ${{ steps.determine.outputs.device-builder }}
       esp32-platformio: ${{ steps.determine.outputs.esp32-platformio }}
       esp32-platformio-components: ${{ steps.determine.outputs.esp32-platformio-components }}
+      esp8266-native: ${{ steps.determine.outputs.esp8266-native }}
+      esp8266-native-components: ${{ steps.determine.outputs.esp8266-native-components }}
       changed-components: ${{ steps.determine.outputs.changed-components }}
       changed-components-with-tests: ${{ steps.determine.outputs.changed-components-with-tests }}
       directly-changed-components-with-tests: ${{ steps.determine.outputs.directly-changed-components-with-tests }}
@@ -165,6 +167,8 @@ jobs:
           echo "device-builder=$(echo "$output" | jq -r '.device_builder')" >> $GITHUB_OUTPUT
           echo "esp32-platformio=$(echo "$output" | jq -r '.esp32_platformio')" >> $GITHUB_OUTPUT
           echo "esp32-platformio-components=$(echo "$output" | jq -r '.esp32_platformio_components')" >> $GITHUB_OUTPUT
+          echo "esp8266-native=$(echo "$output" | jq -r '.esp8266_native')" >> $GITHUB_OUTPUT
+          echo "esp8266-native-components=$(echo "$output" | jq -r '.esp8266_native_components')" >> $GITHUB_OUTPUT
           echo "changed-components=$(echo "$output" | jq -c '.changed_components')" >> $GITHUB_OUTPUT
           echo "changed-components-with-tests=$(echo "$output" | jq -c '.changed_components_with_tests')" >> $GITHUB_OUTPUT
           echo "directly-changed-components-with-tests=$(echo "$output" | jq -c '.directly_changed_components_with_tests')" >> $GITHUB_OUTPUT
@@ -183,6 +187,32 @@ jobs:
           path: .temp/components_graph.json
           key: components-graph-${{ hashFiles('esphome/components/**/*.py') }}
 
+  seed-esp8266-native-cache:
+    name: Seed the esp8266 native toolchain cache
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+    # PR-branch cache saves are invisible to other PRs, so dev pushes seed
+    # the shared entry test-esp8266-native restores. Only dev: the composite
+    # action saves nowhere else, so a beta/release push would download the
+    # toolchain and discard it.
+    if: github.event_name == 'push' && github.ref == 'refs/heads/dev'
+    timeout-minutes: 15
+    steps:
+      - name: Check out code from GitHub
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+      - name: Restore Python
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+      - name: Cache the native toolchain
+        uses: ./.github/actions/cache-arduino8266
+      - name: Install the native toolchain
+        run: |
+          . venv/bin/activate
+          python -c "from esphome.arduino8266.framework import check_and_install; from esphome.components.esp8266 import RECOMMENDED_ARDUINO_FRAMEWORK_VERSION; check_and_install(RECOMMENDED_ARDUINO_FRAMEWORK_VERSION)"
+
   ci-custom:
     name: Run script/ci-custom
     runs-on: ubuntu-24.04
@@ -1228,7 +1258,7 @@ jobs:
 
           # compile validates config first, so a separate config pass is
           # redundant for this smoke test. ESP-IDF framework via PlatformIO:
-          python3 script/test_build_components.py -e compile -t esp32-idf -c "$TEST_COMPONENTS" -f --toolchain platformio
+          python3 script/test_build_components.py -e compile -t esp32-idf -c "$TEST_COMPONENTS" -f --toolchain platformio --fail-on-no-tests
 
           echo ""
           echo "ESP-IDF-via-PlatformIO build passed! Starting Arduino smoke test..."
@@ -1237,6 +1267,40 @@ jobs:
           # Arduino framework via PlatformIO (only components with an esp32-ard test are built):
           python3 script/test_build_components.py -e compile -t esp32-ard -c "$TEST_COMPONENTS" -f --toolchain platformio
 
+  test-esp8266-native:
+    name: Test esp8266 components with the native toolchain
+    runs-on: ubuntu-24.04
+    needs:
+      - common
+      - determine-jobs
+    if: github.event_name == 'pull_request' && needs.determine-jobs.outputs.esp8266-native == 'true'
+    env:
+      # Computed by script/determine-jobs.py (ESP8266_NATIVE_TEST_COMPONENTS)
+      TEST_COMPONENTS: ${{ needs.determine-jobs.outputs.esp8266-native-components }}
+    steps:
+      - name: Check out code from GitHub
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
+
+      - name: Restore Python
+        uses: ./.github/actions/restore-python
+        with:
+          python-version: ${{ env.DEFAULT_PYTHON }}
+          cache-key: ${{ needs.common.outputs.cache-key }}
+
+      - name: Cache the native toolchain
+        uses: ./.github/actions/cache-arduino8266
+
+      - name: Run native toolchain compile test
+        run: |
+          . venv/bin/activate
+
+          echo "Testing components: $TEST_COMPONENTS"
+          echo ""
+
+          # ESP8266 Arduino built directly (no PlatformIO); compile validates
+          # config first, so a separate config pass is redundant.
+          python3 script/test_build_components.py -e compile -t esp8266-ard -c "$TEST_COMPONENTS" -f --toolchain arduino --fail-on-no-tests
+
   device-builder:
     name: Test downstream esphome/device-builder
     runs-on: ubuntu-24.04
@@ -1599,6 +1663,7 @@ jobs:
     needs:
       - common
       - seed-apt-cache
+      - seed-esp8266-native-cache
       - determine-jobs
       - ci-custom
       - pylint
@@ -1614,6 +1679,7 @@ jobs:
       - clang-tidy-esp32-variants
       - test-build-components-split
       - test-esp32-platformio
+      - test-esp8266-native
       - device-builder
       - memory-impact-target-branch
       - memory-impact-pr-branch

esphome/__main__.py

@@ -817,7 +817,9 @@ def write_cpp_file() -> int:
         from esphome.build_gen import espidf
 
         espidf.write_project()
-    else:
+    elif not CORE.using_native_toolchain:
+        # Other native builds generate their project at compile time;
+        # never write a platformio.ini for them
         from esphome.build_gen import platformio
 
         platformio.write_project()
@@ -859,20 +861,14 @@ def compile_program(args: ArgsProtocol, config: ConfigType) -> int:
         toolchain.create_factory_bin()
         toolchain.create_ota_bin()
         toolchain.create_elf_copy()
-        from esphome.build_helpers.idedata import IDEDATA_BEST_EFFORT_ERRORS
+        from esphome.build_helpers.idedata import warn_if_idedata_missing
 
-        try:
-            if toolchain.get_idedata() is None:
-                _LOGGER.warning("No idedata was generated for this build")
-        except IDEDATA_BEST_EFFORT_ERRORS as err:
-            # The firmware already built; an idedata failure must not fail
-            # a successful build.
-            _LOGGER.warning(
-                "Could not generate idedata: %s (IDE, clang-tidy, and "
-                "memory-analysis data will be unavailable for this build)",
-                err,
-            )
-            _LOGGER.debug("Idedata failure detail", exc_info=True)
+        warn_if_idedata_missing(toolchain.get_idedata)
+    elif CORE.using_native_toolchain:
+        raise EsphomeError(
+            f"Toolchain '{CORE.toolchain.value}' resolved but no platform "
+            "backend claimed the build"
+        )
     else:
         from esphome.platformio import toolchain
 
@@ -975,12 +971,15 @@ def upload_using_esptool(
 
     if file is not None:
         flash_images = [FlashImage(path=file, offset="0x0")]
-    elif CORE.using_toolchain_esp_idf:
-        from esphome.espidf import toolchain
-
-        flash_images = [
-            FlashImage(path=toolchain.get_factory_firmware_path(), offset="0x0")
-        ]
+    elif (native := _native_toolchain_module()) is not None:
+        # Every native backend supplies its own 0x0 flash image (bootloader
+        # and partitions included where the target needs them)
+        image = native.get_factory_firmware_path()
+        if not image.is_file():
+            raise EsphomeError(
+                f"{image} does not exist; compile the configuration first"
+            )
+        flash_images = [FlashImage(path=image, offset="0x0")]
     else:
         from esphome.platformio import toolchain
 
@@ -1950,15 +1949,39 @@ def command_update_all(args: ArgsProtocol) -> int | None:
     return run_multiple_configs(files, build_command)
 
 
+# Native build backend per (target platform, toolchain). Keyed here rather
+# than through a platform hook so the serial upload/logs fast path never
+# imports the platform component package (see the esp32 variant comment in
+# upload_using_esptool); the platform half comes from CORE.data the same way.
+_NATIVE_TOOLCHAIN_MODULES = {
+    ("esp32", Toolchain.ESP_IDF): "esphome.espidf.toolchain",
+    ("esp8266", Toolchain.ARDUINO): "esphome.arduino8266.toolchain",
+}
+
+
+def _native_toolchain_module():
+    """The native build backend module for the resolved toolchain."""
+    if not CORE.using_native_toolchain:
+        return None
+    key = (CORE.target_platform, CORE.toolchain)
+    if (module_path := _NATIVE_TOOLCHAIN_MODULES.get(key)) is None:
+        # Degrading to the PlatformIO path would build with the wrong backend
+        raise EsphomeError(
+            f"Toolchain '{CORE.toolchain.value}' has no native build backend "
+            f"module for platform {CORE.target_platform}"
+        )
+    return importlib.import_module(module_path)
+
+
 def command_idedata(args: ArgsProtocol, config: ConfigType) -> int:
     import json
 
-    if CORE.using_toolchain_esp_idf:
-        # Native ESP-IDF derives idedata from the build's compile_commands.json,
-        # so the configuration must already be compiled.
-        from esphome.espidf import toolchain as espidf_toolchain
+    native_toolchain = _native_toolchain_module()
 
-        idedata = espidf_toolchain.get_idedata()
+    if native_toolchain is not None:
+        # Native toolchains derive idedata from the build's
+        # compile_commands.json, so the configuration must already be compiled.
+        idedata = native_toolchain.get_idedata()
         if idedata is None:
             _LOGGER.error(
                 "No idedata available; compile the configuration first",
@@ -1997,6 +2020,17 @@ def command_analyze_memory(args: ArgsProtocol, config: ConfigType) -> int:
     from esphome.analyze_memory.cli import MemoryAnalyzerCLI
     from esphome.analyze_memory.ram_strings import RamStringsAnalyzer
 
+    # Refuse an unsupported toolchain before paying for a full compile
+    native_toolchain = _native_toolchain_module()
+    if native_toolchain is None and not CORE.using_toolchain_platformio:
+        _LOGGER.error(
+            "analyze-memory is not supported with the '%s' toolchain on %s; "
+            "re-run with --toolchain platformio",
+            CORE.toolchain.value if CORE.toolchain else "unresolved",
+            CORE.target_platform,
+        )
+        return 1
+
     # Always compile to ensure fresh data (fast if no changes - just relinks)
     exit_code = write_cpp(config)
     if exit_code != 0:
@@ -2008,13 +2042,31 @@ def command_analyze_memory(args: ArgsProtocol, config: ConfigType) -> int:
 
     # Get idedata for analysis
     idedata = None
-    if CORE.using_toolchain_esp_idf:
-        from esphome.espidf import toolchain
+    if native_toolchain is not None:
+        objdump = native_toolchain.get_objdump_path()
+        readelf = native_toolchain.get_readelf_path()
+        for tool in (objdump, readelf):
+            if not tool.is_file():
+                # The analyzer would silently fall back to host binutils,
+                # which cannot read the target ELF. clean-all is heavy for
+                # ESP-IDF, so suggest a recompile first.
+                _LOGGER.error(
+                    "%s is missing; the toolchain install may be incomplete "
+                    "(recompile, or run 'esphome clean-all' if it persists)",
+                    tool,
+                )
+                return 1
+        objdump_path = str(objdump)
+        readelf_path = str(readelf)
 
-        objdump_path = str(toolchain.get_objdump_path())
-        readelf_path = str(toolchain.get_readelf_path())
-
-        firmware_elf = toolchain.get_elf_path()
+        firmware_elf = native_toolchain.get_elf_path()
+        if not firmware_elf.is_file():
+            # The analyzer swallows tool failures, so a missing ELF would
+            # produce an exit-0 zeroed report
+            _LOGGER.error(
+                "%s is missing; compile the configuration first", firmware_elf
+            )
+            return 1
     else:
         from esphome.platformio import toolchain
 

esphome/arduino8266/framework.py

@@ -19,6 +19,7 @@ from typing import NamedTuple
 
 from esphome.build_helpers.ccache import ccache_defaults_env
 from esphome.build_helpers.ninja import find_ninja
+from esphome.build_helpers.pch import ccache_pch_env
 from esphome.build_helpers.tools_cache import ARDUINO8266_TOOLS_CACHE, tools_cache_path
 from esphome.core import EsphomeError, Version
 from esphome.framework_helpers import str_to_lst_of_str
@@ -156,4 +157,6 @@ def ccache_env(ccache: str | None) -> dict[str, str]:
     """
     if ccache is None:
         return {}
-    return ccache_defaults_env(get_arduino8266_tools_path() / "ccache")
+    env = ccache_defaults_env(get_arduino8266_tools_path() / "ccache")
+    env.update(ccache_pch_env())
+    return env

esphome/arduino8266/toolchain.py

@@ -0,0 +1,329 @@
+"""Native Arduino ESP8266 build driver (the PlatformIO ``run`` equivalent)."""
+
+from __future__ import annotations
+
+import json
+import logging
+from pathlib import Path
+import subprocess
+from typing import Any
+
+from esphome.arduino8266 import framework
+from esphome.build_helpers.ccache import resolve_ccache_path
+from esphome.const import (
+    CONF_COMPILE_PROCESS_LIMIT,
+    CONF_ESPHOME,
+    KEY_CORE,
+    KEY_FRAMEWORK_VERSION,
+)
+from esphome.core import CORE, EsphomeError
+from esphome.helpers import write_file_if_changed
+from esphome.types import ConfigType
+
+_LOGGER = logging.getLogger(__name__)
+
+# ESP8266 user RAM (matches upload.maximum_ram_size in every board manifest)
+_MAX_RAM_SIZE = 81920
+
+
+def _warn_ignored_platformio_options() -> None:
+    """Warn for component-added platformio options the native build drops.
+
+    The consumed set is exported by core/config.py next to the routing that
+    stores these options, so the two cannot drift; YAML upload_speed never
+    reaches CORE.platformio_options here.
+    """
+    from esphome.core.config import NATIVE_ARDUINO_CONSUMED_PIO_OPTIONS
+
+    consumed = NATIVE_ARDUINO_CONSUMED_PIO_OPTIONS
+    for key in sorted(CORE.platformio_options or {}):
+        if key not in consumed:
+            _LOGGER.warning(
+                "platformio_options->%s is ignored when building with the "
+                "native 'arduino' toolchain",
+                key,
+            )
+
+
+_RAM_SECTIONS = (".data", ".rodata", ".bss")
+_FLASH_SECTIONS = (".irom0.text", ".text", ".text1", ".data", ".rodata")
+
+
+def get_build_dir() -> Path:
+    return CORE.relative_pioenvs_path(CORE.name)
+
+
+def get_elf_path() -> Path:
+    return get_build_dir() / "firmware.elf"
+
+
+def _toolchain_tool(name: str) -> Path:
+    return framework.toolchain_tool(framework.get_toolchain_path(), name)
+
+
+def get_factory_firmware_path() -> Path:
+    """The image to serial-flash at 0x0 (same bytes as firmware.bin: the
+    8266 factory copy exists for artifact-contract parity, not content)."""
+    return get_build_dir() / "firmware.factory.bin"
+
+
+def get_addr2line_path() -> Path:
+    return _toolchain_tool("addr2line")
+
+
+def get_objdump_path() -> Path:
+    return _toolchain_tool("objdump")
+
+
+def get_readelf_path() -> Path:
+    return _toolchain_tool("readelf")
+
+
+def run_compile(config: ConfigType, verbose: bool) -> int:
+    from esphome.build_gen import arduino8266 as build_gen
+
+    _warn_ignored_platformio_options()
+    paths = framework.check_and_install(CORE.data[KEY_CORE][KEY_FRAMEWORK_VERSION])
+    # Resolved once per build: the resolution probes PATH and spawns the
+    # runnability check, and three consumers need the same answer
+    ccache = resolve_ccache_path()
+    ninja_changed = build_gen.write_project(paths, ccache)
+
+    build_dir = get_build_dir()
+    env = framework.get_build_env(paths.toolchain, ccache)
+
+    # Regenerate the compile DB before the build (a pure function of
+    # build.ninja); skip only when it is at least as fresh as build.ninja
+    # (an interrupted previous run may have rewritten the manifest without
+    # regenerating the DB).
+    compdb = build_dir / "compile_commands.json"
+    compdb_stamp = build_dir / ".compile_commands.stamp"
+    ninja_file = build_dir / "build.ninja"
+    # Freshness rides a stamp: the DB itself is written through
+    # write_file_if_changed (its mtime feeds get_idedata's cache), so a
+    # regeneration with identical content would stay "stale" forever
+    if (
+        ninja_changed
+        or not compdb.is_file()
+        or not compdb_stamp.is_file()
+        or compdb_stamp.stat().st_mtime < ninja_file.stat().st_mtime
+    ):
+        _write_compile_commands(paths.ninja, build_dir, env)
+        compdb_stamp.touch()
+
+    cmd = [str(paths.ninja)]
+    if verbose:
+        cmd.append("-v")
+    if jobs := config[CONF_ESPHOME].get(CONF_COMPILE_PROCESS_LIMIT):
+        cmd += ["-j", str(jobs)]
+    # Explicit targets, not the default statement: a generator defect that
+    # drops them fails loudly with "unknown target" instead of a green
+    # no-op run that leaves stale artifacts in place
+    targets = ["firmware.factory.bin", "firmware.ota.bin"]
+    cmd += targets
+
+    # A dry-run probe keeps a no-op rebuild quiet: ninja would only print
+    # "no work to do". A freshly rewritten manifest all but guarantees work,
+    # so skip the probe (and its full stat pass) on that path. cwd instead
+    # of -C also drops the "Entering directory" banner on real builds.
+    skip_build = False
+    if not ninja_changed:
+        probe = subprocess.run(
+            [str(paths.ninja), "-n", *targets],
+            cwd=build_dir,
+            env=env,
+            capture_output=True,
+            text=True,
+            check=False,
+            close_fds=False,
+        )
+        if probe.stderr.strip():
+            # A load-time diagnostic (e.g. "multiple rules generate X")
+            # flags a generator bug; the skip branch would otherwise
+            # swallow it forever
+            _LOGGER.warning("ninja: %s", probe.stderr.strip())
+        if probe.returncode != 0:
+            # An unknown target here is the defective-manifest case; fall
+            # through to the real build so the error prints attributably
+            _LOGGER.debug("ninja probe failed; running the full build")
+        skip_build = probe.returncode == 0 and "no work to do" in probe.stdout
+    if skip_build:
+        _LOGGER.debug("ninja: nothing to rebuild")
+    else:
+        _LOGGER.debug("Running: %s", " ".join(cmd))
+        rc = subprocess.run(
+            cmd, cwd=build_dir, env=env, check=False, close_fds=False
+        ).returncode
+        if rc != 0:
+            return rc
+
+    # ninja already refused a manifest missing the explicit targets above;
+    # existence covers the remaining hole (a rule that ran but wrote
+    # elsewhere). The factory/ota copies are what upload and OTA consume.
+    build_dir_artifacts = (
+        get_elf_path(),
+        build_dir / "firmware.bin",
+        get_factory_firmware_path(),
+        build_dir / "firmware.ota.bin",
+    )
+    for artifact in build_dir_artifacts:
+        if not artifact.is_file():
+            _LOGGER.error("Build produced no %s", artifact)
+            return 1
+
+    if not _print_size_summary(build_dir, paths):
+        # The cause was already warned; name the consequence so a build
+        # contributing no RAM/Flash metric is visible to CI harnesses
+        _LOGGER.warning("Firmware size summary unavailable for this build")
+    from esphome.build_helpers.idedata import warn_if_idedata_missing
+
+    warn_if_idedata_missing(lambda: get_idedata(ccache))
+    return 0
+
+
+def _write_compile_commands(
+    ninja_path: Path, build_dir: Path, env: dict[str, str]
+) -> None:
+    compdb = build_dir / "compile_commands.json"
+    result = subprocess.run(
+        [str(ninja_path), "-C", str(build_dir), "-t", "compdb", "c", "cxx", "asm"],
+        env=env,
+        capture_output=True,
+        text=True,
+        check=False,
+        close_fds=False,
+    )
+    if result.returncode != 0:
+        # Drop any stale database so consumers (IDE integration, clang-tidy,
+        # the memory analyzer) can't silently read outdated data.
+        compdb.unlink(missing_ok=True)
+        raise EsphomeError(f"Could not generate compile_commands.json: {result.stderr}")
+    try:
+        entries = json.loads(result.stdout)
+    except ValueError as err:
+        compdb.unlink(missing_ok=True)
+        raise EsphomeError(
+            f"ninja produced an unparsable compile database: {err} "
+            f"(output starts {result.stdout[:120]!r})"
+        ) from err
+    if not entries:
+        # compdb exits 0 with [] for unknown rule names; a renamed compile
+        # rule must fail the build, not silently strand every consumer
+        compdb.unlink(missing_ok=True)
+        raise EsphomeError(
+            "ninja produced an empty compile database; the generator's rule "
+            "names no longer match"
+        )
+    # write_file_if_changed keeps the mtime stable on no-op builds so the
+    # idedata cache in get_idedata() stays valid.
+    write_file_if_changed(compdb, result.stdout)
+
+
+def _parse_app_size(build_dir: Path, paths: framework.InstalledPaths) -> int | None:
+    """Read the app flash budget (irom0_0_seg length) from the linker script."""
+    from esphome.build_gen.arduino8266 import get_flash_ld_path
+    from esphome.components.esp8266.build_surgery import segment_length
+
+    # Warnings, not debug: without the app size the Flash summary line is
+    # dropped and CI's memory-impact extraction loses its flash metric.
+    ld_path = get_flash_ld_path(build_dir, paths)
+    try:
+        ld_text = ld_path.read_text(encoding="utf-8")
+    except (OSError, UnicodeDecodeError) as err:
+        # UnicodeDecodeError: a truncated/corrupt script must degrade to
+        # the same warning, never abort an already-linked build
+        _LOGGER.warning("Cannot read linker script for the Flash summary: %s", err)
+        return None
+    app_size = segment_length(ld_text, "irom0_0_seg")
+    if app_size is None:
+        _LOGGER.warning("irom0_0_seg not found in %s; skipping Flash summary", ld_path)
+        return None
+    if app_size == 0:
+        _LOGGER.warning(
+            "irom0_0_seg has zero length in %s; skipping Flash summary", ld_path
+        )
+        return None
+    return app_size
+
+
+def _print_size_summary(build_dir: Path, paths: framework.InstalledPaths) -> bool:
+    """Print the PlatformIO-shaped RAM/Flash lines; False when skipped.
+
+    The exact shape (including the bar) is parsed by
+    ``script/ci_memory_impact_extract.py``; ``print_size_line`` matches it.
+    """
+    from esphome.build_helpers.size_summary import print_size_line
+
+    size_tool = _toolchain_tool("size")
+    try:
+        result = subprocess.run(
+            [str(size_tool), "-A", "-d", str(get_elf_path())],
+            capture_output=True,
+            text=True,
+            check=False,
+            close_fds=False,
+        )
+    except OSError as err:
+        # The summary is a bonus artifact like idedata; a truncated
+        # toolchain extraction must not discard an already-linked build
+        _LOGGER.warning("Could not summarize firmware size: %s", err)
+        return False
+    if result.returncode != 0:
+        _LOGGER.warning("Could not summarize firmware size: %s", result.stderr)
+        return False
+    sections: dict[str, int] = {}
+    for line in result.stdout.splitlines():
+        parts = line.split()
+        if len(parts) >= 2 and parts[0].startswith("."):
+            try:
+                sections[parts[0]] = int(parts[1])
+            except ValueError:
+                # An unparsed RAM/Flash section trips the missing-sections
+                # guard below, so no total is built on a dropped value
+                _LOGGER.warning("Unparsable size output for section %s", parts[0])
+    if missing := set(_RAM_SECTIONS + _FLASH_SECTIONS) - set(sections):
+        # A defaulted 0 would print a confidently wrong total for CI's metric
+        _LOGGER.warning(
+            "Size output is missing section(s) %s; skipping the size summary",
+            ", ".join(sorted(missing)),
+        )
+        return False
+    # Resolve the flash budget before printing anything: a RAM line without
+    # its Flash line would let CI's memory-impact extraction sum the two
+    # metrics over different build counts (_parse_app_size already warned).
+    app_size = _parse_app_size(build_dir, paths)
+    if not app_size:
+        return False
+    ram = sum(sections[s] for s in _RAM_SECTIONS)
+    flash = sum(sections[s] for s in _FLASH_SECTIONS)
+    print_size_line("RAM", ram, _MAX_RAM_SIZE)
+    print_size_line("Flash", flash, app_size)
+    return True
+
+
+# Sentinel: "resolve for me"; None is a real value meaning disabled.
+_CCACHE_UNRESOLVED: Any = object()
+
+
+def get_idedata(ccache: str | None = _CCACHE_UNRESOLVED) -> dict | None:
+    """Derive idedata from the build's compile_commands.json.
+
+    Same contract as ``espidf.toolchain.get_idedata``: the fields IDE
+    integrations, clang-tidy, and the memory analyzer expect.
+    """
+    from esphome.build_helpers.idedata import load_or_build_idedata
+
+    if ccache is _CCACHE_UNRESOLVED:
+        # Deliberately uncached: env/PATH can change between builds in a
+        # long-lived host process
+        ccache = resolve_ccache_path()
+    return load_or_build_idedata(
+        get_build_dir() / "compile_commands.json",
+        get_elf_path(),
+        # Suffixed so a platformio->arduino->platformio round trip on one
+        # config never serves the other toolchain's cache shape
+        CORE.relative_internal_path("idedata", f"{CORE.name}.arduino.json"),
+        # The compile DB's commands carry the same ccache prefix the ninja
+        # rules were generated with
+        launcher=str(ccache) if ccache else None,
+    )

esphome/build_gen/arduino8266.py

@@ -1,11 +1,11 @@
-"""Build specification for the native ESP8266 Arduino toolchain.
+"""Native ninja build generator for the ESP8266 Arduino core.
 
 Transliterates the PlatformIO build spec for the Arduino ESP8266 framework
 (``framework-arduinoespressif8266/tools/platformio-build.py`` plus
-``platform-espressif8266/builder/main.py``): the flag sets, defines, and
-linker-script generation deliberately match what PlatformIO produces so the
-binaries stay near-identical between the two toolchains. The ninja emission
-(``write_project``) builds on these pieces.
+``platform-espressif8266/builder/main.py``) into a ``build.ninja`` under
+``.pioenvs/<name>/``. The flag sets, defines, link line, linker-script
+generation, and ``elf2bin`` invocation deliberately match what PlatformIO
+produces so the binaries stay near-identical between the two toolchains.
 
 The ``PIO_FRAMEWORK_ARDUINO_*`` knob defines (lwIP variant, NONOS SDK
 version, MMU layout, exceptions, waveform phase) keep working: they are read
@@ -14,6 +14,7 @@ from the build flags with the same precedence as the PlatformIO builder.
 
 from __future__ import annotations
 
+from collections.abc import Collection
 from dataclasses import dataclass
 import hashlib
 import logging
@@ -22,20 +23,61 @@ from pathlib import Path
 import re
 import shlex
 import subprocess
+import sys
 from typing import TYPE_CHECKING, NamedTuple
 
 from esphome.arduino8266.framework import toolchain_tool
-from esphome.build_helpers.ninja import shell_token as _shell_token
+from esphome.build_helpers.ccache import effective_ccache_basedir
+from esphome.build_helpers.idedata import is_joined_include
+from esphome.build_helpers.ninja import (
+    escape as _e,
+    quote_path as _q,
+    shell_token as _shell_token,
+)
+from esphome.build_helpers.pch import (
+    PCH_CORE_HEADER,
+    PCH_HEADER_NAME,
+    mark_pch_emitted,
+    pch_checksum,
+    pch_enabled,
+    pch_header_text,
+)
 from esphome.components.esp8266 import build_surgery
+from esphome.components.esp8266.boards import (
+    BOARDS,
+    ESP8266_BOARD_BUILD,
+    board_ld_script,
+)
+from esphome.components.esp8266.const import (
+    BUILD_FLASH_MODES,
+    KEY_BOARD,
+    KEY_ESP8266,
+    KEY_FLASH_SIZE,
+    KEY_SCANF_FLOAT,
+)
 from esphome.core import CORE, EsphomeError
+from esphome.framework_helpers import (
+    get_project_cxx_compile_flags,
+    strip_win_long_path_prefix,
+)
 from esphome.helpers import mkdir_p, write_file_if_changed
-from esphome.platformio.library import lex_build_flags
+from esphome.platformio.library import SOURCE_KIND_FOR_SUFFIX, lex_build_flags
 
 if TYPE_CHECKING:
     from esphome.arduino8266.framework import InstalledPaths
 
 _LOGGER = logging.getLogger(__name__)
 
+# Always excluded from the core build: ESPHome uses its own native OTA
+# backend, so the Arduino Updater (and its 228-byte global) never links.
+_CORE_EXCLUDE_ALWAYS = {"Updater.cpp"}
+# Excluded when no component called require_waveform(); waveform_stubs.cpp
+# supplies the stopWaveform()/_stopPWM() stubs digitalWrite needs.
+_CORE_EXCLUDE_WAVEFORM = {
+    "core_esp8266_waveform_pwm.cpp",
+    "core_esp8266_waveform_phase.cpp",
+}
+
 # Values that land unquoted on generated command lines are shape-checked
 # against these before use. re.ASCII: a Unicode digit or word character
 # (Arabic-Indic numerals) would pass \d/\w and defeat the named error
@@ -77,6 +119,11 @@ def _apply_surgery(fn, *args: object) -> str:
         raise EsphomeError(str(err)) from err
 
 
+# Every supported board's f_flash is 40 MHz; re-check on a platform bump
+# board_flash_mode's closed set, shared with cv.one_of's validation
+_FLASH_MODES = frozenset(BUILD_FLASH_MODES)
+_FLASH_FREQ_MHZ = 40
+
 # From platformio-build.py. Knob suffix -> SDK define; the first entry is
 # the default (dicts preserve insertion order). With multiple SDK knobs set
 # (a pathological config) ties break by table order, since upstream's
@@ -408,6 +455,52 @@ def _resolve_build_config(defines: dict[str, str]) -> _BuildConfig:
     )
 
 
+_INCOMPLETE_INSTALL = "Arduino toolchain install is incomplete"
+
+
+def _filter_link_flags(unflags: set[str]) -> list[str]:
+    """_LINKFLAGS minus ``unflags``, pair-aware: unflagging a symbol also
+    drops the ``-u`` that carried it, so no dangling operand-less flag
+    reaches ld as the next token's consumer."""
+    out: list[str] = []
+    it = iter(_LINKFLAGS)
+    for tok in it:
+        if tok == "-u":
+            symbol = next(it)
+            if symbol not in unflags:
+                out += [tok, symbol]
+        elif tok not in unflags:
+            out.append(tok)
+    return out
+
+
+def _active_flash_ld_name(flash_ld_name: str) -> str:
+    """The flash linker-script filename the link uses (testing mode renames
+    the surgically patched copy)."""
+    return (
+        f"{_TESTING_LD_PREFIX}{flash_ld_name}" if CORE.testing_mode else flash_ld_name
+    )
+
+
+def _flash_ld_name(board: str) -> str:
+    """The flash linker script: the board's, or a routed user override.
+
+    Published configs override board_build.ldscript to reserve a
+    filesystem region or correct a board's assumed flash size; a bare
+    name is required because the script resolves via the -L search path.
+    """
+    override = _pio_option("board_build.ldscript", "")
+    if not override:
+        # The same shared rule the PlatformIO path pins (layout
+        # preservation, see boards.board_ld_script)
+        return board_ld_script(BOARDS[board])
+    if Path(override).name != override:
+        raise EsphomeError(
+            f"board_build.ldscript must be a bare script name, got {override!r}"
+        )
+    return override
+
+
 def _pio_option(key: str, default: str) -> str:
     """A platformio_options value the native build honors (str-normalized).
 
@@ -551,11 +644,23 @@ _PLAIN_LINKER_FLAGS = (
     "-nostdlib",
     "-rdynamic",
 )
+# The subset whose next token is an operand; unflagging the bare flag
+# would strand the operand. Operand-less members of the list above filter
+# whole-token from both the compile and link lines, as PlatformIO allows.
+_PLAIN_LINKER_OPERAND_FLAGS = ("-u", "-e")
 _PLAIN_LINKER_PREFIXES = ("-T", "-Xlinker")
 # Driver options, not ld options: -Wl, has no equivalent for these
 _PLAIN_DRIVER_LINK_PREFIXES = ("-fuse-ld=", "--specs=", "-specs=")
 
 
+def _collect_sources(root: Path, exclude: Collection[str] = frozenset()) -> list[Path]:
+    return sorted(
+        p
+        for p in root.rglob("*")
+        if p.suffix in SOURCE_KIND_FOR_SUFFIX and p.name not in exclude
+    )
+
+
 def _stat_sig(path: Path) -> str:
     """Size and mtime cache-stamp signature for one input file.
 
@@ -771,3 +876,461 @@ def _generate_testing_flash_ld(
     write_file_if_changed(
         ld_dir / f"{_TESTING_LD_PREFIX}{flash_ld_name}", patched_flash_ld
     )
+
+
+def _ninja_compile_edges(
+    lines: list[str],
+    sources: list[Path],
+    root: Path,
+    group: str,
+    flags: str = "",
+    cxx_override: tuple[str, str] | None = None,
+) -> list[str]:
+    """Emit compile edges for ``sources``; return the object paths.
+
+    ``cxx_override`` is a (flags, implicit-dep) pair applied to C++ edges
+    only, replacing ``flags`` (used for the precompiled header).
+    """
+    objects = []
+    for src in sources:
+        rel = src.relative_to(root).as_posix()
+        obj = f"obj/{group}/{rel}.o"
+        escaped_obj = _e(obj)
+        kind = SOURCE_KIND_FOR_SUFFIX[src.suffix]
+        override = cxx_override if kind == "cxx" and cxx_override else None
+        implicit = f" | {override[1]}" if override else ""
+        lines.append(f"build {escaped_obj}: {kind} {_e(src)}{implicit}")
+        edge_flags = override[0] if override else flags
+        if edge_flags:
+            lines.append(f"  flags = {edge_flags}")
+        # Escaped once here: the returned paths only ever appear in build
+        # statements (archive/link inputs), which use ninja escaping.
+        objects.append(escaped_obj)
+    return objects
+
+
+def _common_parent(paths: list[Path]) -> Path:
+    return Path(os.path.commonpath([str(p.parent) for p in paths]))
+
+
+def write_project(paths: InstalledPaths, ccache: str | None) -> bool:
+    """Write the ninja build for the current configuration.
+
+    ``ccache`` is the caller's already-resolved binary (None when disabled)
+    so one build never pays the runnability probe per consumer. Returns
+    True when ``build.ninja`` changed, so the caller can skip work derived
+    purely from it (the compile database) on unchanged builds.
+    """
+    from esphome.arduino.library import resolve_libraries
+
+    framework = paths.framework
+    toolchain_bin = paths.toolchain / "bin"
+    build_dir = CORE.relative_pioenvs_path(CORE.name)
+    mkdir_p(build_dir)
+
+    unflags = _unflag_tokens()
+    # Lexed once so a malformed entry warns once, not per consumer
+    build_tokens = _lexed_build_flags()
+    flag_defines = _flag_defines(unflags, build_tokens)
+    config = _resolve_build_config(flag_defines)
+    esp8266_data = CORE.data[KEY_ESP8266]
+    board = esp8266_data[KEY_BOARD]
+    # Config validation already gates boards;
+    # kept as defense-in-depth for direct calls, since CONF_BOARD itself is
+    # a free-form string
+    if board not in ESP8266_BOARD_BUILD:
+        raise EsphomeError(f"Board '{board}' is not supported by the native toolchain")
+    board_build = ESP8266_BOARD_BUILD[board]
+    # From the same producer the PlatformIO path reads (one source)
+    flash_mode = _pio_option("board_build.flash_mode", "dout")
+    if flash_mode not in _FLASH_MODES:
+        # Lands unquoted in the elf2bin command and a -D body; validation
+        # (cv.one_of on board_flash_mode) already gates it, defense-in-depth
+        raise EsphomeError(f"Invalid flash mode {flash_mode!r}")
+    flash_ld_name = _flash_ld_name(board)
+
+    generate_ld_scripts(paths, config, flash_ld_name)
+
+    sdk = framework / "tools" / "sdk"
+    core_dir = framework / "cores" / "esp8266"
+    variant_dir = framework / "variants" / board_build["variant"]
+    src_dir = CORE.relative_src_path()
+
+    libraries = resolve_libraries(
+        framework,
+        pio_platform="espressif8266",
+        board_mcu="esp8266",
+        cache_key="arduino8266",
+    )
+
+    if not src_dir.is_dir():
+        # Generated project state, not install state: clean-all would not help
+        raise EsphomeError(f"Generated source directory {src_dir} is missing")
+    # A missing install directory would otherwise surface as a wall of
+    # include errors; failing here names the path instead.
+    include_dirs = [
+        src_dir,
+        sdk / "include",
+        core_dir,
+        paths.toolchain / "include",
+        sdk / "lwip2" / "include",
+        variant_dir,
+    ]
+    for required in include_dirs[1:]:
+        if not required.is_dir():
+            raise EsphomeError(
+                f"{_INCOMPLETE_INSTALL}: missing {required}; {_CLEAN_HINT}"
+            )
+    # The elf2bin edge runs after the full compile and link; a
+    # half-extracted package must fail here, not an hour of wall-clock later
+    for required_file in (
+        framework / "tools" / "elf2bin.py",
+        framework / "bootloaders" / "eboot" / "eboot.elf",
+    ):
+        if not required_file.is_file():
+            raise EsphomeError(
+                f"{_INCOMPLETE_INSTALL}: missing {required_file}; {_CLEAN_HINT}"
+            )
+    for lib in libraries:
+        include_dirs += lib.include_dirs
+
+    (
+        project_compile_flags,
+        project_link_flags,
+        project_lib_dirs,
+        project_libs,
+    ) = _project_flags(unflags, build_tokens)
+    defines = _defines_flags(config, flash_mode, board, board_build["defines"])
+    includes = [f"-I{_q(d)}" for d in include_dirs]
+
+    common = _CCFLAGS + defines + includes + project_compile_flags
+    cflags = _CFLAGS + common
+    cpp_standard = CORE.cpp_standard or "gnu++17"
+    cxxflags = (
+        ["-fno-rtti", f"-std={cpp_standard}"]
+        + ["-fexceptions" if config.exceptions else "-fno-exceptions"]
+        + common
+        + [_shell_token(f) for f in get_project_cxx_compile_flags()]
+    )
+    # PlatformIO's ASPPCOM passes only -D/-I user flags to assembly; match
+    # it (tokens arrive shell-quoted, hence the lstrip)
+    asflags = (
+        _ASFLAGS
+        + defines
+        + includes
+        + [f for f in project_compile_flags if f.lstrip("\"'").startswith(("-D", "-I"))]
+    )
+
+    # build_unflags applies to the framework flag sets too (compile and link),
+    # as under PlatformIO (a silently ignored ``build_unflags: -Os`` would
+    # diverge between the toolchains).
+    # Matching is whole-token, so an unflag that hits nothing in the user
+    # flags or any framework set (a typo, or -DUSE_FOO against -DUSE_FOO=1)
+    # must be visible: the user believes the flag is gone while it still
+    # drives the compile line and the knob selection
+    flag_universe = set(build_tokens)
+    for flags in (cflags, cxxflags, asflags, _LINKFLAGS):
+        flag_universe.update(flags)
+    if unmatched := sorted(unflags - flag_universe):
+        _LOGGER.warning(
+            "build_unflags entries matched no build flag: %s", ", ".join(unmatched)
+        )
+    # _LINKFLAGS stores -u and its operand as two tokens; unflagging the
+    # bare -u would strip all seven and leave the operands as ld "input
+    # files" with an error pointing nowhere near build_unflags
+    if plain := sorted(
+        u
+        for u in unflags
+        if u in _PLAIN_LINKER_OPERAND_FLAGS or u.startswith(_PLAIN_LINKER_PREFIXES)
+    ):
+        raise EsphomeError(
+            f"build_unflags cannot remove plain linker flag(s) "
+            f"{', '.join(plain)}; unflag the full -Wl, form or the symbol"
+        )
+    cflags, cxxflags, asflags = (
+        [f for f in flags if f not in unflags] for flags in (cflags, cxxflags, asflags)
+    )
+    link_flags = _filter_link_flags(unflags)
+    if esp8266_data[KEY_SCANF_FLOAT]:
+        link_flags += ["-u", "_scanf_float"]
+    link_flags += project_link_flags
+    link_flags += [_shell_token(flag) for lib in libraries for flag in lib.link_flags]
+    flash_ld = _active_flash_ld_name(flash_ld_name)
+    # A user-overridden script name re-quotes like every other user token
+    link_flags += ["-T", _shell_token(flash_ld)]
+
+    lib_dirs = [Path("ld"), sdk / "lib", sdk / "ld", sdk / "lib" / config.nonosdk]
+    lib_dirs += project_lib_dirs
+    for lib in libraries:
+        lib_dirs += lib.link_dirs
+    system_libs = (
+        _SYSTEM_LIBS_PRE_LWIP
+        + [config.lwip_lib]
+        + _SYSTEM_LIBS_POST_LWIP
+        + project_libs
+        + [lib_name for lib in libraries for lib_name in lib.link_libs]
+        + ["stdc++-exc" if config.exceptions else "stdc++", "m", "c", "gcc"]
+    )
+
+    build_tool = Path(__file__).parent / "build_tool.py"
+
+    # $in/$out stay unquoted: ninja escapes its built-in path variables
+    # itself; only literal paths need _q().
+    lines = [
+        "# Auto-generated by ESPHome",
+        "ninja_required_version = 1.5",
+        f"cc = {_q(toolchain_tool(paths.toolchain, 'gcc'))}",
+        f"cxx = {_q(toolchain_tool(paths.toolchain, 'g++'))}",
+        # The NSIS launcher starts Python with a \\?\ extended-length path
+        # that cmd.exe cannot spawn; same strip every other emitted binary
+        # path gets
+        f"python = {_q(strip_win_long_path_prefix(sys.executable))}",
+        f"buildtool = {_q(build_tool)}",
+        f"ccache = {_q(ccache) if ccache else ''}",
+        "",
+        # Rule names match SOURCE_KIND_FOR_SUFFIX values (c, cxx, asm, aspp)
+        "rule c",
+        "  command = $ccache $cc -MMD -MF $out.d $cflags $flags -c $in -o $out",
+        "  depfile = $out.d",
+        "  deps = gcc",
+        "  description = CC $out",
+        "rule cxx",
+        "  command = $ccache $cxx -MMD -MF $out.d $cxxflags $flags -c $in -o $out",
+        "  depfile = $out.d",
+        "  deps = gcc",
+        "  description = CXX $out",
+        "rule aspp",
+        "  command = $ccache $cc -MMD -MF $out.d -x assembler-with-cpp $asflags $flags -c $in -o $out",
+        "  depfile = $out.d",
+        "  deps = gcc",
+        "  description = AS $out",
+        # No $ccache: the .gch is compiled once per build dir and ccache
+        # cannot cache it usefully (its bytes embed build-dir paths)
+        "rule pch",
+        "  command = $cxx -MMD -MF $out.d -x c++-header $cxxflags $flags -c $in -o $out",
+        "  depfile = $out.d",
+        "  deps = gcc",
+        "  description = PCH $out",
+        # Plain assembler, as SCons's ASCOM: no preprocessor, so no
+        # depfile and no $flags (defines/includes) either
+        "rule asm",
+        "  command = $ccache $cc -x assembler $asflags -c $in -o $out",
+        "  description = AS $out",
+        "rule ar",
+        f"  command = $python $buildtool ar {_q(toolchain_tool(paths.toolchain, 'ar'))} $out $out.rsp",
+        "  rspfile = $out.rsp",
+        "  rspfile_content = $in_newline",
+        "  description = AR $out",
+        "rule link",
+        "  command = $cxx -o $out $linkflags @$out.rsp $libdirflags -Wl,--start-group $archives $libflags -Wl,--end-group",
+        "  rspfile = $out.rsp",
+        "  rspfile_content = $in_newline",
+        "  description = LINK $out",
+        "rule elf2bin",
+        # --flash_size deliberately stays board-derived, as under
+        # PlatformIO (which reads upload.maximum_size, not the ldscript).
+        # -W: the framework's own elf2bin.py trips SyntaxWarning on 3.12+.
+        f"  command = $python -W ignore::SyntaxWarning {_q(framework / 'tools' / 'elf2bin.py')} --eboot {_q(framework / 'bootloaders' / 'eboot' / 'eboot.elf')} --app $in --flash_mode {flash_mode} --flash_freq {_FLASH_FREQ_MHZ} --flash_size {_flash_size_str(BOARDS[board][KEY_FLASH_SIZE])} --path {_q(toolchain_bin)} --out $out",
+        "  description = BIN $out",
+        "rule copy",
+        "  command = $python $buildtool copy $in $out",
+        "  description = COPY $out",
+        "",
+        f"cflags = {' '.join(cflags)}",
+        f"cxxflags = {' '.join(cxxflags)}",
+        f"asflags = {' '.join(asflags)}",
+        f"linkflags = {' '.join(link_flags)}",
+        f"libdirflags = {' '.join(f'-L{_q(d)}' for d in lib_dirs)}",
+        f"libflags = {' '.join(_shell_token(f'-l{lib}') for lib in system_libs)}",
+        "",
+    ]
+
+    core_exclude = set(_CORE_EXCLUDE_ALWAYS)
+    if "USE_ESP8266_WAVEFORM_STUBS" in flag_defines:
+        core_exclude |= _CORE_EXCLUDE_WAVEFORM
+
+    archives = []
+    direct_objs: list[str] = []
+    # variant_dir existence was already enforced with the include dirs
+    variant_sources = _collect_sources(variant_dir)
+    if variant_sources:
+        objs = _ninja_compile_edges(lines, variant_sources, variant_dir, "variant")
+        lines.append(f"build libFrameworkArduinoVariant.a: ar {' '.join(objs)}")
+        archives.append("libFrameworkArduinoVariant.a")
+
+    core_objs = _ninja_compile_edges(
+        lines, _collect_sources(core_dir, core_exclude), core_dir, "core"
+    )
+    if not core_objs:
+        # An empty archive would link into a wall of undefined references
+        # (app_entry, the exception vectors) far from the cause
+        raise EsphomeError(
+            f"{_INCOMPLETE_INSTALL}: no core sources in {core_dir}; {_CLEAN_HINT}"
+        )
+    lines.append(f"build libFrameworkArduino.a: ar {' '.join(core_objs)}")
+    archives.append("libFrameworkArduino.a")
+
+    for lib in libraries:
+        if not lib.sources:
+            # Header-only libraries are legitimate; the log makes an empty
+            # srcFilter or broken tree traceable before link errors do.
+            _LOGGER.debug(
+                "Library %s has no source files; contributing includes only",
+                lib.name,
+            )
+            continue
+        lib_root = _common_parent(lib.sources)
+        objs = _ninja_compile_edges(
+            lines,
+            lib.sources,
+            lib_root,
+            f"lib/{lib.name}",
+            flags=" ".join(_shell_token(f) for f in lib.flags),
+        )
+        if not lib.lib_archive:
+            # libArchive: false / dot_a_linkage=false: hand the objects to
+            # the linker directly so unreferenced-but-required symbols
+            # (exception handlers, weak overrides) survive
+            direct_objs.extend(objs)
+            continue
+        archive = f"lib{lib.name}.a"
+        lines.append(f"build {_e(archive)}: ar {' '.join(objs)}")
+        archives.append(archive)
+
+    # One source of truth with the PlatformIO path: esp8266/__init__ pins
+    # build_src_flags (the throw_stubs force-include); -include paths
+    # resolve against the source root
+    src_other: list[str] = []
+    src_includes: list[str] = []
+    src_it = iter(
+        lex_build_flags(_pio_option("build_src_flags", ""), "build_src_flags")
+    )
+    for tok in src_it:
+        if tok == "-include":
+            header = next(src_it, "")
+            if not header:
+                raise EsphomeError(
+                    "build_src_flags has a trailing '-include' with no header"
+                )
+            src_includes.append(header)
+        elif is_joined_include(tok):
+            # Left in src_other it would precede the pch include and
+            # silently defeat the .gch
+            src_includes.append(tok[len("-include") :])
+        else:
+            src_other.append(_shell_token(tok))
+    include_flags = [f"-include {_q(src_dir / h)}" for h in src_includes]
+    # One shared variable instead of repeating the flags line on every src
+    # edge (hundreds of edges in a real project)
+    lines.append(f"srcflags = {' '.join(src_other + include_flags)}")
+    src_cxx_override = None
+    if pch_enabled() and any(tok.startswith("-include") for tok in cxxflags):
+        # $cxxflags expands first, so a user -include there means GCC would
+        # never load the .gch
+        _LOGGER.warning(
+            "A -include in build_flags prevents the precompiled header from "
+            "loading; compiling without it"
+        )
+    elif pch_enabled():
+        # C++ src edges swap the force-includes for one precompiled prefix
+        # header (same content plus defines.h); C/assembly keep srcflags
+        pch_header = build_dir / PCH_HEADER_NAME
+        pch_includes = (*src_includes, PCH_CORE_HEADER)
+        pch_text = pch_header_text(pch_includes)
+        checksum = None
+        try:
+            if ccache:
+                # The .sum exists only for CCACHE_PCH_EXTSUM; ninja's depfile
+                # handles staleness. Strip resolved and raw build paths
+                # (symlinks) so identical configs share cache entries
+                flags_id = (
+                    " ".join(cxxflags)
+                    .replace(effective_ccache_basedir(), "")
+                    .replace(str(CORE.build_path), "")
+                )
+                # The header text covers include order
+                checksum = pch_checksum(
+                    src_dir,
+                    pch_includes,
+                    (
+                        pch_text,
+                        str(paths.framework),
+                        str(paths.toolchain),
+                        flags_id,
+                    ),
+                )
+        except (OSError, UnicodeError) as err:
+            # Identity unknown: a stale cache entry must never be served
+            _LOGGER.warning(
+                "Could not establish the pch identity; compiling without it: %s", err
+            )
+        else:
+            _LOGGER.info(
+                "Compiling with a precompiled header "
+                "(set ESPHOME_PCH_ENABLE=0 to disable)"
+            )
+            write_file_if_changed(pch_header, pch_text)
+            sum_path = build_dir / f"{PCH_HEADER_NAME}.gch.sum"
+            if checksum is not None:
+                # Generate-time stamp: a hand-run ninja can rebuild the .gch
+                # while this .sum lags
+                write_file_if_changed(sum_path, checksum + "\n")
+            else:
+                # A stale .sum from an earlier ccache run must not survive
+                sum_path.unlink(missing_ok=True)
+            gch = _e(f"{PCH_HEADER_NAME}.gch")
+            lines.append(f"build {gch}: pch {_e(pch_header)}")
+            if src_other:
+                lines.append(f"  flags = {' '.join(src_other)}")
+            # Relative -include: absolute would break cross-device ccache.
+            # -Wno-error keeps a rejected .gch a warning under user -Werror
+            cxx_parts = src_other + [
+                f"-Winvalid-pch -Wno-error=invalid-pch -include {PCH_HEADER_NAME}"
+            ]
+            lines.append(f"srccxxflags = {' '.join(cxx_parts)}")
+            src_cxx_override = ("$srccxxflags", gch)
+            mark_pch_emitted()
+    src_objs = _ninja_compile_edges(
+        lines,
+        _collect_sources(src_dir),
+        src_dir,
+        "src",
+        flags="$srcflags",
+        cxx_override=src_cxx_override,
+    )
+
+    ld_deps = [f"ld/{_COMMON_LD_NAME}"]
+    if CORE.testing_mode:
+        ld_deps.append(f"ld/{flash_ld}")
+    lines.append(
+        f"build firmware.elf: link {' '.join(src_objs + direct_objs)} | "
+        f"{' '.join(_e(a) for a in archives)} {' '.join(_e(d) for d in ld_deps)}"
+    )
+    lines.append(f"  archives = {' '.join(_shell_token(a) for a in archives)}")
+    lines.append("build firmware.bin: elf2bin firmware.elf")
+    lines.append("build firmware.factory.bin: copy firmware.bin")
+    lines.append("build firmware.ota.bin: copy firmware.bin")
+    lines.append("default firmware.factory.bin firmware.ota.bin")
+    lines.append("")
+
+    return write_file_if_changed(build_dir / "build.ninja", "\n".join(lines))
+
+
+def get_flash_ld_path(build_dir: Path, paths: InstalledPaths) -> Path:
+    """The flash linker script the link actually uses (for size reporting).
+
+    Reads the same install the ninja file linked against instead of
+    re-resolving the framework version. A user-shipped override living in a
+    custom -L dir resolves to a nonexistent path here; the size consumer
+    warns and skips the Flash summary then.
+    """
+    name = _active_flash_ld_name(_flash_ld_name(CORE.data[KEY_ESP8266][KEY_BOARD]))
+    if CORE.testing_mode:
+        return build_dir / "ld" / name
+    return paths.framework / "tools" / "sdk" / "ld" / name
+
+
+def _flash_size_str(flash_size: int) -> str:
+    """Flash size argument for elf2bin (e.g. ``4M``, ``512K``)."""
+    mb = 1024 * 1024
+    return f"{flash_size // mb}M" if flash_size >= mb else f"{flash_size // 1024}K"

esphome/build_helpers/ccache.py

@@ -87,6 +87,19 @@ def ccache_defaults_env(cache_dir: Path) -> dict[str, str]:
         "CCACHE_DIR": str(cache_dir),
         "CCACHE_NOHASHDIR": "true",
         "CCACHE_DEPEND": "1",
-        "CCACHE_BASEDIR": str(Path(CORE.build_path).resolve()),
+        # A user value wins via the filter below
+        "CCACHE_BASEDIR": effective_ccache_basedir(),
     }
     return {k: v for k, v in defaults.items() if k not in os.environ}
+
+
+def effective_ccache_basedir() -> str:
+    """The prefix ccache rewrites out of hashed paths: a user CCACHE_BASEDIR
+    wins, else the resolved build path (matching ccache_defaults_env)."""
+    from esphome.core import CORE
+
+    raw = os.environ.get("CCACHE_BASEDIR")
+    if raw is not None and Path(raw).is_absolute() and len(Path(raw).parts) > 1:
+        return raw
+    # Unset or degenerate ("", "/", relative): fall back to the build path
+    return str(Path(CORE.build_path).resolve())

esphome/build_helpers/idedata.py

@@ -11,6 +11,7 @@ consumers (IDE integration, clang-tidy) expect:
 
 from __future__ import annotations
 
+from collections.abc import Callable
 import json
 import logging
 import os
@@ -21,6 +22,8 @@ import subprocess
 from esphome.core import EsphomeError
 from esphome.helpers import write_file
 
+_LOGGER = logging.getLogger(__name__)
+
 # Everything idedata generation may raise after a successful link; idedata
 # is a bonus artifact, so consumers warn instead of failing the build
 IDEDATA_BEST_EFFORT_ERRORS = (
@@ -31,13 +34,36 @@ IDEDATA_BEST_EFFORT_ERRORS = (
     ValueError,
 )
 
-_LOGGER = logging.getLogger(__name__)
 
-# C++ translation-unit suffixes used to identify ESPHome source files.
-_CXX_SUFFIXES = (".cpp", ".cc")
+def warn_if_idedata_missing(get_idedata: Callable[[], dict | None]) -> None:
+    """Run an idedata generator, downgrading any failure to a warning.
+
+    Shared by the native backends: the firmware already built, so a missing
+    or broken idedata must not fail a successful build.
+    """
+    try:
+        if get_idedata() is None:
+            _LOGGER.warning("No idedata was generated for this build")
+    except IDEDATA_BEST_EFFORT_ERRORS as err:
+        _LOGGER.warning(
+            "Could not generate idedata: %s (IDE, clang-tidy, and "
+            "memory-analysis data will be unavailable for this build)",
+            err,
+        )
+        if isinstance(err, (EsphomeError, OSError)):
+            # Routine environmental failures keep the detail at debug
+            _LOGGER.debug("Idedata failure detail", exc_info=True)
+        else:
+            # LookupError/ValueError/RuntimeError smell like a parsing bug;
+            # a permanently masked traceback would hide it on every build
+            _LOGGER.warning("Idedata failure detail", exc_info=True)
+
+
+# C++ translation-unit suffixes.
+CXX_SOURCE_SUFFIXES = (".cpp", ".cc", ".cxx")
 # Suffixes of input/output files that appear bare on the command line (and so
 # must not be mistaken for compiler flags).
-_INPUT_FILE_SUFFIXES = (*_CXX_SUFFIXES, ".c", ".o", ".S", ".s")
+_INPUT_FILE_SUFFIXES = (*CXX_SOURCE_SUFFIXES, ".c", ".o", ".S", ".s")
 # Path marker identifying an ESPHome source translation unit.
 _ESPHOME_SRC_MARKER = "/src/esphome/"
 
@@ -46,7 +72,7 @@ def _is_esphome_src(file: str) -> bool:
     """Whether ``file`` is an ESPHome C++ translation unit; normalized to
     ``/`` first since Windows compile DBs use backslashes."""
     return _ESPHOME_SRC_MARKER in file.replace("\\", "/") and file.endswith(
-        _CXX_SUFFIXES
+        CXX_SOURCE_SUFFIXES
     )
 
 
@@ -121,7 +147,7 @@ def _pick_entry(entries: list[dict]) -> dict:
         if _is_esphome_src(entry["file"]):
             return entry
     for entry in entries:
-        if entry["file"].endswith(_CXX_SUFFIXES):
+        if entry["file"].endswith(CXX_SOURCE_SUFFIXES):
             return entry
     raise ValueError("no C++ translation unit found in compile_commands.json")
 
@@ -135,6 +161,15 @@ def _is_launcher(token: str) -> bool:
     return Path(token).stem.lower() in _LAUNCHER_STEMS
 
 
+def is_joined_include(tok: str) -> bool:
+    """The joined ``-includefoo.h`` spelling; excludes clang's -include-pch."""
+    return (
+        tok.startswith("-include")
+        and tok != "-include"
+        and not tok.startswith("-include-")
+    )
+
+
 def parse_entry(
     entry: dict, launcher: str | None = None
 ) -> tuple[str, list[str], list[str], list[str]]:
@@ -169,11 +204,23 @@ def parse_entry(
     defines: list[str] = []
     includes: list[str] = []
     cxx_flags: list[str] = []
+    unresolved_force_includes: list[str] = []
 
     it = iter(tokens[1:])
     for tok in it:
         if tok in ("-c", "-o"):
             next(it, None)  # drop the flag and its argument (input/output)
+        elif tok == "-include" or is_joined_include(tok):
+            # Re-anchor only names next to the compile (the pch); a name
+            # meant for the -I chain must stay untouched
+            raw = next(it, "") if tok == "-include" else tok[len("-include") :]
+            if not raw:
+                _LOGGER.warning("Dropping -include with no argument")
+            elif Path(resolved := _include(raw)).is_file():
+                cxx_flags.extend(("-include", resolved))
+            else:
+                unresolved_force_includes.append(raw)
+                cxx_flags.extend(("-include", raw))
         elif tok.startswith("-D"):
             # ``.strip()`` handles tokens like ``-D CONFIGURED=1`` (a single
             # quoted arg with a space after -D) that some flags arrive as.
@@ -192,6 +239,14 @@ def parse_entry(
             pass  # input/output files
         else:
             cxx_flags.append(tok)
+    for raw in unresolved_force_includes:
+        # A deleted build artifact would otherwise surface only downstream
+        if not any((Path(inc) / raw).is_file() for inc in includes):
+            _LOGGER.warning(
+                "-include %s found neither next to the compile nor on the "
+                "include path; cached idedata may not resolve it",
+                raw,
+            )
     return cxx_path, defines, includes, cxx_flags
 
 

esphome/build_helpers/pch.py

@@ -0,0 +1,178 @@
+"""Shared precompiled-header policy for the build backends.
+
+The prefix either mirrors the TUs' own force-includes (ESP8266) or is a
+curated core-header set (ESP-IDF). ``esphome: includes:`` sources receive
+it too; Arduino.h visibility there is intended (esphome#8693).
+"""
+
+from __future__ import annotations
+
+from collections.abc import Iterable
+from dataclasses import dataclass
+import hashlib
+import logging
+import os
+from pathlib import Path
+import posixpath
+import re
+import stat
+
+from esphome.build_helpers.ccache import parse_enable_env
+
+_DOMAIN = "pch"
+
+
+@dataclass
+class _PCHData:
+    emitted: bool = False
+
+
+def _pch_data() -> _PCHData:
+    from esphome.core import CORE
+
+    if _DOMAIN not in CORE.data:
+        CORE.data[_DOMAIN] = _PCHData()
+    return CORE.data[_DOMAIN]
+
+
+def mark_pch_emitted() -> None:
+    """Record that this build's consumers reference the pch."""
+    _pch_data().emitted = True
+
+
+_LOGGER = logging.getLogger(__name__)
+
+# The header and its .gch/.sum sidecars live in the build directory.
+PCH_HEADER_NAME = "esphome_pch.h"
+
+# Every artifact the pch machinery can leave behind, for cleanup.
+PCH_ARTIFACT_NAMES = (
+    PCH_HEADER_NAME,
+    f"{PCH_HEADER_NAME}.gch",
+    f"{PCH_HEADER_NAME}.gch.sum",
+    f"{PCH_HEADER_NAME}.gch.failed",
+)
+
+# The core defines header every backend anchors its prefix on.
+PCH_CORE_HEADER = "esphome/core/defines.h"
+
+# ccache cannot hash through a .gch; CCACHE_PCH_EXTSUM makes it hash the
+# .sum sidecar instead of the .gch bytes, which are not reproducible.
+# Keep in sync with the literals in platformio/pch.py.script.
+_CCACHE_PCH_ENV = {
+    "CCACHE_SLOPPINESS": "pch_defines,time_macros",
+    "CCACHE_PCH_EXTSUM": "true",
+}
+
+# Both include forms: an angle include resolving under src/ must enter the
+# digest too; ones that do not resolve simply end the walk
+_INCLUDE_RE = re.compile(rb'^\s*#\s*include\s+["<]([^">]+)[">]', re.MULTILINE)
+
+
+def pch_enabled() -> bool:
+    """Precompiled-header knob: default on, ``ESPHOME_PCH_ENABLE=0`` opts out."""
+    return parse_enable_env("ESPHOME_PCH_ENABLE") is not False
+
+
+def ccache_pch_env() -> dict[str, str]:
+    """Settings ccache needs to cache compiles that consume the .gch;
+    empty unless this build actually emitted one. User-set values win.
+    Native backends export these process-wide; only time_macros affects
+    non-pch TUs."""
+    if not (pch_enabled() and _pch_data().emitted):
+        return {}
+    extsum = os.environ.get("CCACHE_PCH_EXTSUM")
+    if extsum is not None and extsum.strip().lower() not in ("1", "true", "yes", "on"):
+        # ccache then hashes the non-reproducible .gch bytes: permanent misses
+        _LOGGER.warning("CCACHE_PCH_EXTSUM=%s disables pch caching", extsum)
+    env = {k: v for k, v in _CCACHE_PCH_ENV.items() if k not in os.environ}
+    user_sloppiness = os.environ.get("CCACHE_SLOPPINESS")
+    if user_sloppiness is not None and (
+        missing := [
+            t
+            for t in ("pch_defines", "time_macros")
+            if t not in {tok.strip() for tok in user_sloppiness.split(",")}
+        ]
+    ):
+        # Without these ccache declines every pch-consuming compile
+        env["CCACHE_SLOPPINESS"] = ",".join((user_sloppiness, *missing))
+        _LOGGER.warning(
+            "Adding %s to CCACHE_SLOPPINESS so ccache can cache compiles "
+            "that use the precompiled header",
+            ",".join(missing),
+        )
+    return env
+
+
+def pch_extra_scripts() -> list[str]:
+    """The extra_scripts entries a PlatformIO platform registers for the
+    pch; empty when disabled (the script itself has no enable check)."""
+    return ["post:pch.py"] if pch_enabled() else []
+
+
+def pch_header_text(include_headers: Iterable[str]) -> str:
+    """The prefix-header source: exactly these includes, in order."""
+    return "".join(f'#include "{name}"\n' for name in include_headers)
+
+
+def _resolves(path: Path) -> bool:
+    """False when missing; other stat failures propagate (identity unknown,
+    unlike is_file(), which would silently drop the header)."""
+    try:
+        return stat.S_ISREG(path.stat().st_mode)
+    except (FileNotFoundError, NotADirectoryError):
+        return False
+
+
+def _include_closure(src_dir: Path, roots: Iterable[str]) -> dict[str, bytes]:
+    """Include closure of ``roots``: src-relative name -> contents.
+
+    Resolution mirrors the compiler (includer's dir, then src root); names
+    outside ``src_dir`` end the walk and are versioned by the caller. No
+    #ifdef evaluation: over-approximating is the safe direction.
+    """
+    seen: dict[str, bytes] = {}
+    stack: list[tuple[str, str]] = [(name, "") for name in roots]
+    while stack:
+        name, from_dir = stack.pop()
+        for candidate in (f"{from_dir}/{name}" if from_dir else name, name):
+            rel = posixpath.normpath(candidate)
+            if not rel.startswith("..") and _resolves(src_dir / rel):
+                break
+        else:
+            continue
+        if rel in seen:
+            continue
+        try:
+            data = (src_dir / rel).read_bytes()
+        except OSError as err:
+            # A marker would truncate the transitive walk; fail closed
+            _LOGGER.warning("Could not read %s for the pch checksum: %s", rel, err)
+            raise
+        seen[rel] = data
+        parent = posixpath.dirname(rel)
+        stack.extend(
+            # surrogateescape: a non-UTF-8 name just fails to resolve
+            (inc.decode(errors="surrogateescape"), parent)
+            for inc in _INCLUDE_RE.findall(data)
+        )
+    return seen
+
+
+def pch_checksum(
+    src_dir: Path, include_headers: Iterable[str], extra: Iterable[str]
+) -> str:
+    """Digest standing in for the .gch in ccache's hash: the include closure
+    of the prefix header plus caller-supplied identity strings (versioned
+    install paths, flags). Raises OSError when a header's identity cannot
+    be established at all; callers must then compile without a pch."""
+    digest = hashlib.sha256()
+    closure = _include_closure(src_dir, include_headers)
+    for name in sorted(closure):
+        digest.update(name.encode(errors="surrogateescape"))
+        digest.update(closure[name])
+        digest.update(b"\0")
+    for item in extra:
+        digest.update(item.encode(errors="surrogateescape"))
+        digest.update(b"\0")
+    return digest.hexdigest()

esphome/components/esp8266/__init__.py

@@ -3,8 +3,10 @@ from pathlib import Path
 import platform
 import re
 import subprocess
+import time
 from typing import Any
 
+from esphome.build_helpers.pch import pch_extra_scripts
 import esphome.codegen as cg
 import esphome.config_validation as cv
 from esphome.const import (
@@ -14,6 +16,7 @@ from esphome.const import (
     CONF_FRAMEWORK,
     CONF_PLATFORM_VERSION,
     CONF_SOURCE,
+    CONF_TOOLCHAIN,
     CONF_VERSION,
     KEY_CORE,
     KEY_FRAMEWORK_VERSION,
@@ -21,6 +24,7 @@ from esphome.const import (
     KEY_TARGET_PLATFORM,
     PLATFORM_ESP8266,
     ThreadModel,
+    Toolchain,
 )
 from esphome.core import (
     CORE,
@@ -31,12 +35,13 @@ from esphome.core import (
 )
 from esphome.core.config import BOARD_MAX_LENGTH
 from esphome.helpers import IS_MACOS, copy_file_if_changed
-from esphome.platformio.toolchain import copy_ccache_script
+from esphome.platformio.toolchain import copy_ccache_script, copy_pch_script
 from esphome.storage_json import StorageJSON
 from esphome.types import ConfigType
 
-from .boards import BOARDS, board_ld_script
+from .boards import BOARDS, ESP8266_BOARD_BUILD, board_ld_script
 from .const import (
+    BUILD_FLASH_MODES,
     CONF_EARLY_PIN_INIT,
     CONF_ENABLE_SERIAL,
     CONF_ENABLE_SERIAL1,
@@ -44,6 +49,7 @@ from .const import (
     KEY_BOARD,
     KEY_ESP8266,
     KEY_PIN_INITIAL_STATES,
+    KEY_SCANF_FLOAT,
     KEY_SERIAL1_REQUIRED,
     KEY_SERIAL_REQUIRED,
     KEY_WAVEFORM_REQUIRED,
@@ -103,6 +109,53 @@ def set_core_data(config: ConfigType) -> ConfigType:
     return config
 
 
+_TOOLCHAINS = (Toolchain.PLATFORMIO, Toolchain.ARDUINO)
+_validate_toolchain = cv.toolchain_enum(_TOOLCHAINS)
+_resolve_toolchain = cv.resolve_toolchain("ESP8266", _TOOLCHAINS, Toolchain.PLATFORMIO)
+
+
+def _validate_native_toolchain(config: ConfigType) -> ConfigType:
+    """Constraints of the native (non-PlatformIO) Arduino toolchain."""
+    if not CORE.using_toolchain_arduino:
+        return config
+    from esphome.arduino8266.framework import MIN_FRAMEWORK_VERSION
+
+    conf = config[CONF_FRAMEWORK]
+    version = cv.Version.parse(conf[CONF_VERSION])
+    if version < MIN_FRAMEWORK_VERSION:
+        raise cv.Invalid(
+            "'toolchain: arduino' requires framework version "
+            f"{MIN_FRAMEWORK_VERSION} or newer"
+        )
+    # platform_version is a PlatformIO concept; drop it (as esp32's native
+    # toolchain does), warning when a custom pin is discarded. The floor
+    # above guarantees the schema-derived default is the ARDUINO_4 spec.
+    if (
+        conf.pop(CONF_PLATFORM_VERSION, _ARDUINO_4_PLATFORM_SPEC)
+        != _ARDUINO_4_PLATFORM_SPEC
+    ):
+        _LOGGER.warning(
+            "'platform_version' is ignored by 'toolchain: arduino'; the native "
+            "toolchain downloads the framework and compiler directly"
+        )
+    if conf[CONF_SOURCE] != _format_framework_arduino_version(version):
+        raise cv.Invalid(
+            "'toolchain: arduino' does not support a custom framework source; "
+            "use 'toolchain: platformio'"
+        )
+    # BOARDS is a subset of ESP8266_BOARD_BUILD today; the second clause is
+    # a drift guard for the independently regenerated tables
+    if (
+        config[CONF_BOARD] not in BOARDS
+        or config[CONF_BOARD] not in ESP8266_BOARD_BUILD
+    ):
+        raise cv.Invalid(
+            f"Board '{config[CONF_BOARD]}' is not supported by "
+            "'toolchain: arduino'; use 'toolchain: platformio'"
+        )
+    return config
+
+
 def get_download_types(storage_json: StorageJSON) -> list[dict[str, str]]:
     """Binary-download entries for a built ESP8266 firmware.
 
@@ -132,7 +185,7 @@ def _format_framework_arduino_version(ver: cv.Version) -> str:
     # a PIO platformio/framework-arduinoespressif8266 value
     # List of package versions: https://api.registry.platformio.org/v3/packages/platformio/tool/framework-arduinoespressif8266
     # Same encoding the native toolchain uses for its package download, so a
-    # version bump cannot drift between the two paths.
+    # custom-source check against this value cannot drift from what it fetches.
     from esphome.arduino8266.framework import framework_package_version
 
     try:
@@ -194,7 +247,7 @@ def _arduino_check_versions(value: ConfigType) -> ConfigType:
     platform_version = value.get(CONF_PLATFORM_VERSION)
     if platform_version is None:
         if version >= cv.Version(3, 1, 0):
-            platform_version = _parse_platform_version(str(ARDUINO_4_PLATFORM_VERSION))
+            platform_version = _ARDUINO_4_PLATFORM_SPEC
         else:
             platform_version = _parse_platform_version(str(ARDUINO_3_PLATFORM_VERSION))
     value[CONF_PLATFORM_VERSION] = platform_version
@@ -217,6 +270,10 @@ def _parse_platform_version(value: Any) -> str:
         return value
 
 
+# The platform_version derived for every core >= 3.1.0 config
+_ARDUINO_4_PLATFORM_SPEC = _parse_platform_version(str(ARDUINO_4_PLATFORM_VERSION))
+
+
 ARDUINO_FRAMEWORK_SCHEMA = cv.All(
     cv.Schema(
         {
@@ -233,7 +290,6 @@ ARDUINO_FRAMEWORK_SCHEMA = cv.All(
 )
 
 
-BUILD_FLASH_MODES = ["qio", "qout", "dio", "dout"]
 CONFIG_SCHEMA = cv.All(
     cv.Schema(
         {
@@ -250,15 +306,30 @@ CONFIG_SCHEMA = cv.All(
             cv.Optional(CONF_ENABLE_SERIAL1): cv.boolean,
             cv.Optional(CONF_ENABLE_FULL_PRINTF, default=False): cv.boolean,
             cv.Optional(CONF_ENABLE_SCANF_FLOAT): cv.boolean,
+            cv.Optional(
+                CONF_TOOLCHAIN, visibility=cv.Visibility.ADVANCED
+            ): _validate_toolchain,
         }
     ),
-    # Until the native toolchain lands, PlatformIO is the only backend;
-    # reject a --toolchain this platform cannot serve yet.
-    cv.require_platformio_toolchain("ESP8266"),
+    _resolve_toolchain,
+    _validate_native_toolchain,
     set_core_data,
 )
 
 
+def native_toolchain_module():
+    """The native build backend for the resolved toolchain, if any.
+
+    ``__main__`` dispatches from its own toolchain-keyed table; this helper
+    serves the component's internal callers.
+    """
+    if not CORE.using_toolchain_arduino:
+        return None
+    from esphome.arduino8266 import toolchain
+
+    return toolchain
+
+
 def check_rosetta() -> None:
     """Fail fast when the x86_64 ESP8266 toolchain cannot run on this Mac.
 
@@ -294,12 +365,13 @@ def _choose_ld_script(board: str) -> str:
 
 @coroutine_with_priority(CoroPriority.PLATFORM)
 async def to_code(config: ConfigType) -> None:
+    use_platformio = CORE.using_toolchain_platformio
     cg.add(esp8266_ns.setup_preferences())
 
-    cg.add_platformio_option("lib_ldf_mode", "off")
-    cg.add_platformio_option("lib_compat_mode", "strict")
-
-    cg.add_platformio_option("board", config[CONF_BOARD])
+    if use_platformio:
+        cg.add_platformio_option("lib_ldf_mode", "off")
+        cg.add_platformio_option("lib_compat_mode", "strict")
+        cg.add_platformio_option("board", config[CONF_BOARD])
     cg.add_build_flag("-DUSE_ESP8266")
     cg.set_cpp_standard("gnu++20")
     cg.add_define("ESPHOME_BOARD", config[CONF_BOARD])
@@ -315,28 +387,33 @@ async def to_code(config: ConfigType) -> None:
             "enabling scanf float support (~8KB flash)"
         )
 
-    extra_scripts = [
-        "pre:ccache.py",
-        "pre:testing_mode.py",
-        "pre:exclude_updater.py",
-        "pre:exclude_waveform.py",
-        "pre:relocate_ratetable.py",
-    ]
-    if not enable_scanf_float:
-        extra_scripts.append("pre:remove_float_scanf.py")
-    extra_scripts.append("post:post_build.py")
-    cg.add_platformio_option("extra_scripts", extra_scripts)
+    # The native generator reads the same decision (KEY_SCANF_FLOAT)
+    CORE.data[KEY_ESP8266][KEY_SCANF_FLOAT] = bool(enable_scanf_float)
+    if use_platformio:
+        extra_scripts = [
+            "pre:ccache.py",
+            "pre:testing_mode.py",
+            "pre:exclude_updater.py",
+            "pre:exclude_waveform.py",
+            "pre:relocate_ratetable.py",
+        ]
+        if not enable_scanf_float:
+            extra_scripts.append("pre:remove_float_scanf.py")
+        extra_scripts.extend(pch_extra_scripts())
+        extra_scripts.append("post:post_build.py")
+        cg.add_platformio_option("extra_scripts", extra_scripts)
 
     conf = config[CONF_FRAMEWORK]
-    cg.add_platformio_option("framework", "arduino")
     cg.add_build_flag("-DUSE_ARDUINO")
     cg.add_build_flag("-DUSE_ESP8266_FRAMEWORK_ARDUINO")
     cg.add_build_flag("-Wno-nonnull-compare")
-    cg.add_platformio_option("platform", conf[CONF_PLATFORM_VERSION])
-    cg.add_platformio_option(
-        "platform_packages",
-        [f"platformio/framework-arduinoespressif8266@{conf[CONF_SOURCE]}"],
-    )
+    if use_platformio:
+        cg.add_platformio_option("framework", "arduino")
+        cg.add_platformio_option("platform", conf[CONF_PLATFORM_VERSION])
+        cg.add_platformio_option(
+            "platform_packages",
+            [f"platformio/framework-arduinoespressif8266@{conf[CONF_SOURCE]}"],
+        )
 
     # Default for platformio is LWIP2_LOW_MEMORY with:
     #  - MSS=536
@@ -374,7 +451,8 @@ async def to_code(config: ConfigType) -> None:
     # Force-include inline std::__throw_* overrides so GCC dead-strips the unused
     # libstdc++ error message strings (e.g. "basic_string::_M_create") from DRAM.
     # See throw_stubs.h for details. Must be prepended before <string>, so this
-    # uses build_src_flags with -include.
+    # uses build_src_flags with -include. Unconditional: the native build
+    # generator reads the same option, keeping one source of truth.
     cg.add_platformio_option(
         "build_src_flags", "-include esphome/components/esp8266/throw_stubs.h"
     )
@@ -404,6 +482,8 @@ async def to_code(config: ConfigType) -> None:
     # implementation in the Arduino ESP8266 core.
     cg.add_build_flag("-Wl,--wrap=millis")
 
+    # Unconditional: the native build generator reads the same option,
+    # keeping one source of truth
     cg.add_platformio_option("board_build.flash_mode", config[CONF_BOARD_FLASH_MODE])
 
     ver: cv.Version = CORE.data[KEY_CORE][KEY_FRAMEWORK_VERSION]
@@ -412,7 +492,7 @@ async def to_code(config: ConfigType) -> None:
         cg.RawExpression(f"VERSION_CODE({ver.major}, {ver.minor}, {ver.patch})"),
     )
 
-    if config[CONF_BOARD] in BOARDS:
+    if use_platformio and config[CONF_BOARD] in BOARDS:
         cg.add_platformio_option(
             "board_build.ldscript", _choose_ld_script(config[CONF_BOARD])
         )
@@ -453,8 +533,24 @@ async def finalize_serial_config() -> None:
         cg.add_build_flag("-DNO_GLOBAL_SERIAL1")
 
 
+# Called by __main__.compile_program; returning False falls through to the
+# PlatformIO toolchain.
+def run_compile(args, config: ConfigType) -> bool:
+    # Positive check: the native backend only runs when explicitly resolved
+    toolchain = native_toolchain_module()
+    if toolchain is None:
+        return False
+    if toolchain.run_compile(config, CORE.verbose) != 0:
+        raise EsphomeError("ESP8266 native build failed")
+    return True
+
+
 # Called by writer.py
 def copy_files() -> None:
+    # Native builds skip the PlatformIO extra scripts; the build generator
+    # carries their logic
+    if CORE.using_toolchain_arduino:
+        return
     dir = Path(__file__).parent
     for script in (
         "post_build",
@@ -469,6 +565,7 @@ def copy_files() -> None:
             CORE.relative_build_path(f"{script}.py"),
         )
     copy_ccache_script()
+    copy_pch_script()
 
 
 # ESP logs stack trace decoder, based on https://github.com/me-no-dev/EspExceptionDecoder
@@ -511,22 +608,75 @@ ESP8266_EXCEPTION_CODES = {
 }
 
 
-def _decode_pc(config: ConfigType, addr: str) -> None:
-    from esphome.platformio import toolchain
+_DECODE_WARNED_AT: dict[str, float] = {}
 
-    idedata = toolchain.get_idedata(config)
-    if not idedata.addr2line_path or not idedata.firmware_elf_path:
-        _LOGGER.debug("decode_pc no addr2line")
-        return
-    command = [idedata.addr2line_path, "-pfiaC", "-e", idedata.firmware_elf_path, addr]
+
+def _warn_decode_problem(key: str, message: str, *args) -> bool:
+    """Warn, deduplicated briefly so a burst of stack-dump addresses warns
+    once but a later dump warns again; returns whether it warned so the
+    caller can mark suppressed addresses individually."""
+    now = time.monotonic()
+    last = _DECODE_WARNED_AT.get(key)
+    if last is not None and now - last < 30:
+        return False
+    _DECODE_WARNED_AT[key] = now
+    _LOGGER.warning(message, *args)
+    return True
+
+
+def _decode_pc(config: ConfigType, addr: str, *, bulk: bool = False) -> None:
+    """Decode one crash address. ``bulk``: the caller is scanning every
+    8-hex stack word, most of which are not code addresses -- unmappable
+    ones log at debug so real frames are not buried."""
+    if (native_toolchain := native_toolchain_module()) is not None:
+        addr2line = native_toolchain.get_addr2line_path()
+        elf = native_toolchain.get_elf_path()
+        for path in (addr2line, elf):
+            if not path.is_file():
+                _warn_decode_problem(
+                    str(path), "Cannot decode crash addresses: %s missing", path
+                )
+                # The detailed warning names no address; mark named
+                # registers, but bulk stack words at debug (~150 per dump)
+                log = _LOGGER.debug if bulk else _LOGGER.warning
+                log("Not decoded %s (toolchain file missing)", addr)
+                return
+        addr2line, elf = str(addr2line), str(elf)
+    else:
+        from esphome.platformio import toolchain
+
+        idedata = toolchain.get_idedata(config)
+        if not idedata.addr2line_path or not idedata.firmware_elf_path:
+            _warn_decode_problem(
+                "no-addr2line",
+                "Cannot decode crash addresses: no addr2line or ELF in idedata",
+            )
+            log = _LOGGER.debug if bulk else _LOGGER.warning
+            log("Not decoded %s (no addr2line or ELF)", addr)
+            return
+        addr2line, elf = idedata.addr2line_path, idedata.firmware_elf_path
+    command = [addr2line, "-pfiaC", "-e", elf, addr]
     try:
         translation = subprocess.check_output(command, close_fds=False).decode().strip()
-    except Exception:  # noqa: BLE001  # pylint: disable=broad-except
+    except Exception as err:  # noqa: BLE001  # pylint: disable=broad-except
+        # Warn, not debug: a failing addr2line must be visible. The warning
+        # is rate-limited across a dump, so mark every undecoded address
+        # inline or the rest read as merely unmappable
+        if not _warn_decode_problem(
+            "addr2line-failed", "Could not decode crash address %s (%s)", addr, err
+        ):
+            # The detailed warning already named this address; mark only
+            # the rate-limited ones, and bulk stack words at debug
+            log = _LOGGER.debug if bulk else _LOGGER.warning
+            log("Not decoded %s (addr2line failed)", addr)
         _LOGGER.debug("Caught exception for command %s", command, exc_info=1)
         return
 
     if "?? ??:0" in translation:
-        # Nothing useful
+        # A named register that fails to decode is confusing silence; a
+        # bulk stack word failing is the expected common case
+        log = _LOGGER.debug if bulk else _LOGGER.warning
+        log("Not decoded %s (address not in %s)", addr, elf)
         return
     translation = translation.replace(" at ??:?", "").replace(":?", "")
     _LOGGER.warning("Decoded %s", translation)
@@ -596,6 +746,6 @@ def process_stacktrace(config: ConfigType, line: str, backtrace_state: bool) ->
 
     if backtrace_state:
         for addr in re.finditer(STACKTRACE_ESP8266_BACKTRACE_PC_RE, line):
-            _decode_pc(config, addr.group())
+            _decode_pc(config, addr.group(), bulk=True)
 
     return backtrace_state

esphome/components/esp8266/const.py

@@ -16,7 +16,6 @@ KEY_WAVEFORM_REQUIRED = "waveform_required"
 KEY_SERIAL_REQUIRED = "serial_required"
 KEY_SERIAL1_REQUIRED = "serial1_required"
 # Set for the native (non-PlatformIO) toolchain's build generator
-KEY_FLASH_MODE = "flash_mode"
 KEY_SCANF_FLOAT = "scanf_float"
 # Per-board flash-layout override consumed by board_ld_script()
 KEY_LDSCRIPT = "ldscript"
@@ -73,3 +72,6 @@ def enable_serial1() -> None:
             enable_serial1()
     """
     CORE.data.setdefault(KEY_ESP8266, {})[KEY_SERIAL1_REQUIRED] = True
+
+
+BUILD_FLASH_MODES = ("qio", "qout", "dio", "dout")

esphome/core/config.py

@@ -555,7 +555,14 @@ NATIVE_ARDUINO_PIO_OPTIONS = frozenset({"board_build.f_cpu", "board_build.ldscri
 # that is stored rather than translated away. Consumed by the esp8266 native
 # backend (later in this chain) for its ignored-option warning; defined here
 # so it stays adjacent to the routing.
-NATIVE_ARDUINO_CONSUMED_PIO_OPTIONS = NATIVE_ARDUINO_PIO_OPTIONS | {"lib_ignore"}
+# build_src_flags and board_build.flash_mode: set unconditionally by
+# esp8266/__init__ and read by the native generator; not user-routable, so
+# not in the set above
+NATIVE_ARDUINO_CONSUMED_PIO_OPTIONS = NATIVE_ARDUINO_PIO_OPTIONS | {
+    "lib_ignore",
+    "build_src_flags",
+    "board_build.flash_mode",
+}
 
 
 @coroutine_with_priority(CoroPriority.FINAL)

esphome/platformio/pch.py.script

@@ -0,0 +1,364 @@
+import hashlib
+import os
+from pathlib import Path
+import posixpath
+import re
+import shlex
+import stat
+import subprocess
+import traceback
+
+# pylint: disable=E0602
+Import("env")  # noqa: F821
+_projenv_error = None
+try:
+    Import("projenv")  # noqa: F821
+except Exception as err:  # noqa: BLE001 -- not exported under -t nobuild
+    projenv = None
+    _projenv_error = err
+
+# Precompile the src force-includes plus defines.h and force-include the
+# result into C++ src compiles only; their preprocessed output is unchanged.
+# Registration is gated host-side (pch_enabled()). Keep the closure, ccache
+# values, and stamp flow in sync with build_helpers/pch.py.
+
+# Compiler failures that clear on their own must not latch the .failed marker
+_TRANSIENT_ERRORS = ("No space left", "Cannot allocate", "Resource temporarily")
+
+_INCLUDE_RE = re.compile(rb'^\s*#\s*include\s+["<]([^">]+)[">]', re.MULTILINE)
+_CORE_HEADER = "esphome/core/defines.h"
+
+
+def _raise_walk_error(err: OSError) -> None:
+    raise err
+
+
+def _resolves_dir(path: Path) -> bool:
+    """False when missing; other stat failures propagate."""
+    try:
+        return stat.S_ISDIR(path.stat().st_mode)
+    except (FileNotFoundError, NotADirectoryError):
+        return False
+
+
+def _resolves(path: Path) -> bool:
+    """False when missing; other stat failures propagate (identity unknown)."""
+    try:
+        return stat.S_ISREG(path.stat().st_mode)
+    except (FileNotFoundError, NotADirectoryError):
+        return False
+
+
+def _include_closure(src_dir: Path, roots: list) -> dict:
+    """Quoted-include closure: src-relative name -> contents (mirror of
+    build_helpers/pch.py)."""
+    seen = {}
+    stack = [(name, "") for name in roots]
+    while stack:
+        name, from_dir = stack.pop()
+        for candidate in (f"{from_dir}/{name}" if from_dir else name, name):
+            rel = posixpath.normpath(candidate)
+            if not rel.startswith("..") and _resolves(src_dir / rel):
+                break
+        else:
+            continue
+        if rel in seen:
+            continue
+        try:
+            data = (src_dir / rel).read_bytes()
+        except OSError as err:
+            # A marker would truncate the transitive walk; fail closed
+            print(f"ESPHome: could not read {rel} for the pch checksum: {err}")
+            raise
+        seen[rel] = data
+        parent = posixpath.dirname(rel)
+        stack.extend(
+            (inc.decode(errors="surrogateescape"), parent)
+            for inc in _INCLUDE_RE.findall(data)
+        )
+    return seen
+
+
+def _shell_arg(element) -> str | None:
+    """One compiler argv from one SCons element, matching the real spawn:
+    spaced elements pass whole, the rest get one shell unquote (skipped on
+    Windows, where shlex would eat path backslashes)."""
+    arg = str(element)
+    if " " in arg or os.name == "nt":
+        return arg.replace('\\"', '"')
+    if not arg.strip():
+        return arg
+    try:
+        tokens = shlex.split(arg)
+    except ValueError as err:
+        print(f"ESPHome: could not lex flag {arg!r} for the pch: {err}")
+        return None
+    if len(tokens) != 1:
+        # A flag the model cannot reproduce would diverge the .gch's flags
+        print(f"ESPHome: cannot model flag {arg!r} for the pch")
+        return None
+    return tokens[0]
+
+
+def _compile_gch(cxx, flags, header: Path, gch: Path, proj_dir: Path):
+    """Compile the .gch, then probe that the toolchain can load it back
+    (GCC 10 on macOS arm64 rejects its own per-process). Returns an error
+    string or None; OSError propagates as transient."""
+    result = subprocess.run(  # noqa: PLW1510
+        [cxx, "-x", "c++-header", *flags, "-c", str(header), "-o", str(gch)],
+        cwd=proj_dir,
+        # C locale keeps diagnostics matchable by _TRANSIENT_ERRORS
+        env={**os.environ, "LC_ALL": "C"},
+        capture_output=True,
+        text=True,
+    )
+    if result.returncode < 0:
+        # Signal-killed (OOM, ^C): route to the transient no-marker path
+        raise OSError(f"compiler killed by signal {-result.returncode}")
+    if result.returncode != 0:
+        return result.stderr
+    # -MF is only legal alongside a dependency flag; pass it solely to
+    # redirect a depfile that -MD/-MMD in the flags would otherwise write
+    dep_redirect = (
+        ["-MF", os.devnull]
+        if any(f in ("-MD", "-MMD", "-M", "-MM") for f in flags)
+        else []
+    )
+    probe = subprocess.run(  # noqa: PLW1510
+        [
+            cxx,
+            *flags,
+            *dep_redirect,
+            "-Winvalid-pch",
+            "-include",
+            str(header),
+            "-fsyntax-only",
+            "-x",
+            "c++",
+            "-",
+        ],
+        cwd=proj_dir,
+        env={**os.environ, "LC_ALL": "C"},
+        input="",
+        capture_output=True,
+        text=True,
+    )
+    if probe.returncode < 0:
+        raise OSError(f"probe killed by signal {-probe.returncode}")
+    if probe.returncode != 0 or ".gch" in probe.stderr:
+        return f"toolchain cannot load the pch: {probe.stderr.strip()}"
+    return None
+
+
+def _read_stamp(path: Path) -> str:
+    """A corrupt sidecar must read as stale, not kill the pch forever."""
+    try:
+        return path.read_text(encoding="utf-8").strip()
+    except (OSError, UnicodeDecodeError):
+        return ""
+
+
+def _setup_pch() -> None:
+    if projenv is None:
+        # Expected under -t nobuild; anything else must leave a trail
+        print(f"ESPHome: projenv unavailable ({_projenv_error}); skipping pch")
+        return
+    # Project root: SCons compiles run here, so the relative -include
+    # resolves; an absolute path would break cross-device ccache sharing.
+    proj_dir = Path(env.subst("$PROJECT_DIR"))  # noqa: F821
+    src_dir = Path(env.subst("$PROJECT_SRC_DIR"))  # noqa: F821
+    header = proj_dir / "esphome_pch.h"
+    gch = Path(f"{header}.gch")
+    sum_path = Path(f"{gch}.sum")
+
+    cxx = projenv.subst("$CXX")  # noqa: F821
+    # The header holds the -include entries itself, so the .gch compile must
+    # not see them; consumers keep theirs, which the .gch then satisfies.
+    flags = []
+    include_headers = []
+    raw_args = [
+        _shell_arg(element)
+        for element in projenv.subst_list("$CXXFLAGS $CCFLAGS $_CCCOMCOM")[0]  # noqa: F821
+    ]
+    if any(arg is None for arg in raw_args):
+        print("ESPHome: skipping precompiled header: unmodelable flag")
+        return
+    flag_it = iter(raw_args)
+    for tok in flag_it:
+        if tok == "-include":
+            include_headers.append(next(flag_it, ""))
+        elif tok.startswith("-include") and not tok.startswith("-include-"):
+            include_headers.append(tok[len("-include") :])
+        else:
+            flags.append(tok)
+    if any(not name for name in include_headers):
+        print("ESPHome: build_src_flags has a trailing -include; skipping pch")
+        return
+    # Fold only relative names resolving under src/: consumers keep their
+    # own -include entries, so folding an unguarded user header would
+    # include it twice; unfolded ones stay consumer-only.
+    try:
+        folded = [
+            name
+            for name in include_headers
+            if not Path(name).is_absolute() and _resolves(src_dir / name)
+        ]
+    except OSError as err:
+        print(f"ESPHome: skipping precompiled header: {err}")
+        return
+    if unfolded := [n for n in include_headers if n not in folded]:
+        print(f"ESPHome: not precompiling non-src force-includes: {unfolded}")
+    content = "".join(f'#include "{name}"\n' for name in (*folded, _CORE_HEADER))
+
+    digest = hashlib.sha256()
+    digest.update(content.encode(errors="surrogateescape"))
+    digest.update(cxx.encode(errors="surrogateescape"))
+    # Mirror CCACHE_BASEDIR: strip the per-device build path so identical
+    # configs produce identical .sum files and share cache entries
+    flags_id = " ".join(flags)
+    if basedir := os.environ.get("CCACHE_BASEDIR"):
+        flags_id = flags_id.replace(basedir, "")
+    digest.update(flags_id.encode(errors="surrogateescape"))
+    # GCC never validates a .gch against its source headers, and PlatformIO
+    # package paths carry no version, so a package bump must invalidate here
+    platform = env.PioPlatform()  # noqa: F821
+    for package in sorted(platform.packages):
+        try:
+            version = platform.get_package_version(package)
+        except KeyError:
+            # An unresolved manifest must not hash as a constant
+            if platform.get_package(package) is not None:
+                print(f"ESPHome: skipping precompiled header: no version for {package}")
+                return
+            version = None  # absent optional package
+        except Exception as err:  # noqa: BLE001
+            # No trustworthy package identity: a stale .gch could survive
+            print(f"ESPHome: skipping precompiled header: {err}")
+            return
+        digest.update(f"{package}={version}".encode())
+        digest.update(b"\0")
+    try:
+        closure = _include_closure(src_dir, [*folded, _CORE_HEADER])
+    except OSError as err:
+        print(f"ESPHome: skipping precompiled header: {err}")
+        return
+    for rel in sorted(closure):
+        digest.update(rel.encode(errors="surrogateescape"))
+        digest.update(closure[rel])
+        digest.update(b"\0")
+    # Project-local -I dirs (e.g. rp2's lwip_override) hold generated
+    # headers the src closure cannot see; hash them too
+    prev = ""
+    try:
+        for tok in flags:
+            inc = tok[2:] if tok.startswith("-I") and len(tok) > 2 else ""
+            if prev == "-I":
+                inc = tok
+            prev = tok
+            if not inc:
+                continue
+            inc_dir = Path(inc)
+            if not (
+                _resolves_dir(inc_dir)
+                and inc_dir.is_relative_to(proj_dir)
+                and not inc_dir.is_relative_to(src_dir)
+                # Library trees never enter the prefix closure; walking them
+                # would read every library file each build
+                and not inc_dir.is_relative_to(proj_dir / ".piolibdeps")
+                and not inc_dir.is_relative_to(proj_dir / ".pioenvs")
+            ):
+                continue
+            local_headers = []
+            # os.walk with onerror: rglob would swallow unlistable subtrees
+            for root, _dirs, files in os.walk(inc_dir, onerror=_raise_walk_error):
+                local_headers.extend(
+                    Path(root) / f
+                    for f in files
+                    if f.endswith((".h", ".hpp", ".hh", ".inc"))
+                )
+            for local in sorted(local_headers):
+                digest.update(str(local.relative_to(proj_dir)).encode())
+                digest.update(local.read_bytes())
+                digest.update(b"\0")
+    except OSError as err:
+        print(f"ESPHome: skipping precompiled header: {err}")
+        return
+    checksum = digest.hexdigest()
+
+    # The ccache .sum sidecar doubles as the freshness stamp
+    if (
+        not header.is_file()
+        or not gch.is_file()
+        or not sum_path.is_file()
+        or (_read_stamp(sum_path) != checksum)
+    ):
+        failed_marker = Path(f"{gch}.failed")
+        if _read_stamp(failed_marker) == checksum:
+            print(
+                "ESPHome: skipping precompiled header (previous attempt "
+                f"failed); delete {failed_marker.name} to retry"
+            )
+            return
+        header.write_text(content, encoding="utf-8")
+        try:
+            error = _compile_gch(cxx, flags, header, gch, proj_dir)
+        except OSError as err:
+            # Transient spawn/IO failure: no marker, retry next build
+            print(f"ESPHome: precompiled header compile did not run: {err}")
+            gch.unlink(missing_ok=True)
+            sum_path.unlink(missing_ok=True)
+            return
+        if error is not None:
+            print("ESPHome: precompiled header failed; compiling without it")
+            print(error)
+            gch.unlink(missing_ok=True)
+            sum_path.unlink(missing_ok=True)
+            if any(m in error for m in _TRANSIENT_ERRORS):
+                # Resource exhaustion clears on its own; retry next build
+                return
+            # Skip retries until a flag/header/platform change alters the checksum
+            failed_marker.write_text(checksum + "\n", encoding="utf-8")
+            return
+        failed_marker.unlink(missing_ok=True)
+        sum_path.write_text(checksum + "\n", encoding="utf-8")
+
+    # Computed first so the flags and env land together: a raise between
+    # them would leave a pch-consuming build without its ccache settings.
+    # projenv["ENV"] aliases os.environ, so these reach all TUs; only
+    # time_macros affects non-pch TUs. User values win.
+    ccache_updates = {
+        key: value
+        for key, value in (
+            ("CCACHE_SLOPPINESS", "pch_defines,time_macros"),
+            ("CCACHE_PCH_EXTSUM", "true"),
+        )
+        if key not in os.environ
+    }
+    sloppiness = os.environ.get("CCACHE_SLOPPINESS")
+    if sloppiness is not None:
+        tokens = {tok.strip() for tok in sloppiness.split(",")}
+        missing = [t for t in ("pch_defines", "time_macros") if t not in tokens]
+        if missing:
+            # Without these ccache declines every pch-consuming compile
+            ccache_updates["CCACHE_SLOPPINESS"] = ",".join((sloppiness, *missing))
+            print(f"ESPHome: adding {','.join(missing)} to CCACHE_SLOPPINESS for the pch")
+    extsum = os.environ.get("CCACHE_PCH_EXTSUM")
+    if extsum is not None and extsum.strip().lower() not in ("1", "true", "yes", "on"):
+        # ccache then hashes the non-reproducible .gch bytes: permanent misses
+        print(f"ESPHome: CCACHE_PCH_EXTSUM={extsum} disables pch caching")
+    # Prepended: GCC only uses a .gch while no other tokens precede it.
+    # The relative name also reaches "pio run -t idedata" output.
+    # -Wno-error: the per-process probe can pass while a later cc1plus
+    # rejects the .gch; that must stay a warning under user -Werror.
+    projenv.Prepend(  # noqa: F821
+        CXXFLAGS=["-Winvalid-pch", "-Wno-error=invalid-pch", "-include", header.name]
+    )
+    projenv["ENV"].update(ccache_updates)  # noqa: F821
+    print("ESPHome: Compiling with precompiled header")
+
+
+try:
+    _setup_pch()
+except Exception:  # noqa: BLE001 -- a speedup must never break the build
+    print("ESPHome: pch internal error; compiling without it")
+    traceback.print_exc()

esphome/platformio/toolchain.py

@@ -289,6 +289,15 @@ def copy_ccache_script() -> None:
     )
 
 
+def copy_pch_script() -> None:
+    """Copy the shared precompiled-header SCons post-script into the build
+    dir; platform components pair it with ``post:pch.py`` in extra_scripts."""
+    copy_file_if_changed(
+        Path(__file__).parent / "pch.py.script",
+        CORE.relative_build_path("pch.py"),
+    )
+
+
 def default_libdeps_dir() -> str:
     """The PLATFORMIO_LIBDEPS_DIR value a pio run defaults to; the package
     prefetch must resolve installed libraries against the same dir."""

esphome/writer.py

@@ -7,6 +7,7 @@ import re
 import time
 
 from esphome import loader
+from esphome.build_helpers.pch import PCH_ARTIFACT_NAMES
 from esphome.compiled_config import save_compiled_config
 from esphome.config import iter_component_configs, iter_components
 from esphome.const import (
@@ -609,11 +610,20 @@ def clean_build(clear_pio_cache: bool = True, *, full: bool = False):
             if idf_path.is_dir():
                 _LOGGER.info("Deleting %s", idf_path)
                 rmtree(idf_path)
+        # The PlatformIO pch artifacts live at the project root so the
+        # relative -include resolves; a partial clean must drop them too
+        for name in PCH_ARTIFACT_NAMES:
+            CORE.relative_build_path(name).unlink(missing_ok=True)
 
-    # The idedata cache is derived from the build but lives under the data dir,
-    # not the build path, so it must be removed separately in both modes.
-    idedata_cache = CORE.relative_internal_path("idedata", f"{CORE.name}.json")
-    if idedata_cache.is_file():
+    # The idedata caches are derived from the build but live under the data
+    # dir, not the build path, so they must be removed separately in both
+    # modes. Globbed (name.json plus name.<backend>.json) so a future
+    # backend suffix cannot silently drift out of clean-all.
+    idedata_dir = CORE.relative_internal_path("idedata")
+    for idedata_cache in (
+        *idedata_dir.glob(f"{CORE.name}.json"),
+        *idedata_dir.glob(f"{CORE.name}.*.json"),
+    ):
         _LOGGER.info("Deleting %s", idedata_cache)
         idedata_cache.unlink()
 

script/determine-jobs.py

@@ -50,6 +50,7 @@ from __future__ import annotations
 
 import argparse
 from collections import Counter
+from collections.abc import Callable
 from enum import StrEnum
 from functools import cache
 import json
@@ -531,37 +532,53 @@ ESP32_PLATFORMIO_TRIGGER_PATH_PREFIXES = ("esphome/platformio/",)
 #   - esphome/build_gen/platformio.py -- the PlatformIO build generator
 #   - script/test_build_components.py -- the harness the job invokes
 #   - .github/workflows/ci.yml -- the job's own definition
-ESP32_PLATFORMIO_TRIGGER_FILES = frozenset(
+# Shared by every toolchain smoke-test job: the harness it invokes and the
+# workflow that defines it
+_SMOKE_HARNESS_TRIGGER_FILES = frozenset(
     {
-        "esphome/build_gen/platformio.py",
         "script/test_build_components.py",
         ".github/workflows/ci.yml",
     }
 )
 
+ESP32_PLATFORMIO_TRIGGER_FILES = _SMOKE_HARNESS_TRIGGER_FILES | {
+    "esphome/build_gen/platformio.py",
+}
+
+
+def _path_or_file_trigger(
+    files: list[str],
+    trigger_files: frozenset[str],
+    trigger_prefixes: tuple[str, ...],
+) -> bool:
+    """Whether any changed file matches the given infrastructure triggers."""
+    return any(
+        file in trigger_files or file.startswith(trigger_prefixes) for file in files
+    )
+
+
+@cache
+def _cached_components_closure(files: tuple[str, ...]) -> frozenset[str]:
+    """Dependency closure of the changed components, from the changed files.
+
+    The walk is expensive and every toolchain smoke-test job asks for the
+    same file list, so compute it once per run."""
+    component_files = [f for f in files if filter_component_and_test_files(f)]
+    return frozenset(get_components_with_dependencies(component_files, True))
+
 
 def _esp32_platformio_path_or_file_trigger(files: list[str]) -> bool:
     """Whether any changed file is a PlatformIO infrastructure / harness trigger."""
-    for file in files:
-        if file in ESP32_PLATFORMIO_TRIGGER_FILES:
-            return True
-        if any(
-            file.startswith(prefix) for prefix in ESP32_PLATFORMIO_TRIGGER_PATH_PREFIXES
-        ):
-            return True
-    return False
+    return _path_or_file_trigger(
+        files, ESP32_PLATFORMIO_TRIGGER_FILES, ESP32_PLATFORMIO_TRIGGER_PATH_PREFIXES
+    )
 
 
 def _esp_idf_infra_changed(files: list[str]) -> bool:
     """Whether any changed file is ESP-IDF build/runner infrastructure."""
-    for file in files:
-        if file in ESP_IDF_INFRA_TRIGGER_FILES:
-            return True
-        if any(
-            file.startswith(prefix) for prefix in ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES
-        ):
-            return True
-    return False
+    return _path_or_file_trigger(
+        files, ESP_IDF_INFRA_TRIGGER_FILES, ESP_IDF_INFRA_TRIGGER_PATH_PREFIXES
+    )
 
 
 def esp32_platformio_components_to_test(branch: str | None = None) -> list[str]:
@@ -599,15 +616,23 @@ def esp32_platformio_components_to_test(branch: str | None = None) -> list[str]:
     Returns:
         Sorted list of component names to compile.
     """
+    return _toolchain_components_to_test(
+        branch, ESP32_PLATFORMIO_TEST_COMPONENTS, _esp32_platformio_path_or_file_trigger
+    )
+
+
+def _toolchain_components_to_test(
+    branch: str | None,
+    test_set: frozenset[str],
+    infra_trigger: Callable[[list[str]], bool],
+) -> list[str]:
+    """The shared narrowing rule for the per-toolchain smoke-test jobs."""
     files = changed_files(branch)
 
-    if core_changed(files) or _esp32_platformio_path_or_file_trigger(files):
-        return sorted(ESP32_PLATFORMIO_TEST_COMPONENTS)
+    if core_changed(files) or infra_trigger(files):
+        return sorted(test_set)
 
-    component_files = [f for f in files if filter_component_and_test_files(f)]
-    changed = get_components_with_dependencies(component_files, True)
-
-    return sorted(ESP32_PLATFORMIO_TEST_COMPONENTS & set(changed))
+    return sorted(test_set & _cached_components_closure(tuple(files)))
 
 
 def should_run_esp32_platformio(branch: str | None = None) -> bool:
@@ -628,6 +653,83 @@ def should_run_esp32_platformio(branch: str | None = None) -> bool:
     return bool(esp32_platformio_components_to_test(branch))
 
 
+# The `--toolchain arduino` smoke-test set: covers the core, the bundled and
+# converted registry libraries, and the waveform path.
+ESP8266_NATIVE_TEST_COMPONENTS = frozenset(
+    {
+        "esp8266",
+        "api",
+        "web_server",
+        "captive_portal",
+        "mqtt",
+        "esp8266_pwm",
+        "neopixelbus",
+        "bme280_i2c",
+        "uart",
+    }
+)
+
+# Infrastructure whose changes always trigger the native ESP8266 compile
+# test. esphome/build_helpers/ holds the idedata and size-summary helpers
+# the backend shares with the native ESP-IDF build.
+ESP8266_NATIVE_TRIGGER_PATH_PREFIXES = (
+    "esphome/arduino8266/",
+    "esphome/arduino/",
+    "esphome/build_helpers/",
+)
+# Shared library-conversion modules every native build imports; espidf-only
+# infra (build_gen/espidf.py) deliberately stays out of the esp8266 set.
+_NATIVE_SHARED_TRIGGER_FILES = frozenset(
+    {
+        "esphome/framework_helpers.py",
+        "esphome/platformio/library.py",
+        "esphome/platformio/extra_script.py",
+    }
+)
+# Tripwire: the shared modules must stay in the ESP-IDF trigger set too
+# (now defined in clang_tidy_hash), or its smoke test silently skips them
+assert _NATIVE_SHARED_TRIGGER_FILES <= ESP_IDF_INFRA_TRIGGER_FILES
+ESP8266_NATIVE_TRIGGER_FILES = (
+    _NATIVE_SHARED_TRIGGER_FILES
+    | _SMOKE_HARNESS_TRIGGER_FILES
+    | {
+        "esphome/build_gen/arduino8266.py",
+        "esphome/build_gen/build_tool.py",
+        "esphome/components/esp8266/build_surgery.py",
+        "esphome/components/esp8266/boards.py",
+        "esphome/platformio/registry.py",
+        # esp8266/__init__.py imports copy_ccache_script from it
+        "esphome/platformio/toolchain.py",
+        ".github/actions/cache-arduino8266/action.yml",
+    }
+)
+
+
+def _esp8266_native_path_or_file_trigger(files: list[str]) -> bool:
+    """Whether any changed file is native-ESP8266 infrastructure / harness."""
+    # base_python_changed covers the top-level esphome/*.py modules the
+    # native backend imports directly (framework_helpers, helpers, writer,
+    # __main__); without it a change there would silently skip this job.
+    # base_python_changed is deliberately broad (any top-level esphome/*.py)
+    # as belt-and-braces while the backend is new; narrow it to the modules
+    # the backend imports once the toolchain has soaked a few releases
+    return base_python_changed(files) or _path_or_file_trigger(
+        files, ESP8266_NATIVE_TRIGGER_FILES, ESP8266_NATIVE_TRIGGER_PATH_PREFIXES
+    )
+
+
+def esp8266_native_components_to_test(branch: str | None = None) -> list[str]:
+    """Subset of ``ESP8266_NATIVE_TEST_COMPONENTS`` the job needs to compile.
+
+    Same narrowing logic as ``esp32_platformio_components_to_test``: the full
+    list on core or infrastructure changes, otherwise the intersection with
+    the changed-component dependency closure (empty list skips the job).
+    """
+    return _toolchain_components_to_test(
+        branch, ESP8266_NATIVE_TEST_COMPONENTS, _esp8266_native_path_or_file_trigger
+    )
+
+
 def determine_cpp_unit_tests(
     branch: str | None = None,
 ) -> tuple[bool, list[str]]:
@@ -1226,6 +1328,8 @@ def main() -> None:
         run_device_builder = True
         esp32_platformio_components = sorted(ESP32_PLATFORMIO_TEST_COMPONENTS)
         run_esp32_platformio = True
+        esp8266_native_components = sorted(ESP8266_NATIVE_TEST_COMPONENTS)
+        run_esp8266_native = True
     else:
         integration_run_all, integration_test_files = determine_integration_tests(
             args.branch
@@ -1237,6 +1341,8 @@ def main() -> None:
         run_device_builder = should_run_device_builder(args.branch)
         esp32_platformio_components = esp32_platformio_components_to_test(args.branch)
         run_esp32_platformio = bool(esp32_platformio_components)
+        esp8266_native_components = esp8266_native_components_to_test(args.branch)
+        run_esp8266_native = bool(esp8266_native_components)
     run_integration, integration_test_buckets = _compute_integration_test_buckets(
         integration_run_all, integration_test_files
     )
@@ -1432,6 +1538,8 @@ def main() -> None:
         "device_builder": run_device_builder,
         "esp32_platformio": run_esp32_platformio,
         "esp32_platformio_components": ",".join(esp32_platformio_components),
+        "esp8266_native": run_esp8266_native,
+        "esp8266_native_components": ",".join(esp8266_native_components),
         "changed_components": changed_components,
         "changed_components_with_tests": changed_components_with_tests,
         "directly_changed_components_with_tests": list(directly_changed_with_tests),

script/test_build_components.py

@@ -1027,6 +1027,7 @@ def test_components(
     isolated_components: set[str] | None = None,
     base_only: bool = False,
     toolchain: str | None = None,
+    fail_on_no_tests: bool = False,
 ) -> int:
     """Test components with optional intelligent grouping.
 
@@ -1061,20 +1062,34 @@ def test_components(
     # toolchain build.
     include_validate = esphome_command != "compile"
 
-    # Find all component tests
+    # Find all component tests; remember which components each pattern
+    # (wildcards included) matched, for the deferred no-tests accounting
     all_tests = {}
+    pattern_components: dict[str, set[str]] = {}
     for pattern in component_patterns:
         # Skip empty patterns (happens when components list is empty string)
         if not pattern:
             continue
-        all_tests.update(
-            find_component_tests(
-                tests_dir, pattern, base_only, include_validate=include_validate
-            )
+        found = find_component_tests(
+            tests_dir, pattern, base_only, include_validate=include_validate
         )
+        pattern_components[pattern] = set(found)
+        all_tests.update(found)
+
+    # The flag's contract is "no test matched fails": a fully blank pattern
+    # list would otherwise slide into the reference-baseline fallback and
+    # exit green while building nothing a caller asked for
+    if fail_on_no_tests and not any(component_patterns):
+        print("No components requested (blank component list)")
+        return 1
+
+    if fail_on_no_tests and not all_tests:
+        # Nothing matched at all: fail before the synthetic baseline build,
+        # which would spend a compile reporting success on nothing. Partial
+        # matches defer to the per-pattern accounting after the summary.
+        print(f"No components found matching: {component_patterns}")
+        return 1
 
-    # If no components found, build a reference configuration for baseline comparison
-    # Create a synthetic "empty" component test that will build just the base config
     if not all_tests:
         print(f"No components found matching: {component_patterns}")
         print(
@@ -1178,6 +1193,26 @@ def test_components(
                         toolchain=toolchain,
                     )
 
+    silent: list[str] = []
+    if fail_on_no_tests:
+        # A green run that built nothing for a requested pattern (renamed
+        # fixture, missing base file, version-suffix mismatch, a wildcard
+        # matching no component) must not pass CI. Per pattern: an
+        # all-or-nothing check would let one silent pattern hide behind the
+        # others. Opt-in: some legs (the esp32-ard smoke subset)
+        # legitimately match nothing. Failing is deferred past the summary
+        # so a real failure's reproduce commands still print.
+        built = {c for r in test_results for c in r.components}
+        # A pattern is silent when it matched no fixture, or when none of
+        # its matched components produced a build (wildcards included)
+        silent = [
+            p
+            for p in component_patterns
+            if p and not (pattern_components.get(p, set()) & built)
+        ]
+        if silent:
+            print(f"No tests ran for requested pattern(s): {', '.join(silent)}")
+
     # Separate results into passed and failed
     passed_results = [r for r in test_results if r.success]
     failed_results = [r for r in test_results if not r.success]
@@ -1209,7 +1244,7 @@ def test_components(
     if os.environ.get("GITHUB_STEP_SUMMARY"):
         write_github_summary(test_results, toolchain=toolchain)
 
-    if failed_results:
+    if failed_results or silent:
         return 1
 
     return 0
@@ -1264,6 +1299,12 @@ def main() -> int:
         "--toolchain",
         help="Select toolchain for compiling.",
     )
+    parser.add_argument(
+        "--fail-on-no-tests",
+        action="store_true",
+        help="Exit non-zero when no test matched (for CI legs whose "
+        "components must all have fixtures)",
+    )
 
     args = parser.parse_args()
 
@@ -1282,6 +1323,7 @@ def main() -> int:
         continue_on_fail=args.continue_on_fail,
         enable_grouping=not args.no_grouping,
         isolated_components=isolated_components,
+        fail_on_no_tests=args.fail_on_no_tests,
         base_only=args.base_only,
         toolchain=args.toolchain,
     )

tests/script/test_determine_jobs.py

@@ -78,6 +78,17 @@ def mock_esp32_platformio_components_to_test() -> Generator[Mock, None, None]:
         yield mock
 
 
+@pytest.fixture
+def mock_esp8266_native_components_to_test() -> Generator[Mock, None, None]:
+    """Mock esp8266_native_components_to_test from determine_jobs.
+
+    main() drives both the ``esp8266_native`` boolean output and the
+    ``esp8266_native_components`` CSV from this one function.
+    """
+    with patch.object(determine_jobs, "esp8266_native_components_to_test") as mock:
+        yield mock
+
+
 @pytest.fixture
 def mock_determine_cpp_unit_tests() -> Generator[Mock, None, None]:
     """Mock determine_cpp_unit_tests from helpers."""
@@ -106,6 +117,7 @@ def clear_determine_jobs_caches() -> None:
     """Clear all cached functions before each test."""
     determine_jobs._is_clang_tidy_full_scan.cache_clear()
     determine_jobs._component_has_tests.cache_clear()
+    determine_jobs._cached_components_closure.cache_clear()
 
 
 def test_main_all_tests_should_run(
@@ -116,6 +128,7 @@ def test_main_all_tests_should_run(
     mock_should_run_import_time: Mock,
     mock_should_run_device_builder: Mock,
     mock_esp32_platformio_components_to_test: Mock,
+    mock_esp8266_native_components_to_test: Mock,
     mock_changed_files: Mock,
     mock_determine_cpp_unit_tests: Mock,
     capsys: pytest.CaptureFixture[str],
@@ -132,6 +145,7 @@ def test_main_all_tests_should_run(
     mock_should_run_import_time.return_value = True
     mock_should_run_device_builder.return_value = True
     mock_esp32_platformio_components_to_test.return_value = ["api", "esp32"]
+    mock_esp8266_native_components_to_test.return_value = ["api", "logger"]
     mock_determine_cpp_unit_tests.return_value = (False, ["wifi", "api", "sensor"])
 
     # Mock changed_files to return non-component files (to avoid memory impact)
@@ -208,6 +222,8 @@ def test_main_all_tests_should_run(
     assert output["device_builder"] is True
     assert output["esp32_platformio"] is True
     assert output["esp32_platformio_components"] == "api,esp32"
+    assert output["esp8266_native"] is True
+    assert output["esp8266_native_components"] == "api,logger"
     assert output["changed_components"] == ["wifi", "api", "sensor"]
     # changed_components_with_tests will only include components that actually have test files
     assert "changed_components_with_tests" in output
@@ -244,6 +260,7 @@ def test_main_no_tests_should_run(
     mock_should_run_import_time: Mock,
     mock_should_run_device_builder: Mock,
     mock_esp32_platformio_components_to_test: Mock,
+    mock_esp8266_native_components_to_test: Mock,
     mock_changed_files: Mock,
     mock_determine_cpp_unit_tests: Mock,
     capsys: pytest.CaptureFixture[str],
@@ -260,6 +277,7 @@ def test_main_no_tests_should_run(
     mock_should_run_import_time.return_value = False
     mock_should_run_device_builder.return_value = False
     mock_esp32_platformio_components_to_test.return_value = []
+    mock_esp8266_native_components_to_test.return_value = []
     mock_determine_cpp_unit_tests.return_value = (False, [])
 
     # Mock changed_files to return no component files
@@ -302,6 +320,8 @@ def test_main_no_tests_should_run(
     assert output["device_builder"] is False
     assert output["esp32_platformio"] is False
     assert output["esp32_platformio_components"] == ""
+    assert output["esp8266_native"] is False
+    assert output["esp8266_native_components"] == ""
     assert output["changed_components"] == []
     assert output["changed_components_with_tests"] == []
     assert output["component_test_count"] == 0
@@ -3151,6 +3171,78 @@ def test_memory_impact_elf_layouts_are_found(tmp_path: Path) -> None:
         assert find_elf_path(build_path) == elf, f"{platform} ELF not found"
 
 
+@pytest.mark.parametrize(
+    "changed",
+    [
+        "esphome/arduino8266/framework.py",
+        "esphome/build_gen/arduino8266.py",
+        "esphome/components/esp8266/build_surgery.py",
+        # Shared modules the native build depends on
+        "esphome/build_helpers/idedata.py",
+        "esphome/platformio/library.py",
+        # Top-level esphome/*.py modules the backend imports directly
+        "esphome/framework_helpers.py",
+        "esphome/writer.py",
+        # esp8266/__init__.py imports copy_ccache_script from it
+        "esphome/platformio/toolchain.py",
+        # The composite cache action must not ship unexercised
+        ".github/actions/cache-arduino8266/action.yml",
+    ],
+)
+def test_esp8266_native_components_full_list_on_infra_change(changed: str) -> None:
+    """Native-ESP8266 infrastructure changes run the full test list."""
+    with (
+        patch.object(determine_jobs, "changed_files", return_value=[changed]),
+        patch.object(
+            determine_jobs,
+            "get_components_with_dependencies",
+            return_value=["wifi"],
+        ),
+    ):
+        result = determine_jobs.esp8266_native_components_to_test()
+        assert result == sorted(determine_jobs.ESP8266_NATIVE_TEST_COMPONENTS)
+
+
+@pytest.mark.parametrize(
+    ("changed_files", "dependency_closure", "expected"),
+    [
+        # Tested component changed -- narrow to the intersection.
+        (
+            ["esphome/components/mqtt/mqtt_client.cpp"],
+            ["mqtt", "json"],
+            ["mqtt"],
+        ),
+        # Components outside the test set return an empty list (job skipped).
+        (
+            ["esphome/components/wifi/wifi_component.cpp"],
+            ["wifi", "network"],
+            [],
+        ),
+        # espidf infrastructure is not an esp8266-native trigger; the
+        # native backend depends on esphome/build_helpers/ instead.
+        (["esphome/build_gen/espidf.py"], [], []),
+        (["esphome/espidf/toolchain.py"], [], []),
+        (["README.md"], [], []),
+    ],
+)
+def test_esp8266_native_components_to_test_narrowing(
+    changed_files: list[str],
+    dependency_closure: list[str],
+    expected: list[str],
+) -> None:
+    """Component changes narrow the native-ESP8266 test list."""
+    with (
+        patch.object(determine_jobs, "changed_files", return_value=changed_files),
+        patch.object(
+            determine_jobs,
+            "get_components_with_dependencies",
+            return_value=dependency_closure,
+        ),
+    ):
+        result = determine_jobs.esp8266_native_components_to_test()
+        assert result == expected
+
+
 def test_compute_integration_test_buckets_no_durations_full_fanout() -> None:
     """Without recorded durations the fan-out stays at the maximum."""
     files = [f"tests/integration/test_{i:03d}.py" for i in range(15)]

tests/script/test_test_build_components.py

@@ -236,3 +236,98 @@ def test_run_grouped_test_closes_group_when_subprocess_raises(
         )
 
     assert "::endgroup::" in capsys.readouterr().out
+
+
+def test_components_empty_match_fails_with_flag(
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """Under --fail-on-no-tests, a real component filtered to a platform
+    with no matching test file must not pass CI as a green zero-component
+    compile."""
+    rc = tbc.test_components(
+        ["logger"],
+        "zz-none",
+        "compile",
+        False,
+        enable_grouping=False,
+        fail_on_no_tests=True,
+    )
+    assert rc == 1
+    assert "No tests ran for requested pattern(s): logger" in (capsys.readouterr().out)
+
+
+def test_components_component_with_no_base_file_fails_with_flag(
+    capsys: pytest.CaptureFixture[str],
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """A component whose fixture matches the platform but whose platform has
+    no base file builds nothing; under the flag that silent zero fails by
+    component name instead of hiding behind other components."""
+    monkeypatch.setattr(tbc, "get_platform_base_files", lambda base_dir: {})
+    rc = tbc.test_components(
+        ["logger"],
+        "esp8266-ard",
+        "compile",
+        False,
+        enable_grouping=False,
+        fail_on_no_tests=True,
+    )
+    assert rc == 1
+    assert "No tests ran for requested pattern(s): logger" in (capsys.readouterr().out)
+
+
+def test_components_blank_list_fails_with_flag(
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """A fully blank component list must not slide into the baseline
+    fallback and exit green under the flag."""
+    rc = tbc.test_components(
+        [""], "esp8266-ard", "compile", False, fail_on_no_tests=True
+    )
+    assert rc == 1
+    assert "blank component list" in capsys.readouterr().out
+
+
+def test_components_wildcard_no_match_fails_with_flag(
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """A wildcard matching nothing must not degrade to the synthetic
+    baseline build and exit green under the flag."""
+    rc = tbc.test_components(
+        ["zz_no_such*"],
+        "esp8266-ard",
+        "compile",
+        False,
+        enable_grouping=False,
+        fail_on_no_tests=True,
+    )
+    assert rc == 1
+    assert "No components found matching" in capsys.readouterr().out
+
+
+def test_components_empty_match_tolerated_without_flag() -> None:
+    """The esp32-ard smoke leg deliberately builds only the subset with a
+    matching fixture; without the flag an empty match stays green."""
+    assert (
+        tbc.test_components(
+            ["logger"], "zz-none", "compile", False, enable_grouping=False
+        )
+        == 0
+    )
+
+
+def test_components_unknown_component_fails_with_flag(
+    capsys: pytest.CaptureFixture[str],
+) -> None:
+    """A renamed smoke-test component must shrink coverage loudly, not fall
+    into the reference-baseline build."""
+    rc = tbc.test_components(
+        ["no_such_component_xyz"],
+        "esp8266-ard",
+        "compile",
+        False,
+        enable_grouping=False,
+        fail_on_no_tests=True,
+    )
+    assert rc == 1
+    assert "No components found matching" in capsys.readouterr().out

tests/unit_tests/build_gen/test_arduino8266.py

@@ -1,38 +1,34 @@
-"""Drift tests for the native ESP8266 Arduino build spec.
+"""Drift tests for the native ESP8266 Arduino build generator.
 
-These pin the ESPHome side of the transliteration (the knob-define
-precedence, the define/flag sets, and the linker-script generation) against
-literals audited from the PlatformIO builder
-(framework-arduinoespressif8266/tools/platformio-build.py and
-platform-espressif8266/builder/main.py). They catch an accidental edit on
-this side; an upstream change in a new framework release is caught by the
-A/B byte-identical build check on a version bump, not by these tests.
+Pin the transliterated flag/define/link sets against literals audited from
+the PlatformIO builder. Upstream drift is caught by the A/B build check on
+version bumps, not here.
 """
 
 from __future__ import annotations
 
 from collections.abc import Generator
+import logging
 import os
 from pathlib import Path
+import shutil
 from unittest.mock import MagicMock, patch
 
 import pytest
 
+from esphome.arduino.library import ArduinoLibrary
 from esphome.arduino8266.framework import InstalledPaths, toolchain_tool
 from esphome.build_gen import arduino8266
 from esphome.build_gen.arduino8266 import (
     _defines_flags,
     _flag_defines,
+    _flash_size_str,
     _resolve_build_config,
+    get_flash_ld_path,
 )
-from esphome.components.esp8266.boards import ESP8266_BOARD_BUILD
+from esphome.components.esp8266.boards import BOARDS, ESP8266_BOARD_BUILD
 from esphome.components.esp8266.build_surgery import RATETABLE_RULE
-from esphome.components.esp8266.const import (
-    KEY_BOARD,
-    KEY_ESP8266,
-    KEY_FLASH_MODE,
-    KEY_SCANF_FLOAT,
-)
+from esphome.components.esp8266.const import KEY_BOARD, KEY_ESP8266, KEY_SCANF_FLOAT
 import esphome.config_validation as cv
 from esphome.const import KEY_CORE, KEY_FRAMEWORK_VERSION
 from esphome.core import CORE, EsphomeError
@@ -47,9 +43,13 @@ def _setup_core(tmp_path: Path) -> Generator[None]:
     CORE.data[KEY_CORE] = {KEY_FRAMEWORK_VERSION: cv.Version(3, 1, 2)}
     CORE.data[KEY_ESP8266] = {
         KEY_BOARD: "nodemcuv2",
-        KEY_FLASH_MODE: "dout",
         KEY_SCANF_FLOAT: False,
     }
+    # The producers esp8266/__init__ pins unconditionally
+    CORE.platformio_options = {
+        "board_build.flash_mode": "dout",
+        "build_src_flags": "-include esphome/components/esp8266/throw_stubs.h",
+    }
     yield
     # CORE.reset() (the suite-wide autouse fixture) does not clear this flag
     CORE.testing_mode = False
@@ -59,6 +59,11 @@ def _set_flags(*flags: str) -> None:
     CORE.build_flags = set(flags)
 
 
+def _shq(tok: str) -> str:
+    """The platform's shell_token quote wrapper (argv rule on Windows)."""
+    return f'"{tok}"' if os.name == "nt" else f"'{tok}'"
+
+
 def _resolve(*flags: str) -> arduino8266._BuildConfig:
     """Set the build flags and resolve the knob config in one step."""
     _set_flags(*flags)
@@ -206,12 +211,284 @@ def _make_framework(tmp_path: Path) -> InstalledPaths:
     for sub in ("include", "ld", "lwip2/include", "lib"):
         (framework / "tools" / "sdk" / sub).mkdir(parents=True)
     (framework / "libraries").mkdir()
+    (framework / "tools" / "elf2bin.py").write_text("")
+    eboot = framework / "bootloaders" / "eboot"
+    eboot.mkdir(parents=True)
+    (eboot / "eboot.elf").write_text("")
     toolchain = tmp_path / "toolchain"
     (toolchain / "bin").mkdir(parents=True)
     (toolchain / "include").mkdir()
     return InstalledPaths(framework=framework, toolchain=toolchain, ninja=Path("ninja"))
 
 
+def _write_ninja(
+    paths: InstalledPaths,
+    libraries: list | None = None,
+    ccache: str | None = None,
+) -> str:
+    src = CORE.relative_src_path()
+    (src / "esphome" / "components" / "esp8266").mkdir(parents=True, exist_ok=True)
+    (src / "main.cpp").write_text("")
+    (src / "esphome" / "vendor.c").write_text("")
+
+    with (
+        patch.object(arduino8266, "generate_ld_scripts"),
+        patch(
+            "esphome.arduino.library.resolve_libraries",
+            return_value=libraries or [],
+        ),
+    ):
+        arduino8266.write_project(paths, ccache)
+    return (CORE.relative_pioenvs_path(CORE.name) / "build.ninja").read_text()
+
+
+def test_write_project_rejects_bad_flash_mode(tmp_path: Path) -> None:
+    """A flash mode outside the closed set fails by name before landing
+    unquoted in the elf2bin command."""
+    paths = _make_framework(tmp_path)
+    CORE.platformio_options["board_build.flash_mode"] = "dout; rm -rf /"
+    with pytest.raises(EsphomeError, match="Invalid flash mode"):
+        _write_ninja(paths)
+
+
+def test_write_project_flash_mode_reaches_define_and_elf2bin(
+    tmp_path: Path,
+) -> None:
+    """A non-default board_flash_mode lands in both the compile define and
+    the elf2bin image header, never silently falling back to dout."""
+    paths = _make_framework(tmp_path)
+    CORE.platformio_options["board_build.flash_mode"] = "dio"
+    content = _write_ninja(paths)
+    assert "-DFLASHMODE_DIO" in content
+    assert "--flash_mode dio" in content
+    assert "FLASHMODE_DOUT" not in content
+
+
+def test_write_project_trailing_include_raises(tmp_path: Path) -> None:
+    """A dangling -include must fail by name, not become -include <src_dir>."""
+    paths = _make_framework(tmp_path)
+    CORE.platformio_options["build_src_flags"] = "-include"
+    with pytest.raises(EsphomeError, match="trailing '-include'"):
+        _write_ninja(paths)
+
+
+def test_write_project_passes_other_src_flags_through(tmp_path: Path) -> None:
+    """Non-include build_src_flags tokens are requoted onto the src edges."""
+    paths = _make_framework(tmp_path)
+    CORE.platformio_options["build_src_flags"] = (
+        "-include esphome/components/esp8266/throw_stubs.h -DSRC_ONLY=1"
+    )
+    content = _write_ninja(paths)
+    assert "throw_stubs.h" in content
+    assert "-DSRC_ONLY=1" in content
+
+
+def test_write_project_link_line_and_exclusions(tmp_path: Path) -> None:
+    paths = _make_framework(tmp_path)
+    _set_flags(
+        "-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH",
+        "-DUSE_ESP8266_WAVEFORM_STUBS",
+        "-Wl,--wrap=millis",
+        "-Wl,--wrap=printf",
+        "-Wno-nonnull-compare",
+        "-L/opt/blobs",
+        "-luser_blob",
+        "-L /spc/blobs -l spaced_blob",
+    )
+    content = _write_ninja(paths)
+
+    # Base link flags from the PlatformIO builder
+    for flag in (
+        "-Wl,--no-check-sections",
+        "-Wl,-static",
+        "-Wl,--gc-sections",
+        "-Wl,-wrap,system_restart_local",
+        "-Wl,-wrap,spi_flash_read",
+        "-u app_entry",
+        "-u _printf_float",
+        "-u _DebugExceptionVector",
+        "-u _DoubleExceptionVector",
+        "-u _KernelExceptionVector",
+        "-u _NMIExceptionVector",
+        "-u _UserExceptionVector",
+    ):
+        assert flag in content
+    # ESPHome's link flags and the board linker script
+    assert "-Wl,--wrap=millis" in content
+    assert "-Wl,--wrap=printf" in content
+    assert "-T eagle.flash.4m.ld" in content
+    # scanf float disabled: the forced-link flag must not appear
+    assert "_scanf_float" not in content
+    # $in/$out must stay unquoted; ninja escapes its own path variables
+    assert "-c $in -o $out" in content
+    assert "--app $in --flash_mode" in content
+    assert '"$in"' not in content
+    assert '"$out"' not in content
+    # -L/-l from esphome build_flags reach the link line, not the compiles;
+    # spaced forms ("-L /path") are shell-lexed the way PlatformIO does.
+    # str(Path(...)) so the separator matches the host platform.
+    opt_blobs = str(Path("/opt/blobs"))
+    spc_blobs = str(Path("/spc/blobs"))
+    assert f"-L{_shq(opt_blobs)}" in content
+    assert "-luser_blob" in content
+    assert f"-L{_shq(spc_blobs)}" in content
+    assert "-lspaced_blob" in content
+    for line in content.splitlines():
+        if line.split(" = ")[0] in ("cflags", "cxxflags", "asflags"):
+            assert "user_blob" not in line
+            assert opt_blobs not in line
+            assert "spaced_blob" not in line
+            assert spc_blobs not in line
+    # System libraries with the selected lwIP variant, in the builder's order
+    assert (
+        "-lhal -lphy -lpp -lnet80211 -llwip2-1460 -lwpa -lcrypto -lmain -lwps "
+        "-lbearssl -lespnow -lsmartconfig -lairkiss -lwpa2 -lspaced_blob "
+        "-luser_blob "
+        "-lstdc++ -lm -lc -lgcc" in content
+    )
+    # Core exclusions: native OTA backend and waveform stubs
+    assert "Updater.cpp" not in content
+    assert "core_esp8266_waveform_pwm.cpp" not in content
+    assert "core_esp8266_waveform_phase.cpp" not in content
+    assert "core_esp8266_main.cpp.o" in content
+    # Assembly and C sources compile through their own rules
+    assert "cont.S.o: aspp" in content
+    assert "abi.c.o: c" in content
+    # throw_stubs is force-included for ESPHome sources only, via one shared
+    # srcflags variable rather than a copy of the flags line per edge
+    src_lines = [line for line in content.splitlines() if "obj/src/" in line]
+    assert any("main.cpp.o: cxx" in line for line in src_lines)
+    assert content.count("throw_stubs.h") == 1
+    assert "srcflags = -include" in content
+    flags_lines = [
+        line for line in content.splitlines() if line.startswith("  flags = ")
+    ]
+    assert flags_lines
+    # C++ src edges consume the precompiled header; C/assembly keep srcflags
+    assert set(flags_lines) == {"  flags = $srcflags", "  flags = $srccxxflags"}
+
+
+def test_write_project_pch(tmp_path: Path) -> None:
+    paths = _make_framework(tmp_path)
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    content = _write_ninja(paths, ccache="/usr/bin/ccache")
+    build_dir = CORE.relative_pioenvs_path(CORE.name)
+    assert "rule pch" in content
+    assert "build esphome_pch.h.gch: pch" in content
+    for line in content.splitlines():
+        # C++ edges wait on the .gch; the C edge must not reference it
+        if line.startswith("build obj/src/main.cpp.o:"):
+            assert line.endswith("| esphome_pch.h.gch")
+        if line.startswith("build obj/src/esphome/vendor.c.o:"):
+            assert "esphome_pch" not in line
+    assert (build_dir / "esphome_pch.h").read_text().splitlines() == [
+        '#include "esphome/components/esp8266/throw_stubs.h"',
+        '#include "esphome/core/defines.h"',
+    ]
+    assert (build_dir / "esphome_pch.h.gch.sum").read_text().strip()
+    # Emission recorded so the framework env exports the ccache settings
+    from esphome.build_helpers.pch import ccache_pch_env
+
+    assert "CCACHE_PCH_EXTSUM" in ccache_pch_env()
+
+
+def test_write_project_pch_sum_only_with_ccache(tmp_path: Path) -> None:
+    """The .sum sidecar exists solely for ccache; skip it when disabled."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    content = _write_ninja(paths)
+    build_dir = CORE.relative_pioenvs_path(CORE.name)
+    assert "build esphome_pch.h.gch: pch" in content
+    assert not (build_dir / "esphome_pch.h.gch.sum").exists()
+
+
+def test_write_project_pch_identity_unknown_skips_pch(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """An OSError from the checksum means the .sum cannot vouch for the
+    .gch: the build must fall back to plain srcflags."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    with patch(
+        "esphome.build_gen.arduino8266.pch_checksum",
+        side_effect=OSError("stat failed"),
+    ):
+        content = _write_ninja(paths, ccache="/usr/bin/ccache")
+    assert "esphome_pch" not in content
+    assert "srccxxflags" not in content
+    assert "  flags = $srcflags" in content
+    assert "Could not establish the pch identity" in caplog.text
+
+
+def test_write_project_pch_folds_joined_src_force_include(
+    tmp_path: Path,
+) -> None:
+    """-includefoo.h in build_src_flags must fold into the pch like the
+    separated spelling, not precede and defeat it."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    CORE.platformio_options["build_src_flags"] = "-includeesphome/core/defines.h"
+    content = _write_ninja(paths, ccache="/usr/bin/ccache")
+    assert "build esphome_pch.h.gch: pch" in content
+    assert "srccxxflags" in content
+    assert "-includeesphome" not in content
+
+
+def test_write_project_pch_skipped_for_joined_force_include_spelling(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """GCC also accepts -includefoo.h as one token; the guard must see it."""
+    paths = _make_framework(tmp_path)
+    _set_flags(
+        "-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH", "-includefoo.h"
+    )
+    content = _write_ninja(paths, ccache="/usr/bin/ccache")
+    assert "esphome_pch" not in content
+    assert "prevents the precompiled header" in caplog.text
+
+
+def test_write_project_pch_skipped_when_user_force_include_precedes(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A -include in build_flags lands ahead of the pch include, so GCC
+    would never load the .gch; skip it and say so."""
+    paths = _make_framework(tmp_path)
+    _set_flags(
+        "-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH", "-include foo.h"
+    )
+    content = _write_ninja(paths, ccache="/usr/bin/ccache")
+    assert "esphome_pch" not in content
+    assert "srccxxflags" not in content
+    assert "prevents the precompiled header" in caplog.text
+    # No pch emitted: the ccache relaxation must stay off
+    from esphome.build_helpers.pch import ccache_pch_env
+
+    assert ccache_pch_env() == {}
+
+
+def test_write_project_pch_disabled(
+    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
+) -> None:
+    monkeypatch.setenv("ESPHOME_PCH_ENABLE", "0")
+    paths = _make_framework(tmp_path)
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    content = _write_ninja(paths)
+    assert "esphome_pch" not in content
+    assert "srccxxflags" not in content
+    assert "  flags = $srcflags" in content
+
+
+def test_write_project_scanf_float_and_waveform_kept(tmp_path: Path) -> None:
+    paths = _make_framework(tmp_path)
+    CORE.data[KEY_ESP8266][KEY_SCANF_FLOAT] = True
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+    content = _write_ninja(paths)
+    assert "-u _scanf_float" in content
+    # Waveform not stubbed out: both implementations stay in the archive
+    assert "core_esp8266_waveform_pwm.cpp.o" in content
+    assert "core_esp8266_waveform_phase.cpp.o" in content
+
+
 @pytest.mark.parametrize(
     ("knob", "lib", "mss", "features", "ipv6"),
     [
@@ -360,8 +637,8 @@ def test_generate_ld_scripts(tmp_path: Path) -> None:
         _run_generate_ld_scripts(paths)
     mock_run.assert_not_called()
 
-    # A surgery edit invalidates the stamp via the fingerprint (a stale
-    # linker script would otherwise persist until an esphome clean)
+    # An edit to the surgery constants invalidates the stamp (a stale linker
+    # script would otherwise persist until an esphome clean)
     with (
         patch.object(
             arduino8266.build_surgery, "surgery_fingerprint", return_value="changed"
@@ -401,8 +678,6 @@ def test_generate_ld_scripts_testing_mode(tmp_path: Path) -> None:
 
     paths = _make_framework(tmp_path)
     (paths.framework / "tools" / "sdk" / "ld" / "eagle.flash.4m.ld").write_text(
-        # Real flash ld scripts carry no iram1_0_seg (that lives in the
-        # generated common ld); the surgery rejects one it was not asked about
         "MEMORY\n{\n"
         "  dram0_0_seg :    org = 0x3FFE8000, len = 0x14000\n"
         "  irom0_0_seg :    org = 0x40201010, len = 0xfeff0\n"
@@ -416,6 +691,96 @@ def test_generate_ld_scripts_testing_mode(tmp_path: Path) -> None:
     assert "len = 0x2000000" in patched
 
 
+def test_write_project_libraries_and_variant(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+
+    paths = _make_framework(tmp_path)
+    variant_src = paths.framework / "variants" / "nodemcu" / "variant.cpp"
+    variant_src.write_text("")
+
+    lib_dir = tmp_path / "libsrc"
+    lib_dir.mkdir()
+    (lib_dir / "lib.cpp").write_text("")
+    (lib_dir / "impl.cc").write_text("")
+    headers_only = ArduinoLibrary(name="HeadersOnly", include_dirs=[lib_dir])
+    library = ArduinoLibrary(
+        name="MyLib",
+        sources=[lib_dir / "impl.cc", lib_dir / "lib.cpp"],
+        include_dirs=[lib_dir],
+        flags=["-DMYLIB=1"],
+        link_dirs=[lib_dir / "blobs"],
+        link_libs=["algobsec"],
+        link_flags=["-Wl,--wrap=malloc"],
+    )
+    _set_flags("-DPIO_FRAMEWORK_ARDUINO_ENABLE_EXCEPTIONS")
+
+    with caplog.at_level(logging.DEBUG, logger="esphome.build_gen.arduino8266"):
+        content = _write_ninja(
+            paths, libraries=[library, headers_only], ccache="/cc/ccache"
+        )
+
+    assert "build libFrameworkArduinoVariant.a: ar" in content
+    assert "build libMyLib.a: ar" in content
+    # A headers-only library contributes includes but no archive, with a
+    # debug log distinguishing it from a resolution failure
+    assert "libHeadersOnly.a" not in content
+    assert "Library HeadersOnly has no source files" in caplog.text
+    assert "  flags = -DMYLIB=1" in content
+    assert "-lalgobsec" in content
+    # Library link flags reach the firmware link line; .cc compiles as C++
+    assert "-Wl,--wrap=malloc" in content
+    assert "impl.cc.o: cxx" in content
+    assert f"-L{_shq(str(lib_dir / 'blobs'))}" in content
+    # Exceptions knob: -fexceptions and the exception-enabled stdc++
+    assert "-fexceptions" in content
+    assert "-lstdc++-exc" in content
+    assert f"ccache = {_shq('/cc/ccache')}" in content
+
+
+def test_get_flash_ld_path(tmp_path: Path) -> None:
+
+    paths = InstalledPaths(
+        framework=tmp_path / "framework",
+        toolchain=tmp_path / "toolchain",
+        ninja=Path("ninja"),
+    )
+    CORE.testing_mode = True
+    assert get_flash_ld_path(tmp_path, paths) == (
+        tmp_path / "ld" / "testing_eagle.flash.4m.ld"
+    )
+
+    CORE.testing_mode = False
+    # Reads the same install the ninja file linked against; no re-resolve
+    assert get_flash_ld_path(tmp_path, paths) == (
+        tmp_path / "framework" / "tools" / "sdk" / "ld" / "eagle.flash.4m.ld"
+    )
+
+
+def test_flash_size_str() -> None:
+    assert _flash_size_str(4 * 1024 * 1024) == "4M"
+    assert _flash_size_str(512 * 1024) == "512K"
+
+
+def test_write_project_testing_mode(tmp_path: Path) -> None:
+    paths = _make_framework(tmp_path)
+    CORE.testing_mode = True
+    _set_flags()
+    content = _write_ninja(paths)
+    assert "-T testing_eagle.flash.4m.ld" in content
+    assert "ld/testing_eagle.flash.4m.ld" in content
+
+
+def test_write_project_missing_framework_dir_raises(tmp_path: Path) -> None:
+    """An incomplete framework install fails naming the missing path."""
+
+    paths = _make_framework(tmp_path)
+    shutil.rmtree(paths.framework / "tools" / "sdk" / "lwip2")
+    _set_flags()
+    with pytest.raises(EsphomeError, match="incomplete.*lwip2"):
+        _write_ninja(paths)
+
+
 def test_generate_ld_scripts_testing_mode_missing_flash_ld_raises(
     tmp_path: Path,
 ) -> None:
@@ -440,6 +805,61 @@ def test_build_config_nonosdk_precedence() -> None:
     assert _resolve_build_config(_defines()).nonosdk == "NONOSDK221"
 
 
+def test_write_project_plain_asm_rule_skips_preprocessor(tmp_path: Path) -> None:
+    """A lowercase .s source assembles plain (SCons AS), never through the
+    preprocessor rule that a .S source gets."""
+    paths = _make_framework(tmp_path)
+    core_dir = paths.framework / "cores" / "esp8266"
+    (core_dir / "lowlevel.s").write_text("nop\n")
+    _set_flags()
+    content = _write_ninja(paths)
+    assert "lowlevel.s.o: asm " in content
+    assert "rule asm\n  command = $ccache $cc -x assembler $asflags -c $in -o $out" in (
+        content
+    )
+
+
+def test_write_project_unflags_operandless_linker_flag(tmp_path: Path) -> None:
+    """build_unflags: -nostdlib filters whole-token from both lines, as
+    PlatformIO allows; only operand-taking flags hard-error."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    CORE.build_unflags = {"-nostdlib"}
+    content = _write_ninja(paths)
+    assert "-nostdlib" not in content
+
+
+def test_write_project_unflagged_symbol_takes_its_dash_u(tmp_path: Path) -> None:
+    """Unflagging a -u symbol drops the -u that carried it; a dangling -u
+    would consume the next token and hand ld a symbol as an input file."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    CORE.build_unflags = {"_printf_float"}
+    content = _write_ninja(paths)
+    link_line = next(
+        line for line in content.splitlines() if line.startswith("linkflags = ")
+    )
+    assert "_printf_float" not in link_line
+    assert "-u -u" not in link_line
+    # The neighbors survive as intact pairs
+    assert "-u app_entry" in link_line
+    assert "-u _DebugExceptionVector" in link_line
+
+
+def test_write_project_build_unflags_apply_to_framework_flags(tmp_path: Path) -> None:
+    """build_unflags removes flags from the framework sets, as PlatformIO does."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    CORE.build_unflags = {"-fipa-pta", "-Wl,--gc-sections"}
+    content = _write_ninja(paths)
+    for line in content.splitlines():
+        key = line.split(" = ")[0]
+        if key in ("cflags", "cxxflags", "asflags"):
+            assert "-fipa-pta" not in line
+        if key == "linkflags":
+            assert "-Wl,--gc-sections" not in line
+
+
 def test_project_flags_trailing_bare_linker_flag_warns(
     caplog: pytest.LogCaptureFixture,
 ) -> None:
@@ -494,10 +914,18 @@ def test_project_flags_requotes_lexed_defines() -> None:
     compile_flags, _link, _dirs, _libs = _split_flags()
     # shlex folds the quotes (as PIO's ParseFlags does); _shell_token
     # re-quotes the spaced token so the shell passes one argv element
-    expected = (
-        '"-DGREETING=hello world"' if os.name == "nt" else "'-DGREETING=hello world'"
-    )
-    assert compile_flags == [expected]
+    assert compile_flags == [_shq("-DGREETING=hello world")]
+
+
+def test_write_project_empty_core_raises(tmp_path: Path) -> None:
+    """A framework tree with no core sources fails at generation, not link."""
+    paths = _make_framework(tmp_path)
+    core = paths.framework / "cores" / "esp8266"
+    for f in core.iterdir():
+        f.unlink()
+    _set_flags()
+    with pytest.raises(EsphomeError, match="no core sources"):
+        _write_ninja(paths)
 
 
 def test_flag_defines_joins_spaced_define() -> None:
@@ -508,6 +936,75 @@ def test_flag_defines_joins_spaced_define() -> None:
     assert "" not in defines
 
 
+def test_ninja_path_escaping() -> None:
+    """Build-statement paths and command-line paths escape differently."""
+    assert arduino8266._e("a b:$c") == "a$ b$:$$c"
+    assert arduino8266._q("/a b/$x") == _shq("/a b/$$x")
+
+
+def test_write_project_asm_excludes_non_define_user_flags(tmp_path: Path) -> None:
+    """The ASPPCOM command under PlatformIO never sees CCFLAGS, so only -D/-I user flags
+    reach assembly compiles."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DUSER_KNOB=1", "-Wno-volatile")
+    content = _write_ninja(paths)
+    asflags = next(line for line in content.splitlines() if line.startswith("asflags"))
+    assert "-DUSER_KNOB=1" in asflags
+    assert "-Wno-volatile" not in asflags
+    cxxflags = next(
+        line for line in content.splitlines() if line.startswith("cxxflags")
+    )
+    assert "-Wno-volatile" in cxxflags
+
+
+def test_write_project_returns_changed(tmp_path: Path) -> None:
+    """The documented contract: True when build.ninja changed, False on an
+    identical regeneration (pins byte-stable output too)."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    src = CORE.relative_src_path()
+    (src / "esphome" / "components" / "esp8266").mkdir(parents=True, exist_ok=True)
+    (src / "main.cpp").write_text("")
+    with (
+        patch.object(arduino8266, "generate_ld_scripts"),
+        patch("esphome.arduino.library.resolve_libraries", return_value=[]),
+    ):
+        assert arduino8266.write_project(paths, None) is True
+        assert arduino8266.write_project(paths, None) is False
+
+
+def test_write_project_missing_elf2bin_raises(tmp_path: Path) -> None:
+    """A half-extracted package must fail by name at generation, not after
+    the full compile at the elf2bin edge."""
+    paths = _make_framework(tmp_path)
+    (paths.framework / "tools" / "elf2bin.py").unlink()
+    _set_flags()
+    src = CORE.relative_src_path()
+    (src / "main.cpp").parent.mkdir(parents=True, exist_ok=True)
+    (src / "main.cpp").write_text("")
+    with (
+        patch.object(arduino8266, "generate_ld_scripts"),
+        patch("esphome.arduino.library.resolve_libraries", return_value=[]),
+        pytest.raises(EsphomeError, match="elf2bin"),
+    ):
+        arduino8266.write_project(paths, None)
+
+
+def test_write_project_missing_src_dir_raises(tmp_path: Path) -> None:
+    """A missing generated source tree is its own error, not an install one."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    with (
+        patch.object(arduino8266, "generate_ld_scripts"),
+        patch("esphome.arduino.library.resolve_libraries", return_value=[]),
+        patch.object(
+            arduino8266.CORE, "relative_src_path", return_value=tmp_path / "nope"
+        ),
+        pytest.raises(EsphomeError, match="source directory"),
+    ):
+        arduino8266.write_project(paths, None)
+
+
 def test_build_config_custom_mmu_without_knob_raises() -> None:
     """Custom MMU sizes without the CUSTOM knob are refused."""
     with pytest.raises(EsphomeError, match="PIO_FRAMEWORK_ARDUINO_MMU_CUSTOM"):
@@ -600,6 +1097,49 @@ def test_generate_ld_scripts_missing_compiler_is_clean(tmp_path: Path) -> None:
         _run_generate_ld_scripts(paths)
 
 
+def test_write_project_asm_keeps_quoted_defines(tmp_path: Path) -> None:
+    """A spaced -D/-I user flag arrives shell-quoted; assembly must still
+    receive it."""
+    paths = _make_framework(tmp_path)
+    _set_flags('-DGREETING="hello world"', "-Wno-volatile")
+    content = _write_ninja(paths)
+    asflags = next(line for line in content.splitlines() if line.startswith("asflags"))
+    assert _shq("-DGREETING=hello world") in asflags
+    assert "-Wno-volatile" not in asflags
+
+
+def test_write_project_unarchived_library_links_objects(tmp_path: Path) -> None:
+    """A libArchive:false library's objects reach the link directly."""
+
+    paths = _make_framework(tmp_path)
+    lib_src = tmp_path / "gdb" / "src"
+    lib_src.mkdir(parents=True)
+    (lib_src / "GDBStub.cpp").write_text("")
+    _set_flags()
+    lib = ArduinoLibrary(
+        name="GDBStub",
+        sources=[lib_src / "GDBStub.cpp"],
+        include_dirs=[lib_src],
+        lib_archive=False,
+    )
+    content = _write_ninja(paths, libraries=[lib])
+    assert "libGDBStub.a" not in content
+    link_line = next(
+        line for line in content.splitlines() if line.startswith("build firmware.elf")
+    )
+    assert "GDBStub.cpp.o" in link_line
+
+
+def test_write_project_unknown_board_fails_by_name(tmp_path: Path) -> None:
+    """A caller bypassing config validation gets the board named, not a
+    KeyError."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    CORE.data[KEY_ESP8266][KEY_BOARD] = "not_a_board"
+    with pytest.raises(EsphomeError, match="'not_a_board' is not supported"):
+        _write_ninja(paths)
+
+
 def test_unflag_tokens_join_spaced_entries() -> None:
     """Spaced build_unflags entries ("-D FOO") match the joined token."""
     CORE.build_unflags = {"-D FOO", "-l bar"}
@@ -1072,6 +1612,31 @@ def test_generate_ld_scripts_surgery_failure_is_named(tmp_path: Path) -> None:
         _run_generate_ld_scripts(paths)
 
 
+def test_write_project_unmatched_unflag_warns(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """An unflag that removes nothing is named; a matching one is silent."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DUSE_FOO=1")
+    CORE.build_unflags = {"-DUSE_FOO", "-Os"}
+    content = _write_ninja(paths)
+    assert "matched no build flag: -DUSE_FOO" in caplog.text
+    assert "-Os" not in caplog.text.split("matched no build flag")[-1].splitlines()[0]
+    # The matching -Os unflag really removed the framework flag
+    cflags = next(line for line in content.splitlines() if line.startswith("cflags"))
+    assert " -Os " not in cflags
+
+
+def test_write_project_lexes_build_flags_once(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A malformed build_flags entry warns once per generation."""
+    paths = _make_framework(tmp_path)
+    _set_flags("-DFOO=1 -l")
+    _write_ninja(paths)
+    assert caplog.text.count("Ignoring trailing '-l'") == 1
+
+
 def test_build_config_mmu_conflict_names_the_variant_knob_with_custom() -> None:
     """With MMU_CUSTOM also set, the actionable fix is dropping the variant
     knob, not setting the knob the user already set."""
@@ -1171,6 +1736,11 @@ def test_generate_ld_scripts_unreadable_header_forces_regeneration(
         mock_run.assert_called_once()
 
 
+def test_board_tables_are_equal() -> None:
+    """BOARDS and ESP8266_BOARD_BUILD must stay exactly in sync."""
+    assert set(BOARDS) == set(ESP8266_BOARD_BUILD)
+
+
 def test_bare_include_and_define_dropped(
     caplog: pytest.LogCaptureFixture,
 ) -> None:
@@ -1208,3 +1778,67 @@ def test_defines_flags_honors_f_cpu_override() -> None:
     CORE.platformio_options = {"board_build.f_cpu": "160000000L"}
     defines = _defines_flags(config, "dout", "nodemcuv2", board_build["defines"])
     assert "-DF_CPU=160000000L" in defines
+
+
+def test_flash_ld_name_honors_ldscript_override(tmp_path: Path) -> None:
+    """board_build.ldscript (filesystem reservation, corrected flash size)
+    replaces the board default; a path is rejected since the name resolves
+    via the -L search path."""
+    assert arduino8266._flash_ld_name("nodemcuv2") == "eagle.flash.4m.ld"
+    CORE.platformio_options = {"board_build.ldscript": "eagle.flash.4m2m.ld"}
+    assert arduino8266._flash_ld_name("nodemcuv2") == "eagle.flash.4m2m.ld"
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    content = _write_ninja(paths)
+    assert "-T eagle.flash.4m2m.ld" in content
+    CORE.platformio_options = {"board_build.ldscript": "../evil.ld"}
+    with pytest.raises(EsphomeError, match="bare script name"):
+        arduino8266._flash_ld_name("nodemcuv2")
+
+
+def test_unflagging_a_plain_linker_flag_raises(tmp_path: Path) -> None:
+    """build_unflags: -u would strip all seven -u tokens and leave the
+    operands as ld input files; refuse by name instead."""
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    CORE.build_unflags = {"-u _printf_float"}
+    with pytest.raises(EsphomeError, match="cannot remove plain linker"):
+        _write_ninja(paths)
+
+
+def test_d1_wroom_02_keeps_its_shipped_flash_layout() -> None:
+    """The board joined BOARDS late; the flash-size default (2m.ld) would
+    move _FS_end and the preferences sector on existing devices."""
+    assert arduino8266._flash_ld_name("d1_wroom_02") == "eagle.flash.2m64.ld"
+    assert arduino8266._flash_ld_name("nodemcuv2") == "eagle.flash.4m.ld"
+
+
+def test_write_project_rejects_spaced_ldscript_override(tmp_path: Path) -> None:
+    """A spaced override never reaches the link line: generate_ld_scripts
+    rejects the name first (the -T _shell_token quoting behind it is
+    defence-in-depth)."""
+    CORE.platformio_options = {"board_build.ldscript": "my script.ld"}
+    paths = _make_framework(tmp_path)
+    _set_flags()
+    with pytest.raises(EsphomeError, match="Invalid flash linker script name"):
+        arduino8266.write_project(paths, None)
+
+
+def test_write_project_pch_no_device_path_poison(tmp_path: Path) -> None:
+    """Regression: the -include stays relative and the .sum carries no
+    per-device path, or cross-device ccache sharing breaks."""
+    paths = _make_framework(tmp_path / "shared")
+    sums = []
+    for name in ("dev_a", "dev_b"):
+        CORE.name = name
+        CORE.build_path = tmp_path / name
+        _set_flags("-DPIO_FRAMEWORK_ARDUINO_LWIP2_HIGHER_BANDWIDTH_LOW_FLASH")
+        content = _write_ninja(paths, ccache="/usr/bin/ccache")
+        assert (
+            "srccxxflags = -Winvalid-pch -Wno-error=invalid-pch "
+            "-include esphome_pch.h" in content
+        )
+        sums.append(
+            (CORE.relative_pioenvs_path(name) / "esphome_pch.h.gch.sum").read_text()
+        )
+    assert sums[0] == sums[1]

tests/unit_tests/build_helpers/test_ccache.py

@@ -10,6 +10,7 @@ from unittest.mock import patch
 import pytest
 
 from esphome.build_helpers import ccache
+from esphome.core import CORE
 
 
 def test_resolve_opt_out() -> None:
@@ -120,3 +121,17 @@ def test_parse_enable_env_spelling_tables(
     """cv.boolean's spelling tables plus the 1/0 env convention."""
     monkeypatch.setenv("ESPHOME_CCACHE_ENABLE", raw)
     assert ccache.parse_enable_env("ESPHOME_CCACHE_ENABLE") is expected
+
+
+def test_effective_ccache_basedir_prefers_user_value(tmp_path: Path) -> None:
+    CORE.build_path = tmp_path
+    # Drive-qualified on Windows: "/custom/base" is not absolute there
+    base = "C:\\custom\\base" if os.name == "nt" else "/custom/base"
+    with patch.dict(os.environ, {"CCACHE_BASEDIR": base}, clear=True):
+        assert ccache.effective_ccache_basedir() == base
+    with patch.dict(os.environ, {}, clear=True):
+        assert ccache.effective_ccache_basedir() == str(tmp_path.resolve())
+    # Degenerate values would strip substrings ccache never rewrites
+    for bad in ("", "/", "a/b"):
+        with patch.dict(os.environ, {"CCACHE_BASEDIR": bad}, clear=True):
+            assert ccache.effective_ccache_basedir() == str(tmp_path.resolve())

tests/unit_tests/build_helpers/test_idedata.py

@@ -85,6 +85,87 @@ def test_parse_entry_resolves_relative_includes() -> None:
     assert all(Path(inc).is_absolute() for inc in includes)
 
 
+def test_parse_entry_resolves_force_include_path(tmp_path: Path) -> None:
+    """The pch -include is emitted relative to the build dir; idedata must
+    resolve it so cached flags work from any cwd."""
+    (tmp_path / "esphome_pch.h").write_text("")
+    entry = _entry(
+        str(tmp_path),
+        f"{tmp_path}/src/esphome/x.cpp",
+        "g++ -include esphome_pch.h -c x.cpp",
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    idx = cxx_flags.index("-include")
+    resolved = cxx_flags[idx + 1]
+    assert Path(resolved).is_absolute()
+    assert resolved == str(tmp_path / "esphome_pch.h").replace("\\", "/")
+
+
+def test_parse_entry_resolves_joined_force_include(tmp_path: Path) -> None:
+    """The joined -includefoo.h spelling takes the same resolve path."""
+    (tmp_path / "esphome_pch.h").write_text("")
+    entry = _entry(
+        str(tmp_path),
+        f"{tmp_path}/src/esphome/x.cpp",
+        "g++ -includeesphome_pch.h -c x.cpp",
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    resolved = cxx_flags[cxx_flags.index("-include") + 1]
+    assert resolved == str(tmp_path / "esphome_pch.h").replace("\\", "/")
+
+
+def test_parse_entry_keeps_search_chain_force_include(tmp_path: Path) -> None:
+    """-include names resolved via the -I chain (libretiny's Arduino.h) must
+    not be re-anchored to a nonexistent build-dir path."""
+    entry = _entry(
+        str(tmp_path),
+        f"{tmp_path}/src/esphome/x.cpp",
+        "g++ -include Arduino.h -c x.cpp",
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    assert cxx_flags[cxx_flags.index("-include") + 1] == "Arduino.h"
+
+
+def test_parse_entry_warns_on_vanished_force_include(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A build-dir force-include deleted by clean_build must leave a trail;
+    a name resolvable via the -I chain must not warn."""
+    inc = tmp_path / "inc"
+    inc.mkdir()
+    (inc / "Arduino.h").write_text("")
+    entry = _entry(
+        str(tmp_path),
+        f"{tmp_path}/src/esphome/x.cpp",
+        f"g++ -I{inc} -include Arduino.h -include esphome_pch.h -c x.cpp",
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    assert "Arduino.h" in cxx_flags
+    assert "esphome_pch.h" in caplog.text
+    assert "Arduino.h" not in caplog.text
+
+
+def test_parse_entry_drops_trailing_force_include(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    entry = _entry(
+        str(tmp_path), f"{tmp_path}/src/esphome/x.cpp", "g++ -c x.cpp -include"
+    )
+
+    _, _, _, cxx_flags = idedata.parse_entry(entry)
+
+    assert "-include" not in cxx_flags
+    assert "no argument" in caplog.text
+
+
 def test_parse_entry_skips_dependency_flags() -> None:
     """Dependency-generation flags (and their args) are dropped."""
     entry = _entry(

tests/unit_tests/build_helpers/test_pch.py

@@ -0,0 +1,218 @@
+"""Tests for esphome.build_helpers.pch."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from unittest.mock import patch
+
+import pytest
+
+from esphome.build_helpers import pch
+
+
+def _write(src_dir: Path, name: str, content: str) -> None:
+    path = src_dir / name
+    path.parent.mkdir(parents=True, exist_ok=True)
+    path.write_text(content)
+
+
+@pytest.mark.parametrize(
+    ("value", "expected"),
+    [
+        (None, True),
+        ("1", True),
+        ("0", False),
+        ("false", False),
+        ("", False),
+    ],
+)
+def test_pch_enabled(value: str | None, expected: bool) -> None:
+    env = {} if value is None else {"ESPHOME_PCH_ENABLE": value}
+    with patch.dict(os.environ, env, clear=True):
+        assert pch.pch_enabled() is expected
+
+
+def test_ccache_pch_env_empty_until_emitted() -> None:
+    """No sloppiness relaxation for a build that skipped the pch."""
+    with patch.dict(os.environ, {}, clear=True):
+        assert pch.ccache_pch_env() == {}
+
+
+def test_ccache_pch_env_enabled() -> None:
+    pch.mark_pch_emitted()
+    with patch.dict(os.environ, {}, clear=True):
+        env = pch.ccache_pch_env()
+    assert env["CCACHE_SLOPPINESS"] == "pch_defines,time_macros"
+    assert env["CCACHE_PCH_EXTSUM"] == "true"
+
+
+def test_ccache_pch_env_disabled() -> None:
+    with patch.dict(os.environ, {"ESPHOME_PCH_ENABLE": "0"}, clear=True):
+        assert pch.ccache_pch_env() == {}
+
+
+def test_ccache_pch_env_token_check_is_membership_not_substring(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A token merely containing ours must not suppress the union."""
+    pch.mark_pch_emitted()
+    with patch.dict(os.environ, {"CCACHE_SLOPPINESS": "pch_defines_extra"}, clear=True):
+        env = pch.ccache_pch_env()
+    assert env["CCACHE_SLOPPINESS"] == "pch_defines_extra,pch_defines,time_macros"
+
+
+def test_ccache_pch_env_unions_user_sloppiness(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """Without pch_defines/time_macros ccache declines every pch-consuming
+    compile, so missing tokens are unioned onto the user's value."""
+    pch.mark_pch_emitted()
+    with patch.dict(os.environ, {"CCACHE_SLOPPINESS": "locale"}, clear=True):
+        env = pch.ccache_pch_env()
+    assert env["CCACHE_SLOPPINESS"] == "locale,pch_defines,time_macros"
+    assert env["CCACHE_PCH_EXTSUM"] == "true"
+    assert "Adding pch_defines,time_macros" in caplog.text
+    caplog.clear()
+    with patch.dict(
+        os.environ, {"CCACHE_SLOPPINESS": "pch_defines,time_macros"}, clear=True
+    ):
+        env = pch.ccache_pch_env()
+    assert "CCACHE_SLOPPINESS" not in env
+    assert not caplog.records
+
+
+def test_pch_header_text_preserves_order() -> None:
+    text = pch.pch_header_text(["b.h", "a.h"])
+    assert text == '#include "b.h"\n#include "a.h"\n'
+
+
+def test_include_closure_resolves_relative_and_root(tmp_path: Path) -> None:
+    """Sibling includes resolve against the includer's directory first,
+    full paths against the src root; unresolvable names end the walk."""
+    _write(tmp_path, "esphome/components/x/a.h", '#include "b.h"\n')
+    _write(
+        tmp_path,
+        "esphome/components/x/b.h",
+        '#include "esphome/core/deep.h"\n#include <system.h>\n#include "missing.h"\n',
+    )
+    _write(tmp_path, "esphome/core/deep.h", "")
+    closure = pch._include_closure(tmp_path, ["esphome/components/x/a.h"])
+    assert sorted(closure) == [
+        "esphome/components/x/a.h",
+        "esphome/components/x/b.h",
+        "esphome/core/deep.h",
+    ]
+
+
+def test_include_closure_handles_cycles(tmp_path: Path) -> None:
+    _write(tmp_path, "a.h", '#include "b.h"\n')
+    _write(tmp_path, "b.h", '#include "a.h"\n')
+    assert sorted(pch._include_closure(tmp_path, ["a.h"])) == ["a.h", "b.h"]
+
+
+def test_include_closure_blocks_parent_escape(tmp_path: Path) -> None:
+    _write(tmp_path / "src", "a.h", '#include "../outside.h"\n')
+    (tmp_path / "outside.h").write_text("")
+    assert sorted(pch._include_closure(tmp_path / "src", ["a.h"])) == ["a.h"]
+
+
+def test_pch_checksum_tracks_closure_content(tmp_path: Path) -> None:
+    """A transitive header edit or an extra-identity change must change the
+    digest; unrelated files must not."""
+    _write(tmp_path, "root.h", '#include "nested.h"\n')
+    _write(tmp_path, "nested.h", "int a;\n")
+    _write(tmp_path, "unrelated.h", "int u;\n")
+    base = pch.pch_checksum(tmp_path, ["root.h"], ["id"])
+    assert base == pch.pch_checksum(tmp_path, ["root.h"], ["id"])
+    assert base != pch.pch_checksum(tmp_path, ["root.h"], ["other-id"])
+    _write(tmp_path, "unrelated.h", "int changed;\n")
+    assert base == pch.pch_checksum(tmp_path, ["root.h"], ["id"])
+    _write(tmp_path, "nested.h", "int b;\n")
+    assert base != pch.pch_checksum(tmp_path, ["root.h"], ["id"])
+
+
+@pytest.mark.skipif(
+    os.name == "nt" or os.geteuid() == 0, reason="chmod is ineffective here"
+)
+def test_include_closure_fails_closed_on_unreadable(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A marker would truncate the transitive walk; the OSError propagates
+    so callers compile without a pch."""
+    _write(tmp_path, "a.h", '#include "locked.h"\n')
+    locked = tmp_path / "locked.h"
+    locked.write_text("")
+    locked.chmod(0)
+    try:
+        with pytest.raises(OSError):
+            pch._include_closure(tmp_path, ["a.h"])
+    finally:
+        locked.chmod(0o644)
+    assert "Could not read locked.h" in caplog.text
+
+
+def test_pch_extra_scripts_gated(monkeypatch: pytest.MonkeyPatch) -> None:
+    with patch.dict(os.environ, {}, clear=True):
+        assert pch.pch_extra_scripts() == ["post:pch.py"]
+    monkeypatch.setenv("ESPHOME_PCH_ENABLE", "0")
+    assert pch.pch_extra_scripts() == []
+
+
+def test_include_closure_raises_when_identity_unknown(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """An unreadable header propagates; callers compile without a pch."""
+
+    class _BadFile:
+        def stat(self):  # noqa: ANN202 -- regular-file mode only
+            import os
+            import stat as stat_mod
+
+            return os.stat_result((stat_mod.S_IFREG | 0o644,) + (0,) * 9)
+
+        def read_bytes(self) -> bytes:
+            raise OSError("read failed")
+
+    class _FakeSrcDir:
+        def __truediv__(self, rel: str) -> _BadFile:
+            return _BadFile()
+
+    with pytest.raises(OSError, match="read failed"):
+        pch._include_closure(_FakeSrcDir(), ["a.h"])
+    assert "Could not read a.h" in caplog.text
+
+
+def test_include_closure_survives_non_utf8_include_name(tmp_path: Path) -> None:
+    """A non-UTF-8 quoted include must not abort the build; it simply does
+    not resolve and ends the walk."""
+    (tmp_path / "a.h").write_bytes(b'#include "bad\xff.h"\n#include "b.h"\n')
+    (tmp_path / "b.h").write_text("")
+    closure = pch._include_closure(tmp_path, ["a.h"])
+    assert set(closure) == {"a.h", "b.h"}
+
+
+def test_pch_checksum_survives_surrogate_extra(tmp_path: Path) -> None:
+    """Install paths from non-UTF-8 filesystems carry surrogates; hashing
+    them must not raise past the caller's identity-unknown guard."""
+    assert pch.pch_checksum(tmp_path, [], ["/opt/bad\udcff/framework"])
+
+
+def test_include_closure_walks_angle_includes_under_src(tmp_path: Path) -> None:
+    """An angle include resolving under src/ must enter the digest; one
+    that does not simply ends the walk."""
+    _write(tmp_path, "a.h", "#include <local.h>\n#include <Arduino.h>\n")
+    (tmp_path / "local.h").write_text("")
+    closure = pch._include_closure(tmp_path, ["a.h"])
+    assert set(closure) == {"a.h", "local.h"}
+
+
+def test_ccache_pch_env_warns_on_falsy_extsum(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A user CCACHE_PCH_EXTSUM=false makes ccache hash the .gch bytes."""
+    pch.mark_pch_emitted()
+    with patch.dict(os.environ, {"CCACHE_PCH_EXTSUM": "false"}, clear=True):
+        env = pch.ccache_pch_env()
+    assert "CCACHE_PCH_EXTSUM" not in env
+    assert "disables pch caching" in caplog.text

tests/unit_tests/components/esp8266/test_toolchain_validation.py

@@ -0,0 +1,193 @@
+"""Tests for the native (non-PlatformIO) toolchain config validation."""
+
+from __future__ import annotations
+
+from collections.abc import Generator
+from pathlib import Path
+from types import SimpleNamespace
+from unittest.mock import patch
+
+import pytest
+
+from esphome.components import esp8266
+from esphome.components.esp8266 import (
+    ARDUINO_FRAMEWORK_SCHEMA,
+    _resolve_toolchain,
+    _validate_native_toolchain,
+)
+import esphome.config_validation as cv
+from esphome.const import (
+    CONF_BOARD,
+    CONF_FRAMEWORK,
+    CONF_PLATFORM_VERSION,
+    CONF_SOURCE,
+    CONF_TOOLCHAIN,
+    CONF_VERSION,
+    Toolchain,
+)
+from esphome.core import CORE, EsphomeError
+from esphome.types import ConfigType
+
+
+@pytest.fixture(autouse=True)
+def _arduino_toolchain() -> Generator[None]:
+    # The suite-wide reset_core fixture clears CORE.toolchain after each test
+    CORE.toolchain = Toolchain.ARDUINO
+    esp8266._DECODE_WARNED_AT.clear()
+    yield
+    esp8266._DECODE_WARNED_AT.clear()
+
+
+def _config(
+    board: str = "nodemcuv2",
+    version: str = "3.1.2",
+    source: str | None = None,
+    platform_version: str | None = None,
+) -> ConfigType:
+    framework: dict[str, str] = {CONF_VERSION: version}
+    if source is not None:
+        framework[CONF_SOURCE] = source
+    if platform_version is not None:
+        framework[CONF_PLATFORM_VERSION] = platform_version
+    # The real schema fills the source/platform_version defaults, so these
+    # tests validate against what config validation actually emits
+    return {
+        CONF_FRAMEWORK: ARDUINO_FRAMEWORK_SCHEMA(framework),
+        CONF_BOARD: board,
+    }
+
+
+def test_valid_config_passes() -> None:
+    config = _config()
+    assert _validate_native_toolchain(config) is config
+
+
+def test_platformio_toolchain_skips_checks() -> None:
+    # 3.0.2 is pio-legal (>= the global 3.0.0 floor) but below the native
+    # toolchain's own 3.1.1 floor; the bogus board only the native path checks
+    CORE.toolchain = Toolchain.PLATFORMIO
+    config = _config(board="not_a_board", version="3.0.2")
+    assert _validate_native_toolchain(config) is config
+
+
+def test_version_below_floor_rejected() -> None:
+    # 3.1.0 has no registry package, so the native floor is 3.1.1
+    with pytest.raises(cv.Invalid, match="3.1.1 or newer"):
+        _validate_native_toolchain(_config(version="3.1.0"))
+
+
+def test_version_at_floor_accepted() -> None:
+    _validate_native_toolchain(_config(version="3.1.1"))
+
+
+def test_custom_platform_version_warns_and_is_dropped(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    config = _config(platform_version="platformio/espressif8266@4.0.1")
+    _validate_native_toolchain(config)
+    assert "'platform_version' is ignored" in caplog.text
+    assert CONF_PLATFORM_VERSION not in config[CONF_FRAMEWORK]
+
+
+def test_default_platform_version_does_not_warn(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    config = _config()
+    _validate_native_toolchain(config)
+    assert "'platform_version' is ignored" not in caplog.text
+    assert CONF_PLATFORM_VERSION not in config[CONF_FRAMEWORK]
+
+
+def test_custom_source_rejected() -> None:
+    with pytest.raises(cv.Invalid, match="custom framework source"):
+        _validate_native_toolchain(
+            _config(source="https://github.com/esp8266/Arduino.git")
+        )
+
+
+def test_unsupported_board_rejected() -> None:
+    with pytest.raises(cv.Invalid, match="not supported by"):
+        _validate_native_toolchain(_config(board="not_a_board"))
+
+
+def test_yaml_toolchain_key_resolves() -> None:
+    """The documented `toolchain: arduino` YAML key selects the native path."""
+    CORE.toolchain = None
+    _resolve_toolchain({CONF_TOOLCHAIN: Toolchain.ARDUINO})
+    assert CORE.toolchain == Toolchain.ARDUINO
+    assert CORE.using_toolchain_arduino
+
+
+def test_yaml_toolchain_key_defaults_to_platformio() -> None:
+    CORE.toolchain = None
+    _resolve_toolchain({})
+    assert CORE.toolchain == Toolchain.PLATFORMIO
+
+
+def test_decode_pc_native_missing_tools_warns_once(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A stack dump of many addresses produces one missing-tool warning."""
+
+    with (
+        patch(
+            "esphome.arduino8266.toolchain.get_addr2line_path",
+            return_value=tmp_path / "missing-addr2line",
+        ),
+        patch(
+            "esphome.arduino8266.toolchain.get_elf_path",
+            return_value=tmp_path / "missing.elf",
+        ),
+    ):
+        esp8266._decode_pc({}, "40201234")
+        esp8266._decode_pc({}, "40201238")
+    assert caplog.text.count("Cannot decode crash addresses") == 1
+
+
+def test_decode_pc_platformio_missing_tools_warns_once(
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """The PlatformIO branch reports a missing addr2line/ELF at the same
+    warning level as the native one; raw undecoded addresses with no
+    stated reason are undiagnosable at default log level."""
+
+    CORE.toolchain = Toolchain.PLATFORMIO
+    idedata = SimpleNamespace(addr2line_path=None, firmware_elf_path=None)
+    with patch("esphome.platformio.toolchain.get_idedata", return_value=idedata):
+        esp8266._decode_pc({}, "40201234")
+        esp8266._decode_pc({}, "40201238")
+    assert caplog.text.count("Cannot decode crash addresses") == 1
+
+
+def test_resolve_toolchain_rejects_unsupported() -> None:
+    """ESP8266 rejects a CLI toolchain it cannot serve, like every platform."""
+
+    CORE.toolchain = Toolchain.SDK_NRF
+    with pytest.raises(cv.Invalid, match="Unsupported toolchain 'sdk-nrf'"):
+        _resolve_toolchain({})
+
+
+def test_run_compile_platformio_falls_through() -> None:
+    """Under toolchain: platformio the hook returns False without touching
+    the native backend; this is what keeps existing users on PlatformIO."""
+    CORE.toolchain = Toolchain.PLATFORMIO
+    with patch("esphome.arduino8266.toolchain.run_compile") as mock_native:
+        assert esp8266.run_compile(SimpleNamespace(), {}) is False
+    mock_native.assert_not_called()
+
+
+def test_run_compile_arduino_failure_raises() -> None:
+    """A non-zero native build fails by name instead of returning success."""
+    CORE.verbose = False
+    with (
+        patch("esphome.arduino8266.toolchain.run_compile", return_value=1),
+        pytest.raises(EsphomeError, match="native build failed"),
+    ):
+        esp8266.run_compile(SimpleNamespace(), {})
+
+
+def test_copy_files_native_skips_platformio_scripts(tmp_path: Path) -> None:
+    """The native build writes no PlatformIO extra scripts."""
+    CORE.build_path = tmp_path
+    esp8266.copy_files()
+    assert list(tmp_path.iterdir()) == []

tests/unit_tests/core/test_config.py

@@ -1429,12 +1429,3 @@ async def test_add_platformio_options_native_arduino(
     assert "board_build.ldscript is ignored" in caplog.text
     assert "'arduino' toolchain" in caplog.text
     assert "upload_speed" not in caplog.text
-
-
-def test_esp8266_rejects_unsupported_cli_toolchain() -> None:
-    """Until the native backend lands, ESP8266 serves only PlatformIO."""
-    from esphome.components.esp8266 import CONFIG_SCHEMA
-
-    CORE.toolchain = Toolchain.ARDUINO
-    with pytest.raises(cv.Invalid, match="Unsupported toolchain 'arduino'"):
-        CONFIG_SCHEMA({"board": "nodemcuv2"})

tests/unit_tests/fixtures/lazy_imports/native_upload_fast_path.py

@@ -0,0 +1,65 @@
+"""Run the native-toolchain serial-upload path and report heavy modules.
+
+Executed as a subprocess by test_lazy_imports.py: heavy module names come
+in on argv, the ones found in sys.modules afterwards go out on stdout.
+``upload_using_esptool`` dispatches native toolchains through a
+toolchain-keyed table; if someone routes it back through the platform
+component packages (esp32 or esp8266), this reports the leak.
+"""
+
+import os
+from pathlib import Path
+import sys
+import tempfile
+from unittest.mock import patch
+
+from _leak_report import print_leaked_modules
+
+from esphome.__main__ import upload_using_esptool
+from esphome.const import (
+    CONF_ESPHOME,
+    KEY_CORE,
+    KEY_ESP32,
+    KEY_TARGET_PLATFORM,
+    KEY_VARIANT,
+    Toolchain,
+)
+from esphome.core import CORE
+
+# An ambient ESPHOME_USE_SUBPROCESS would route past the patched
+# run_external_command into run_external_process and confuse the checks.
+os.environ.pop("ESPHOME_USE_SUBPROCESS", None)
+
+config = {CONF_ESPHOME: {"platformio_options": {}}}
+
+with tempfile.TemporaryDirectory() as build_dir:
+    CORE.name = "leaktest"
+    CORE.build_path = build_dir
+
+    for platform, toolchain, backend in (
+        ("esp8266", Toolchain.ARDUINO, "esphome.arduino8266.toolchain"),
+        ("esp32", Toolchain.ESP_IDF, "esphome.espidf.toolchain"),
+    ):
+        CORE.data[KEY_CORE] = {KEY_TARGET_PLATFORM: platform}
+        if platform == "esp32":
+            CORE.data[KEY_ESP32] = {KEY_VARIANT: "ESP32S3"}
+        CORE.toolchain = toolchain
+
+        import importlib
+
+        image = importlib.import_module(backend).get_factory_firmware_path()
+        image.parent.mkdir(parents=True, exist_ok=True)
+        image.write_bytes(b"\x00")
+
+        with patch("esphome.__main__.run_external_command", return_value=0) as mock_run:
+            rc = upload_using_esptool(config, "/dev/ttyUSB0", None, None)
+
+        # Fail loudly if the upload path stopped doing its work; otherwise
+        # an empty leak list could just mean nothing ran.
+        if rc != 0:
+            sys.exit(f"upload_using_esptool({platform}) returned {rc}")
+        cmd = list(mock_run.call_args[0][1:])
+        if str(image) not in [str(Path(c)) for c in cmd]:
+            sys.exit(f"native factory image did not reach esptool: {cmd}")
+
+print_leaked_modules()

tests/unit_tests/test_arduino8266_toolchain.py

@@ -0,0 +1,657 @@
+"""Tests for esphome.arduino8266.toolchain (the ninja build driver)."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+from unittest.mock import MagicMock, patch
+
+import pytest
+
+from esphome.arduino8266 import framework, toolchain
+from esphome.build_helpers.pch import mark_pch_emitted
+import esphome.config_validation as cv
+from esphome.const import (
+    CONF_COMPILE_PROCESS_LIMIT,
+    CONF_ESPHOME,
+    KEY_CORE,
+    KEY_FRAMEWORK_VERSION,
+)
+from esphome.core import CORE, EsphomeError
+
+_SIZE_OUTPUT = """\
+firmware.elf  :
+section       size    addr
+.data         1924    1073643520
+.noinit       56      1073645444
+.text         496     1074790400
+.irom0.text   342804  1075843088
+.text1        27489   1074790896
+.rodata       2588    1073645504
+.bss          26504   1073648096
+Total         401861
+"""
+
+
+@pytest.fixture(autouse=True)
+def _setup_core(tmp_path: Path) -> None:
+    CORE.name = "test8266"
+    CORE.config_path = tmp_path / "test8266.yaml"
+    CORE.build_path = tmp_path
+    CORE.data[KEY_CORE] = {KEY_FRAMEWORK_VERSION: cv.Version(3, 1, 2)}
+    # run_compile verifies the produced artifacts; give every test a build
+    # that "produced" them (tests for the guard delete them again). The
+    # manifest comes first: artifacts must not be older than build.ninja.
+    build_dir = CORE.relative_pioenvs_path("test8266")
+    build_dir.mkdir(parents=True, exist_ok=True)
+    (build_dir / "build.ninja").write_text("# manifest")
+    for artifact in (
+        "firmware.elf",
+        "firmware.bin",
+        "firmware.factory.bin",
+        "firmware.ota.bin",
+    ):
+        (build_dir / artifact).write_bytes(b"")
+
+
+def _paths(tmp_path: Path) -> framework.InstalledPaths:
+    return framework.InstalledPaths(
+        framework=tmp_path / "framework",
+        toolchain=tmp_path / "toolchain",
+        ninja=tmp_path / "ninja",
+    )
+
+
+def test_path_getters(tmp_path: Path) -> None:
+    assert toolchain.get_build_dir() == CORE.relative_pioenvs_path("test8266")
+    assert toolchain.get_elf_path().name == "firmware.elf"
+    # The framework accessor owns the layout and the Windows suffix
+    suffix = ".exe" if os.name == "nt" else ""
+    assert toolchain.get_addr2line_path().name == f"xtensa-lx106-elf-addr2line{suffix}"
+    assert toolchain.get_objdump_path().name == f"xtensa-lx106-elf-objdump{suffix}"
+    assert toolchain.get_readelf_path().name == f"xtensa-lx106-elf-readelf{suffix}"
+
+
+def test_run_compile_build_failure(tmp_path: Path) -> None:
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project"),
+        patch.object(
+            toolchain.subprocess, "run", return_value=MagicMock(returncode=2)
+        ) as mock_run,
+        patch.object(toolchain, "_write_compile_commands") as mock_compdb,
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=True) == 2
+    cmd = mock_run.call_args[0][0]
+    assert "-v" in cmd
+    # The compile database is generated before the build runs, so a failed
+    # build cannot leave a stale database behind.
+    mock_compdb.assert_called_once()
+
+
+def test_run_compile_success(tmp_path: Path) -> None:
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        # An unchanged manifest is what makes the -n probe run
+        patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout="", stderr=""),
+        ) as mock_run,
+        patch.object(toolchain, "_write_compile_commands") as mock_compdb,
+        patch.object(toolchain, "_print_size_summary") as mock_size,
+        patch.object(toolchain, "get_idedata") as mock_idedata,
+    ):
+        rc = toolchain.run_compile(
+            {CONF_ESPHOME: {CONF_COMPILE_PROCESS_LIMIT: 4}}, verbose=False
+        )
+    assert rc == 0
+    # The -n probe runs first, then the real build (cwd, no -C banner)
+    ninja_calls = [c for c in mock_run.call_args_list if "ninja" in str(c[0][0][0])]
+    assert "-n" in ninja_calls[0][0][0]
+    # Explicit targets: a manifest missing them fails as "unknown target"
+    assert ninja_calls[0][0][0][-1] == "firmware.ota.bin"
+    cmd = ninja_calls[1][0][0]
+    assert cmd[-4:] == ["-j", "4", "firmware.factory.bin", "firmware.ota.bin"]
+    assert "-C" not in cmd
+    assert ninja_calls[1][1]["cwd"] is not None
+    mock_compdb.assert_called_once()
+    mock_size.assert_called_once()
+    mock_idedata.assert_called_once()
+
+
+def test_run_compile_noop_skips_the_build_spawn(tmp_path: Path) -> None:
+    """A no-op rebuild stays quiet: the -n probe answers "no work to do"
+    and the real ninja spawn (and its banner) never happens."""
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        # An unchanged manifest is what makes the -n probe run
+        patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(
+                returncode=0, stdout="ninja: no work to do.\n", stderr=""
+            ),
+        ) as mock_run,
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary"),
+        patch.object(toolchain, "get_idedata"),
+    ):
+        rc = toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False)
+    assert rc == 0
+    ninja_calls = [c for c in mock_run.call_args_list if "ninja" in str(c[0][0][0])]
+    assert len(ninja_calls) == 1
+    assert "-n" in ninja_calls[0][0][0]
+    # Explicit targets: a manifest missing them fails as "unknown target"
+    assert ninja_calls[0][0][0][-1] == "firmware.ota.bin"
+
+
+def test_run_compile_regenerates_stale_compdb(tmp_path: Path) -> None:
+    """An interrupted run can leave build.ninja newer than the compile DB;
+    mere existence must not skip regeneration."""
+    build_dir = toolchain.get_build_dir()
+    build_dir.mkdir(parents=True, exist_ok=True)
+    (build_dir / "build.ninja").write_text("")
+    compdb = build_dir / "compile_commands.json"
+    compdb.write_text("[]")
+    os.utime(compdb, ((build_dir / "build.ninja").stat().st_mtime - 5,) * 2)
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(
+                returncode=0, stdout="ninja: no work to do.\n", stderr=""
+            ),
+        ),
+        patch.object(toolchain, "_write_compile_commands") as mock_compdb,
+        patch.object(toolchain, "_print_size_summary"),
+        patch.object(toolchain, "get_idedata"),
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+    mock_compdb.assert_called_once()
+
+
+def test_run_compile_surfaces_probe_diagnostics(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A load-time ninja diagnostic (a generator bug signal) reaches the
+    user even when the no-work branch skips the real spawn."""
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        # An unchanged manifest is what makes the -n probe run
+        patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(
+                returncode=0,
+                stdout="ninja: no work to do.\n",
+                stderr="ninja: warning: multiple rules generate x\n",
+            ),
+        ),
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary"),
+        patch.object(toolchain, "get_idedata"),
+    ):
+        rc = toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False)
+    assert rc == 0
+    assert "multiple rules generate x" in caplog.text
+
+
+def test_run_compile_missing_artifact_fails(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A zero ninja exit that produced no firmware must not be a green
+    build (size summary and idedata only warn)."""
+    (toolchain.get_build_dir() / "firmware.elf").unlink()
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project"),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout="", stderr=""),
+        ),
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary") as mock_size,
+        patch.object(toolchain, "get_idedata"),
+    ):
+        rc = toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False)
+    assert rc == 1
+    assert "Build produced no" in caplog.text
+    mock_size.assert_not_called()
+
+
+def test_run_compile_warns_when_idedata_fails(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A failed idedata generation right after a successful build is visible,
+    not deferred to a misleading error in a later command."""
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project"),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout="", stderr=""),
+        ),
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary"),
+        patch.object(toolchain, "get_idedata", return_value=None),
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+    assert "No idedata was generated for this build" in caplog.text
+
+
+def test_write_compile_commands(tmp_path: Path) -> None:
+    build_dir = tmp_path / "build"
+    build_dir.mkdir()
+    entries = '[{"file": "a.cpp", "command": "cc"}]\n'
+    with patch.object(
+        toolchain.subprocess,
+        "run",
+        return_value=MagicMock(returncode=0, stdout=entries),
+    ):
+        toolchain._write_compile_commands(tmp_path / "ninja", build_dir, {})
+    assert (build_dir / "compile_commands.json").read_text() == entries
+
+
+@pytest.mark.parametrize(
+    ("stdout", "match"),
+    [
+        ("[]\n", "empty compile database"),
+        # A parse failure names its cause, not the rule-name story
+        ("not json", "unparsable compile database.*not json"),
+    ],
+)
+def test_write_compile_commands_bad_db_raises(
+    tmp_path: Path, stdout: str, match: str
+) -> None:
+    """An empty or unparsable compile database fails the build with its
+    actual cause and drops any stale database."""
+    build_dir = tmp_path / "build"
+    build_dir.mkdir()
+    (build_dir / "compile_commands.json").write_text("[stale]")
+    with (
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout=stdout),
+        ),
+        pytest.raises(EsphomeError, match=match),
+    ):
+        toolchain._write_compile_commands(tmp_path / "ninja", build_dir, {})
+    assert not (build_dir / "compile_commands.json").exists()
+
+
+def test_write_compile_commands_failure_removes_stale_db(tmp_path: Path) -> None:
+    """A failed compdb run must not leave a stale database behind."""
+    stale = tmp_path / "compile_commands.json"
+    stale.write_text("[]")
+    with (
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=1, stderr="boom"),
+        ),
+        pytest.raises(EsphomeError, match="compile_commands"),
+    ):
+        toolchain._write_compile_commands(tmp_path / "ninja", tmp_path, {})
+    assert not stale.exists()
+
+
+def test_parse_app_size(tmp_path: Path) -> None:
+    ld = tmp_path / "eagle.flash.4m.ld"
+    ld.write_text("MEMORY\n{\n  irom0_0_seg :  org = 0x40201010, len = 0xfeff0\n}\n")
+    with patch("esphome.build_gen.arduino8266.get_flash_ld_path", return_value=ld):
+        assert toolchain._parse_app_size(tmp_path, _paths(tmp_path)) == 0xFEFF0
+
+    ld.write_text("MEMORY { }\n")
+    with patch("esphome.build_gen.arduino8266.get_flash_ld_path", return_value=ld):
+        assert toolchain._parse_app_size(tmp_path, _paths(tmp_path)) is None
+
+    # A zero-length segment is bad data, not a budget; warn and drop it
+    ld.write_text("MEMORY\n{\n  irom0_0_seg :  org = 0x40201010, len = 0x0\n}\n")
+    with patch("esphome.build_gen.arduino8266.get_flash_ld_path", return_value=ld):
+        assert toolchain._parse_app_size(tmp_path, _paths(tmp_path)) is None
+
+    with patch(
+        "esphome.build_gen.arduino8266.get_flash_ld_path",
+        return_value=tmp_path / "missing.ld",
+    ):
+        assert toolchain._parse_app_size(tmp_path, _paths(tmp_path)) is None
+
+
+def test_print_size_summary(tmp_path: Path, capsys: pytest.CaptureFixture[str]) -> None:
+    with (
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout=_SIZE_OUTPUT),
+        ),
+        patch.object(toolchain, "_parse_app_size", return_value=1044464),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    out = capsys.readouterr().out
+    # Exact PlatformIO shape so script/ci_memory_impact_extract.py can parse it
+    assert "RAM:   [====      ]  37.9% (used 31016 bytes from 81920 bytes)" in out
+    assert "Flash: [====      ]  35.9% (used 375301 bytes from 1044464 bytes)" in out
+
+
+def test_print_size_summary_missing_size_tool_warns(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A missing size binary degrades to a warning; the firmware already
+    linked and must not be discarded."""
+    with patch.object(
+        toolchain.subprocess, "run", side_effect=FileNotFoundError("no size")
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    assert "Could not summarize firmware size" in caplog.text
+
+
+def test_print_size_summary_no_app_size(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    with (
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout=_SIZE_OUTPUT),
+        ),
+        patch.object(toolchain, "_parse_app_size", return_value=None),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    out = capsys.readouterr().out
+    # Both lines are skipped together: a RAM line without Flash would skew
+    # CI's memory-impact sums across builds
+    assert out == ""
+
+
+def test_print_size_summary_size_tool_failure(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    with patch.object(
+        toolchain.subprocess,
+        "run",
+        return_value=MagicMock(returncode=1, stdout="", stderr="bad elf"),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    assert capsys.readouterr().out == ""
+    assert "Could not summarize firmware size" in caplog.text
+
+
+def test_get_idedata_delegates(tmp_path: Path) -> None:
+    with (
+        patch(
+            "esphome.build_helpers.idedata.load_or_build_idedata",
+            return_value={"cc_path": "x"},
+        ) as mock_load,
+        patch.object(toolchain, "resolve_ccache_path", return_value="/cc/ccache"),
+    ):
+        assert toolchain.get_idedata() == {"cc_path": "x"}
+    compile_commands, elf, cache = mock_load.call_args[0]
+    assert compile_commands.name == "compile_commands.json"
+    assert elf.name == "firmware.elf"
+    assert cache.name == "test8266.arduino.json"
+    # The exact configured launcher string is passed for compile DB parsing
+    # (resolve_ccache_path returns a str, untouched on every platform)
+    assert mock_load.call_args.kwargs["launcher"] == "/cc/ccache"
+
+
+def test_get_idedata_no_ccache(tmp_path: Path) -> None:
+    with (
+        patch(
+            "esphome.build_helpers.idedata.load_or_build_idedata", return_value={}
+        ) as mock_load,
+        patch.object(toolchain, "resolve_ccache_path", return_value=None),
+    ):
+        toolchain.get_idedata()
+    assert mock_load.call_args.kwargs["launcher"] is None
+
+
+def test_run_compile_skips_compdb_when_ninja_unchanged(tmp_path: Path) -> None:
+    """An unchanged build.ninja means the compile DB is already current."""
+    build_dir = toolchain.get_build_dir()
+    build_dir.mkdir(parents=True, exist_ok=True)
+    # write_project (stubbed below) always leaves a build.ninja behind
+    (build_dir / "build.ninja").write_text("# manifest")
+
+    def run(regenerate_expected: bool) -> None:
+        with (
+            patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+            patch.object(framework, "get_build_env", return_value={}),
+            patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+            patch.object(
+                toolchain.subprocess,
+                "run",
+                return_value=MagicMock(returncode=0, stdout="", stderr=""),
+            ),
+            patch.object(toolchain, "_write_compile_commands") as mock_compdb,
+            patch.object(toolchain, "_print_size_summary"),
+            patch.object(toolchain, "get_idedata"),
+        ):
+            assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+        assert mock_compdb.called == regenerate_expected
+
+    # Missing compile DB: regenerated even though build.ninja is unchanged
+    run(regenerate_expected=True)
+    # Present compile DB + unchanged build.ninja: skipped
+    (build_dir / "compile_commands.json").write_text("[]")
+    run(regenerate_expected=False)
+
+
+def test_print_size_summary_unparsable_section(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A totals-relevant section that fails to parse must not produce a
+    confident wrong number; an irrelevant one only warns."""
+    bad = _SIZE_OUTPUT.replace(".bss          26504", ".bss          abc")
+    with patch.object(
+        toolchain.subprocess,
+        "run",
+        return_value=MagicMock(returncode=0, stdout=bad),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    assert capsys.readouterr().out == ""
+    assert "Unparsable size output" in caplog.text
+
+    caplog.clear()
+    harmless = _SIZE_OUTPUT + ".broken   abc   0\n"
+    with (
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout=harmless),
+        ),
+        patch.object(toolchain, "_parse_app_size", return_value=1044464),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    assert "RAM:" in capsys.readouterr().out
+    assert "Unparsable size output" in caplog.text
+
+
+def test_print_size_summary_missing_section_skips_summary(
+    tmp_path: Path,
+    capsys: pytest.CaptureFixture[str],
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A totals section absent from the output must not default to zero."""
+    without_bss = "\n".join(
+        line for line in _SIZE_OUTPUT.splitlines() if ".bss" not in line
+    )
+    with patch.object(
+        toolchain.subprocess,
+        "run",
+        return_value=MagicMock(returncode=0, stdout=without_bss),
+    ):
+        toolchain._print_size_summary(tmp_path, _paths(tmp_path))
+    assert capsys.readouterr().out == ""
+    assert "missing section(s) .bss" in caplog.text
+
+
+def test_warn_ignored_platformio_options(caplog: pytest.LogCaptureFixture) -> None:
+    """Component-added options the native build drops are warned by name;
+    the honored ones (lib_ignore, f_cpu, ldscript, build_src_flags,
+    flash_mode) stay quiet."""
+    CORE.platformio_options = {
+        "board_build.ldscript": "eagle.flash.4m2m.ld",
+        "board_build.f_cpu": "160000000L",
+        "board_build.filesystem": "littlefs",
+        "board_build.flash_mode": "dio",
+        "build_src_flags": "-include throw_stubs.h",
+        "lib_ignore": ["Updater"],
+        "upload_speed": "460800",
+    }
+    toolchain._warn_ignored_platformio_options()
+    assert "platformio_options->board_build.filesystem is ignored" in caplog.text
+    assert "native 'arduino' toolchain" in caplog.text
+    assert "board_build.ldscript is ignored" not in caplog.text
+    assert "board_build.f_cpu is ignored" not in caplog.text
+    assert "lib_ignore" not in caplog.text
+    assert "build_src_flags" not in caplog.text
+    assert "flash_mode" not in caplog.text
+    # Component-added upload_speed never gets read under the native
+    # toolchain, so it must warn
+    assert "upload_speed" in caplog.text
+
+
+def test_run_compile_idedata_error_does_not_fail_build(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """An unusable compile DB after a successful build warns, never fails."""
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project"),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout="", stderr=""),
+        ),
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary"),
+        patch.object(
+            toolchain,
+            "get_idedata",
+            side_effect=EsphomeError("compile database is unusable"),
+        ),
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+    assert "Could not generate idedata: compile database is unusable" in caplog.text
+
+
+def test_run_compile_skipped_size_summary_names_consequence(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A skipped RAM/Flash summary warns about the missing metric; the
+    per-cause warnings alone are invisible to CI harnesses."""
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project"),
+        patch.object(
+            toolchain.subprocess,
+            "run",
+            return_value=MagicMock(returncode=0, stdout="", stderr=""),
+        ),
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary", return_value=False),
+        patch.object(toolchain, "get_idedata", return_value=None),
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+    assert "Firmware size summary unavailable for this build" in caplog.text
+
+
+def test_parse_app_size_non_utf8_ld_warns(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A corrupt (non-UTF-8) linker script degrades to the same warning as
+    an unreadable one, never a traceback after a successful link."""
+    paths = _paths(tmp_path)
+    ld = tmp_path / "corrupt.ld"
+    ld.write_bytes(b"\xff\xfe not utf8")
+    with patch("esphome.build_gen.arduino8266.get_flash_ld_path", return_value=ld):
+        assert toolchain._parse_app_size(tmp_path, paths) is None
+    assert "Cannot read linker script" in caplog.text
+
+
+def test_run_compile_failed_probe_runs_full_build(tmp_path: Path) -> None:
+    """A failing -n probe (e.g. unknown target from a defective manifest)
+    falls through to the real build so the error prints attributably."""
+    probe = MagicMock(returncode=1, stdout="", stderr="")
+    ok = MagicMock(returncode=0, stdout="", stderr="")
+
+    def fake_run(cmd, *args, **kwargs):
+        return probe if "-n" in cmd else ok
+
+    with (
+        patch.object(framework, "check_and_install", return_value=_paths(tmp_path)),
+        patch.object(framework, "get_build_env", return_value={}),
+        patch("esphome.build_gen.arduino8266.write_project", return_value=False),
+        patch.object(toolchain.subprocess, "run", side_effect=fake_run) as mock_run,
+        patch.object(toolchain, "_write_compile_commands"),
+        patch.object(toolchain, "_print_size_summary", return_value=True),
+        patch.object(toolchain, "get_idedata", return_value={}),
+    ):
+        assert toolchain.run_compile({CONF_ESPHOME: {}}, verbose=False) == 0
+    # The real build ran after the failed probe
+    assert any(
+        "firmware.ota.bin" in c[0][0] and "-n" not in c[0][0]
+        for c in mock_run.call_args_list
+    )
+
+
+def test_get_idedata_accepts_preresolved_ccache() -> None:
+    """run_compile threads its resolved ccache through; the probe must not
+    run again."""
+    with (
+        patch(
+            "esphome.build_helpers.idedata.load_or_build_idedata",
+            return_value={"ok": True},
+        ) as mock_build,
+        patch.object(toolchain, "resolve_ccache_path") as mock_resolve,
+    ):
+        assert toolchain.get_idedata("/usr/bin/ccache") == {"ok": True}
+    mock_resolve.assert_not_called()
+    assert mock_build.call_args.kwargs["launcher"] == "/usr/bin/ccache"
+
+
+def test_ccache_env_includes_pch_settings() -> None:
+    """The native build exports the ccache settings the pch needs."""
+    mark_pch_emitted()
+    with patch.dict(os.environ, {}, clear=True):
+        env = framework.ccache_env("/usr/bin/ccache")
+    assert env["CCACHE_SLOPPINESS"] == "pch_defines,time_macros"
+    assert env["CCACHE_PCH_EXTSUM"] == "true"
+
+
+def test_ccache_env_pch_disabled() -> None:
+    with patch.dict(os.environ, {"ESPHOME_PCH_ENABLE": "0"}, clear=True):
+        env = framework.ccache_env("/usr/bin/ccache")
+    assert "CCACHE_SLOPPINESS" not in env
+    assert "CCACHE_PCH_EXTSUM" not in env
+
+
+def test_ccache_env_respects_user_sloppiness() -> None:
+    mark_pch_emitted()
+    with patch.dict(os.environ, {"CCACHE_SLOPPINESS": "locale"}, clear=True):
+        env = framework.ccache_env("/usr/bin/ccache")
+    # The user's tokens survive; the ones the pch needs are unioned on
+    assert env["CCACHE_SLOPPINESS"] == "locale,pch_defines,time_macros"
+    assert env["CCACHE_PCH_EXTSUM"] == "true"

tests/unit_tests/test_lazy_imports.py

@@ -175,6 +175,27 @@ def test_esptool_upload_fast_path_does_not_import_heavy_modules(
     )
 
 
+def test_native_upload_fast_path_does_not_import_heavy_modules(
+    fixture_path: Path,
+    probe_env: dict[str, str],
+) -> None:
+    """The native-toolchain serial upload dispatches through a
+    toolchain-keyed backend table; resolving the flash image must not drag
+    in the esp32 or esp8266 component package or the validation stack.
+    """
+    leaked = _leaked_from_fixture(
+        fixture_path,
+        probe_env,
+        "native_upload_fast_path.py",
+        extra=("esphome.components.esp8266",),
+    )
+    assert not leaked, (
+        f"native upload_using_esptool pulls in heavy modules: {leaked}. "
+        "The upload fast path skips validation; importing a platform "
+        "component package executes its codegen module by design."
+    )
+
+
 def test_api_client_does_not_import_heavy_modules() -> None:
     """``esphome.api_client`` is on the logs fast path and must stay light.
 

tests/unit_tests/test_main.py

@@ -13,7 +13,7 @@ import sys
 import time
 from types import SimpleNamespace
 from typing import Any, Self
-from unittest.mock import AsyncMock, MagicMock, Mock, patch
+from unittest.mock import AsyncMock, MagicMock, Mock, PropertyMock, patch
 
 import pytest
 from pytest import CaptureFixture
@@ -25,6 +25,7 @@ from esphome.__main__ import (
     Purpose,
     _get_configured_xtal_freq,
     _make_crystal_freq_callback,
+    _native_toolchain_module,
     _redact_with_legacy_fallback,
     _resolve_network_devices,
     _should_subscribe_states,
@@ -7030,7 +7031,7 @@ def test_command_run_rp2040_bootsel_redetects_serial_port() -> None:
 
 def test_command_idedata_esp_idf_prints_json(capsys: CaptureFixture) -> None:
     """Under the native ESP-IDF toolchain, idedata is emitted as JSON."""
-    setup_core()
+    setup_core(platform=PLATFORM_ESP32)
     CORE.toolchain = Toolchain.ESP_IDF
     data = {"cxx_path": "g++", "prog_path": "/build/firmware.elf"}
 
@@ -7044,7 +7045,7 @@ def test_command_idedata_esp_idf_prints_json(capsys: CaptureFixture) -> None:
 
 def test_command_idedata_esp_idf_no_build_errors() -> None:
     """Under ESP-IDF, a missing build (no idedata) returns an error, not a crash."""
-    setup_core()
+    setup_core(platform=PLATFORM_ESP32)
     CORE.toolchain = Toolchain.ESP_IDF
 
     with patch("esphome.espidf.toolchain.get_idedata", return_value=None):
@@ -7218,6 +7219,201 @@ def test_warn_source_tree_mismatch_falls_back_when_stat_fails(
     assert not caplog.text
 
 
+def test_upload_using_esptool_arduino_toolchain(
+    tmp_path: Path,
+    mock_run_external_command_main: Mock,
+) -> None:
+    """The native ESP8266 Arduino toolchain flashes its factory image at
+    0x0, resolved from the toolchain-keyed backend table (deliberately not
+    the platform hook: that import would break the upload fast path)."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path, name="test")
+    CORE.toolchain = Toolchain.ARDUINO
+    from esphome.arduino8266 import toolchain as native
+
+    factory = native.get_factory_firmware_path()
+    factory.parent.mkdir(parents=True, exist_ok=True)
+    factory.touch()
+
+    config = {CONF_ESPHOME: {"platformio_options": {}}}
+    result = upload_using_esptool(config, "/dev/ttyUSB0", None, None)
+
+    assert result == 0
+    cmd_list = list(mock_run_external_command_main.call_args[0][1:])
+    firmware_offset_idx = cmd_list.index("write-flash") + 4
+    assert cmd_list[firmware_offset_idx] == "0x0"
+    assert cmd_list[firmware_offset_idx + 1] == str(factory)
+
+
+@pytest.mark.parametrize(
+    ("toolchain", "pio_project_written"),
+    [
+        # The native toolchain generates its project at compile time, so
+        # write_cpp_file must not write a platformio.ini; the default
+        # toolchain writes the PlatformIO project files.
+        (Toolchain.ARDUINO, False),
+        (None, True),
+    ],
+)
+def test_write_cpp_file_project_generation_follows_toolchain(
+    tmp_path: Path, toolchain: Toolchain | None, pio_project_written: bool
+) -> None:
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path, name="test")
+    CORE.toolchain = toolchain
+
+    with (
+        patch("esphome.writer.write_cpp") as mock_write_cpp,
+        patch("esphome.build_gen.platformio.write_project") as mock_pio_project,
+        patch.object(
+            type(CORE), "cpp_main_section", new_callable=PropertyMock
+        ) as mock_section,
+    ):
+        mock_section.return_value = ""
+        assert main.write_cpp_file() == 0
+
+    mock_write_cpp.assert_called_once()
+    assert mock_pio_project.called is pio_project_written
+
+
+def test_command_idedata_arduino_prints_json(
+    tmp_path: Path, capsys: CaptureFixture
+) -> None:
+    """Under the native ESP8266 Arduino toolchain, idedata is emitted as JSON."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path)
+    CORE.toolchain = Toolchain.ARDUINO
+    data = {"cxx_path": "g++", "prog_path": "/build/firmware.elf"}
+
+    with patch(
+        "esphome.arduino8266.toolchain.get_idedata", return_value=data
+    ) as mock_get:
+        result = command_idedata(MagicMock(), CORE.config)
+
+    assert result == 0
+    mock_get.assert_called_once_with()
+    assert json.loads(capsys.readouterr().out) == data
+
+
+def test_command_idedata_arduino_no_build_errors(tmp_path: Path) -> None:
+    """A missing native build (no idedata) returns an error, not a crash."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path)
+    CORE.toolchain = Toolchain.ARDUINO
+
+    with patch("esphome.arduino8266.toolchain.get_idedata", return_value=None):
+        result = command_idedata(MagicMock(), CORE.config)
+
+    assert result == 1
+
+
+@pytest.mark.parametrize(
+    ("platform", "toolchain", "module"),
+    [
+        (PLATFORM_ESP8266, Toolchain.ARDUINO, "esphome.arduino8266.toolchain"),
+        (PLATFORM_ESP32, Toolchain.ESP_IDF, "esphome.espidf.toolchain"),
+    ],
+)
+def test_command_analyze_memory_native_toolchains(
+    tmp_path: Path,
+    mock_write_cpp: Mock,
+    mock_compile_program: Mock,
+    mock_get_esphome_components: Mock,
+    mock_memory_analyzer_cli: Mock,
+    mock_ram_strings_analyzer: Mock,
+    platform: str,
+    toolchain: Toolchain,
+    module: str,
+) -> None:
+    """analyze-memory uses the native toolchain's binutils instead of
+    falling into the PlatformIO branch."""
+    setup_core(platform=platform, tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = toolchain
+
+    config = {CONF_ESPHOME: {CONF_NAME: "test_device"}}
+    # The tools must exist: a missing binutils now fails by name instead of
+    # silently falling back to host tools
+    objdump = tmp_path / "objdump"
+    readelf = tmp_path / "readelf"
+    objdump.write_text("")
+    readelf.write_text("")
+    # The ELF must exist too: the analyzer swallows tool failures, so a
+    # missing image would report zeroes with exit 0
+    firmware_elf = tmp_path / "firmware.elf"
+    firmware_elf.write_text("")
+    with (
+        patch(f"{module}.get_objdump_path", return_value=objdump),
+        patch(f"{module}.get_readelf_path", return_value=readelf),
+        patch(f"{module}.get_elf_path", return_value=firmware_elf),
+    ):
+        result = command_analyze_memory(MockArgs(), config)
+
+    assert result == 0
+    mock_memory_analyzer_cli.assert_called_once_with(
+        str(firmware_elf),
+        str(objdump),
+        str(readelf),
+        set(),
+        idedata=None,
+    )
+
+
+def test_command_analyze_memory_native_missing_elf_fails(
+    tmp_path: Path,
+    mock_write_cpp: Mock,
+    mock_compile_program: Mock,
+    mock_get_esphome_components: Mock,
+    mock_memory_analyzer_cli: Mock,
+    mock_ram_strings_analyzer: Mock,
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A missing firmware.elf fails by name instead of an exit-0 zeroed
+    report."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = Toolchain.ARDUINO
+
+    config = {CONF_ESPHOME: {CONF_NAME: "test_device"}}
+    objdump = tmp_path / "objdump"
+    readelf = tmp_path / "readelf"
+    objdump.write_text("")
+    readelf.write_text("")
+    module = "esphome.arduino8266.toolchain"
+    with (
+        patch(f"{module}.get_objdump_path", return_value=objdump),
+        patch(f"{module}.get_readelf_path", return_value=readelf),
+        patch(f"{module}.get_elf_path", return_value=tmp_path / "missing.elf"),
+    ):
+        result = command_analyze_memory(MockArgs(), config)
+
+    assert result == 1
+    assert "compile the configuration first" in caplog.text
+    mock_memory_analyzer_cli.assert_not_called()
+
+
+def test_command_analyze_memory_missing_binutils_fails_by_name(
+    tmp_path: Path, caplog: pytest.LogCaptureFixture
+) -> None:
+    """A truncated toolchain install fails naming the missing tool instead
+    of silently analyzing with host binutils."""
+    setup_core(platform="esp8266", tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = Toolchain.ARDUINO
+    config = {CONF_ESPHOME: {CONF_NAME: "test_device"}}
+    module = "esphome.arduino8266.toolchain"
+    with (
+        patch(f"{module}.get_objdump_path", return_value=tmp_path / "missing-objdump"),
+        patch(f"{module}.get_readelf_path", return_value=tmp_path / "readelf"),
+        patch("esphome.__main__.write_cpp", return_value=0),
+        patch("esphome.__main__.compile_program", return_value=0),
+    ):
+        assert command_analyze_memory(MockArgs(), config) == 1
+    assert "missing-objdump" in caplog.text
+    assert "toolchain install may be incomplete" in caplog.text
+
+
+def test_command_idedata_incompatible_toolchain(tmp_path: Path) -> None:
+    """A non-native, non-platformio toolchain errors out cleanly."""
+    setup_core(platform=PLATFORM_ESP32, tmp_path=tmp_path)
+    CORE.toolchain = Toolchain.SDK_NRF
+
+    assert command_idedata(MagicMock(), CORE.config) == 1
+
+
 @pytest.mark.parametrize(
     "error",
     [
@@ -7294,6 +7490,46 @@ def test_compile_program_espidf_idedata_none_warns(
     assert "No idedata was generated" in caplog.text
 
 
+def test_native_toolchain_table_serves_every_native_toolchain() -> None:
+    """Every member of NATIVE_TOOLCHAINS has a backend entry; a gap would
+    surface as a targeted EsphomeError on the one affected config, and this
+    pin keeps the table from drifting when a toolchain is added."""
+    from esphome.const import NATIVE_TOOLCHAINS
+
+    assert {tc for _, tc in main._NATIVE_TOOLCHAIN_MODULES} == set(NATIVE_TOOLCHAINS)
+
+
+def test_native_toolchain_module_missing_backend_raises(tmp_path: Path) -> None:
+    """A native toolchain missing from the backend table is a bug and must
+    fail, not silently degrade to the PlatformIO path."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = Toolchain.ARDUINO
+    with (
+        patch.dict(main._NATIVE_TOOLCHAIN_MODULES, clear=True),
+        pytest.raises(EsphomeError, match="no native build backend"),
+    ):
+        _native_toolchain_module()
+
+
+def test_command_analyze_memory_unsupported_toolchain(
+    tmp_path: Path,
+    mock_write_cpp: Mock,
+    mock_compile_program: Mock,
+    caplog: pytest.LogCaptureFixture,
+) -> None:
+    """A hook-less non-PlatformIO toolchain is refused by name, never routed
+    into the PlatformIO branch."""
+    setup_core(platform=PLATFORM_NRF52, tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = Toolchain.SDK_NRF
+    mock_write_cpp.return_value = 0
+    mock_compile_program.return_value = 0
+
+    result = command_analyze_memory(MockArgs(), {CONF_ESPHOME: {CONF_NAME: "t"}})
+
+    assert result == 1
+    assert "analyze-memory is not supported" in caplog.text
+
+
 def test_cli_toolchain_skips_the_validated_config_cache(tmp_path: Path) -> None:
     """An explicit --toolchain must run the per-platform validators, so the
     upload/logs fast path becomes a cache miss."""
@@ -7309,6 +7545,29 @@ def test_cli_toolchain_skips_the_validated_config_cache(tmp_path: Path) -> None:
     mock_read.assert_called_once()
 
 
+def test_upload_using_esptool_native_missing_firmware_raises(
+    tmp_path: Path,
+) -> None:
+    """A stale or absent firmware.bin fails by name instead of flashing air."""
+    setup_core(platform=PLATFORM_ESP8266, tmp_path=tmp_path, name="test")
+    CORE.toolchain = Toolchain.ARDUINO
+    with pytest.raises(EsphomeError, match="compile the configuration first"):
+        upload_using_esptool(
+            {CONF_ESPHOME: {"platformio_options": {}}}, "/dev/ttyUSB0", None, None
+        )
+
+
+def test_compile_program_unclaimed_native_toolchain_raises(
+    tmp_path: Path,
+) -> None:
+    """A resolved native toolchain no platform backend claims must fail,
+    never fall through to the PlatformIO project path."""
+    setup_core(platform=PLATFORM_ESP32, tmp_path=tmp_path, name="test_device")
+    CORE.toolchain = Toolchain.ARDUINO  # esp32 has no arduino-native backend
+    with pytest.raises(EsphomeError, match="no platform backend claimed"):
+        compile_program(MockArgs(), {})
+
+
 def test_cli_toolchain_still_refreshes_the_validated_config_cache(
     tmp_path: Path,
 ) -> None:

tests/unit_tests/test_platformio_pch_script.py

@@ -0,0 +1,460 @@
+"""Tests for esphome/platformio/pch.py.script against a fake SCons env."""
+
+from __future__ import annotations
+
+import os
+from pathlib import Path
+import stat
+from unittest.mock import patch
+
+import pytest
+
+from esphome.platformio import toolchain
+
+pytestmark = pytest.mark.skipif(
+    os.name == "nt", reason="the fake compiler is a POSIX shell script"
+)
+
+_SCRIPT = Path(toolchain.__file__).parent / "pch.py.script"
+
+
+class _FakePlatform:
+    packages = {"framework-x": {}, "toolchain-y": {}}
+
+    def get_package_version(self, name: str) -> str:
+        if name == "toolchain-y":
+            raise KeyError(name)
+        return "1.2.3"
+
+    def get_package(self, name: str) -> object | None:
+        return None
+
+
+class _BrokenPlatform(_FakePlatform):
+    def get_package_version(self, name: str) -> str:
+        raise RuntimeError("manifest parse error")
+
+
+class _UnresolvedPlatform(_FakePlatform):
+    """KeyError from a package that IS installed: unresolved identity."""
+
+    def get_package_version(self, name: str) -> str:
+        raise KeyError(name)
+
+    def get_package(self, name: str) -> object:
+        return object()
+
+
+class _FakeSConsEnv(dict):
+    """Just enough of a SCons construction environment for pch.py."""
+
+    def __init__(
+        self,
+        proj_dir: Path,
+        src_dir: Path,
+        cxx: str,
+        flags: list[str],
+        platform_cls: type[_FakePlatform] = _FakePlatform,
+    ):
+        super().__init__(ENV={})
+        self._subst = {
+            "$PROJECT_DIR": str(proj_dir),
+            "$PROJECT_SRC_DIR": str(src_dir),
+            "$CXX": cxx,
+        }
+        self._flags = flags
+        self._platform_cls = platform_cls
+        self.prepended: list[str] = []
+
+    def subst(self, expr: str) -> str:  # noqa: N802
+        return self._subst[expr]
+
+    def subst_list(self, expr: str) -> list[list[str]]:  # noqa: N802
+        return [self._flags]
+
+    def PioPlatform(self) -> _FakePlatform:  # noqa: N802
+        return self._platform_cls()
+
+    def Prepend(self, CXXFLAGS: list[str]) -> None:  # noqa: N802, N803
+        self.prepended = CXXFLAGS
+
+
+def _fake_cxx(
+    tmp_path: Path,
+    fail: bool = False,
+    fail_msg: str | None = None,
+    reject_pch: bool = False,
+    probe_exit: int = 0,
+) -> Path:
+    """A compiler stand-in that records its argv and writes the -o target.
+
+    With reject_pch it builds the .gch fine but, like GCC 10 on macOS arm64,
+    warns on any consuming compile that the .gch cannot be loaded; probe_exit
+    sets the exit code of non-header compiles (the load probe).
+    """
+    cxx = tmp_path / "fake-gxx"
+    body = (
+        'printf -- ---call---\\\\n >> "$0.argv"; printf \'%s\\n\' "$@" >> "$0.argv"\n'
+    )
+    if fail:
+        body += f"echo {fail_msg or 'boom'} >&2\nexit 1\n"
+    else:
+        # Only the c++-header compile has a -o; the load probe has none
+        body += 'out=""; prev=""; mf=0; dep=0\nfor a in "$@"; do [ "$prev" = "-o" ] && out="$a"; prev="$a"; [ "$a" = "-MF" ] && mf=1; case "$a" in -M|-MM|-MD|-MMD) dep=1;; esac; done\n'
+        # Real cc1plus rejects -MF without a dependency flag
+        body += 'if [ "$mf" = 1 ] && [ "$dep" = 0 ]; then echo "cc1plus: error: to generate dependencies you must specify either \x27-M\x27 or \x27-MM\x27" >&2; exit 1; fi\n'
+        body += '[ -n "$out" ] && echo gch > "$out"\n'
+        if reject_pch:
+            body += 'case " $* " in *c++-header*) ;; *) echo "warning: esphome_pch.h.gch: had text segment at different address" >&2;; esac\n'
+        body += f'case " $* " in *c++-header*) exit 0;; *) exit {probe_exit};; esac\n'
+    cxx.write_text("#!/bin/sh\n" + body)
+    cxx.chmod(cxx.stat().st_mode | stat.S_IEXEC)
+    return cxx
+
+
+def _run_script(
+    tmp_path: Path,
+    flags: list[str] | None = None,
+    fail: bool = False,
+    fail_msg: str | None = None,
+    reject_pch: bool = False,
+    probe_exit: int = 0,
+    missing_cxx: bool = False,
+    env_vars: dict[str, str] | None = None,
+    name: str = "dev",
+    platform_cls: type[_FakePlatform] = _FakePlatform,
+) -> _FakeSConsEnv:
+    proj = tmp_path / name
+    src = proj / "src"
+    (src / "esphome" / "core").mkdir(parents=True, exist_ok=True)
+    (src / "esphome" / "core" / "defines.h").write_text("#define USE_X\n")
+    cxx = _fake_cxx(
+        tmp_path,
+        fail=fail,
+        fail_msg=fail_msg,
+        reject_pch=reject_pch,
+        probe_exit=probe_exit,
+    )
+    if missing_cxx:
+        cxx = tmp_path / "no-such-gxx"
+    args = (proj, src, str(cxx), flags or ["-DX=1"], platform_cls)
+    # Distinct objects: the -include flags must land on projenv only
+    global_env = _FakeSConsEnv(*args)
+    projenv = _FakeSConsEnv(*args)
+    projenv.global_env = global_env
+    source = _SCRIPT.read_text()
+    with patch.dict(os.environ, env_vars or {}, clear=True):
+        exec(  # noqa: S102
+            compile(source, "pch.py", "exec"),
+            {"Import": lambda *_names: None, "env": global_env, "projenv": projenv},
+        )
+    return projenv
+
+
+def test_pch_script_builds_and_prepends_relative_include(tmp_path: Path) -> None:
+    scons_env = _run_script(tmp_path)
+    proj = tmp_path / "dev"
+    assert (proj / "esphome_pch.h").read_text().endswith('"esphome/core/defines.h"\n')
+    assert (proj / "esphome_pch.h.gch").is_file()
+    assert len((proj / "esphome_pch.h.gch.sum").read_text().strip()) == 64
+    # Relative include: an absolute path would poison ccache keys
+    assert scons_env.prepended == [
+        "-Winvalid-pch",
+        "-Wno-error=invalid-pch",
+        "-include",
+        "esphome_pch.h",
+    ]
+    # In production projenv["ENV"] aliases os.environ; only the -include
+    # flags are genuinely scoped to projenv (src compiles)
+    assert scons_env["ENV"]["CCACHE_SLOPPINESS"] == "pch_defines,time_macros"
+    assert scons_env["ENV"]["CCACHE_PCH_EXTSUM"] == "true"
+    assert scons_env.global_env.prepended == []
+
+
+def test_pch_script_preserves_spaced_flag_elements(tmp_path: Path) -> None:
+    """One SCons element stays one compiler argv; -include pairs are
+    stripped from the .gch compile."""
+    spaced = tmp_path / "My Configs"
+    spaced.mkdir()
+    (tmp_path / "dev" / "src").mkdir(parents=True, exist_ok=True)
+    (tmp_path / "dev" / "src" / "other.h").write_text("")
+    flags = ['-DUSB_PRODUCT=\\"Pico 2W\\"', "-I", str(spaced), "-include", "other.h"]
+    _run_script(tmp_path, flags=flags)
+    calls = (tmp_path / "fake-gxx.argv").read_text().split("---call---\n")
+    gch_call = next(c for c in calls if "c++-header" in c).splitlines()
+    assert '-DUSB_PRODUCT="Pico 2W"' in gch_call
+    assert str(spaced) in gch_call
+    assert "-include" not in gch_call
+    # The stripped src-resolvable -include is folded into the prefix header
+    pch = (tmp_path / "dev" / "esphome_pch.h").read_text()
+    assert pch.splitlines()[0] == '#include "other.h"'
+
+
+def test_pch_script_folds_joined_force_include_spelling(tmp_path: Path) -> None:
+    """-includefoo.h folds like the separated form, matching the native path."""
+    (tmp_path / "dev" / "src").mkdir(parents=True, exist_ok=True)
+    (tmp_path / "dev" / "src" / "other.h").write_text("")
+    _run_script(tmp_path, flags=["-DX=1", "-includeother.h"])
+    pch = (tmp_path / "dev" / "esphome_pch.h").read_text()
+    assert pch.splitlines()[0] == '#include "other.h"'
+
+
+def test_pch_script_leaves_absolute_force_includes_unfolded(
+    tmp_path: Path,
+) -> None:
+    """An absolute -include resolves through src_dir / name; it must still
+    stay consumer-only or the host path enters the .sum."""
+    outside = tmp_path / "outside.h"
+    outside.write_text("")
+    _run_script(tmp_path, flags=["-DX=1", "-include", str(outside)])
+    pch = (tmp_path / "dev" / "esphome_pch.h").read_text()
+    assert "outside.h" not in pch
+
+
+def test_pch_script_leaves_non_src_force_includes_unfolded(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """A user -include outside src/ must not enter the prefix header:
+    consumers keep their own copy, so folding an unguarded header would
+    include it twice."""
+    _run_script(tmp_path, flags=["-DX=1", "-include", "user_extra.h"])
+    pch = (tmp_path / "dev" / "esphome_pch.h").read_text()
+    assert "user_extra.h" not in pch
+    assert pch.splitlines()[-1] == '#include "esphome/core/defines.h"'
+    assert "not precompiling non-src force-includes" in capsys.readouterr().out
+
+
+def test_pch_script_sum_is_device_independent(tmp_path: Path) -> None:
+    """Regression: identical configs in different dirs share cache keys."""
+    sums = []
+    for name in ("dev_a", "dev_b"):
+        proj = tmp_path / name
+        _run_script(
+            tmp_path,
+            flags=["-DX=1", "-I", str(proj / "include")],
+            env_vars={"CCACHE_BASEDIR": str(proj)},
+            name=name,
+        )
+        sums.append((proj / "esphome_pch.h.gch.sum").read_text())
+        (tmp_path / "fake-gxx").unlink()
+        (tmp_path / "fake-gxx.argv").unlink(missing_ok=True)
+    assert sums[0] == sums[1]
+
+
+def test_pch_script_transient_compiler_failure_does_not_latch(
+    tmp_path: Path,
+) -> None:
+    """ENOSPC-style failures clear on their own; no .failed marker."""
+    scons_env = _run_script(tmp_path, fail=True, fail_msg="No space left on device")
+    proj = tmp_path / "dev"
+    assert not (proj / "esphome_pch.h.gch.failed").exists()
+    assert scons_env.prepended == []
+
+
+def test_pch_script_failure_marker_suppresses_retry(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    scons_env = _run_script(tmp_path, fail=True)
+    proj = tmp_path / "dev"
+    assert (proj / "esphome_pch.h.gch.failed").is_file()
+    assert not (proj / "esphome_pch.h.gch.sum").exists()
+    assert scons_env.prepended == []
+    # Second run: same checksum, no compile attempt, but says so
+    attempts = (tmp_path / "fake-gxx.argv").read_text().count("c++-header")
+    _run_script(tmp_path, fail=True)
+    out = capsys.readouterr().out
+    assert (tmp_path / "fake-gxx.argv").read_text().count("c++-header") == attempts
+    assert "delete esphome_pch.h.gch.failed to retry" in out
+
+
+def test_pch_script_probe_rejection_falls_back(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """A toolchain that cannot load its own .gch (GCC 10 on macOS arm64)
+    must not leave consumers paying for a pch every compile rejects."""
+    scons_env = _run_script(tmp_path, reject_pch=True)
+    proj = tmp_path / "dev"
+    assert not (proj / "esphome_pch.h.gch").exists()
+    assert not (proj / "esphome_pch.h.gch.sum").exists()
+    assert (proj / "esphome_pch.h.gch.failed").is_file()
+    assert scons_env.prepended == []
+    assert "toolchain cannot load the pch" in capsys.readouterr().out
+
+
+def test_pch_script_spawn_failure_is_transient(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """A spawn failure must not latch a .failed marker (matches espidf)."""
+    scons_env = _run_script(tmp_path, missing_cxx=True)
+    proj = tmp_path / "dev"
+    assert not (proj / "esphome_pch.h.gch.failed").exists()
+    assert not (proj / "esphome_pch.h.gch.sum").exists()
+    assert scons_env.prepended == []
+    assert "did not run" in capsys.readouterr().out
+
+
+def test_pch_script_probe_nonzero_exit_falls_back(tmp_path: Path) -> None:
+    """A probe failure whose stderr never mentions .gch must still count."""
+    scons_env = _run_script(tmp_path, probe_exit=1)
+    proj = tmp_path / "dev"
+    assert not (proj / "esphome_pch.h.gch").exists()
+    assert (proj / "esphome_pch.h.gch.failed").is_file()
+    assert scons_env.prepended == []
+
+
+def test_pch_script_unresolved_package_version_skips_pch(tmp_path: Path) -> None:
+    """A KeyError for an installed package is unresolved identity, not absence."""
+    scons_env = _run_script(tmp_path, platform_cls=_UnresolvedPlatform)
+    assert not (tmp_path / "dev" / "esphome_pch.h.gch").exists()
+    assert scons_env.prepended == []
+
+
+def test_pch_script_package_version_error_skips_pch(tmp_path: Path) -> None:
+    """Without trustworthy package identity a stale .gch could survive an
+    upgrade, so the script must not build one at all."""
+    scons_env = _run_script(tmp_path, platform_cls=_BrokenPlatform)
+    proj = tmp_path / "dev"
+    assert not (proj / "esphome_pch.h.gch").exists()
+    assert scons_env.prepended == []
+
+
+def test_pch_script_corrupt_sidecar_reads_as_stale(tmp_path: Path) -> None:
+    """A truncated/corrupt .failed marker must not disable the pch forever."""
+    _run_script(tmp_path, fail=True)
+    proj = tmp_path / "dev"
+    (proj / "esphome_pch.h.gch.failed").write_bytes(b"\xff\xfe corrupt")
+    _run_script(tmp_path)
+    assert (proj / "esphome_pch.h.gch").is_file()
+    assert (proj / "esphome_pch.h.gch.sum").is_file()
+
+
+def test_pch_script_rebuilds_when_header_missing(tmp_path: Path) -> None:
+    _run_script(tmp_path)
+    proj = tmp_path / "dev"
+    (proj / "esphome_pch.h").unlink()
+    _run_script(tmp_path)
+    assert (proj / "esphome_pch.h").is_file()
+
+
+def test_copy_pch_script(tmp_path: Path) -> None:
+    from esphome.core import CORE
+
+    CORE.build_path = tmp_path
+    toolchain.copy_pch_script()
+    assert (tmp_path / "pch.py").read_text() == _SCRIPT.read_text()
+
+
+def test_pch_script_unions_user_sloppiness(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """A user CCACHE_SLOPPINESS without the pch tokens gets them unioned on,
+    mirroring ccache_pch_env, or every src TU is a permanent miss."""
+    scons_env = _run_script(tmp_path, env_vars={"CCACHE_SLOPPINESS": "locale"})
+    assert scons_env["ENV"]["CCACHE_SLOPPINESS"] == "locale,pch_defines,time_macros"
+    assert "adding pch_defines,time_macros" in capsys.readouterr().out
+
+
+def test_pch_script_unmodelable_flag_skips_pch(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """Unbalanced quotes and multi-token elements cannot be reproduced as
+    one argv; the pch is skipped rather than built with diverging flags."""
+    for bad in ("-DFOO='bar", "-DA=1\t-DB=2"):
+        scons_env = _run_script(tmp_path, flags=["-DX=1", bad])
+        assert scons_env.prepended == []
+        assert not (tmp_path / "dev" / "esphome_pch.h.gch").exists()
+    assert "unmodelable flag" in capsys.readouterr().out
+
+
+@pytest.mark.skipif(
+    getattr(os, "geteuid", lambda: -1)() == 0, reason="root ignores file modes"
+)
+def test_pch_script_unlistable_include_dir_skips_pch(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """An unlistable subtree must not silently drop out of the digest."""
+    proj = tmp_path / "dev"
+    override = proj / "lwip_override"
+    hidden = override / "hidden"
+    hidden.mkdir(parents=True)
+    (hidden / "gen.h").write_text("")
+    hidden.chmod(0)
+    try:
+        scons_env = _run_script(tmp_path, flags=["-DX=1", "-I", str(override)])
+    finally:
+        hidden.chmod(0o755)
+    assert scons_env.prepended == []
+    assert "skipping precompiled header" in capsys.readouterr().out
+
+
+def test_pch_script_nobuild_without_projenv_is_noop(tmp_path: Path) -> None:
+    """-t nobuild never exports projenv; the script must not abort."""
+    proj = tmp_path / "dev"
+    (proj / "src").mkdir(parents=True)
+
+    def strict_import(*names: str) -> None:
+        if "projenv" in names:
+            raise RuntimeError("Import of non-existent variable 'projenv'")
+
+    env = _FakeSConsEnv(proj, proj / "src", "g++", ["-DX=1"])
+    exec(  # noqa: S102
+        compile(_SCRIPT.read_text(), "pch.py", "exec"),
+        {"Import": strict_import, "env": env},
+    )
+    assert not (proj / "esphome_pch.h").exists()
+
+
+def test_pch_script_ignores_library_trees_and_non_headers(tmp_path: Path) -> None:
+    """.piolibdeps and non-header files must not enter the digest (or be
+    read at all); package versions already cover library identity."""
+    proj = tmp_path / "dev"
+    libdeps = proj / ".piolibdeps" / "lib" / "src"
+    libdeps.mkdir(parents=True)
+    (libdeps / "lib.h").write_text("#define A 1\n")
+    override = proj / "lwip_override"
+    override.mkdir(parents=True)
+    (override / "lwipopts.h").write_text("#define TCP_MSS 1460\n")
+    (override / "notes.txt").write_text("v1\n")
+    flags = ["-DX=1", "-I", str(libdeps), "-I", str(override)]
+    _run_script(tmp_path, flags=flags)
+    first = (proj / "esphome_pch.h.gch.sum").read_text()
+    (libdeps / "lib.h").write_text("#define A 2\n")
+    (override / "notes.txt").write_text("v2\n")
+    (tmp_path / "fake-gxx.argv").unlink(missing_ok=True)
+    _run_script(tmp_path, flags=flags)
+    assert (proj / "esphome_pch.h.gch.sum").read_text() == first
+
+
+def test_pch_script_hashes_project_local_include_dirs(tmp_path: Path) -> None:
+    """Generated headers in project-local -I dirs (e.g. rp2's lwip_override)
+    must invalidate the checksum when they change."""
+    proj = tmp_path / "dev"
+    override = proj / "lwip_override"
+    override.mkdir(parents=True)
+    (override / "lwipopts.h").write_text("#define TCP_MSS 1460\n")
+    flags = ["-DX=1", "-I", str(override)]
+    _run_script(tmp_path, flags=flags)
+    first = (proj / "esphome_pch.h.gch.sum").read_text()
+    (override / "lwipopts.h").write_text("#define TCP_MSS 536\n")
+    (tmp_path / "fake-gxx.argv").unlink(missing_ok=True)
+    _run_script(tmp_path, flags=flags)
+    assert (proj / "esphome_pch.h.gch.sum").read_text() != first
+
+
+@pytest.mark.skipif(
+    getattr(os, "geteuid", lambda: -1)() == 0, reason="root ignores file modes"
+)
+def test_pch_script_unreadable_local_header_skips_pch(
+    tmp_path: Path, capsys: pytest.CaptureFixture[str]
+) -> None:
+    """An unreadable generated header means unknown identity: no pch."""
+    proj = tmp_path / "dev"
+    override = proj / "lwip_override"
+    override.mkdir(parents=True)
+    secret = override / "lwipopts.h"
+    secret.write_text("#define TCP_MSS 1460\n")
+    secret.chmod(0)
+    scons_env = _run_script(tmp_path, flags=["-DX=1", "-I", str(override)])
+    assert not (proj / "esphome_pch.h.gch.sum").exists()
+    assert scons_env.prepended == []
+    assert "skipping precompiled header" in capsys.readouterr().out

tests/unit_tests/test_writer.py

@@ -516,10 +516,13 @@ def test_clean_build(
     dependencies_lock = tmp_path / "dependencies.lock"
     dependencies_lock.write_text("lock file")
 
-    # idedata cache lives under the data dir, not the build path.
+    # idedata caches live under the data dir, not the build path; the
+    # .arduino.json variant is the native esp8266 toolchain's.
     idedata_cache = tmp_path / "idedata" / "test.json"
     idedata_cache.parent.mkdir()
     idedata_cache.write_text("{}")
+    arduino_idedata_cache = tmp_path / "idedata" / "test.arduino.json"
+    arduino_idedata_cache.write_text("{}")
 
     # Native ESP-IDF toolchain artifacts.
     idf_build_dir = tmp_path / "build"
@@ -580,6 +583,7 @@ def test_clean_build(
     assert not piolibdeps_dir.exists()
     assert not dependencies_lock.exists()
     assert not idedata_cache.exists()
+    assert not arduino_idedata_cache.exists()
     assert not idf_build_dir.exists()
     assert not managed_components_dir.exists()
     assert not pio_components_dir.exists()
@@ -673,6 +677,32 @@ def test_clean_build_partial_exists(
     assert "dependencies.lock" not in caplog.text
 
 
+@patch("esphome.writer.CORE")
+def test_clean_build_partial_removes_pch_artifacts(
+    mock_core: MagicMock,
+    tmp_path: Path,
+) -> None:
+    """The PlatformIO pch sidecars live at the project root and must go in
+    a partial clean, like the native backend's under .pioenvs."""
+    names = (
+        "esphome_pch.h",
+        "esphome_pch.h.gch",
+        "esphome_pch.h.gch.sum",
+        "esphome_pch.h.gch.failed",
+    )
+    for name in names:
+        (tmp_path / name).write_text("x")
+    mock_core.relative_pioenvs_path.return_value = tmp_path / ".pioenvs"
+    mock_core.relative_piolibdeps_path.return_value = tmp_path / ".piolibdeps"
+    mock_core.relative_build_path.side_effect = lambda name: tmp_path / name
+    mock_core.relative_internal_path.side_effect = tmp_path.joinpath
+
+    clean_build()
+
+    for name in names:
+        assert not (tmp_path / name).exists()
+
+
 @patch("esphome.writer.CORE")
 def test_clean_build_nothing_exists(
     mock_core: MagicMock,