pureboot: the classic megas and the word-addressed 1284P groundwork

Device: boot-section detection probes SPMCR beside SPMCSR, the link
picks any hardware USART through the instance-aware lookups (URSEL
chips included), WDRF reads MCUSR-or-MCUCSR, and the >64 KiB shape
lands — word-addressed wire flash (info flag bit 1, page byte 0 means
256, base as a word address), far reads through flash_load_far, a
single 32-bit byte-cursor page walk (the 256-byte page wraps its low
byte exactly), and slot arithmetic in words (the return address
already is one). Host: addresses stay bytes internally and scale at
the wire, the boot-fuse decode becomes a per-signature table (byte
index + BOOTSZ ladder — the m168A's lives in EXTENDED), and the
planner tests pin every chip's ladder plus the word-addressed info
decode. Tests: the device runner serves every mega over the USART pty,
pbapp banners over the right link, the update rehearsal synthesizes
its assumed fuses from the tool's own table, and the PI lint tracks
the renamed info symbol. All six classic-mega/168A targets pass the
full suite (size, PI, planner, protocol, reloc, self-update) at
466–504 B; the 1284P builds await a libavr far-path slimming to make
its 512.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
2026-07-21 12:00:19 +02:00
parent a1ff032c87
commit 0fa53e1cad
12 changed files with 755 additions and 126 deletions

View File

@@ -23,7 +23,7 @@ trampoline (below).
| Chip | Serial | Baud | Clock assumed |
|---|---|---|---|
| ATmega328P | USART0, RXD/TXD = PD0/PD1 | 115200 8N1 | 16 MHz crystal |
| every ATmega (8/8A, 16, 32/32A, 168A, 328P, 1284P) | the hardware USART (USART0), RXD/TXD per pinout | 115200 8N1 | 16 MHz crystal |
| ATtiny85 | software UART, RX = PB0, TX = PB1 | 57600 8N1 | 8 MHz internal RC |
| ATtiny13A | software UART, RX = PB0, TX = PB1 | 57600 8N1 | 9.6 MHz internal RC |
@@ -55,6 +55,10 @@ write to finish and sends the prompt `+` (0x2b) — the prompt is therefore
also the completion ack of the previous command. A session is: await `+`,
send a command, read its reply, repeat.
On chips whose flash exceeds 64 KiB (the 1284P — info-block flag bit 1) the
`R`/`W` flash addresses are **word** addresses; everywhere else they are byte
addresses. EEPROM addresses are always bytes, counts always bytes.
| Cmd | Arguments | Reply |
|---|---|---|
| `b` | — | the 12-byte info block |
@@ -88,18 +92,29 @@ The info block (`b`):
|---|---|
| 02 | `'P'`, `'B'`, protocol version (1) |
| 35 | device signature |
| 6 | SPM page size in bytes |
| 78 | loader base — application flash ends here |
| 6 | SPM page size in bytes (0 means 256) |
| 78 | loader base — application flash ends here (a word address when bit 1 is set) |
| 910 | EEPROM size |
| 11 | bit 0 set: host must patch the reset vector (no hardware boot section) |
| 11 | bit 0: host must patch the reset vector (no hardware boot section); bit 1: flash wire addresses are word addresses |
Composites are the host's job: verify = read back and compare, erase =
write `0xff` (per page for flash, per byte for EEPROM).
## Deployment
**ATmega328P**: program the loader at 0x7e00 with an external programmer.
Two fuse profiles, same binary:
**Megas**: program the loader at `flash 512` with an external programmer.
Every mega's smallest-but-one BOOTSZ puts the boot-section start exactly at
the loader base (512 B — the m8/16/168A reach it at their second-smallest
step, the m32/328P at their smallest), so the ATmega328P profiles below
apply to all of them with their own addresses; the per-chip BOOTSZ ladders
live in the host tool (`BOOT_FUSE`). The **ATmega1284P** is the exception:
its smallest boot section is 1 KB, so the standalone profile does not exist
— BOOTSZ = 512 words always, and with BOOTRST programmed reset lands at
0x1f800, one erased slot below the loader (the loader-first walk behavior
below, built in). Its staging slot sits inside that same 1 KB section, so
self-update needs no fuse change.
ATmega328P profiles (addresses for its 32 KiB):
| BOOTSZ | BOOTRST | Behavior |
|---|---|---|

View File

@@ -53,14 +53,35 @@ consteval avr::hertz_t clock()
using dev = avr::device<{.clock = clock()}>;
// The watchdog reset flag's home: MCUSR, or the classic megas' MCUCSR.
consteval std::int16_t wdrf_field()
{
auto reg = std::string_view{avr::hw::db.regs[static_cast<std::size_t>(avr::power::detail::reset_reg())].name};
return avr::hw::db.field_index(reg, "WDRF");
}
// Geometry: the resident loader owns the top 512 bytes of flash; the word
// below it is the trampoline (the application's relocated reset vector) on
// chips without a hardware boot section. The RWWSRE bit marks a separate
// boot section — on classic AVR the two capabilities coincide.
// boot section — on classic AVR the two capabilities coincide (the m8/m32
// packs spell its register SPMCR).
constexpr std::uint16_t boot_bytes = 512;
constexpr std::uint16_t base = static_cast<std::uint16_t>(spm::flash_bytes - boot_bytes);
constexpr std::uint32_t base = spm::flash_bytes - boot_bytes;
constexpr std::uint16_t page = spm::page_bytes;
constexpr bool boot_section = avr::hw::db.field_index("SPMCSR", "RWWSRE") >= 0;
constexpr bool boot_section = [] {
for (auto reg : {"SPMCSR", "SPMCR"})
if (avr::hw::db.field_index(reg, "RWWSRE") >= 0)
return true;
return false;
}();
// Past 64 KiB a byte address no longer fits the wire's 16 bits, so on the
// large chips every flash address on the wire — and all slot arithmetic —
// is a word address instead ('J' always was one). Slots stay 512 bytes =
// 256 words, so a slot is one high byte in either unit.
constexpr bool word_flash = spm::flash_bytes > 65536;
constexpr std::uint16_t wire_base = word_flash ? static_cast<std::uint16_t>(base / 2) : static_cast<std::uint16_t>(base);
constexpr std::uint16_t wire_page_mask = word_flash ? (page / 2 - 1) : (page - 1);
// The activation window, in seconds, is a compile-time constant (the build
// may override it): the whole EEPROM belongs to the application, and
@@ -71,8 +92,10 @@ constexpr bool boot_section = avr::hw::db.field_index("SPMCSR", "RWWSRE") >= 0;
constexpr std::uint8_t timeout_seconds = PUREBOOT_TIMEOUT;
// The 12-byte info block the host reads with the 'b' command; flash-resident
// (there is no crt to copy a .data image).
inline constexpr std::array<std::uint8_t, 12> info_data = {
// (there is no crt to copy a .data image), word-aligned so its wire (word)
// address is exact on the large chips. The page byte is the wire count
// convention: 0 means 256.
[[gnu::progmem]] alignas(2) inline constexpr std::array<std::uint8_t, 12> info_data = {
'P',
'B',
1, // magic, protocol version
@@ -80,13 +103,14 @@ inline constexpr std::array<std::uint8_t, 12> info_data = {
avr::hw::db.signature[1],
avr::hw::db.signature[2],
static_cast<std::uint8_t>(page),
base & 0xff,
base >> 8, // app flash ends here; resident loader base
wire_base & 0xff,
wire_base >> 8, // app flash ends here; resident loader base (a word address on large chips)
avr::hw::db.mem.eeprom_size & 0xff,
avr::hw::db.mem.eeprom_size >> 8,
boot_section ? 0 : 1, // bit 0: host must patch the reset vector (no hardware boot section)
// bit 0: host must patch the reset vector (no hardware boot section);
// bit 1: flash wire addresses are word addresses
static_cast<std::uint8_t>((boot_section ? 0 : 1) | (word_flash ? 2 : 0)),
};
using info = avr::flash_table<info_data>;
// The serial link: the hardware USART where the chip has one, the polled
// software UART (no vector — the table belongs to the application) on PB0/PB1
@@ -98,19 +122,19 @@ using info = avr::flash_table<info_data>;
template <avr::hertz_t C>
consteval std::int16_t rxc_field()
{
return avr::hw::db.field_index("UCSR0A", "RXC0");
return avr::uart::detail::ufield<'0', "UCSR#A", "RXC#">();
}
template <avr::hertz_t C>
consteval std::int16_t txc_field()
{
return avr::hw::db.field_index("UCSR0A", "TXC0");
return avr::uart::detail::ufield<'0', "UCSR#A", "TXC#">();
}
template <avr::hertz_t C>
consteval std::int16_t status_reg()
{
return avr::hw::db.reg_index("UCSR0A");
return avr::uart::detail::ureg<'0', "UCSR#A">();
}
template <avr::hertz_t C>
@@ -190,7 +214,8 @@ struct software_link {
}
};
using link = std::conditional_t<avr::hw::db.has_reg("UDR0"), hardware_link<dev::clock>, software_link<dev::clock>>;
using link = std::conditional_t<avr::hw::db.has_instance("USART0") || avr::hw::db.has_instance("USART"),
hardware_link<dev::clock>, software_link<dev::clock>>;
// The application's entry, an absolute address the linker pins (--defsym in
// CMakeLists.txt): 0x0000 on the mega (word 0 stays the application's own
@@ -245,19 +270,22 @@ std::uint16_t rx16()
return static_cast<std::uint16_t>(low | (link::rx() << 8));
}
const std::uint8_t *flash_ptr(std::uint16_t address)
{
return reinterpret_cast<const std::uint8_t *>(address);
}
// The streamers take the count in the wire's 8-bit form: 0 means 256.
// send_flash stays out of line: its two callers ('b' and 'R') otherwise each
// inline a private copy of the loop.
// inline a private copy of the loop. On the large chips the address is a
// word address and the read goes through ELPM (flash_load_far).
[[gnu::noinline]] void send_flash(std::uint16_t address, std::uint8_t count)
{
do
link::tx(avr::flash_load(flash_ptr(address++)));
while (--count);
if constexpr (word_flash) {
auto byte_address = static_cast<std::uint32_t>(address) << 1;
do
link::tx(avr::flash_load_far<std::uint8_t>(byte_address++));
while (--count);
} else {
do
link::tx(avr::flash_load(reinterpret_cast<const std::uint8_t *>(address++)));
while (--count);
}
}
void send_eeprom(std::uint16_t address, std::uint8_t count)
@@ -287,7 +315,7 @@ void store_eeprom(std::uint16_t address, std::uint8_t count)
// copy flashed one slot lower may rewrite the slot above it — how pureboot
// updates itself. On the mega the RWW section is re-enabled so reads work
// immediately.
void program_flash(std::uint16_t address, std::uint8_t slot_high)
void program_flash(std::uint16_t wire_address, std::uint8_t slot_high)
{
// A buffer word cannot be loaded twice without an erase (§26.2.1), so a
// refused page's drained data must not linger for the next write:
@@ -297,19 +325,40 @@ void program_flash(std::uint16_t address, std::uint8_t slot_high)
spm::rww_enable<off>();
else
spm::clear_buffer<off>();
// The address is the loop's only state: pages are aligned, so the walk
// ends when the offset bits wrap back to zero.
do {
std::uint8_t low = link::rx();
std::uint8_t high = link::rx();
spm::fill<off>(address, static_cast<std::uint16_t>(low | (high << 8)));
address += 2;
} while (static_cast<std::uint8_t>(address) & (page - 1));
address -= 2; // back inside the page — erase and write ignore the word bits
const std::uint8_t page_high = static_cast<std::uint8_t>(address >> 8) & 0xfe;
// One induction either way. On the byte-addressed chips the wire address
// itself walks the page (aligned, so the offset bits wrap to zero); on
// the word-addressed large chips the wire word address becomes a 32-bit
// byte cursor once, and their 256-byte page makes its low byte the whole
// in-page offset. The slot index is one high byte of the wire address —
// two values on byte-addressed chips (the & ~1), bits 16:9 re-packed on
// the large ones.
spm::flash_address_t address;
std::uint8_t page_high;
if constexpr (word_flash) {
address = static_cast<spm::flash_address_t>(static_cast<std::uint32_t>(wire_address) << 1);
const auto start = address;
do {
std::uint8_t low = link::rx();
std::uint8_t high = link::rx();
spm::fill<off>(address, static_cast<std::uint16_t>(low | (high << 8)));
address += 2;
} while (static_cast<std::uint8_t>(address));
address = start;
page_high = static_cast<std::uint8_t>(static_cast<std::uint16_t>(address >> 8) >> 1);
} else {
address = static_cast<spm::flash_address_t>(wire_address);
do {
std::uint8_t low = link::rx();
std::uint8_t high = link::rx();
spm::fill<off>(address, static_cast<std::uint16_t>(low | (high << 8)));
address += 2;
} while (static_cast<std::uint8_t>(address) & (page - 1));
address -= 2; // back inside the page — erase and write ignore the word bits
page_high = static_cast<std::uint8_t>(address >> 8) & 0xfe;
}
if (page_high != slot_high) {
// The tinies halt the CPU through the erase and the write, so only
// the mega — running on while its RWW section programs — waits.
// the megas — running on while their RWW section programs — wait.
spm::erase_page<off>(address);
if constexpr (boot_section)
spm::wait();
@@ -336,18 +385,20 @@ void send_fuses()
{
// A watchdog reset belongs to the application (whose watchdog stays
// forced on until it clears WDRF) — no activation window in its way.
if (avr::hw::mcusr::wdrf.test())
// The flag register is MCUSR, or the classic megas' MCUCSR.
if (avr::hw::field_impl<wdrf_field()>::test())
run_app();
link::init();
// The high byte of the 512-byte-aligned base this copy runs at: the word
// return address's high byte is the byte address >> 9 (the slot index),
// doubled back into address terms. program_flash refuses this one slot
// and the info block is addressed from it, so both follow wherever the
// code was flashed.
// The high byte of the 512-byte-aligned base this copy runs at: the
// return address is a word address, whose high byte is the 256-word slot
// index — on byte-addressed chips doubled back into byte terms.
// program_flash refuses this one slot and the info block is addressed
// from it, so both follow wherever the code was flashed.
const std::uint16_t ra_words = reinterpret_cast<std::uint16_t>(__builtin_return_address(0));
const std::uint8_t slot_high =
static_cast<std::uint8_t>((reinterpret_cast<std::uint16_t>(__builtin_return_address(0)) >> 8) << 1);
word_flash ? static_cast<std::uint8_t>(ra_words >> 8) : static_cast<std::uint8_t>((ra_words >> 8) << 1);
// The knock: 'p' then 'b', each under a fresh window; any other byte is
// line noise and waits again. Falling out of a window runs the app.
@@ -364,11 +415,15 @@ void send_fuses()
case 'b': { // info block, read relative to the running slot
// The block sits in the image's first 256 bytes (the build lint
// asserts it), and slots are 512-aligned — so the low byte of its
// link address is its offset in any slot, and the high byte of
// its runtime address is the running slot's. Built as a byte
// pair so no absolute 16-bit address is ever materialized.
const std::uint8_t low = static_cast<std::uint8_t>(reinterpret_cast<std::uint16_t>(info::storage.data()));
send_flash(std::bit_cast<std::uint16_t>(std::array{low, slot_high}), info::size());
// link address (in wire units: bytes, or words on the large
// chips) is its offset in any slot, and the high byte of its
// runtime address is the running slot's. Built as a byte pair so
// no absolute address is ever materialized.
const auto link_low = reinterpret_cast<std::uint16_t>(info_data.data());
const std::uint8_t low =
word_flash ? static_cast<std::uint8_t>(link_low >> 1) : static_cast<std::uint8_t>(link_low);
send_flash(std::bit_cast<std::uint16_t>(std::array{low, slot_high}),
static_cast<std::uint8_t>(info_data.size()));
break;
}
case 'J': { // jump to a wire word address: hand-over and staging transfer

View File

@@ -270,10 +270,14 @@ class Info:
raise Error(f"protocol version {raw[2]}, tool speaks {PROTOCOL_VERSION}")
self.raw = bytes(raw)
self.signature = raw[3:6]
self.page = raw[6]
self.base = raw[7] | (raw[8] << 8)
self.eeprom_size = raw[9] | (raw[10] << 8)
self.page = raw[6] or 256 # the wire count convention: 0 means 256
self.patch_vector = bool(raw[11] & 1)
# Large chips speak word addresses for flash (bit 1); the host keeps
# every address in bytes and converts at the wire.
self.word_flash = bool(raw[11] & 2)
scale = 2 if self.word_flash else 1
self.base = (raw[7] | (raw[8] << 8)) * scale
self.eeprom_size = raw[9] | (raw[10] << 8)
self.flash_size = self.base + SLOT
self.stage = self.base - SLOT # where a staging copy of the loader goes
# The hand-over target, as the word address 'J' takes: the trampoline
@@ -331,25 +335,33 @@ class Loader:
self._expect_prompt(timeout)
return reply
def _stream_read(self, command, address, count):
def _stream_read(self, command, address, count, address_scale=1):
data = b""
while count:
chunk = min(count, 256)
head = bytes((ord(command), address & 0xFF, address >> 8, chunk & 0xFF))
wire = address // address_scale
head = bytes((ord(command), wire & 0xFF, wire >> 8, chunk & 0xFF))
data += self._command(head, chunk, 5.0)
address += chunk
count -= chunk
return data
def read_flash(self, address, count):
return self._stream_read("R", address, count)
if not self.info.word_flash:
return self._stream_read("R", address, count)
# Word-addressed wire: widen to even bounds, read, trim.
start = address & ~1
span = (address + count + 1 & ~1) - start
data = self._stream_read("R", start, span, address_scale=2)
return data[address - start : address - start + count]
def read_eeprom(self, address, count):
return self._stream_read("r", address, count)
def write_page(self, address, data):
assert len(data) == self.info.page and address % self.info.page == 0
head = bytes((ord("W"), address & 0xFF, address >> 8))
wire = address // (2 if self.info.word_flash else 1)
head = bytes((ord("W"), wire & 0xFF, wire >> 8))
self._command(head + data, 0, 2.0)
def write_eeprom(self, address, data):
@@ -495,14 +507,32 @@ def covered(pages, info, skip_blank):
# ----------------------------------------------------------------- fuses ---
def mega_boot(high_fuse):
"""Decode the ATmega328P high fuse's boot configuration (DS40002061B
§27.3, Table 27-13/27-16): BOOTSZ1:0 in bits 2:1 select the boot-section
words, BOOTRST in bit 0 (programmed = 0) re-vectors reset to its start.
Returns (bootrst_programmed, boot_section_start_byte)."""
bootsz = (high_fuse >> 1) & 0x03
words = {0b11: 256, 0b10: 512, 0b01: 1024, 0b00: 2048}[bootsz]
return (high_fuse & 1) == 0, 0x8000 - words * 2
# Per-chip boot fuse geometry, keyed by the signature's family/part bytes:
# which byte of the 'F' reply (low, lock, extended, high) carries BOOTSZ/
# BOOTRST, and the BOOTSZ->words ladder. Sources: Atmel-2486/2466/2503
# (HIGH fuse), Atmel-8271 (m168A: EXTENDED; m328P: HIGH), Atmel-42719.
BOOT_FUSE = {
bytes((0x93, 0x07)): (3, {0b11: 128, 0b10: 256, 0b01: 512, 0b00: 1024}), # m8/8A
bytes((0x94, 0x03)): (3, {0b11: 128, 0b10: 256, 0b01: 512, 0b00: 1024}), # m16
bytes((0x95, 0x02)): (3, {0b11: 256, 0b10: 512, 0b01: 1024, 0b00: 2048}), # m32/32A
bytes((0x94, 0x06)): (2, {0b11: 128, 0b10: 256, 0b01: 512, 0b00: 1024}), # m168A
bytes((0x95, 0x0F)): (3, {0b11: 256, 0b10: 512, 0b01: 1024, 0b00: 2048}), # m328P
bytes((0x97, 0x05)): (3, {0b11: 512, 0b10: 1024, 0b01: 2048, 0b00: 4096}), # 1284P
}
def mega_boot(info, fuse_bytes):
"""Decode a mega's boot configuration from its fuses (the byte and the
BOOTSZ ladder are per chip): BOOTSZ1:0 in bits 2:1 select the
boot-section words, BOOTRST in bit 0 (programmed = 0) re-vectors reset
to its start. Returns (bootrst_programmed, boot_section_start_byte)."""
entry = BOOT_FUSE.get(bytes(info.signature[1:3]))
if entry is None:
raise Error(f"unknown mega signature {info.signature.hex()} — no boot fuse map")
which, ladder = entry
fuse = fuse_bytes[which]
words = ladder[(fuse >> 1) & 0x03]
return (fuse & 1) == 0, info.flash_size - words * 2
# ---------------------------------------------------------- loader update ---
@@ -557,12 +587,11 @@ def update_preflight(image, info, fuse_bytes):
if not info.patch_vector:
if fuse_bytes is None:
raise Error("a loader update on this chip needs its fuses — unreadable? pass --assume-fuses")
high = fuse_bytes[3]
bootrst, bls_start = mega_boot(high)
bootrst, bls_start = mega_boot(info, fuse_bytes)
if info.stage < bls_start:
raise Error(
f"cannot self-update: the staging slot {info.stage:#06x} lies below the "
f"boot section ({bls_start:#06x}, high fuse {high:#04x}) where SPM is disabled "
f"boot section ({bls_start:#06x}) where SPM is disabled "
f"— a boot section of at least 1 KB (BOOTSZ) is required, and only an "
f"external programmer can change fuses"
)
@@ -710,7 +739,7 @@ def check_walk_region(pages, info, fuse_bytes, force):
Only checkable when the fuses are known (--fuses or --assume-fuses)."""
if info.patch_vector or fuse_bytes is None:
return
bootrst, bls_start = mega_boot(fuse_bytes[3])
bootrst, bls_start = mega_boot(info, fuse_bytes)
if not bootrst or bls_start >= info.base:
return
overlap = [a for a in sorted(pages) if a >= bls_start and pages[a].count(0xFF) != len(pages[a])]