opencode.siil.arti.ee/tms-c-compiler
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs: M1 minimal backend and test convention

Browse Source
This commit is contained in:
arti committed 2026-11-11 09:53:23 +00:00
1 parent 755b3cb7b2
commit 56d6a71cb7
2 files changed
+233 -72
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

README.md
+153 -56
View File
@@ -3,13 +3,14 @@
A modern C toolchain for the DEC VT1200 / TI TMS34010, built as a small fork of
upstream LLVM.
**Status: M0.** Clang understands the `tms34010-unknown-none` target, the TI
data model, and bit-addressed pointers. There is **no machine-code backend
yet**: M0 stops at LLVM IR. The backend, TI ABI, ROM calls, Xlib and real
clients are M1–M5 and are separate plans (see the design spec).
**Status: M1.** `llc -march=tms34010` emits GSPA-compatible assembly for a
small set of leaf functions, and that assembly assembles, links and runs in
MAME. M1 is an **assembly-only, test-only** snapshot: it uses an A0–A3/A0
register convention, not the TI ABI, and has no object writer, calls or runtime.
- Design: [`docs/superpowers/specs/2026-11-10-tms34010-llvm-design.md`](docs/superpowers/specs/2026-11-10-tms34010-llvm-design.md)
- M0 plan: [`docs/superpowers/plans/2026-11-10-tms34010-m0-fork-and-bit-address-semantics.md`](docs/superpowers/plans/2026-11-10-tms34010-m0-fork-and-bit-address-semantics.md)
- M1 plan: [`docs/superpowers/plans/2026-11-11-tms34010-m1-minimal-backend.md`](docs/superpowers/plans/2026-11-11-tms34010-m1-minimal-backend.md)
- Spike: [`spike/REPORT-23.md`](spike/REPORT-23.md) (current, LLVM 23.1.5);
[`spike/REPORT.md`](spike/REPORT.md) (historical, 19.1.7)
@@ -17,42 +18,67 @@ clients are M1–M5 and are separate plans (see the design spec).
- **Compiler fork:** <https://git.wut.ee/opencode.siil.arti.ee/llvm-tms34010>
(branch `tms34010`) — the full buildable LLVM 23.1.5 tree with the TMS34010
target and clang bit-address changes.
target, clang bit-address changes and the M1 backend.
- **This project:** <https://git.wut.ee/opencode.siil.arti.ee/tms-c-compiler>
— plans/specs, VT1200 integration tooling, `tms-bitaddr`, `tms-clang`,
symbols and the spike.
symbols, the M1 harness and the spike.
The compiler fork is the canonical source for LLVM changes. `patches/llvm`
here is a lightweight mirror used by `scripts/build-llvm.sh`, so the project
can be built without cloning 3 GB of LLVM; keep it in sync with the fork's
`tms34010` branch.
The compiler fork is the canonical source for all target code. `patches/llvm`
here is a **frozen M0 mirror** used by the default upstream+patches build; M1
does not change it.
## What M0 provides
## What M1 provides
- `clang --target=tms34010-unknown-none`: 32-bit bit-addressed pointers,
8-bit signed `char`, 16-bit `short`, 32-bit `int`/`long`, packed structs,
`sizeof(struct { char; int; void*; }) == 9`.
- Pointer arithmetic scaled by 8; pointer difference divided by 8
(`char *` → `/8`, `int *` → `/32`), while `(int)(p+1) - (int)p` stays an
integer subtraction.
- `tools/tms-bitaddr`: GEP `--scale` and late `--normalize` passes, including
constant-expression GEPs.
- `scripts/tms-clang`: the M0 pipeline
`clang -O0` → `tms-bitaddr --scale` → `opt -O2` → `tms-bitaddr --normalize`.
- Lit tests for the triple, data model and pointer difference; an integration
test over a real Xlib client IR.
M1 fills in the M0 target skeleton and stops at assembly:
- `llc -march=tms34010 -filetype=asm` for leaf 32-bit functions:
integer/pointer arguments in `A0`–`A3`, result in `A0`, return via `RETS 0`.
- `ADD` for `i32`, `MOVI` for constants, `MOVE ..., 1` (FS1) for 32-bit
load/store, `MOVE ..., 0` (FS0) for `i16`, and sign-extending `MOVB` for
`i8` load/store.
- Constant global addresses as `MOVI _sym+bits`; a surviving GEP offset is a
bit offset (`ADDI 24`, not `ADDI 192`).
- Hard errors, not silent fallbacks, for calls (`M1 has no calls; M2 adds the
TI ABI`), stack frames (`M1 has no stack frame`), more than four register
arguments, and unsupported types.
- Nine `llvm-lit`/FileCheck tests under
`llvm/test/CodeGen/TMS34010/m1-*.ll` in the compiler fork.
- An end-to-end path: `scripts/tms-clang` IR → `llc -O0 -filetype=asm` →
**1987 GSPA/GSPLNK 2.00 under DOSBox** → COFF → the VT1200 loader in MAME.
The run checks `add(20, 22) == 42`, `load(&word) == 0x12345678`,
`second() == &arr[1]` (a `symbol+32` bit relocation), `loadb` of a byte
`0x81` returns the sign-extended `0xFFFFFF81`, and `storeb` leaves `0x81`
in memory.
- A `dis34` round-trip that reads the linked `.text` from the COFF and checks
the instructions GSPA encoded at their linked bit addresses.
## M1 is not the TI ABI
| | M1 test convention | TI ABI (M2) |
|---|---|---|
| Integer/pointer arguments | `A0`–`A3`, four maximum | program stack (via `A14` STK) |
| Return value | `A0` | `A8` |
| Frame | none; leaf functions only | `A13` FP, `A14` STK, `A15` system stack |
| Return instruction | `RETS 0` | `RETS 2` |
| Calls, callee-saves, stack args | rejected | required |
| Field select | harness sets FS1/FS0 before the leaf | `FS1 = 32` preserved across calls |
M1 assembly is not ABI-compatible with TI's compiler or the ROM Xlib. M2
replaces the convention; do not build on the M1 register assignment.
## Repository layout
```text
build/llvm-src/ local fork (ignored; created by build-llvm.sh or manually)
build/llvm-build/ Ninja build of the fork (ignored)
build/llvm-clean*/ clean CI clone/build (ignored)
patches/llvm/ the complete fork delta against llvmorg-23.1.5
build/llvm-clean/ compiler fork checkout used for M1 (ignored)
build/llvm-clean-build/ its Ninja build tree (ignored)
build/llvm-src/ M0 working fork (ignored)
patches/llvm/ frozen M0 patch series against llvmorg-23.1.5
tools/tms-bitaddr/ GEP scale/normalize tool
scripts/tms-clang M0 pipeline wrapper
scripts/tms-clang IR pipeline wrapper (clang + tms-bitaddr + opt)
scripts/m1/ M1 harness: C inputs, entry.asm, link.cmd, checks
scripts/tests/m1/ M1 gates (run.sh, dis-roundtrip.sh)
scripts/tests/m0/ M0 integration tests
scripts/build-llvm.sh clean clone + patches + build
scripts/build-llvm.sh fork or upstream+patches build
scripts/fetch-vt1200.sh fetch the VT1200 RE tree at its pinned commit
spike/ measurement corpus and reports
symbols/ per-ROM import tables (v2.2 subset)
@@ -63,10 +89,13 @@ docs/ design and plans
## Prerequisites
`cmake`, `ninja-build`, `ccache`, a C++17 host compiler (GCC 14 is used here),
Python 3, and about 20 GB of disk for the LLVM source and build. On Debian:
Python 3, and about 20 GB of disk for the LLVM source and build. The M1 gates
additionally need `dosbox`, `mtools` (`mcopy`), the VT1200 SDK image, the
`dwt` COFF loader, and a built MAME `vt1200` binary
(`build/mame-src/vt1200`). On Debian:
```sh
sudo apt-get install -y cmake ninja-build ccache
sudo apt-get install -y cmake ninja-build ccache dosbox mtools
```
## Fetch the VT1200 assets
@@ -78,54 +107,122 @@ Xlib call sequence, `as34`, `dis34`, SDKs, manuals). It is not vendored:
./scripts/fetch-vt1200.sh # -> vendor/vt1200, pinned 17dfd24...
```
## Build the pinned fork
## Build the compiler
### Compiler fork (M1, the normal path)
The fork checkout already exists at `build/llvm-clean` on `tms34010`. Build it
in place:
```sh
LLVM_SRC=$PWD/build/llvm-clean LLVM_BRANCH=tms34010 ./scripts/build-llvm.sh
# binaries: build/llvm-clean-build/bin/{llc,clang,opt,llvm-config,FileCheck}
# lit driver: build/llvm-clean-build/bin/llvm-lit
```
To clone the fork first instead of using an existing checkout:
```sh
LLVM_REPO=gitea@git.wut.ee:opencode.siil.arti.ee/llvm-tms34010.git \
LLVM_BRANCH=tms34010 ./scripts/build-llvm.sh
```
Fork mode **never** checks out, resets, cleans or patches the source tree; it
only runs CMake and Ninja. Local commits on the branch are safe. Use
`LLVM_BUILD=...` to keep a separate build directory. It builds `llc`, clang,
opt, `llvm-config`, FileCheck and the test helpers with X86 (host) and
TMS34010; a first build takes roughly 20–40 minutes.
### Upstream + patches (frozen M0 path)
```sh
./scripts/build-llvm.sh
# binaries: build/llvm-clean-build/bin/{clang,opt,llvm-config,FileCheck,llvm-lit}
```
The script clones `llvmorg-23.1.5`, applies `patches/llvm/*.patch`, and builds
`clang`/`opt` with X86 (host) and TMS34010. A first build takes roughly
20–40 minutes; rebuilds are incremental.
This mode clones `llvmorg-23.1.5`, applies `patches/llvm/*.patch`, and builds
`build/llvm-clean-build`. It owns and resets only a plain `llvm-project` clone:
if `build/llvm-clean` has some other origin (for example the M1 fork), it
refuses rather than destroying it. Use fork mode for that checkout.
For development, `build/llvm-src` is the working fork and `build/llvm-build` is
its build tree. `scripts/export-patches.sh` regenerates the committed patch
series with `git diff` (including new files via `git add -N`).
`scripts/export-patches.sh` regenerates the committed M0 patch series with
`git diff` (including new files via `git add -N`); it is not used for M1.
## Build the M0 tool and run the tests
## Build the M0 support tools
The M1 pipeline still uses `tms-bitaddr` and `tms-clang`:
```sh
LLVM_CONFIG=$PWD/build/llvm-clean-build/bin/llvm-config \
tools/tms-bitaddr/build.sh
```
M0 regression test (optional in M1):
```sh
TMS_CLANG=$PWD/build/llvm-clean-build/bin/clang \
TMS_OPT=$PWD/build/llvm-clean-build/bin/opt \
TMS_BITADDR=$PWD/tools/tms-bitaddr/tms-bitaddr \
./scripts/tests/m0/run.sh
# -> M0 pipeline: PASS
build/llvm-clean-build/bin/llvm-lit -v \
build/llvm-clean-build/tools/clang/test --filter='TMS34010'
# -> 3 passed
```
## Use
## Run the M1 gates
These are the documented commands; all three must pass:
```sh
scripts/tms-clang -O2 -I spike -o /tmp/client.ll spike/xclient.c
# 1. LLVM codegen tests: 9 passed
build/llvm-clean-build/bin/llvm-lit -v \
build/llvm-clean-build/test/CodeGen/TMS34010 --filter='m1-'
# 2. Compile, assemble, link and run in MAME
./scripts/tests/m1/run.sh
# -> M1 MAME: PASS
# 3. Rebuild the COFF and check the GSPA encodings with dis34
./scripts/tests/m1/dis-roundtrip.sh
# -> M1 dis34 round-trip: PASS
```
The output is LLVM IR in which every address is a bit address. It is not
assembly or an object file; `llc` support is M1.
`run.sh` compiles five C inputs with `scripts/tms-clang -O2`, runs
`llc -march=tms34010 -O0 -filetype=asm` on the normalized IR, concatenates
the result with `scripts/m1/entry.asm` and `scripts/m1/char-test.asm`,
assembles/links with GSPA/GSPLNK 2.00 under DOSBox, then loads the COFF in
MAME. `dis-roundtrip.sh` runs the same build and then disassembles the linked
`.text` at the symbol addresses from the COFF symbol table. Paths are
overridable with `TMS_CLANG`, `TMS_OPT`, `TMS_BITADDR`, `TMS_LLC`, `MAME`,
`DOSBOX`, `SDK_IMG`, `M1_WORK`, `M1_COFF` and `M1_SDK`.
## What is deliberately missing
## Emit assembly by hand
- TMS34010 instruction selection, registers, frame lowering, asm/MC output.
- The TI ABI: A8 return, A13/A14/A15 dual stack, `RETS 2`, `FS1 = 32`.
- ROM import symbols beyond the subset in `symbols/`.
- Xlib headers/clients on the target, `memcpy`/`byval`/varargs/float/bitfield
lowering, native COFF.
```sh
TMS_CLANG=$PWD/build/llvm-clean-build/bin/clang \
TMS_OPT=$PWD/build/llvm-clean-build/bin/opt \
TMS_BITADDR=$PWD/tools/tms-bitaddr/tms-bitaddr \
scripts/tms-clang -O2 -o /tmp/add.ll scripts/m1/add.c
These are M1–M5 in the design spec, written as separate plans because their
tasks depend on what M0/M1 measure.
build/llvm-clean-build/bin/llc -march=tms34010 -O0 -filetype=asm \
/tmp/add.ll -o /tmp/add.s
```
Use `-O0` on the IR that `tms-clang` already optimized and scaled; feeding
raw clang IR to `llc` double-scales GEP offsets. The lit tests use
`llc -march=tms34010 -O2 -filetype=asm` on hand-written IR to exercise the
selector pipeline directly. Output is assembly only: `llc` cannot emit an
object file or a usable COFF for this target.
## Not yet supported (M2 and later)
- **Object emission:** no `MCCodeEmitter`, object writer, `llc -filetype=obj`,
`clang -c`, or native COFF. GSPA assembles and GSPLNK links; the existing
COFF loader loads.
- **TI ABI:** `A8` return, `A13`/`A14`/`A15` dual stack, `RETS 2`, `FS1 = 32`
across calls, callee-saves, stack arguments, more than four arguments.
- **C features:** calls, locals/`alloca`, structures/`byval`, floats, varargs,
bitfields, 64-bit arithmetic, `memcpy`/`alloca` lowering.
- **Runtime and clients:** ROM/Xlib imports, `strlen`/`malloc`, X11 headers,
real VT1200 clients, debug info.
- **2ine:** M1 uses the proven 1987 GSPA/GSPLNK under DOSBox, not the external
2ine r5.01 tools; 2ine is an M2 speed-up.
These are M2–M5 in the design spec, each with its own plan.
scripts/build-llvm.sh
+80 -16
View File
@@ -1,28 +1,92 @@
#!/bin/sh
# Build the pinned LLVM 23.1.5 fork from a clean clone and the committed
# patch series. Intended for CI and for verifying that patches/llvm is the
# complete fork delta.
# Build the TMS34010 LLVM compiler.
#
# Two modes:
#
# upstream + patches (M0, default)
# ./scripts/build-llvm.sh
# Clones llvmorg-23.1.5 into build/llvm-clean, applies patches/llvm and
# builds it. This mode owns its clone and resets it, but refuses to touch
# a checkout whose origin is not llvm-project.
#
# compiler fork (M1 and later)
# LLVM_SRC=$PWD/build/llvm-clean LLVM_BRANCH=tms34010 \
# ./scripts/build-llvm.sh
# LLVM_REPO=gitea@git.wut.ee:opencode.siil.arti.ee/llvm-tms34010.git \
# LLVM_BRANCH=tms34010 ./scripts/build-llvm.sh
# Build the fork branch in place. Fork mode never checks out, resets,
# cleans or patches the source tree; it only configures and builds it.
#
# Environment:
# LLVM_SRC source tree (default build/llvm-clean)
# LLVM_BUILD build tree (default build/llvm-clean-build)
# LLVM_REPO git URL to clone when LLVM_SRC does not exist
# LLVM_BRANCH branch to clone/build; setting this (or LLVM_REPO) selects
# fork mode
set -eu
ROOT=$(cd "$(dirname "$0")/.." && pwd)
cd "$ROOT"
mkdir -p build
SRC=${LLVM_SRC:-build/llvm-clean}
BUILD=${LLVM_BUILD:-build/llvm-clean-build}
REPO=${LLVM_REPO:-}
BRANCH=${LLVM_BRANCH:-}
TAG=llvmorg-23.1.5
if [ ! -d "$SRC/.git" ]; then
git clone --depth 1 --branch "$TAG" \
https://github.com/llvm/llvm-project.git "$SRC"
fi
git -C "$SRC" checkout --detach "$TAG"
git -C "$SRC" reset --hard
git -C "$SRC" clean -fdx
origin_url() {
git -C "$1" config --get remote.origin.url 2>/dev/null || true
}
PATCHROOT=$(pwd)/patches/llvm
for p in "$PATCHROOT"/*.patch; do
[ -e "$p" ] || continue
git -C "$SRC" apply "$p"
done
if [ -n "$REPO" ] || [ -n "$BRANCH" ]; then
echo "== compiler fork mode: $SRC${BRANCH:+ (branch $BRANCH)}"
if [ ! -e "$SRC/.git" ]; then
[ -n "$REPO" ] || {
echo "build-llvm: $SRC is not a git checkout; set LLVM_REPO to clone it" >&2
exit 1
}
if [ -n "$BRANCH" ]; then
git clone --branch "$BRANCH" "$REPO" "$SRC"
else
git clone "$REPO" "$SRC"
fi
fi
if [ -n "$BRANCH" ]; then
CURRENT=$(git -C "$SRC" rev-parse --abbrev-ref HEAD)
if [ "$CURRENT" != "$BRANCH" ]; then
echo "build-llvm: $SRC is on '$CURRENT', not '$BRANCH'" >&2
echo "build-llvm: check the branch out first; fork mode never" >&2
echo "build-llvm: modifies the source tree" >&2
exit 1
fi
fi
else
echo "== upstream + patches mode: $SRC (tag $TAG)"
if [ -e "$SRC/.git" ]; then
URL=$(origin_url "$SRC")
case "$URL" in
*llvm-project*) ;;
*)
echo "build-llvm: refusing to reset $SRC: origin is" >&2
echo "build-llvm: '${URL:-<none>}', not llvm-project." >&2
echo "build-llvm: use LLVM_REPO/LLVM_BRANCH for a compiler fork" >&2
echo "build-llvm: checkout (default mode must not reset it)" >&2
exit 1
;;
esac
else
git clone --depth 1 --branch "$TAG" \
https://github.com/llvm/llvm-project.git "$SRC"
fi
git -C "$SRC" checkout --detach "$TAG"
git -C "$SRC" reset --hard
git -C "$SRC" clean -fdx
PATCHROOT=$(pwd)/patches/llvm
for p in "$PATCHROOT"/*.patch; do
[ -e "$p" ] || continue
git -C "$SRC" apply "$p"
done
fi
cmake -S "$SRC/llvm" -B "$BUILD" -G Ninja \
-DCMAKE_BUILD_TYPE=Release -DLLVM_ENABLE_ASSERTIONS=ON \
@@ -30,4 +94,4 @@ cmake -S "$SRC/llvm" -B "$BUILD" -G Ninja \
-DLLVM_TARGETS_TO_BUILD="X86;TMS34010" \
-DLLVM_INCLUDE_TESTS=ON -DLLVM_INCLUDE_EXAMPLES=OFF \
-DLLVM_INCLUDE_BENCHMARKS=OFF
ninja -C "$BUILD" clang opt llvm-config FileCheck count not
ninja -C "$BUILD" clang opt llc llvm-config FileCheck count not