Files
SplitBit-Emulator/makefile
T
AnachronautandClaude Opus 5 4c3eac8d9c Widen the memory controller's path to sixteen bits
The controller now reaches bank memory two bytes at a time, so a transfer whose source,
destination and length are all even moves two bytes a cycle between banks and one within
a bank - twice what each was. A 256 byte block between banks falls from 257 cycles to 129.

Alignment is required all three ways because a word is read at an even address and written
at an even address; an odd anything would mean shifting bytes across word boundaries to
line them up, which is a different design. A misaligned transfer falls back to the byte a
cycle it cost before, so nothing already written got slower.

THE CPU DOES NOT CHANGE. It still sees eight bits, a Data Pointer still addresses a byte,
and no instruction means anything different. This is a peripheral getting faster, which is
why it is worth doing now rather than after more is built on top of it.

The rule is deliberately visible rather than smoothed over: aligning a buffer costs nothing
and halves what moving it costs, and a cost a program cannot see is a cost it cannot avoid.

Tests/cycles.sh is new, and is the test the Test Manual has always said this kind of change
would need - run.sh strips the cycle count from every recorded result, so nothing else in
the suite can see any of this. It pins the RATE rather than a total: each case runs twice
from programs whose instructions are identical but for the byte written to the Command
port, once asking for the transfer and once for GuardOff, which costs nothing beyond the
port write. The difference is the transfer and nothing else. Verified by disabling the
widening, which failed exactly the three aligned cases and left the five misaligned ones
passing.

The Programming Manual gains a section saying what a transfer costs, which it never said at
all - it only promised a transfer does not wait, which is a different claim and could be
read as promising it is free.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E2JrLzFvuFX9fgi1LDRjrW
2026-08-28 21:30:12 -04:00

220 lines
8.0 KiB
Makefile

# SplitBit Emulator and Assembler Makefile
# Anachronaut
# 10/16/2024
# Compiler and flags
CC ?= gcc
CFLAGS ?= -Wall -Os
PREFIX ?= /usr/local
# Have the compiler write out which headers each object depends on, so that
# editing a header rebuilds everything that includes it.
DEPFLAGS = -MMD -MP
# The emulator and the assembler use POSIX interfaces that ISO C does not have:
# realpath, clock_gettime,
# strdup, dirname and getopt. Asking for POSIX.1-2008 by name means the build does
# not rely on the compiler happening to default to a mode where those are visible,
# and it survives someone overriding CFLAGS, which is why it is kept separate.
#
# _XOPEN_SOURCE=700 IS POSIX.1-2008, plus the XSI extensions. The plain
# _POSIX_C_SOURCE=200809L was here and is not quite enough: realpath is an XSI interface,
# so under -std=c11 -pedantic it went undeclared and the assembler would not compile. The
# ordinary build never noticed, because without -std=c11 the compiler's own default
# already declares it. The README makes a claim about the strict build, so 'make strict'
# below settles it rather than leaving it to be discovered.
POSIXFLAGS = -D_XOPEN_SOURCE=700
# Directories
SRC_DIR_EMU = Source/Emulator
SRC_DIR_ASM = Source/Assembler
SRC_DIR_DSK = Source/DiskTool
SRC_DIR_LINT = Source/Linter
OBJ_DIR = Object
# Source files
EMU_SRCS = emulator.c io.c controller.c utility.c cpu.c bootstrap.c assembly.c rom.c
ASM_SRCS = Assembler.c assembly.c firstPass.c Assm-util.c secondPass.c
DSK_SRCS = SplitDisk.c
LINT_SRCS = Linter.c
EMU_OBJS = $(EMU_SRCS:%.c=$(OBJ_DIR)/%.o)
ASM_OBJS = $(ASM_SRCS:%.c=$(OBJ_DIR)/%.o)
DSK_OBJS = $(DSK_SRCS:%.c=$(OBJ_DIR)/%.o)
LINT_OBJS = $(LINT_SRCS:%.c=$(OBJ_DIR)/%.o) $(OBJ_DIR)/assembly.o
# Output binary names
EMU_TARGET = SplitBit
ASM_TARGET = Assembler
DSK_TARGET = SplitDisk
LINT_TARGET = SplitLint
# Default target: build the emulator and its three host-side tools.
all: $(EMU_TARGET) $(ASM_TARGET) $(DSK_TARGET) $(LINT_TARGET)
# ---- The ROM the machine wakes up in ----
#
# Generated from the assembly rather than committed, because a copy of a program kept
# beside the program is a copy that goes stale. It needs the assembler, which is built
# first; that is a real dependency and saying so is better than hiding it.
#
# od and awk rather than xxd, which is not everywhere, and rather than python, which the
# README does not ask anybody to install to build this.
$(SRC_DIR_EMU)/rom.c: Programs/Boot/stage1.asm $(ASM_TARGET)
@mkdir -p $(OBJ_DIR)
@./$(ASM_TARGET) Programs/Boot/stage1.asm -o $(OBJ_DIR)/stage1.bin > /dev/null
@{ \
echo '// rom.c'; \
echo '// GENERATED from Programs/Boot/stage1.asm by the makefile. Do not edit.'; \
echo '//'; \
echo '// The first thing the machine runs, and the only part of it that is not on'; \
echo '// the disk. On hardware this is a chip; here it is an array, placed into'; \
echo '// Program Memory at reset the way a shadowed ROM is.'; \
echo ''; \
echo '#include "rom.h"'; \
echo ''; \
echo 'const unsigned char bootROM[] = {'; \
od -v -An -tu1 $(OBJ_DIR)/stage1.bin | awk '{ printf " "; for (i = 1; i <= NF; i++) printf " %s,", $$i; print "" }'; \
echo '};'; \
echo ''; \
echo 'const unsigned long bootROMBytes = sizeof(bootROM);'; \
} > $@
$(OBJ_DIR)/rom.o: $(SRC_DIR_EMU)/rom.c $(SRC_DIR_EMU)/rom.h
# Emulator binary
$(EMU_TARGET): $(EMU_OBJS)
$(CC) $(CFLAGS) -o $(EMU_TARGET) $(EMU_OBJS)
# Assembler binary
$(ASM_TARGET): $(ASM_OBJS)
$(CC) $(CFLAGS) -o $(ASM_TARGET) $(ASM_OBJS)
# Compile emulator source files to object files
$(OBJ_DIR)/%.o: $(SRC_DIR_EMU)/%.c
mkdir -p $(OBJ_DIR)
$(CC) $(CFLAGS) $(POSIXFLAGS) $(DEPFLAGS) -c $< -o $@
# Disk tool binary
$(DSK_TARGET): $(DSK_OBJS)
$(CC) $(CFLAGS) -o $(DSK_TARGET) $(DSK_OBJS)
# Assembly source linter. The instruction table is shared with the assembler and
# emulator so that adding an opcode cannot leave the linter with a private copy.
$(LINT_TARGET): $(LINT_OBJS)
$(CC) $(CFLAGS) -o $(LINT_TARGET) $(LINT_OBJS)
$(OBJ_DIR)/Linter.o: $(SRC_DIR_LINT)/Linter.c
@mkdir -p $(OBJ_DIR)
$(CC) $(CFLAGS) $(POSIXFLAGS) $(DEPFLAGS) -c $< -o $@
# Compile disk tool source files to object files
$(OBJ_DIR)/%.o: $(SRC_DIR_DSK)/%.c
@mkdir -p $(OBJ_DIR)
$(CC) $(CFLAGS) $(POSIXFLAGS) $(DEPFLAGS) -c $< -o $@
# Compile assembler source files to object files
$(OBJ_DIR)/%.o: $(SRC_DIR_ASM)/%.c
mkdir -p $(OBJ_DIR)
$(CC) $(CFLAGS) $(POSIXFLAGS) $(DEPFLAGS) -c $< -o $@
# Pull in the header dependencies written out by the compiler above.
-include $(EMU_OBJS:.o=.d) $(ASM_OBJS:.o=.d) $(DSK_OBJS:.o=.d) $(LINT_OBJS:.o=.d)
# ---- The strict build the README promises ----
#
# "The sources are ISO C and build clean under -std=c11 -pedantic with -Wall -Wextra."
# That is a claim somebody may check by typing it, so the suite checks it first. It was
# false when this target was written: realpath went undeclared under a feature test macro
# that did not reach far enough, which the ordinary -Os build never saw.
STRICT = -std=c11 -pedantic -Wall -Wextra -Werror $(POSIXFLAGS)
strict:
@for source in $(SRC_DIR_EMU)/*.c $(SRC_DIR_ASM)/*.c $(SRC_DIR_DSK)/*.c $(SRC_DIR_LINT)/*.c; do \
$(CC) $(STRICT) -c $$source -o /dev/null || exit 1; \
done
@echo "The sources build clean under -std=c11 -pedantic."
# Run the test suite against the programs in Programs/
test: $(EMU_TARGET) $(ASM_TARGET) $(DSK_TARGET) $(LINT_TARGET) strict
@./Tests/lint.sh
@echo
@echo
@./Tests/run.sh
@echo
@./Tests/disk.sh
@echo
@./Tests/cycles.sh
@echo
@./Tests/terminal.sh
@echo
@./Tests/native.sh
@echo
@./Tests/agree.sh
@echo
@./Tests/docs.sh
# Rebuild all three tools with the address and undefined behaviour sanitizers and run
# the test suite under them. Slower than 'make test', and worth running before a
# release or after anything that touches memory handling.
#
# THE WHOLE SUITE, which it did not used to be: it built all three tools sanitized and
# then ran only run.sh and terminal.sh, so SplitDisk was compiled with the sanitizers and
# never exercised, and native.sh - which drives the assembler and the emulator harder than
# anything else here - was skipped. Those are the parts where block arithmetic on disk
# images and buffer indexing in two assemblers live, which is exactly what the sanitizers
# are for. Adding the three of them costs about six seconds.
#
# The sanitizers catch reads and writes off the end of an array, use after free,
# leaks, and undefined arithmetic. What they do NOT usefully catch here is uninitialised
# memory: AddressSanitizer's junk fill is a toolchain default this build does not
# configure, there are six heap allocations in the repository and the largest is a
# deliberate calloc, and the machine's own memories are static arrays it never touches.
#
# If the suite fails, the sanitizer binaries are deliberately left in place so
# that the failing case can be run again by hand. 'make' puts the normal ones back.
SANITIZE_FLAGS = -Wall -Wextra -g -O1 -fsanitize=address,undefined -fno-omit-frame-pointer
sanitize:
@$(MAKE) --no-print-directory clean
@$(MAKE) --no-print-directory CFLAGS="$(SANITIZE_FLAGS)"
@echo "Running the test suite under AddressSanitizer and UndefinedBehaviorSanitizer."
@./Tests/lint.sh
@echo
@./Tests/run.sh
@echo
@./Tests/disk.sh
@echo
@./Tests/cycles.sh
@echo
@./Tests/terminal.sh
@echo
@./Tests/native.sh
@echo
@./Tests/agree.sh
@echo
@./Tests/docs.sh
@$(MAKE) --no-print-directory clean
@$(MAKE) --no-print-directory
@echo "Sanitizer run finished cleanly. Normal binaries rebuilt."
# Record the current output of every test as the expected result.
# Only do this when the current output is known to be correct.
bless: $(EMU_TARGET) $(ASM_TARGET)
@./Tests/run.sh --bless
# Clean up object and binary files
clean:
rm -rf $(OBJ_DIR)
rm -rf Tests/build
rm -f $(SRC_DIR_EMU)/rom.c
rm -f $(EMU_TARGET) $(ASM_TARGET) $(DSK_TARGET) $(LINT_TARGET)
# Install compiled binaries
install: $(EMU_TARGET) $(ASM_TARGET) $(DSK_TARGET) $(LINT_TARGET)
mkdir -p "$(PREFIX)/bin"
install -m 755 $^ "$(PREFIX)/bin/"
# Phony targets
.PHONY: all clean install test bless sanitize