uxndebug

commit 81418dfc8eca7f85e745b31e2d2d15b73058a7fb
parent b97e22a90af15408ca33c0b9f30e1235881474cc
Author: aabacchus <ben@bvnf.space>
Date:   Sun,  5 Jun 2022 20:19:32 +0100

naive debug file generation, add tool to parse debug file, add file format manpage

Diffstat:
A
uxndebug.5
 | 
35
+++++++++++++++++++++++++++++++++++
M
uxndebug.c
 | 
28
++++++++++++++++++++++++++++
A
uxnsolve
 | 
50
++++++++++++++++++++++++++++++++++++++++++++++++++

3 files changed, 113 insertions(+), 0 deletions(-)
diff --git a/uxndebug.5 b/uxndebug.5
@@ -0,0 +1,35 @@
+.Dd June 5, 2022
+.Dt UXNDEBUG 5
+.Os
+.Sh NAME
+.Nm uxndebug ,
+.Nm .tal.debug
+.Nd uxntal debug info file format
+.Sh DESCRIPTION
+The format of debug information for an uxntal rom file is documented in this file.
+.Pp
+Each line of the debug file is terminated by a single
+.Aq newline
+character, and contains a single hexadecimal number.
+The line number of the file corresponds to the
+.Em byte
+of the rom,
+and the number written on the line signifies which
+.Em token
+in the source files was assembled into that
+.Em byte .
+The name of the debug file is used to identify which source file contains the relevant
+.Em token Ns s .
+.Sh FILES
+For a source file
+.Pa example.tal ,
+the corresponding rom
+.Pa example.rom
+and debug information file
+.Pa example.tal.debug
+are created by running:
+.Bd -literal -offset indent
+uxndebug example.tal example.rom
+.Ed
+.Sh SEE ALSO
+.Xr uxndebug 1
diff --git a/uxndebug.c b/uxndebug.c
@@ -48,6 +48,9 @@ static int litlast = 0;
 static int jsrlast = 0;
 
 static int token_number = 0;
+static int debug_lineno = 0;
+static char *debug_filename = "debug.tal.debug";
+FILE *debug_file = NULL;
 
 /* clang-format off */
 
@@ -191,6 +194,17 @@ makereference(char *scope, char *label, Uint16 addr)
 	return 1;
 }
 
+static void
+debug_next_byte(void) {
+	while (p.ptr - 0x0100 != debug_lineno) {
+		fprintf(debug_file, "\n");
+		debug_lineno++;
+		/* TODO: better way to handle this is to write to buffer with byte,token pairs, then write to file at end */
+	}
+	fprintf(debug_file, "%d\n", token_number);
+	debug_lineno++;
+}
+
 static int
 writebyte(Uint8 b)
 {
@@ -200,6 +214,7 @@ writebyte(Uint8 b)
 		return error("Writing after the end of RAM", "");
 	else if(p.ptr < p.length)
 		return error("Memory overwrite", "");
+	debug_next_byte();
 	p.data[p.ptr++] = b;
 	p.length = p.ptr;
 	litlast = 0;
@@ -211,11 +226,14 @@ static int
 writeopcode(char *w)
 {
 	Uint8 res;
+	/* disable optimizations which go back
 	if(jsrlast && scmp(w, "JMP2r", 5)) { /* tail-call optimization */
+	/*
 		p.data[p.ptr - 1] = jsrlast == 2 ? findopcode("JMP2") : findopcode("JMP");
 		jsrlast = 0;
 		return 1;
 	}
+	*/
 	res = writebyte(findopcode(w));
 	if(scmp(w, "JSR2", 4))
 		jsrlast = 2;
@@ -235,12 +253,15 @@ writeshort(Uint16 s, int lit)
 static int
 writelitbyte(Uint8 b)
 {
+	/* disable optimizations which go back
 	if(litlast) { /* literals optimization */
+	/*
 		Uint8 hb = p.data[p.ptr - 1];
 		p.ptr -= 2;
 		p.length = p.ptr;
 		return writeshort((hb << 8) + b, 1);
 	}
+	*/
 	if(!writebyte(findopcode("LIT"))) return 0;
 	if(!writebyte(b)) return 0;
 	litlast = 1;
@@ -454,6 +475,12 @@ int
 main(int argc, char *argv[])
 {
 	FILE *src, *dst;
+	debug_file = fopen(debug_filename, "w");
+	if (debug_file == NULL) {
+		perror(debug_filename);
+		return 1;
+	}
+
 	if(argc < 3)
 		return !error("usage", "input.tal output.rom");
 	if(!(src = fopen(argv[1], "r")))
@@ -466,5 +493,6 @@ main(int argc, char *argv[])
 		return !error("Assembly", "Output rom is empty.");
 	fwrite(p.data + TRIM, p.length - TRIM, 1, dst);
 	review(argv[2]);
+	fclose(debug_file);
 	return 0;
 }
diff --git a/uxnsolve b/uxnsolve
@@ -0,0 +1,50 @@
+#!/bin/sh
+
+# desired behaviour
+# shellcheck disable=SC2015
+[ "$#" -eq 2 ] && [ -r "$1" ] || {
+    printf "usage: %s file.tal.debug address\n" "$0" >&2
+    exit 1
+}
+
+case "$1" in
+    *.tal.debug) ;;
+    *)
+        printf "%s: does not end in .tal.debug, can't work out source file\n" "$1" >&2
+        exit 1
+    ;;
+esac
+
+debug_file="$1"
+source_file="${1%.debug}"
+
+address="$(printf "%s" "$2" | tr 'a-f' 'A-F')"
+dec_address="$(printf "16iAo%sp\n" "$address" | dc || kill 0)"
+# add 1 because the zeroth byte is on the first line
+dec_address="$((dec_address + 1))"
+# TODO: should we subtract the #0100 offset or request the user to do that?
+
+is_integer() {
+    printf %d "$1" >/dev/null 2>&1
+}
+
+token="$(sed -n "${dec_address}p" "$debug_file")"
+: "${token:?can\'t find integer from debug file at given byte}"
+is_integer "$token" || {
+    printf "can't find an integer token number in %s for byte %X\n" "$debug_file" "$address" >&2
+    exit 1
+}
+
+i=0
+lineno=0
+while read -r line; do
+    lineno="$((lineno + 1))"
+    # word splitting is desired.
+    # shellcheck disable=SC2086
+    set -- $line
+    i="$((i + $#))"
+    [ "$i" -ge "$token" ] && {
+        printf "line %d: %s\n" "$lineno" "$line"
+        break
+    }
+done < "$source_file"