Add deloopify, a trace compactor

Also some makefile fettling

4 files changed, 198 insertions, 44 deletions

diff --git a/README b/README
index 4fef5ab..5d84783 100644
--- a/README
+++ b/README
@@ -29,6 +29,9 @@ this shell script will run vbtracetool in such a way to do a POST
 
 try either of these in x and lose.
 
+Bonus trace processor:
+"./deloopify TRACENAME" will compact loops in your trace
+
 Killing:
 do not interrupt, once started -- you will likely earn yourself a hung machine
 if you really really don't like what's happening, turn the power off
diff --git a/deloopify.c b/deloopify.c
new file mode 100644
index 0000000..c9d8a5f
--- /dev/null
+++ b/deloopify.c
@@ -0,0 +1,181 @@
+/*
+ * deloopify, a loop compactor for vbtracetool generated traces
+ *
+ * Copyright 2007 Stuart Bennett <sb476@cam.ac.uk>
+ *
+ * This program is released under the terms of the GNU General Public License, version 2
+ */
+
+#include <pci/pci.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+/* length of sliding history of line beginnings */
+#define MAX_LOOP_LEN 2000
+
+bool loop_concluded(FILE *tracef, uint16_t loopendip)
+{
+	/* from the current position in the file (assumed to be a loop
+	 * instruction), see if the next ip is 2 bigger (implies we fell out
+	 * of the loop)
+	 *
+	 * if we hit eof, behave as if loop ended, and reset eof, so end of
+	 * file gets read and written out by the caller
+	 */
+	fpos_t loopendpos;
+	char line[180];
+
+	fgetpos(tracef, &loopendpos);
+	do {
+		if (!fgets(line, 180, tracef))
+			break;	// eof
+	} while (line[4] != ':');
+	fsetpos(tracef, &loopendpos);
+	clearerr(tracef);
+
+	if (strtoul(&line[5], NULL, 16) != loopendip + 2)	// we looped
+		return false;
+
+	return true;
+}
+
+int main(int argc, char *argv[])
+{
+	FILE *tracef, *outf;
+	long lineno = -1;
+	char sig[MAX_LOOP_LEN][100];
+	char loopsig[] = "Loop";
+
+	char outname[1024];
+	strcpy(outname, argv[1]);
+	strcat(outname, "-deloopd");
+	if (!(tracef = fopen(argv[1], "r"))) {
+		printf("File open failed\n");
+		exit(EXIT_FAILURE);
+	}
+	if (!(outf = fopen(outname, "w"))) {
+		printf("File open failed\n");
+		exit(EXIT_FAILURE);
+	}
+
+	memset(sig, 0, sizeof(sig));
+	while (!feof(tracef)) {
+		char line[180];
+
+		if (!fgets(line, 180, tracef))
+			break;	// eof
+		fwrite(line, strlen(line), 1, outf);
+		lineno += 1;
+		/* we maintain a sliding history of previous line beginnings in sig,
+		 * so we can search for the cs:ip of where the loop began
+		 */
+		strncpy(sig[lineno % MAX_LOOP_LEN], line, 9);
+		/* this is only necessary if you want intelligible debug printing */
+//sig[lineno % MAX_LOOP_LEN][9] = 0;
+
+		if (strstr(line, "LOOP")) {
+			uint16_t loopendip = strtoul(&line[5], NULL, 16);
+
+			/* if the trace doesn't loop, there's nothing to do */
+			if (loop_concluded(tracef, loopendip))
+				goto next;
+
+//fprintf(outf, "loop\n");
+			/* adapt the loop ip detection for LOOP, LOOPE and LOOPNE opcodes */
+			int jumpipofs = 0;
+			switch (line[34]) {
+			case 'E':
+				jumpipofs += 1;
+				break;
+			case 'N':
+				jumpipofs += 2;
+				break;
+			}
+
+			uint16_t loopcs = strtoul(&line[0], NULL, 16);
+			uint16_t loopstartip;
+			/* for some reason, the loop ip is occasionally 8 bytes */
+			if ((strtoul(&line[35], NULL, 16) & 0xffff0000) == 0xffff0000)
+				loopstartip = strtoul(&line[39 + jumpipofs], NULL, 16);
+			else
+				loopstartip = strtoul(&line[35 + jumpipofs], NULL, 16);
+			/* the 'signature' (cs:ip) of where the loop started */
+			char startsig[10];
+			sprintf(startsig, "%04x:%04x", loopcs, loopstartip);
+
+			int looplen = 1;
+			int bonus = 0;
+			for ( ; looplen < MAX_LOOP_LEN; looplen++) {
+				/* on our way back through the sliding sig list, retrive the
+				 * uncontracted length of any previously compacted loops
+				 */
+				if (!strncmp(loopsig, sig[(lineno - looplen) % MAX_LOOP_LEN], 4)) {
+					int len, mult;
+					sscanf(sig[(lineno - looplen) % MAX_LOOP_LEN], "Loop of last %d lines repeats a further %d times", &len, &mult);
+					if (mult)
+						bonus += len * mult - 1;
+				}
+				/* when we find the loop start, stop */
+				if (!strncmp(startsig, sig[(lineno - looplen) % MAX_LOOP_LEN], 9))
+					break;
+			}
+
+			if (looplen == MAX_LOOP_LEN) {	/* fail */
+				printf("Didn't find start of loop -- try increasing MAX_LOOP_LEN?\n");
+				exit(EXIT_FAILURE);
+			}
+
+//fprintf(outf, "loop sig %s, startsig %s, found %s, looplen %d, bonus %d\n", sig[lineno % MAX_LOOP_LEN], startsig, sig[(lineno - looplen - 4) % MAX_LOOP_LEN], looplen, bonus);
+			/* Assume loop instr adds 4 lines */
+			looplen += 4;
+
+			fpos_t loopendpos;
+			int loopcount = 0;
+			/* slurp until the loop terminates */
+			while (!loop_concluded(tracef, loopendip)) {
+				fgetpos(tracef, &loopendpos);
+				/* we know the loop length from finding the signature, so move
+				 * forward the same length plus the length of any previous
+				 * contractions (as stuff further into the file will not be
+				 * contracted) and see if we're at another loop instruction
+				 */
+				for (int i = 0; i < looplen + bonus; i++)
+					if (!fgets(line, 180, tracef)) {	// eof
+						clearerr(tracef);
+						goto dissimilar;
+					}
+				/* if we are, treat this loop as identical, and carry on */
+				if (!strncmp(line, sig[lineno % MAX_LOOP_LEN], 9)) {
+					loopcount++;
+					continue;
+				}
+dissimilar:
+				/* if we're not, this loop is not identical, and cannot be compacted */
+				fsetpos(tracef, &loopendpos);
+//fprintf(outf, "loop changed size -- looking for %s, got %s\n", sig[lineno % MAX_LOOP_LEN], line);
+				break;
+			}
+			/* if we found one or more identical length loops, replace them in the
+			 * output with a contraction message. Also add this message to the
+			 * sliding signature log, so that it can be counted in, if this loop
+			 * is inside another loop
+			 */
+			if (loopcount) {
+				char loopstr[100];
+				snprintf(loopstr, 100, "Loop of last %d lines repeats a further %d times", looplen, loopcount);
+				fprintf(outf, "\n%s\n\n\n", loopstr);
+				lineno += 1;
+				strncpy(sig[lineno % MAX_LOOP_LEN], loopstr, 100);
+			}
+next:
+			continue;
+		}
+	}
+
+	return 0;
+}
diff --git a/makefile b/makefile
index 5019daf..ddf3f71 100644
--- a/makefile
+++ b/makefile
@@ -1,40 +1,8 @@
-#############################################################################
-#
-#						Realmode X86 Emulator Library
-#
-#            	Copyright (C) 1996-1999 SciTech Software, Inc.
-#
-#  ========================================================================
-#
-#  Permission to use, copy, modify, distribute, and sell this software and
-#  its documentation for any purpose is hereby granted without fee,
-#  provided that the above copyright notice appear in all copies and that
-#  both that copyright notice and this permission notice appear in
-#  supporting documentation, and that the name of the authors not be used
-#  in advertising or publicity pertaining to distribution of the software
-#  without specific, written prior permission.  The authors makes no
-#  representations about the suitability of this software for any purpose.
-#  It is provided "as is" without express or implied warranty.
-#
-#  THE AUTHORS DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
-#  INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO
-#  EVENT SHALL THE AUTHORS BE LIABLE FOR ANY SPECIAL, INDIRECT OR
-#  CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF
-#  USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR
-#  OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
-#  PERFORMANCE OF THIS SOFTWARE.
-#
-#  ========================================================================
-#
-# Descripton:   Linux specific makefile for the x86emu library.
-#
-#############################################################################
+LIBX86EMU = libx86emu.a
+VBTT=vbtracetool
+DL=deloopify
 
-TARGETLIB = libx86emu.a
-
-PROG=vbtracetool
-
-OBJS=\
+LIBX86EMUOBJS=\
 debug.o \
 decode.o \
 fpu.o \
@@ -50,13 +18,16 @@ vbtracetool.o \
 thunk.o \
 x86-common.o
 
-all: $(PROG)
+all: $(VBTT) $(DL)
+
+$(LIBX86EMU): $(LIBX86EMUOBJS)
+	ar r $(LIBX86EMU) $(LIBX86EMUOBJS)
 
-$(PROG): $(TARGETLIB) $(VBTTOBJS)
-	gcc $(VBTTOBJS) $(TARGETLIB) -lpci -lz -s -o $(PROG)
+$(VBTT): $(LIBX86EMU) $(VBTTOBJS)
+	gcc $(VBTTOBJS) $(LIBX86EMU) -lpci -lz -s -o $(VBTT)
 
-$(TARGETLIB): $(OBJS)
-	ar r $(TARGETLIB) $(OBJS)
+$(DL):
+	gcc $(CFLAGS) -std=gnu99 $(shell getconf LFS_CFLAGS) $(DL).c -o $(DL)
 
 INCS   = -I. -Ix86emu
 CFLAGS = -DDEBUG -Wall -O3
@@ -67,6 +38,6 @@ CFLAGS = -DDEBUG -Wall -O3
 distclean: clean
 
 clean:
-	rm -f *.a *.o *~ $(PROG)
+	rm -f *.a *.o *~ $(VBTT) $(DL)
 
 install:
diff --git a/vbtracetool.c b/vbtracetool.c
index 2b6472a..84ddf80 100644
--- a/vbtracetool.c
+++ b/vbtracetool.c
@@ -5,8 +5,7 @@
  *
  * Based on vbetool.c, Copyright Matthew Garrett <mjg59@srcf.ucam.org>
  *
- * This program is released under the terms of the GNU General Public License,
- * version 2
+ * This program is released under the terms of the GNU General Public License, version 2
  */
 
 #include <pci/pci.h>