Use separate buffers in fsort.c and msort.c. Fixes a problem with

buffer corruptions for some very large files.  Also fixes some
TRECHEADER vs. RECHEADER mismatches.  Back in after unlock.

4 files changed, 35 insertions, 45 deletions

diff --git a/usr.bin/sort/fsort.c b/usr.bin/sort/fsort.c
index cd25ab559a4..d706dd44ab5 100644
--- a/usr.bin/sort/fsort.c
+++ b/usr.bin/sort/fsort.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: fsort.c,v 1.17 2007/03/11 00:38:01 deraadt Exp $	*/
+/*	$OpenBSD: fsort.c,v 1.18 2007/03/13 17:33:58 millert Exp $	*/
 
 /*-
  * Copyright (c) 1993
@@ -36,7 +36,7 @@
 #if 0
 static char sccsid[] = "@(#)fsort.c	8.1 (Berkeley) 6/6/93";
 #else
-static char rcsid[] = "$OpenBSD: fsort.c,v 1.17 2007/03/11 00:38:01 deraadt Exp $";
+static char rcsid[] = "$OpenBSD: fsort.c,v 1.18 2007/03/13 17:33:58 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -53,10 +53,9 @@ static char rcsid[] = "$OpenBSD: fsort.c,v 1.17 2007/03/11 00:38:01 deraadt Exp
 #include <stdlib.h>
 #include <string.h>
 
-u_char *buffer = NULL, *bufend = NULL, *linebuf = NULL;
-size_t bufsize = BUFSIZE, linebuf_size = MAXLLEN;
+u_char *linebuf;
+size_t linebuf_size = MAXLLEN;
 struct tempfile fstack[MAXFCT];
-extern char toutpath[];
 #define FSORTMAX 4
 int PANIC = FSORTMAX;
 
@@ -64,8 +63,9 @@ void
 fsort(int binno, int depth, union f_handle infiles, int nfiles, FILE *outfp,
     struct field *ftbl)
 {
-	u_char *weights, **keypos, *tmpbuf;
-	static u_char **keylist;
+	u_char *weights, **keypos, *bufend, *tmpbuf;
+	static u_char *buffer, **keylist;
+	static size_t bufsize;
 	int ntfiles, mfct = 0, total, i, maxb, lastb, panic = 0;
 	int c, nelem;
 	long sizes[NBINS+1];
@@ -85,10 +85,13 @@ fsort(int binno, int depth, union f_handle infiles, int nfiles, FILE *outfp,
 		tfield[0].weights = ascii;
 	tfield[0].icol.num = 1;
 	weights = ftbl[0].weights;
-	if (keylist == NULL) {
-		if ((keylist = calloc(MAXNUM, sizeof(u_char *))) == NULL)
+	if (buffer == NULL) {
+		bufsize = BUFSIZE;
+		if ((buffer = malloc(bufsize)) == NULL ||
+		    (keylist = calloc(MAXNUM, sizeof(u_char *))) == NULL)
 			err(2, NULL);
 	}
+	bufend = buffer + bufsize - 1;
 	if (binno >= 0) {
 		tfiles.top = infiles.top + nfiles;
 		get = getnext;
diff --git a/usr.bin/sort/fsort.h b/usr.bin/sort/fsort.h
index 6b85bf05e23..89913c6a199 100644
--- a/usr.bin/sort/fsort.h
+++ b/usr.bin/sort/fsort.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: fsort.h,v 1.9 2007/03/11 00:38:01 deraadt Exp $	*/
+/*	$OpenBSD: fsort.h,v 1.10 2007/03/13 17:33:58 millert Exp $	*/
 
 /*-
  * Copyright (c) 1993
@@ -42,8 +42,8 @@
 #define MAXFCT 1000
 #define MAXLLEN ((1 << min(POW-4, 16)) - 14)
 
-extern u_char *buffer, *bufend, *linebuf;
-extern size_t bufsize, linebuf_size;
+extern u_char *linebuf;
+extern size_t linebuf_size;
 
 /* temp files in the stack have a file descriptor, a largest bin (maxb)
  * which becomes the last non-empty bin (lastb) when the actual largest
diff --git a/usr.bin/sort/msort.c b/usr.bin/sort/msort.c
index 39407e25109..cfd223c4570 100644
--- a/usr.bin/sort/msort.c
+++ b/usr.bin/sort/msort.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: msort.c,v 1.19 2007/03/11 00:38:01 deraadt Exp $	*/
+/*	$OpenBSD: msort.c,v 1.20 2007/03/13 17:33:58 millert Exp $	*/
 
 /*-
  * Copyright (c) 1993
@@ -36,7 +36,7 @@
 #if 0
 static char sccsid[] = "@(#)msort.c	8.1 (Berkeley) 6/6/93";
 #else
-static char rcsid[] = "$OpenBSD: msort.c,v 1.19 2007/03/11 00:38:01 deraadt Exp $";
+static char rcsid[] = "$OpenBSD: msort.c,v 1.20 2007/03/13 17:33:58 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -56,13 +56,10 @@ typedef struct mfile {
 	short flno;
 	RECHEADER rec[1];
 } MFILE;
-typedef struct tmfile {
-	u_char *end;
-	short flno;
-	TRECHEADER rec[1];
-} TMFILE;
-u_char *wts, *wts1 = 0;
-struct mfile *cfilebuf;
+static u_char *wts, *wts1;
+static struct mfile *cfilebuf;
+static void *buffer;
+static size_t bufsize;
 
 static int cmp(RECHEADER *, RECHEADER *);
 static int insert(struct mfile **, struct mfile **, int, int);
@@ -80,21 +77,17 @@ fmerge(int binno, union f_handle files, int nfiles,
 	wts = ftbl->weights;
 	if (!UNIQUE && SINGL_FLD && (ftbl->flags & F))
 		wts1 = (ftbl->flags & R) ? Rascii : ascii;
-	if (!cfilebuf) {
-		cfilebuf = malloc(MAXLLEN + sizeof(TMFILE));
+	if (cfilebuf == NULL) {
+		cfilebuf = malloc(MAXLLEN + sizeof(MFILE));
 		if (cfilebuf == NULL)
 			errx(2, "cannot allocate memory");
 	}
 
-	i = min(16, nfiles) * LALIGN(MAXLLEN+sizeof(TMFILE));
+	i = min(16, nfiles) * LALIGN(MAXLLEN + sizeof(MFILE));
 	if (i > bufsize) {
-		do {
-			bufsize *= 2;
-		} while  (i > bufsize);
-		buffer = realloc(buffer, bufsize);
-		if (!buffer)
-			errx(2, "cannot allocate memory");
-		bufend = buffer + bufsize - 1;
+		bufsize = i;
+		if ((buffer = realloc(buffer, bufsize)) == NULL)
+			err(2, NULL);
 	}
 
 	if (binno >= 0)
@@ -148,7 +141,7 @@ merge(int infl0, int nfiles,
 
 	for (i = j = 0; i < nfiles; i++) {
 		cfile = (MFILE *) (buffer +
-		    i * LALIGN(MAXLLEN + sizeof(TMFILE)));
+		    i * LALIGN(MAXLLEN + sizeof(MFILE)));
 		cfile->flno = j + infl0;
 		cfile->end = cfile->rec->data + MAXLLEN;
 		for (c = 1; c == 1;) {
@@ -253,18 +246,15 @@ order(union f_handle infile,
 	int c;
 	RECHEADER *crec, *prec, *trec;
 
-	if (bufsize < 2 * ALIGN(MAXLLEN + sizeof(TRECHEADER))) {
-		do {
-			bufsize *= 2;
-		} while (bufsize < 2 * ALIGN(MAXLLEN + sizeof(TRECHEADER)));
+	if (bufsize < 2 * ALIGN(MAXLLEN + sizeof(RECHEADER))) {
+		bufsize = 2 * ALIGN(MAXLLEN + sizeof(RECHEADER));
 		if ((buffer = realloc(buffer, bufsize)) == NULL)
 			err(2, NULL);
-		bufend = buffer + bufsize - 1;
 	}
 	crec = (RECHEADER *) buffer;
-	crec_end = ((char *)crec) + ALIGN(MAXLLEN + sizeof(TRECHEADER));
+	crec_end = ((char *)crec) + ALIGN(MAXLLEN + sizeof(RECHEADER));
 	prec = (RECHEADER *) crec_end;
-	prec_end = ((char *)prec) + ALIGN(MAXLLEN + sizeof(TRECHEADER));
+	prec_end = ((char *)prec) + ALIGN(MAXLLEN + sizeof(RECHEADER));
 	wts = ftbl->weights;
 	if (SINGL_FLD && (ftbl->flags & F))
 		wts1 = ftbl->flags & R ? Rascii : ascii;
diff --git a/usr.bin/sort/sort.c b/usr.bin/sort/sort.c
index 48f869b63e4..5205c2304d7 100644
--- a/usr.bin/sort/sort.c
+++ b/usr.bin/sort/sort.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: sort.c,v 1.33 2007/03/11 00:38:01 deraadt Exp $	*/
+/*	$OpenBSD: sort.c,v 1.34 2007/03/13 17:33:58 millert Exp $	*/
 
 /*-
  * Copyright (c) 1993
@@ -42,7 +42,7 @@ static char copyright[] =
 #if 0
 static char sccsid[] = "@(#)sort.c	8.1 (Berkeley) 6/6/93";
 #else
-static char rcsid[] = "$OpenBSD: sort.c,v 1.33 2007/03/11 00:38:01 deraadt Exp $";
+static char rcsid[] = "$OpenBSD: sort.c,v 1.34 2007/03/13 17:33:58 millert Exp $";
 #endif
 #endif /* not lint */
 
@@ -254,10 +254,7 @@ main(int argc, char *argv[])
 	else
 		get = makekey;
 
-	if ((buffer = malloc(bufsize)) == NULL)
-		err(2, NULL);
-	 bufend = buffer + bufsize - 1;
-	 if (!SINGL_FLD) {
+	if (!SINGL_FLD) {
 		if ((linebuf = malloc(linebuf_size)) == NULL)
 			err(2, NULL);
 	}