libfetch: Reduce copying

Reduce the amount of copying we do when performing buffered reads.

MFC after:	1 week
Reviewed by:	op
Differential Revision:	https://reviews.freebsd.org/D58113

lib/libfetch/common.c

@@ -1437,12 +1437,23 @@ fetch_getln(conn_t *conn)
 	/* allocate initial buffer */
 	if (conn->buf == NULL) {
 		if ((conn->buf = malloc(MIN_BUF_SIZE)) == NULL)
-			return (-1);
+			goto fail;
+		conn->line = conn->buf;
 		conn->bufsize = MIN_BUF_SIZE;
 		conn->buflen = 0;
 		conn->pos = 0;
 	}
 
+	/* look at the data we already have */
+	if (conn->pos < conn->buflen) {
+		conn->line = conn->buf + conn->pos;
+		while (conn->pos < conn->buflen)
+			if (conn->buf[conn->pos++] == '\n')
+				goto found;
+		/* reset for the upcoming memmove() */
+		conn->pos = conn->line - conn->buf;
+	}
+
 	/* move residual data up */
 	if (conn->buflen > 0) {
 		if (conn->pos < conn->buflen) {
@@ -1451,20 +1462,16 @@ fetch_getln(conn_t *conn)
 		}
 		conn->buflen -= conn->pos;
 		conn->pos -= conn->pos;
+		conn->line = conn->buf;
 	}
 
-	/* do we have a complete line? */
-	while (conn->pos < conn->buflen)
-		if (conn->buf[conn->pos++] == '\n')
-			goto found;
-
 	for (;;) {
 		/* read as much as we can right now */
 		rlen = fetch_read(conn, conn->buf + conn->buflen,
 		    conn->bufsize - conn->buflen - 1);
 		/* error */
 		if (rlen < 0)
-			return (-1);
+			goto fail;
 		/* advance and terminate */
 		conn->buflen += rlen;
 		conn->buf[conn->buflen] = '\0';
@@ -1480,21 +1487,25 @@ fetch_getln(conn_t *conn)
 			tmp = conn->buf;
 			tmpsize = conn->bufsize * 2;
 			if ((tmp = realloc(tmp, tmpsize)) == NULL)
-				return (-1);
-			conn->buf = tmp;
+				goto fail;
+			conn->line = conn->buf = tmp;
 			conn->bufsize = tmpsize;
 		}
 	}
 	/* connection closed, return what's left */
 	conn->pos = conn->buflen;
 found:
-	rlen = conn->pos;
-	if (rlen > 0 && conn->buf[rlen - 1] == '\n')
-		conn->buf[--rlen] = '\0';
-	if (rlen > 0 && conn->buf[rlen - 1] == '\r')
-		conn->buf[--rlen] = '\0';
-	DEBUGF("<<< %.*s\n", (int)rlen, conn->buf);
-	return (rlen);
+	conn->linelen = (conn->buf + conn->pos) - conn->line;
+	if (conn->linelen > 0 && conn->line[conn->linelen - 1] == '\n')
+		conn->line[--conn->linelen] = '\0';
+	if (conn->linelen > 0 && conn->line[conn->linelen - 1] == '\r')
+		conn->line[--conn->linelen] = '\0';
+	DEBUGF("<<< %.*s\n", (int)conn->linelen, conn->line);
+	return (conn->linelen);
+fail:
+	conn->line = NULL;
+	conn->linelen = 0;
+	return (-1);
 }
 
 
@@ -1520,6 +1531,8 @@ fetch_bufread(conn_t *conn, void *buf, size_t len)
 		conn->buflen = 0;
 		conn->pos = 0;
 	}
+	conn->line = NULL;
+	conn->linelen = 0;
 
 	/* return residual data first */
 	if (conn->buflen > conn->pos) {

lib/libfetch/common.h

@@ -49,8 +49,10 @@ typedef struct fetchconn conn_t;
 struct fetchconn {
 	int		 sd;		/* socket descriptor */
 	char		*buf;		/* buffer */
+	char		*line;		/* line (within buffer) */
 	size_t		 bufsize;	/* buffer size */
 	size_t		 buflen;	/* length of buffer contents */
+	size_t		 linelen;	/* length of line */
 	size_t		 pos;		/* current position in buffer */
 	int		 err;		/* last protocol reply code */
 #ifdef WITH_SSL

lib/libfetch/ftp.c

@@ -148,8 +148,9 @@ ftp_chkerr(conn_t *conn)
 		fetch_syserr();
 		return (-1);
 	}
-	if (isftpinfo(conn->buf)) {
-		while (conn->buflen && !isftpreply(conn->buf)) {
+
+	if (isftpinfo(conn->line)) {
+		while (rlen && !isftpreply(conn->line)) {
 			if ((rlen = fetch_getln(conn)) < 0) {
 				fetch_syserr();
 				return (-1);
@@ -157,14 +158,14 @@ ftp_chkerr(conn_t *conn)
 		}
 	}
 
-	if (!isftpreply(conn->buf)) {
+	if (!isftpreply(conn->line)) {
 		ftp_seterr(FTP_PROTOCOL_ERROR);
 		return (-1);
 	}
 
-	conn->err = (conn->buf[0] - '0') * 100
-	    + (conn->buf[1] - '0') * 10
-	    + (conn->buf[2] - '0');
+	conn->err = (conn->line[0] - '0') * 100
+	    + (conn->line[1] - '0') * 10
+	    + (conn->line[2] - '0');
 
 	return (conn->err);
 }
@@ -236,8 +237,8 @@ ftp_pwd(conn_t *conn, char *pwd, size_t pwdlen)
 	if (conn->err != FTP_WORKING_DIRECTORY &&
 	    conn->err != FTP_FILE_ACTION_OK)
 		return (FTP_PROTOCOL_ERROR);
-	end = conn->buf + conn->buflen;
-	src = conn->buf + 4;
+	end = conn->line + conn->linelen;
+	src = conn->line + 4;
 	if (src >= end || *src++ != '"')
 		return (FTP_PROTOCOL_ERROR);
 	for (q = 0, dst = pwd; src < end && pwdlen--; ++src) {
@@ -417,7 +418,7 @@ ftp_stat(conn_t *conn, const char *file, struct url_stat *us)
 		ftp_seterr(e);
 		return (-1);
 	}
-	for (ln = conn->buf + 4; *ln && isspace((unsigned char)*ln); ln++)
+	for (ln = conn->line + 4; *ln && isspace((unsigned char)*ln); ln++)
 		/* nothing */ ;
 	for (us->size = 0; *ln && isdigit((unsigned char)*ln); ln++)
 		us->size = us->size * 10 + *ln - '0';
@@ -435,7 +436,7 @@ ftp_stat(conn_t *conn, const char *file, struct url_stat *us)
 		ftp_seterr(e);
 		return (-1);
 	}
-	for (ln = conn->buf + 4; *ln && isspace((unsigned char)*ln); ln++)
+	for (ln = conn->line + 4; *ln && isspace((unsigned char)*ln); ln++)
 		/* nothing */ ;
 	switch (strspn(ln, "0123456789")) {
 	case 14:
@@ -688,7 +689,7 @@ ftp_transfer(conn_t *conn, const char *oper, const char *file,
 		 * Find address and port number. The reply to the PASV command
 		 * is IMHO the one and only weak point in the FTP protocol.
 		 */
-		ln = conn->buf;
+		ln = conn->line;
 		switch (e) {
 		case FTP_PASSIVE_MODE:
 		case FTP_LPASSIVE_MODE:

lib/libfetch/http.c

@@ -126,8 +126,7 @@
  * I/O functions for decoding chunked streams
  */
 
-struct httpio
-{
+struct httpio {
 	conn_t		*conn;		/* connection */
 	int		 chunked;	/* chunked mode */
 	char		*buf;		/* chunk buffer */
@@ -148,17 +147,14 @@ struct httpio
 static int
 http_new_chunk(struct httpio *io)
 {
-	unsigned char *p, *eol;
+	char *p;
 
 	if (fetch_getln(io->conn) == -1)
 		return (-1);
 
-	p = (unsigned char *)io->conn->buf;
-	if (io->conn->pos < 2 || !isxdigit(*p))
-		return (-1);
-
-	eol = (unsigned char *)io->conn->buf + io->conn->pos;
-	while (p < eol && *p && !isspace(*p) && *p != ';') {
+	for (p = io->conn->line;
+	     *p != '\0' && !isspace((unsigned char)*p) && *p != ';';
+	     p++) {
 		io->chunksize <<= 4;
 		if (*p >= '0' && *p <= '9')
 			io->chunksize += *p - '0';
@@ -168,7 +164,6 @@ http_new_chunk(struct httpio *io)
 			io->chunksize += 10 + *p - 'a';
 		else
 			return (-1);
-		p++;
 	}
 
 #ifndef NDEBUG
@@ -435,9 +430,9 @@ http_get_reply(conn_t *conn)
 	 * on finding one, but if we do, insist on it being 1.0 or 1.1.
 	 * We don't care about the reason phrase.
 	 */
-	if (strncmp(conn->buf, "HTTP", 4) != 0)
+	if (conn->linelen < 4 || strncmp(conn->line, "HTTP", 4) != 0)
 		return (HTTP_PROTOCOL_ERROR);
-	p = conn->buf + 4;
+	p = conn->line + 4;
 	if (*p == '/') {
 		if (p[1] != '1' || p[2] != '.' || (p[3] != '0' && p[3] != '1'))
 			return (HTTP_PROTOCOL_ERROR);
@@ -521,17 +516,17 @@ http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p)
 {
 	unsigned int i, len;
 
-	if (conn->pos == 0 || conn->buf[0] == '\0')
+	if (conn->linelen == 0)
 		return (hdr_end);
 
 	/* Copy the line to the headerbuf */
-	if (hbuf->bufsize < conn->pos + 1) {
-		if ((hbuf->buf = realloc(hbuf->buf, conn->pos + 1)) == NULL)
+	if (hbuf->bufsize <= conn->linelen) {
+		if ((hbuf->buf = realloc(hbuf->buf, conn->linelen + 1)) == NULL)
 			return (hdr_syserror);
-		hbuf->bufsize = conn->pos + 1;
+		hbuf->bufsize = conn->linelen + 1;
 	}
-	strcpy(hbuf->buf, conn->buf);
-	hbuf->buflen = conn->buflen;
+	memcpy(hbuf->buf, conn->line, conn->linelen + 1);
+	hbuf->buflen = conn->linelen;
 
 	/*
 	 * Fetch possible continuation lines. Stop at 1st non-continuation
@@ -545,19 +540,19 @@ http_next_header(conn_t *conn, http_headerbuf_t *hbuf, const char **p)
 		 * Note: we previously considered a pure whitespace line
 		 * equivalent to an empty one.  This was incorrect.
 		 */
-		if (conn->buf[0] != ' ' && conn->buf[0] != "\t"[0])
+		if (conn->line[0] != ' ' && conn->line[0] != '\t')
 			break;
 
 		/* Got a continuation line. Concatenate to previous */
-		len = hbuf->buflen + conn->buflen;
-		if (hbuf->bufsize < len + 1) {
+		len = hbuf->buflen + conn->linelen;
+		if (hbuf->bufsize <= len) {
 			len *= 2;
 			if ((hbuf->buf = realloc(hbuf->buf, len + 1)) == NULL)
 				return (hdr_syserror);
 			hbuf->bufsize = len + 1;
 		}
-		strcpy(hbuf->buf + hbuf->buflen, conn->buf);
-		hbuf->buflen += conn->buflen;
+		memcpy(hbuf->buf + hbuf->buflen, conn->line, conn->linelen + 1);
+		hbuf->buflen += conn->linelen;
 	}
 
 	/*