Update to libpcap 1.9.1 (initial commit, no Nmap-specific patches)

libpcap/missing/asprintf.c

@@ -0,0 +1,101 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+
+#include "portability.h"
+
+/*
+ * vasprintf() and asprintf() for platforms with a C99-compliant
+ * snprintf() - so that, if you format into a 1-byte buffer, it
+ * will return how many characters it would have produced had
+ * it been given an infinite-sized buffer.
+ */
+int
+pcap_vasprintf(char **strp, const char *format, va_list args)
+{
+	char buf;
+	int len;
+	size_t str_size;
+	char *str;
+	int ret;
+
+	/*
+	 * XXX - the C99 standard says, in section 7.19.6.5 "Thes
+	 * nprintf function":
+	 *
+	 *    The snprintf function is equivalent to fprintf, except that
+	 *    the output is written into an array (specified by argument s)
+	 *    rather than to a stream.  If n is zero, nothing is written,
+	 *    and s may be a null pointer.  Otherwise, output characters
+	 *    beyond the n-1st are discarded rather than being written
+	 *    to the array, and a null character is written at the end
+	 *    of the characters actually written into the array.
+	 *
+	 *        ...
+	 *
+	 *    The snprintf function returns the number of characters that
+	 *    would have been written had n been sufficiently large, not
+	 *    counting the terminating null character, or a negative value
+	 *    if an encoding error occurred. Thus, the null-terminated
+	 *    output has been completely written if and only if the returned
+	 *    value is nonnegative and less than n.
+	 *
+	 * That doesn't make it entirely clear whether, if a null buffer
+	 * pointer and a zero count are passed, it will return the number
+	 * of characters that would have been written had a buffer been
+	 * passed.
+	 *
+	 * And, even if C99 *does*, in fact, say it has to work, it
+	 * doesn't work in Solaris 8, for example - it returns -1 for
+	 * NULL/0, but returns the correct character count for a 1-byte
+	 * buffer.
+	 *
+	 * So we pass a one-character pointer in order to find out how
+	 * many characters this format and those arguments will need
+	 * without actually generating any more of those characters
+	 * than we need.
+	 *
+	 * (The fact that it might happen to work with GNU libc or with
+	 * various BSD libcs is completely uninteresting, as those tend
+	 * to have asprintf() already and thus don't even *need* this
+	 * code; this is for use in those UN*Xes that *don't* have
+	 * asprintf().)
+	 */
+	len = vsnprintf(&buf, sizeof buf, format, args);
+	if (len == -1) {
+		*strp = NULL;
+		return (-1);
+	}
+	str_size = len + 1;
+	str = malloc(str_size);
+	if (str == NULL) {
+		*strp = NULL;
+		return (-1);
+	}
+	ret = vsnprintf(str, str_size, format, args);
+	if (ret == -1) {
+		free(str);
+		*strp = NULL;
+		return (-1);
+	}
+	*strp = str;
+	/*
+	 * vsnprintf() shouldn't truncate the string, as we have
+	 * allocated a buffer large enough to hold the string, so its
+	 * return value should be the number of characters written.
+	 */
+	return (ret);
+}
+
+int
+pcap_asprintf(char **strp, const char *format, ...)
+{
+	va_list args;
+	int ret;
+
+	va_start(args, format);
+	ret = pcap_vasprintf(strp, format, args);
+	va_end(args);
+	return (ret);
+}
+