diff --git a/CHANGES b/CHANGES
index 2b5fbcbc1..4a017a72b 100644
--- a/CHANGES
+++ b/CHANGES
@@ -6,8 +6,9 @@
Changelog
-Yang Tse (21 Aug 2008)
-- Fixed a couple of bugs in libcurl's internal curl_m*printf() functions.
+Yang Tse (22 Aug 2008)
+- Improved libcurl's internal curl_m*printf() functions integral data type
+ size and signedness handling.
Daniel Fandrich (20 Aug 2008)
- Added an edited version of Vincent Le Normand's documentation of SFTP quote
diff --git a/RELEASE-NOTES b/RELEASE-NOTES
index e89c4f2a3..67d89c52e 100644
--- a/RELEASE-NOTES
+++ b/RELEASE-NOTES
@@ -55,6 +55,7 @@ This release includes the following bugfixes:
o potential buffer overflows in the MS-DOS command-line port fixed
o --stderr is now honoured with the -v option
o memory leak in libcurl on Windows built with OpenSSL
+ o improved curl_m*printf() integral data type size and signedness handling
This release includes the following known bugs:
diff --git a/lib/mprintf.c b/lib/mprintf.c
index 0623aaea9..2f7f31e29 100644
--- a/lib/mprintf.c
+++ b/lib/mprintf.c
@@ -49,6 +49,10 @@
#include <curl/mprintf.h>
+#include "memory.h"
+/* The last #include file should be: */
+#include "memdebug.h"
+
#ifndef SIZEOF_LONG_DOUBLE
#define SIZEOF_LONG_DOUBLE 0
#endif
@@ -74,9 +78,17 @@
# endif
#endif
-#include "memory.h"
-/* The last #include file should be: */
-#include "memdebug.h"
+/*
+ * Max integer data types that mprintf.c is capable
+ */
+
+#ifdef HAVE_LONG_LONG_TYPE
+# define mp_intmax_t LONG_LONG_TYPE
+# define mp_uintmax_t unsigned LONG_LONG_TYPE
+#else
+# define mp_intmax_t long
+# define mp_uintmax_t unsigned long
+#endif
#define BUFFSIZE 256 /* buffer for long-to-str and float-to-str calcs */
#define MAX_PARAMETERS 128 /* lame static limit */
@@ -146,10 +158,10 @@ typedef struct {
union {
char *str;
void *ptr;
- long num;
-#ifdef HAVE_LONG_LONG_TYPE
- LONG_LONG_TYPE lnum;
-#endif
+ union {
+ mp_intmax_t as_signed;
+ mp_uintmax_t as_unsigned;
+ } num;
double dnum;
} data;
} va_stack_t;
@@ -544,7 +556,7 @@ static long dprintf_Pass1(const char *format, va_stack_t *vto, char **endpos,
/* Width/precision arguments must be read before the main argument
* they are attached to
*/
- vto[i + 1].data.num = va_arg(arglist, int);
+ vto[i + 1].data.num.as_signed = (mp_intmax_t)va_arg(arglist, int);
}
switch (vto[i].type)
@@ -561,15 +573,27 @@ static long dprintf_Pass1(const char *format, va_stack_t *vto, char **endpos,
case FORMAT_INT:
#ifdef HAVE_LONG_LONG_TYPE
- if(vto[i].flags & FLAGS_LONGLONG)
- vto[i].data.lnum = va_arg(arglist, LONG_LONG_TYPE);
+ if((vto[i].flags & FLAGS_LONGLONG) && (vto[i].flags & FLAGS_UNSIGNED))
+ vto[i].data.num.as_unsigned =
+ (mp_uintmax_t)va_arg(arglist, unsigned LONG_LONG_TYPE);
+ else if(vto[i].flags & FLAGS_LONGLONG)
+ vto[i].data.num.as_signed =
+ (mp_intmax_t)va_arg(arglist, LONG_LONG_TYPE);
else
#endif
{
- if(vto[i].flags & FLAGS_LONG)
- vto[i].data.num = va_arg(arglist, long);
+ if((vto[i].flags & FLAGS_LONG) && (vto[i].flags & FLAGS_UNSIGNED))
+ vto[i].data.num.as_unsigned =
+ (mp_uintmax_t)va_arg(arglist, unsigned long);
+ else if(vto[i].flags & FLAGS_LONG)
+ vto[i].data.num.as_signed =
+ (mp_intmax_t)va_arg(arglist, long);
+ else if(vto[i].flags & FLAGS_UNSIGNED)
+ vto[i].data.num.as_unsigned =
+ (mp_uintmax_t)va_arg(arglist, unsigned int);
else
- vto[i].data.num = va_arg(arglist, int);
+ vto[i].data.num.as_signed =
+ (mp_intmax_t)va_arg(arglist, int);
}
break;
@@ -645,13 +669,10 @@ static int dprintf_formatf(
long base;
/* Integral values to be written. */
-#ifdef HAVE_LONG_LONG_TYPE
- unsigned LONG_LONG_TYPE num;
- LONG_LONG_TYPE signed_num;
-#else
- unsigned long num;
- long signed_num;
-#endif
+ mp_uintmax_t num;
+
+ /* Used to convert negative in positive. */
+ mp_intmax_t signed_num;
if(*f != '%') {
/* This isn't a format spec, so write everything out until the next one
@@ -690,13 +711,13 @@ static int dprintf_formatf(
/* pick up the specified width */
if(p->flags & FLAGS_WIDTHPARAM)
- width = vto[p->width].data.num;
+ width = (long)vto[p->width].data.num.as_signed;
else
width = p->width;
/* pick up the specified precision */
if(p->flags & FLAGS_PRECPARAM) {
- prec = vto[p->precision].data.num;
+ prec = (long)vto[p->precision].data.num.as_signed;
param_num++; /* since the precision is extraced from a parameter, we
must skip that to get to the next one properly */
}
@@ -709,12 +730,7 @@ static int dprintf_formatf(
switch (p->type) {
case FORMAT_INT:
-#ifdef HAVE_LONG_LONG_TYPE
- if(p->flags & FLAGS_LONGLONG)
- num = (unsigned LONG_LONG_TYPE)p->data.lnum;
- else
-#endif
- num = (unsigned long)p->data.num;
+ num = p->data.num.as_unsigned;
if(p->flags & FLAGS_CHAR) {
/* Character. */
if(!(p->flags & FLAGS_LEFT))
@@ -747,32 +763,14 @@ static int dprintf_formatf(
/* Decimal integer. */
base = 10;
-#ifdef HAVE_LONG_LONG_TYPE
- if(p->flags & FLAGS_LONGLONG) {
- /* long long */
- is_neg = (p->data.lnum < 0);
- if(is_neg) {
- /* signed long long might fail to hold absolute LLONG_MIN by 1 */
- signed_num = p->data.lnum + (LONG_LONG_TYPE)1;
- num = (unsigned LONG_LONG_TYPE)-signed_num;
- num += (unsigned LONG_LONG_TYPE)1;
- }
- else
- num = p->data.lnum;
- }
- else
-#endif
- {
- is_neg = (p->data.num < 0);
- if(is_neg) {
- /* signed long might fail to hold absolute LONG_MIN by 1 */
- signed_num = p->data.num + (long)1;
- num = (unsigned long)-signed_num;
- num += (unsigned long)1;
- }
- else
- num = p->data.num;
+ is_neg = p->data.num.as_signed < 0;
+ if(is_neg) {
+ /* signed_num might fail to hold absolute negative minimum by 1 */
+ signed_num = p->data.num.as_signed + (mp_intmax_t)1;
+ num = (mp_uintmax_t)-signed_num;
+ num += (mp_uintmax_t)1;
}
+
goto number;
unsigned_number:
@@ -938,13 +936,13 @@ static int dprintf_formatf(
if(p->flags & FLAGS_WIDTH)
width = p->width;
else if(p->flags & FLAGS_WIDTHPARAM)
- width = vto[p->width].data.num;
+ width = (long)vto[p->width].data.num.as_signed;
prec = -1;
if(p->flags & FLAGS_PREC)
prec = p->precision;
else if(p->flags & FLAGS_PRECPARAM)
- prec = vto[p->precision].data.num;
+ prec = (long)vto[p->precision].data.num.as_signed;
if(p->flags & FLAGS_LEFT)
strcat(formatbuf, "-");