Attachment 48283 Details for Bug 11668 – [PATCH 1/2] Cope with platforms whose vsnprintf violates both POSIX and C99

[patch] [PATCH 1/2] Cope with platforms whose vsnprintf violates both POSIX and C99

0001-Cope-with-platforms-whose-vsnprintf-violates-both-PO.patch (text/plain), 3.01 KB, created by Simon McVittie on 2011-06-22 06:36:53 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Simon McVittie

Created: 2011-06-22 06:36:53 UTC

Size: 3.01 KB

patch

obsolete

>From 12b51b91b80b6678b7cd4208516173908edf4c6f Mon Sep 17 00:00:00 2001
>From: Simon McVittie <simon.mcvittie@collabora.co.uk>
>Date: Tue, 21 Jun 2011 14:09:32 +0100
>Subject: [PATCH 1/2] Cope with platforms whose vsnprintf violates both POSIX
> and C99
>
>---
> dbus/dbus-string.c       |    3 ++
> dbus/dbus-sysdeps-unix.c |   52 +++++++++++++++++++++++++++++++++++++++++++--
> 2 files changed, 52 insertions(+), 3 deletions(-)
>
>diff --git a/dbus/dbus-string.c b/dbus/dbus-string.c
>index e2eb93b..ca3a670 100644
>--- a/dbus/dbus-string.c
>+++ b/dbus/dbus-string.c
>@@ -1255,6 +1255,9 @@ _dbus_string_append_printf_valist  (DBusString        *str,
>   /* Measure the message length without terminating nul */
>   len = _dbus_printf_string_upper_bound (format, args);
> 
>+  if (len < 0)
>+    return FALSE;
>+
>   if (!_dbus_string_lengthen (str, len))
>     {
>       /* don't leak the copy */
>diff --git a/dbus/dbus-sysdeps-unix.c b/dbus/dbus-sysdeps-unix.c
>index e1e1728..79d0bb5 100644
>--- a/dbus/dbus-sysdeps-unix.c
>+++ b/dbus/dbus-sysdeps-unix.c
>@@ -2978,14 +2978,60 @@ _dbus_full_duplex_pipe (int        *fd1,
>  *
>  * @param format a printf-style format string
>  * @param args arguments for the format string
>- * @returns length of the given format string and args
>+ * @returns length of the given format string and args, or -1 if no memory
>  */
> int
> _dbus_printf_string_upper_bound (const char *format,
>                                  va_list     args)
> {
>-  char c;
>-  return vsnprintf (&c, 1, format, args);
>+  char static_buf[1024];
>+  int bufsize = sizeof (static_buf);
>+  int len;
>+
>+  len = vsnprintf (static_buf, bufsize, format, args);
>+
>+  /* If vsnprintf() returned non-negative, then either the string fits in
>+   * static_buf, or this OS has the POSIX and C99 behaviour where vsnprintf
>+   * returns the number of characters that were needed, or this OS returns the
>+   * truncated length.
>+   *
>+   * We ignore the possibility that snprintf might just ignore the length and
>+   * overrun the buffer (64-bit Solaris 7), because that's pathological.
>+   * If your libc is really that bad, come back when you have a better one. */
>+  if (len == bufsize)
>+    {
>+      /* This could be the truncated length (Tru64 and IRIX have this bug),
>+       * or the real length could be coincidentally the same. Which is it?
>+       * If vsnprintf returns the truncated length, we'll go to the slow
>+       * path. */
>+      if (vsnprintf (static_buf, 1, format, args) == 1)
>+        len = -1;
>+    }
>+
>+  /* If vsnprintf() returned negative, we have to do more work.
>+   * HP-UX returns negative. */
>+  while (len < 0)
>+    {
>+      char *buf;
>+
>+      bufsize *= 2;
>+
>+      buf = dbus_malloc (bufsize);
>+
>+      if (buf == NULL)
>+        return -1;
>+
>+      len = vsnprintf (buf, bufsize, format, args);
>+      dbus_free (buf);
>+
>+      /* If the reported length is exactly the buffer size, round up to the
>+       * next size, in case vsnprintf has been returning the truncated
>+       * length */
>+      if (len == bufsize)
>+        len = -1;
>+    }
>+
>+  return len;
> }
> 
> /**
>-- 
>1.7.5.4
>

Actions: View | Splinter Review

Attachments on bug 11668: 48283 | 48284 | 49614

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.