Re: %.1s format with vasnprintf reads more than one byte from argument

[Bruno Haible]

Hi Eric,

> I'm now getting failures on Solaris 10, at the same line, and traced it to a 
> call to the native:
> 
> /* second byte uninitialized, and happens to be invalid character */
> wchar_t str[2] = { 'a', 0x65666768 };
> snprintf(buffer, 12, "%.*ls", 1, str);
> 
> with a return of returns -1 with errno EILSEQ, instead of populating buffer 
> with "a" and returning 1.  In short, Solaris parses too far into the wchar_t* 
> array, detects failure in converting str[1], and fails with EILSEQ, even 
> though 
> str[0] was sufficient to provide the requested precision.  Random failures 
> due 
> to reading uninitialized memory are unacceptable, and while this was EILSEQ, 
> it 
> is also possible to trigger SIGSEGV.

This was somehow to be expected... This paragraph is ISO C 99 is hard to read
and hard to implement.

> This patch was sufficient to make the configure test detect the Solaris bug, 
> while still letting Linux defer to the (working) native version; okay to 
> apply, 
> or do you want to touch it up further?

I removed the part of the test that should normally always fail:
    if (sprintf (buf, "%ls", wstring) != -1)
      return 2;
because it does not fail on any known platform. Then refactored the code
to keep different tests separate. Added comments and updated the doc:


2009-02-26  Eric Blake  <address@hidden>
            Bruno Haible  <address@hidden>

        Work around a *printf bug with %ls on Solaris.
        * m4/printf.m4 (gl_PRINTF_DIRECTIVE_LS): Also test whether, when a
        precision is specified, sprintf stops converting the wide string
        argument when the number of bytes that have been produced by this
        conversion equals or exceeds the precision.
        * doc/posix-functions/fprintf.texi: Update.
        * doc/posix-functions/printf.texi: Update.
        * doc/posix-functions/snprintf.texi: Update.
        * doc/posix-functions/sprintf.texi: Update.
        * doc/posix-functions/vfprintf.texi: Update.
        * doc/posix-functions/vprintf.texi: Update.
        * doc/posix-functions/vsnprintf.texi: Update.
        * doc/posix-functions/vsprintf.texi: Update.
        * doc/glibc-functions/obstack_printf.texi: Update.
        * doc/glibc-functions/obstack_vprintf.texi: Update.

--- m4/printf.m4.orig   2009-02-27 02:06:06.000000000 +0100
+++ m4/printf.m4        2009-02-27 01:59:44.000000000 +0100
@@ -1,4 +1,4 @@
-# printf.m4 serial 30
+# printf.m4 serial 31
 dnl Copyright (C) 2003, 2007-2009 Free Software Foundation, Inc.
 dnl This file is free software; the Free Software Foundation
 dnl gives unlimited permission to copy and/or distribute it,
@@ -626,7 +626,10 @@
 ])
 
 dnl Test whether the *printf family of functions supports the %ls format
-dnl directive.
+dnl directive and in particular, when a precision is specified, whether
+dnl the functions stop converting the wide string argument when the number
+dnl of bytes that have been produced by this conversion equals or exceeds
+dnl the precision.
 dnl Result is gl_cv_func_printf_directive_ls.
 
 AC_DEFUN([gl_PRINTF_DIRECTIVE_LS],
@@ -646,19 +649,40 @@
 #include <time.h>
 #include <wchar.h>
 #include <string.h>
-static wchar_t wstring[] = { 'a', 'b', 'c', 0 };
 int main ()
 {
   char buf[100];
-  buf[0] = '\0';
-  sprintf (buf, "%ls", wstring);
-  return strcmp (buf, "abc") != 0;
+  /* Test whether %ls works at all.
+     This test fails on OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.  */
+  {
+    static const wchar_t wstring[] = { 'a', 'b', 'c', 0 };
+    buf[0] = '\0';
+    if (sprintf (buf, "%ls", wstring) < 0
+        || strcmp (buf, "abc") != 0)
+      return 1;
+  }
+  /* Test whether precisions in %ls are supported as specified in ISO C 99
+     section 7.19.6.1:
+       "If a precision is specified, no more than that many bytes are written
+        (including shift sequences, if any), and the array shall contain a
+        null wide character if, to equal the multibyte character sequence
+        length given by the precision, the function would need to access a
+        wide character one past the end of the array."
+     This test fails on Solaris 10.  */
+  {
+    static const wchar_t wstring[] = { 'a', 'b', (wchar_t) 0xfdfdfdfd, 0 };
+    buf[0] = '\0';
+    if (sprintf (buf, "%.2ls", wstring) < 0
+        || strcmp (buf, "ab") != 0)
+      return 1;
+  }
+  return 0;
 }], [gl_cv_func_printf_directive_ls=yes], [gl_cv_func_printf_directive_ls=no],
       [
 changequote(,)dnl
        case "$host_os" in
          openbsd*)        gl_cv_func_printf_directive_ls="guessing no";;
-         solaris2.[1-6]*) gl_cv_func_printf_directive_ls="guessing no";;
+         solaris*)        gl_cv_func_printf_directive_ls="guessing no";;
          irix*)           gl_cv_func_printf_directive_ls="guessing no";;
          beos* | haiku*)  gl_cv_func_printf_directive_ls="guessing no";;
          *)               gl_cv_func_printf_directive_ls="guessing yes";;
--- doc/glibc-functions/obstack_printf.texi.orig        2009-02-27 
02:06:06.000000000 +0100
+++ doc/glibc-functions/obstack_printf.texi     2009-02-27 02:05:25.000000000 
+0100
@@ -38,6 +38,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/glibc-functions/obstack_vprintf.texi.orig       2009-02-27 
02:06:06.000000000 +0100
+++ doc/glibc-functions/obstack_vprintf.texi    2009-02-27 02:05:26.000000000 
+0100
@@ -38,6 +38,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/fprintf.texi.orig       2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/fprintf.texi    2009-02-27 02:04:38.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/printf.texi.orig        2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/printf.texi     2009-02-27 02:04:38.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/snprintf.texi.orig      2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/snprintf.texi   2009-02-27 02:04:37.000000000 +0100
@@ -43,6 +43,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/sprintf.texi.orig       2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/sprintf.texi    2009-02-27 02:04:38.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/vfprintf.texi.orig      2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/vfprintf.texi   2009-02-27 02:04:38.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/vprintf.texi.orig       2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/vprintf.texi    2009-02-27 02:04:37.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/vsnprintf.texi.orig     2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/vsnprintf.texi  2009-02-27 02:04:37.000000000 +0100
@@ -43,6 +43,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.
--- doc/posix-functions/vsprintf.texi.orig      2009-02-27 02:06:06.000000000 
+0100
+++ doc/posix-functions/vsprintf.texi   2009-02-27 02:04:37.000000000 +0100
@@ -32,6 +32,10 @@
 This function does not support the @samp{ls} directive on some platforms:
 OpenBSD 4.0, IRIX 6.5, Solaris 2.6, Haiku.
 @item
+This function does not support precisions in the @samp{ls} directive correctly
+on some platforms:
+Solaris 10.
address@hidden
 This function does not support format directives that access arguments in an
 arbitrary order, such as @code{"%2$s"}, on some platforms:
 NetBSD 3.0, mingw, BeOS.