Using \u2030 in plot axis label -> stack smashing

On Tue, 2006-09-19 at 08:26 +0100, Prof Brian Ripley wrote:

I didn't have access to my FC5 boxes yesterday (electrical testing).

This does need the FC5-specific compilation options set
(-Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector
--param=ssp-buffer-size=4), so it is not surprising it is not
reproducible elsewhere (including under valgrind, BTW).

Ei-ji's patch works (and is incorporated now), but the buffer is used at

     strncpy(s, buf, sizeof(buf) - 1); /* ensure 0-terminated */

and the 's' here should be big enough (\uxxxx can only expand to 3 bytes
in UTF-8, so "\u2030" is four bytes in UTF-8 including the null
terminator).  Can Ei-ji explain?

I can understand how Gavin saw this in the released FC5 RPM.  What I don't
understand is how he saw this in 2.4.0 alpha/R-devel without setting
non-default CFLAGS he did not tell us about.

Thanks Prof. Ripley and Ei-Ji. I should have mentioned that all the
versions I reported for were self-compiled, and I did so with the same
set of flags as the FC5 rpm. Will add that to the list in my head of
things to report.

BTW, just applying this patch will not work: you need to rebuild gram.c
in maintainer mode.

I'm not clear what you mean by maintainer mode - not something I have
come across before. If I update the local source on my machine from the
svn server, and make clean, configure and make again, will this be
sufficient? Or do I need to do something else?

Many thanks,

G

On Tue, 19 Sep 2006, Ei-ji Nakama wrote:

This seems to be the mine which I contrived. m(_|_)m

--- R-alpha.orig/src/main/gram.y        2006-09-04 23:41:33.000000000 +0900
+++ R-alpha/src/main/gram.y     2006-09-19 13:01:41.000000000 +0900
@@ -99,11 +99,12 @@
# endif
#endif
#include <errno.h>
+#define MB_BUF 16

static size_t ucstomb(char *s, wchar_t wc, mbstate_t *ps)
{
    char     tocode[128];
-    char     buf[16];
+    char     buf[MB_BUF];
    void    *cd = NULL ;
    wchar_t  wcs[2];
    char    *inbuf = (char *) wcs;
@@ -1709,7 +1710,7 @@
               error(_("\\uxxxx sequences not supported"));
#else
               wint_t val = 0; int i, ext; size_t res;
-               char buff[5]; Rboolean delim = FALSE;
+               char buff[MB_BUF]; Rboolean delim = FALSE;
               if((c = xxgetc()) == '{') delim = TRUE; else xxungetc(c);
               for(i = 0; i < 4; i++) {
                   c = xxgetc();
@@ -1743,7 +1744,7 @@
#ifdef SUPPORT_MBCS
               else {
                   wint_t val = 0; int i, ext; size_t res;
-                   char buff[9]; Rboolean delim = FALSE;
+                   char buff[MB_BUF]; Rboolean delim = FALSE;
                   if((c = xxgetc()) == '{') delim = TRUE; else xxungetc(c);
                   for(i = 0; i < 8; i++) {
                       c = xxgetc();


2006/9/19, Gregor Gorjanc <gregor.gorjanc at bfro.uni-lj.si>:

Gavin Simpson wrote:

On Mon, 2006-09-18 at 19:02 +0000, Gregor Gorjanc wrote:

Gavin Simpson <gavin.simpson <at> ucl.ac.uk> writes:

Dear List

I just noticed the following behaviour in R 2.3.1 Patched (2006-06-13
r38342) and confirmed similar behaviour in R 2.4.0 alpha (2006-09-18
r39383) & R 2.5.0 (2006-09-18 r39383) - which may actually be the same
thing?, that trying to plot the unicode character \u2030 (which should
be in a ?? [per mille] sign) in an axis label leads to the following
error:

*** stack smashing detected ***: /home/gavin/R/R-devel/build/bin/exec/R
terminated
Aborted

The simplest, reproducible example I have tried is:

plot(1:10, ylab = "\u2030")

I can not reproduce this on my Debian GNU/Linux. I get something like "S
for y label under 2.3.1 2006-06-01 and 2.5.0 2006-09-13 r39292 with the
following locale

[1] "LC_CTYPE=en_GB.UTF-8;LC_NUMERIC=C;LC_TIME=en_GB.UTF-8;
LC_COLLATE=en_GB.UTF-8;LC_MONETARY=en_GB.UTF-8;LC_MESSAGES=en_GB.UTF-8;
LC_PAPER=C;LC_NAME=C;LC_ADDRESS=C;LC_TELEPHONE=C;LC_MEASUREMENT=C;
LC_IDENTIFICATION=C"

It does not change if I set everything into en_GB.UTF-8. Is this valid
unicode code?

Gregor

Cheers for the follow up Gregor,

I was following advice given by Prof. Ripley in a posting on R-Help
about how to get the per mille character:

http://finzi.psych.upenn.edu/R/Rhelp02a/archive/48709.html

It should look like a "%" character but with two circles at the bottom.

Perhaps I do not have appropriate font for this character.

Gregor

--
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595