Обсуждение: Cursor support buffer patch

Here is a small patch for the cursor support which Jan recently added.
The code assumed that there would be a '\0' in buf after storing the
characters in new->refname, but it did nothing to ensure that.

I can't convince myself that this code does not have the possibility
of buffer overflow.  However, I have not tried to fix that.  For that
matter, I see other possibilities for buffer overflow in gram.y, such
as in decl_cursor_arglist.  Buffer overflow of this sort is not good,
as it means that anybody who is permitted to create functions can
completely break security.

Ian

Index: gram.y
===================================================================
RCS file: /home/projects/pgsql/cvsroot/pgsql/src/pl/plpgsql/src/gram.y,v
retrieving revision 1.20
diff -u -p -r1.20 gram.y
--- gram.y    2001/05/31 17:15:40    1.20
+++ gram.y    2001/06/06 06:35:46
@@ -385,7 +385,8 @@ decl_statement    : decl_varname decl_const
                                 *cp2++ = '\\';
                             *cp2++ = *cp1++;
                         }
-                        strcat(buf, "'");
+                        *cp2++ = '\'';
+                        *cp2 = '\0';
                         curname_def->query = strdup(buf);
                         new->default_val = curname_def;

Ian Lance Taylor <ian@airs.com> writes:
> The code assumed that there would be a '\0' in buf after storing the
> characters in new->refname, but it did nothing to ensure that.

Good catch.

> I can't convince myself that this code does not have the possibility
> of buffer overflow.

It obviously does; the fixed-size buffer should be replaced by a
PLpgSQL_dstring, probably.  I don't much like the fixed-size
fieldnames[] buffers elsewhere in that file, either.

            regards, tom lane

Your patch has been added to the PostgreSQL unapplied patches list at:

    http://candle.pha.pa.us/cgi-bin/pgpatches

I will try to apply it within the next 48 hours.

> Here is a small patch for the cursor support which Jan recently added.
> The code assumed that there would be a '\0' in buf after storing the
> characters in new->refname, but it did nothing to ensure that.
>
> I can't convince myself that this code does not have the possibility
> of buffer overflow.  However, I have not tried to fix that.  For that
> matter, I see other possibilities for buffer overflow in gram.y, such
> as in decl_cursor_arglist.  Buffer overflow of this sort is not good,
> as it means that anybody who is permitted to create functions can
> completely break security.
>
> Ian
>
> Index: gram.y
> ===================================================================
> RCS file: /home/projects/pgsql/cvsroot/pgsql/src/pl/plpgsql/src/gram.y,v
> retrieving revision 1.20
> diff -u -p -r1.20 gram.y
> --- gram.y    2001/05/31 17:15:40    1.20
> +++ gram.y    2001/06/06 06:35:46
> @@ -385,7 +385,8 @@ decl_statement    : decl_varname decl_const
>                                  *cp2++ = '\\';
>                              *cp2++ = *cp1++;
>                          }
> -                        strcat(buf, "'");
> +                        *cp2++ = '\'';
> +                        *cp2 = '\0';
>                          curname_def->query = strdup(buf);
>                          new->default_val = curname_def;
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
>
> http://www.postgresql.org/users-lounge/docs/faq.html
>

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026

I see this was already installed by Jan.

> Here is a small patch for the cursor support which Jan recently added.
> The code assumed that there would be a '\0' in buf after storing the
> characters in new->refname, but it did nothing to ensure that.
>
> I can't convince myself that this code does not have the possibility
> of buffer overflow.  However, I have not tried to fix that.  For that
> matter, I see other possibilities for buffer overflow in gram.y, such
> as in decl_cursor_arglist.  Buffer overflow of this sort is not good,
> as it means that anybody who is permitted to create functions can
> completely break security.
>
> Ian
>
> Index: gram.y
> ===================================================================
> RCS file: /home/projects/pgsql/cvsroot/pgsql/src/pl/plpgsql/src/gram.y,v
> retrieving revision 1.20
> diff -u -p -r1.20 gram.y
> --- gram.y    2001/05/31 17:15:40    1.20
> +++ gram.y    2001/06/06 06:35:46
> @@ -385,7 +385,8 @@ decl_statement    : decl_varname decl_const
>                                  *cp2++ = '\\';
>                              *cp2++ = *cp1++;
>                          }
> -                        strcat(buf, "'");
> +                        *cp2++ = '\'';
> +                        *cp2 = '\0';
>                          curname_def->query = strdup(buf);
>                          new->default_val = curname_def;
>
>
> ---------------------------(end of broadcast)---------------------------
> TIP 5: Have you checked our extensive FAQ?
>
> http://www.postgresql.org/users-lounge/docs/faq.html
>

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026