* Re: Util/helpfiles - problem with provided example
2004-07-28 10:01 ` Peter Stephenson
@ 2004-07-28 10:21 ` Peter Stephenson
2004-07-29 15:40 ` Peter Stephenson
1 sibling, 0 replies; 4+ messages in thread
From: Peter Stephenson @ 2004-07-28 10:21 UTC (permalink / raw)
To: zsh-workers
Peter Stephenson wrote:
> Michael Prokop wrote:
> > man () {
> > man $1 # this is the original
> > }
> > $ man zsh
>
> > Segmentation fault
>
> This isn't very nice. Should we turn on --enable-max-function-depth by
> default? The default for --enable-max-function-depth=yes is 4096 which
> isn't going to hurt most people. Or should we make it configurable
> internally with a sensible default?
More investigation reveals a further problem. There's an arbitrary
debugging check when unreferencing the function that it hasn't been
referenced more than 256 times. This needs to be the function depth if
that is defined plus 1 (to allow for the original function struture).
I've added a few on for fuzz. Presumably the test is meaningless if
there is no function depth limit.
If you try this with a the default function depth limit, it's
interesting to note that the error message is printed almost straight
away, while the prompt takes a while to return. I wonder if freeing
things up is slower than necessary?
Index: Src/parse.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/parse.c,v
retrieving revision 1.46
diff -u -r1.46 parse.c
--- Src/parse.c 28 Jun 2004 15:38:14 -0000 1.46
+++ Src/parse.c 28 Jul 2004 10:18:28 -0000
@@ -2168,7 +2168,10 @@
/* paranoia */
DPUTS(p->nref > 0 && (p->flags & EF_HEAP), "Heap EPROG has nref > 0");
DPUTS(p->nref < 0 && !(p->flags & EF_HEAP), "Real EPROG has nref < 0");
- DPUTS(p->nref < -1 || p->nref > 256, "Uninitialised EPROG nref");
+ DPUTS(p->nref < -1, "Uninitialised EPROG nref");
+#ifdef MAX_FUNCTION_DEPTH
+ DPUTS(p->nref > MAX_FUNCTION_DEPTH + 10, "Overlarge EPROG nref");
+#endif
if (p->nref > 0 && !--p->nref) {
for (i = p->npats, pp = p->pats; i--; pp++)
freepatprog(*pp);
--
Peter Stephenson <pws@csr.com> Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK Tel: +44 (0)1223 692070
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.
www.mimesweeper.com
**********************************************************************
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: Util/helpfiles - problem with provided example
2004-07-28 10:01 ` Peter Stephenson
2004-07-28 10:21 ` Peter Stephenson
@ 2004-07-29 15:40 ` Peter Stephenson
1 sibling, 0 replies; 4+ messages in thread
From: Peter Stephenson @ 2004-07-29 15:40 UTC (permalink / raw)
To: zsh-workers
Peter Stephenson wrote:
> Edited down to bare essentials:
>
> Michael Prokop wrote:
> > man () {
> > man $1 # this is the original
> > }
> > $ man zsh
>
> > Segmentation fault
>
> This isn't very nice. Should we turn on --enable-max-function-depth by
> default? The default for --enable-max-function-depth=yes is 4096 which
> isn't going to hurt most people. Or should we make it configurable
> internally with a sensible default?
Here's a patch which turns it on by default.
Index: configure.ac
===================================================================
RCS file: /cvsroot/zsh/zsh/configure.ac,v
retrieving revision 1.18
diff -u -r1.18 configure.ac
--- configure.ac 8 Jun 2004 13:34:12 -0000 1.18
+++ configure.ac 29 Jul 2004 15:39:00 -0000
@@ -301,12 +301,14 @@
AH_TEMPLATE([MAX_FUNCTION_DEPTH],
[Define for function depth limits])
AC_ARG_ENABLE(max-function-depth,
-[ --enable-max-function-depth=MAX limit function depth to MAX],
+[ --enable-max-function-depth=MAX limit function depth to MAX, default 4096],
[if test x$enableval = xyes; then
AC_DEFINE(MAX_FUNCTION_DEPTH, 4096)
-else
+elif test x$enableval != xno; then
AC_DEFINE_UNQUOTED(MAX_FUNCTION_DEPTH, $enableval)
-fi])
+fi],
+[AC_DEFINE(MAX_FUNCTION_DEPTH, 4096)]
+)
dnl Do you want to look for pcre support?
AC_ARG_ENABLE(pcre,
Index: INSTALL
===================================================================
RCS file: /cvsroot/zsh/zsh/INSTALL,v
retrieving revision 1.17
diff -u -r1.17 INSTALL
--- INSTALL 2 Jul 2004 15:59:11 -0000 1.17
+++ INSTALL 29 Jul 2004 15:39:01 -0000
@@ -330,6 +330,19 @@
`make uninstall' or `make uninstall.fns', although the version-specific
directory and its contents will be deleted.
+Function depth
+--------------
+
+Shell functions may be called recursively. In order to detect infinite
+recursion the shell has a limit on the depth to which functions may be
+called: note that this is a single limit for all functions, not a limit
+for each function called recursively. The default for the limit is 4096.
+The limit may be altered to the value MAX by passing the option
+--enable-max-function-depth=MAX to configure. Alternatively, the limit may
+be disabled with --disable-max-function-depth. However, this is not
+recommended as it is likely to cause the shell to crash on an infinite
+recursion.
+
Support for large files and integers
------------------------------------
--
Peter Stephenson <pws@csr.com> Software Engineer
CSR Ltd., Science Park, Milton Road,
Cambridge, CB4 0WH, UK Tel: +44 (0)1223 692070
**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.
This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.
www.mimesweeper.com
**********************************************************************
^ permalink raw reply [flat|nested] 4+ messages in thread