NetBSD Problem Report #20112
Received: (qmail 11088 invoked by uid 605); 29 Jan 2003 20:31:12 -0000
Message-Id: <200301292030.h0TKUa307186@mail.duh.org>
Date: Wed, 29 Jan 2003 15:30:36 -0500 (EST)
From: tv@pobox.com
Sender: gnats-bugs-owner@netbsd.org
Reply-To: tv@pobox.com
To: gnats-bugs@gnats.netbsd.org
Subject: MAXINTERP set far too low for modern scripts
X-Send-Pr-Version: 3.95
>Number: 20112
>Category: kern
>Synopsis: MAXINTERP set far too low for modern scripts
>Confidential: no
>Severity: non-critical
>Priority: medium
>Responsible: kern-bug-people
>State: closed
>Class: change-request
>Submitter-Id: net
>Arrival-Date: Wed Jan 29 20:32:00 +0000 2003
>Closed-Date: Wed Apr 02 01:02:02 +0000 2003
>Last-Modified: Sun Aug 17 13:42:03 +0000 2003
>Originator: Todd Vierling
>Release: NetBSD 1.6_STABLE
>Organization:
DUH.ORG: Pointing out the obvious since 1994.
>Environment:
na
>Description:
#define MAXINTERP 64 /* max interpreter file name length */
The above is in <sys/param.h>. It makes many modern scripts for
commercial packages (typically for binary emulation), which want
to do things like
#!/path/to/customized/perl/interpreter -I/path/to/custom/libs -I/path/to/more/libs
bomb. This number really needs to be kicked up to at least 100 or so.
Or, better yet, kern_subr.c could check that the *interpreter path* is
the length actually checked, and leave its arguments out of the length
check. Even so, interpreter pathnames may still be longer than 64....
>How-To-Repeat:
grep MAXINTERP /usr/include/sys/param.h
>Fix:
Increase the number for MAXINTERP.
>Release-Note:
>Audit-Trail:
From: Todd Vierling <tv@pobox.com>
To: netbsd-bugs@netbsd.org
Cc: gnats-bugs@netbsd.org
Subject: Re: kern/20112: MAXINTERP set far too low for modern scripts
Date: Wed, 29 Jan 2003 15:50:47 -0500 (EST)
: #!/path/to/customized/perl/interpreter -I/path/to/custom/libs -I/path/to/more/libs
I've been requested to provide a better test case, so here it is to
demonstrate the problem very succinctly:
=====
#!/usr/pkg/bin/perl -I/obnoxiously/long/path/to/some/directory/over/64/chars
print "hello world!\n";
=====
With the -I..., the MAXINTERP check fails and the script is run with a
standard shell. Without the -I, it runs fine.
--
-- Todd Vierling <tv@pobox.com>
From: cgd@broadcom.com
To: tv@pobox.com
Cc: netbsd-bugs@netbsd.org, gnats-bugs@netbsd.org
Subject: Re: kern/20112: MAXINTERP set far too low for modern scripts
Date: 29 Jan 2003 13:41:54 -0800
Hmm. I find this diff in my local source tree. it allow interpreter
names (incl arguments) to the the max path length.
I recall sending it off to somebody, i don't remember who, for
review/comment, but don't recall hearing anything back.
Anyway, enjoy; do with it what you will.
cgd
--
Index: kern/exec_conf.c
===================================================================
RCS file: /cvsroot/syssrc/sys/kern/exec_conf.c,v
retrieving revision 1.77
diff -u -p -r1.77 exec_conf.c
--- kern/exec_conf.c 2002/11/13 15:16:27 1.77
+++ kern/exec_conf.c 2002/11/23 05:33:33
@@ -189,7 +189,7 @@ extern const struct emul emul_netbsd_aou
const struct execsw execsw_builtin[] = {
#ifdef EXEC_SCRIPT
/* Shell scripts */
- { MAXINTERP,
+ { SCRIPT_HDR_SIZE,
exec_script_makecmds,
{ NULL },
NULL,
Index: kern/exec_script.c
===================================================================
RCS file: /cvsroot/syssrc/sys/kern/exec_script.c,v
retrieving revision 1.32
diff -u -p -r1.32 exec_script.c
--- kern/exec_script.c 2002/10/29 12:31:23 1.32
+++ kern/exec_script.c 2002/11/23 05:33:33
@@ -96,7 +96,7 @@ exec_script_makecmds(struct proc *p, str
* (The latter requirement means that we have to check
* for both spaces and tabs later on.)
*/
- hdrlinelen = min(epp->ep_hdrvalid, MAXINTERP);
+ hdrlinelen = min(epp->ep_hdrvalid, SCRIPT_HDR_SIZE);
for (cp = hdrstr + EXEC_SCRIPT_MAGICLEN; cp < hdrstr + hdrlinelen;
cp++) {
if (*cp == '\n') {
Index: sys/exec_script.h
===================================================================
RCS file: /cvsroot/syssrc/sys/sys/exec_script.h,v
retrieving revision 1.7
diff -u -p -r1.7 exec_script.h
--- sys/exec_script.h 1997/01/22 07:09:15 1.7
+++ sys/exec_script.h 2002/11/23 05:33:33
@@ -36,6 +36,9 @@
#define EXEC_SCRIPT_MAGIC "#!"
#define EXEC_SCRIPT_MAGICLEN 2
+/* Extra 2 are for possible space between #! and shell name, and newline. */
+#define SCRIPT_HDR_SIZE (EXEC_SCRIPT_MAGICLEN + MAXINTERP + 2)
+
#ifdef _KERNEL
/* the shell script handler's entry in the exec switch */
Index: sys/param.h
===================================================================
RCS file: /cvsroot/syssrc/sys/sys/param.h,v
retrieving revision 1.153
diff -u -p -r1.153 param.h
--- sys/param.h 2002/11/17 22:53:46 1.153
+++ sys/param.h 2002/11/23 05:33:34
@@ -100,7 +100,7 @@
#include <sys/syslimits.h>
#define MAXCOMLEN 16 /* max command name remembered */
-#define MAXINTERP 64 /* max interpreter file name length */
+#define MAXINTERP PATH_MAX /* max interpreter file name length */
/* DEPRECATED: use LOGIN_NAME_MAX instead. */
#define MAXLOGNAME (LOGIN_NAME_MAX - 1) /* max login name length */
#define NCARGS ARG_MAX /* max bytes for an exec function */
From: Todd Vierling <tv@pobox.com>
To: cgd@broadcom.com
Cc: netbsd-bugs@netbsd.org, gnats-bugs@netbsd.org
Subject: Re: kern/20112: MAXINTERP set far too low for modern scripts
Date: Wed, 29 Jan 2003 16:48:31 -0500 (EST)
On Wed, 29 Jan 2003 cgd@broadcom.com wrote:
: Hmm. I find this diff in my local source tree. it allow interpreter
: names (incl arguments) to the the max path length.
:
: I recall sending it off to somebody, i don't remember who, for
: review/comment, but don't recall hearing anything back.
This seems rather sensible as a default ("hint hint"), actually. Note that
this still doesn't address the fact that MAXINTERP is a maximum limit on the
whole line (not just the interpreter pathname), but at least it's more than
the obnoxiously small 64, and it names the macro used in exec_script.c
better (SCRIPT_HDR_LEN).
--
-- Todd Vierling <tv@pobox.com>
From: cgd@broadcom.com
To: "Todd Vierling" <tv@pobox.com>
Cc: netbsd-bugs@netbsd.org, gnats-bugs@netbsd.org
Subject: Re: kern/20112: MAXINTERP set far too low for modern scripts
Date: 29 Jan 2003 13:52:46 -0800
At Wed, 29 Jan 2003 16:48:31 -0500 (EST), Todd Vierling wrote:
> Note that
> this still doesn't address the fact that MAXINTERP is a maximum limit on the
> whole line (not just the interpreter pathname), but at least it's more than
> the obnoxiously small 64, and it names the macro used in exec_script.c
> better (SCRIPT_HDR_LEN).
Yes, exactly.
(The reason why i created this diff is because i actually ran up
against the limit in the executable name, even w/o argument... I
don't remember the exact details but it was a script using an
executable installed way down deep below my home dir, or something.)
cgd
From: "Perry E. Metzger" <perry@netbsd.org>
To: gnats-bugs@gnats.netbsd.org
Cc:
Subject: pr/20112 CVS commit: src/sys/sys
Date: Wed, 2 Apr 2003 03:58:58 +0300 (EEST)
Module Name: src
Committed By: perry
Date: Wed Apr 2 00:58:57 UTC 2003
Modified Files:
src/sys/sys: exec_script.h param.h
Log Message:
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd Vierling
To generate a diff of this commit:
cvs rdiff -r1.7 -r1.8 src/sys/sys/exec_script.h
cvs rdiff -r1.162 -r1.163 src/sys/sys/param.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: "Perry E. Metzger" <perry@netbsd.org>
To: gnats-bugs@gnats.netbsd.org
Cc:
Subject: pr/20112 CVS commit: src/sys/kern
Date: Wed, 2 Apr 2003 04:00:20 +0300 (EEST)
Module Name: src
Committed By: perry
Date: Wed Apr 2 01:00:20 UTC 2003
Modified Files:
src/sys/kern: exec_conf.c exec_script.c
Log Message:
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd Vierling
To generate a diff of this commit:
cvs rdiff -r1.79 -r1.80 src/sys/kern/exec_conf.c
cvs rdiff -r1.32 -r1.33 src/sys/kern/exec_script.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
State-Changed-From-To: open->closed
State-Changed-By: perry
State-Changed-When: Tue Apr 1 17:01:35 PST 2003
State-Changed-Why:
I applied cgd's proposed patch.
From: Todd Vierling <tv@pobox.com>
To: "Perry E. Metzger" <perry@piermont.com>
Cc:
Subject: Re: kern/20112
Date: Mon, 28 Apr 2003 18:12:47 -0400 (EDT)
On Tue, 1 Apr 2003, Perry E. Metzger wrote:
: BTW, please verify that the problem is now gone for you.
: > Synopsis: MAXINTERP set far too low for modern scripts
: >
: > State-Changed-From-To: open->closed
: > I applied cgd's proposed patch.
Looks fine to me. I'll merge to it in when moving to the future 2.0 release
branch.
--
-- Todd Vierling <tv@pobox.com>
From: Matthias Scheler <tron@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: pr/20112 CVS commit: [netbsd-1-6] src/sys/sys
Date: Sun, 17 Aug 2003 11:19:35 +0000 (UTC)
Module Name: src
Committed By: tron
Date: Sun Aug 17 11:19:35 UTC 2003
Modified Files:
src/sys/sys [netbsd-1-6]: param.h
Log Message:
Pull up revision 1.163 (requested by tv in ticket #1423):
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd
Vierling
To generate a diff of this commit:
cvs rdiff -r1.138.4.4 -r1.138.4.5 src/sys/sys/param.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Matthias Scheler <tron@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: pr/20112 CVS commit: [netbsd-1-6] src/sys/sys
Date: Sun, 17 Aug 2003 11:19:42 +0000 (UTC)
Module Name: src
Committed By: tron
Date: Sun Aug 17 11:19:42 UTC 2003
Modified Files:
src/sys/sys [netbsd-1-6]: exec_script.h
Log Message:
Pull up revision 1.8 (requested by tv in ticket #1423):
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd
Vierling
To generate a diff of this commit:
cvs rdiff -r1.7 -r1.7.54.1 src/sys/sys/exec_script.h
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Matthias Scheler <tron@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: pr/20112 CVS commit: [netbsd-1-6] src/sys/kern
Date: Sun, 17 Aug 2003 13:40:54 +0000 (UTC)
Module Name: src
Committed By: tron
Date: Sun Aug 17 13:40:54 UTC 2003
Modified Files:
src/sys/kern [netbsd-1-6]: exec_conf.c
Log Message:
Pull up revision 1.80 (requested by tv in ticket #1424):
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd Vierling
To generate a diff of this commit:
cvs rdiff -r1.72 -r1.72.4.1 src/sys/kern/exec_conf.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
From: Matthias Scheler <tron@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: pr/20112 CVS commit: [netbsd-1-6] src/sys/kern
Date: Sun, 17 Aug 2003 13:41:01 +0000 (UTC)
Module Name: src
Committed By: tron
Date: Sun Aug 17 13:41:00 UTC 2003
Modified Files:
src/sys/kern [netbsd-1-6]: exec_script.c
Log Message:
Pull up revision 1.33 (requested by tv in ticket #1424):
grow #! line length -- patch from cgd, fixes PR kern/20112 from Todd Vierling
To generate a diff of this commit:
cvs rdiff -r1.30 -r1.30.10.1 src/sys/kern/exec_script.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.