No Subject

No Subject

[Philippe De Muyter]

Am I the only one to have (sometimes) the problem of finding a non-executable
with the same name earlier in my path than the program I want to debug ?

Here is a patch.

I also removed the obsolete (now that we have gdb_string.h) strstr
declaration comment.

Mon Apr  6 13:20:21 1998  Philippe De Muyter  <phdm@macqel.be>

	* source.c (openp): Accept only files that match prot.
	* exec.c (exec_file_command): Request openp to find an executable
	file.

--- ./gdb/source.c	Mon Apr  6 11:22:17 1998
+++ ./gdb/source.c	Mon Apr  6 09:49:08 1998
@@ -83,11 +83,6 @@
 
 static void find_source_lines PARAMS ((struct symtab *, int));
 
-/* If we use this declaration, it breaks because of fucking ANSI "const" stuff
-   on some systems.  We just have to not declare it at all, have it default
-   to int, and possibly botch on a few systems.  Thanks, ANSIholes... */
-/* extern char *strstr(); */
-
 /* Path of directories to search for source files.
    Same format as the PATH environment variable's value.  */
 
@@ -521,7 +516,7 @@ openp (path, try_cwd_first, string, mode
     {
       int i;
       filename = string;
-      fd = open (filename, mode, prot);
+      fd = open (filename, mode);
       if (fd >= 0)
 	goto done;
       for (i = 0; string[i]; i++)
@@ -570,8 +565,10 @@ openp (path, try_cwd_first, string, mode
       strcat (filename+len, SLASH_STRING);
       strcat (filename, string);
 
-      fd = open (filename, mode);
-      if (fd >= 0) break;
+      /* Accept only files matching prot. */
+      if ((prot == 0 || access(filename, prot) == 0)
+      	  && (fd = open (filename, mode)) >= 0)
+	break;
     }
 
  done:
--- ./gdb/exec.c	Mon Apr  6 11:22:18 1998
+++ ./gdb/exec.c	Sat Apr  4 09:37:50 1998
@@ -32,6 +32,12 @@
 #endif
 
 #include <fcntl.h>
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+#ifndef X_OK
+#define X_OK	1
+#endif
 #include "gdb_string.h"
 
 #include "gdbcore.h"
@@ -185,8 +191,8 @@ exec_file_command (args, from_tty)
       make_cleanup (free, filename);
       
       scratch_chan = openp (getenv ("PATH"), 1, filename, 
-			    write_files? O_RDWR|O_BINARY: O_RDONLY|O_BINARY, 0,
-			    &scratch_pathname);
+			    write_files? O_RDWR|O_BINARY: O_RDONLY|O_BINARY,
+			    X_OK, &scratch_pathname);
 #if defined(__GO32__) || defined(_WIN32)
       if (scratch_chan < 0)
       {

No Subject

[Stan Shebs]

   Date: Mon, 6 Apr 1998 13:41:37 +0200 (MET DST)
   From: "Philippe De Muyter" <phdm@macqel.be>

   Am I the only one to have (sometimes) the problem of finding a non-executable
   with the same name earlier in my path than the program I want to debug ?

Apparently. :-)  But seriously, I don't think this change is a good idea.
While it would work fine for native Unix debugging, it will lose for
just about everything else.  For both cross-Unix and embedded debugging
you almost certainly want the programs *not* to be marked as executable,
so that your current host doesn't try to execute them.

However, if you set things up so that this test is only made when
using a Unix child_ops, this would be a useful addition.

							Stan

Re:

[Michael Snyder]

Stan Shebs wrote:
> 
>    Date: Mon, 6 Apr 1998 13:41:37 +0200 (MET DST)
>    From: "Philippe De Muyter" <phdm@macqel.be>
> 
>    Am I the only one to have (sometimes) the problem of finding a non-executable
>    with the same name earlier in my path than the program I want to debug ?
> 
> Apparently. :-)  But seriously, I don't think this change is a good idea.
> While it would work fine for native Unix debugging, it will lose for
> just about everything else.  For both cross-Unix and embedded debugging
> you almost certainly want the programs *not* to be marked as executable,
> so that your current host doesn't try to execute them.
> 
> However, if you set things up so that this test is only made when
> using a Unix child_ops, this would be a useful addition.

I don't think I have ever had this problem.  Is it not the case
that GDB always looks in the "current working directory" first, 
and only then looks along your path?  (certainly I do not have
"current working directory" on my path!)

If the above is the case, then under what circumstances could 
you have this problem?  The only one that I can think of is if
you were trying to debug something that is NOT in your current
working directory, yet you did NOT give a path to the binary
that you wanted to debug.

In which case, I would have to ask, "why would you want to do that?"

Michael

Re:

[Michael Meissner]

Stan Shebs wrote:

|    Date: Mon, 6 Apr 1998 13:41:37 +0200 (MET DST)
|    From: "Philippe De Muyter" <phdm@macqel.be>
| 
|    Am I the only one to have (sometimes) the problem of finding a non-executable
|    with the same name earlier in my path than the program I want to debug ?
| 
| Apparently. :-)  But seriously, I don't think this change is a good idea.
| While it would work fine for native Unix debugging, it will lose for
| just about everything else.  For both cross-Unix and embedded debugging
| you almost certainly want the programs *not* to be marked as executable,
| so that your current host doesn't try to execute them.

Maybe on YOUR system you don't want them to be marked executable.  On
my 2.1.xx linux system, I just tell the system to run the appropriate
simulator for the binary (for example big and little endian
PowerPC's), so I definately want the executable bit set.

-- 
Michael Meissner, Cygnus Solutions (Massachusetts office)
4th floor, 955 Massachusetts Avenue, Cambridge, MA 02139, USA
meissner@cygnus.com,	617-354-5416 (office),	617-354-7161 (fax)

Re:

[Andrew Cagney]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1477 bytes --]

DJBARROW@de.ibm.com wrote:
> 
> Hi,
> I forgot in the last note, let Boas & me know if the patches are accepted
> or rejected so that we can
> send the correct md5 checksums & the release letter.

Puzzled expression.

I'm pretty sure that parts of the changes will need more work (that is
without looking at them ;-).  This sort of thing needs to be done
incrementally.  It is simply a part of life.

What exactly do you mean when you refer to an md5 checksum and release
letter?  For this to work, you'll need to assign the changes and then
set up some sort of arangement where by you can continue to submit
patches.

I'm sorry but I'm confused.

	Andrew


> D.J. Barrow Linux for S/390 kernel developer
> eMail: djbarrow@de.ibm.com,barrow_dj@yahoo.com
> Phone: +49-(0)7031-16-2583
> IBM Germany Lab, Schönaicherstr. 220, 71032 Böblingen
From DJBARROW@de.ibm.com Tue Sep 19 02:34:00 2000
From: DJBARROW@de.ibm.com
To: gdb-patches@sourceware.cygnus.com
Subject: Re gdb s390 patches
Date: Tue, 19 Sep 2000 02:34:00 -0000
Message-id: <C125695F.0034815E.00@d12mta09.de.ibm.com>
X-SW-Source: 2000-09/msg00165.html
Content-length: 336

Hi,
I forgot in the last note, let Boas & me know if the patches are accepted
or rejected so that we can
send the correct md5 checksums & the release letter.

D.J. Barrow Linux for S/390 kernel developer
eMail: djbarrow@de.ibm.com,barrow_dj@yahoo.com
Phone: +49-(0)7031-16-2583
IBM Germany Lab, Schönaicherstr. 220, 71032 Böblingen

Re:

[Christoph Arenz]

Hi Denis,

I just want to let you know that the Software Letter has been signed and is
on its way to the FSF.
So the patches on our webpage are covered by this copyright assignment.

http://www10.software.ibm.com/developerworks/opensource/linux390/exp_src.html
     Patch:    gdb-5.1pre-050901-s390.tar.gz (09/11/2001)
     MD5: 886251f3719a754dd65a69df462ceac1

Kind regards,
Christoph

GNU/Linux on zSeries PDT
Phone:   (845) 435-8989, Tie-Line: 295-8989
Fax:        (845) 432-9787

Re:

[Andrew Cagney]

> Hi Denis,
> 
> I just want to let you know that the Software Letter has been signed and is
> on its way to the FSF.
> So the patches on our webpage are covered by this copyright assignment.
> 
> http://www10.software.ibm.com/developerworks/opensource/linux390/exp_src.html
> Patch:    gdb-5.1pre-050901-s390.tar.gz (09/11/2001)
>      MD5: 886251f3719a754dd65a69df462ceac1

Thanks, we now all wait ...

	Andrew

Re:

[Andrew Cagney]

> Hi Denis,
> 
> I just want to let you know that the Software Letter has been signed and is
> on its way to the FSF.
> So the patches on our webpage are covered by this copyright assignment.
> 
> http://www10.software.ibm.com/developerworks/opensource/linux390/exp_src.html
> Patch:    gdb-5.1pre-050901-s390.tar.gz (09/11/2001)
>      MD5: 886251f3719a754dd65a69df462ceac1
> 
> Thanks, we now all wait ...

fyi, it has arrived.

	Andrew