strings, change default to -a?

2014-10-25 Thread Stuart Henderson 
http://seclists.org/oss-sec/2014/q4/445

Any thoughts on changing strings(1) to use -a by default, to avoid
libbfd parsing, and add a new option to allow previous behaviour for
people who want it?

About -a, posix says Scan files in their entirety. If -a is not
specified, it is implementation-defined what portion of each file is
scanned for strings.

http://pubs.opengroup.org/onlinepubs/9699919799/utilities/strings.html#tag_20_121_04



Index: binutils/strings.c
===
RCS file: /cvs/src/gnu/usr.bin/binutils/binutils/strings.c,v
retrieving revision 1.8
diff -u -p -r1.8 strings.c
--- binutils/strings.c  31 Aug 2014 13:40:02 -  1.8
+++ binutils/strings.c  25 Oct 2014 12:01:00 -
@@ -174,17 +174,21 @@ main (int argc, char **argv)
   string_min = -1;
   print_addresses = FALSE;
   print_filenames = FALSE;
-  datasection_only = TRUE;
+  datasection_only = FALSE;
   target = NULL;
   encoding = 's';
 
-  while ((optc = getopt_long (argc, argv, afhHn:ot:e:Vv0123456789,
+  while ((optc = getopt_long (argc, argv, aAfhHn:ot:e:Vv0123456789,
  long_options, (int *) 0)) != EOF)
 {
   switch (optc)
{
case 'a':
  datasection_only = FALSE;
+ break;
+
+   case 'A':
+ datasection_only = TRUE;
  break;
 
case 'f':
Index: binutils/doc/strings.1
===
RCS file: /cvs/src/gnu/usr.bin/binutils/binutils/doc/strings.1,v
retrieving revision 1.1.1.2
diff -u -p -r1.1.1.2 strings.1
--- binutils/doc/strings.1  2 Nov 2004 20:22:07 -   1.1.1.2
+++ binutils/doc/strings.1  25 Oct 2014 12:01:00 -
@@ -133,7 +133,7 @@
 strings \- print the strings of printable characters in files.
 .SH SYNOPSIS
 .IX Header SYNOPSIS
-strings [\fB\-afov\fR] [\fB\-\fR\fImin-len\fR]
+strings [\fB\-Aafov\fR] [\fB\-\fR\fImin-len\fR]
 [\fB\-n\fR \fImin-len\fR] [\fB\-\-bytes=\fR\fImin-len\fR]
 [\fB\-t\fR \fIradix\fR] [\fB\-\-radix=\fR\fIradix\fR]
 [\fB\-e\fR \fIencoding\fR] [\fB\-\-encoding=\fR\fIencoding\fR]
@@ -163,6 +163,10 @@ files.
 .PD
 Do not scan only the initialized and loaded sections of object files;
 scan the whole files.
+This is the default in this version.
+.IP \fB\-A\fR 4
+.IX Item -A
+Scan only the initialized and loaded sections of object files.
 .IP \fB\-f\fR 4
 .IX Item -f
 .PD 0

Re: strings, change default to -a?

2014-10-25 Thread Mark Kettenis 
 Date: Sat, 25 Oct 2014 13:18:04 +0100
 From: Stuart Henderson st...@openbsd.org
 
 http://seclists.org/oss-sec/2014/q4/445
 
 Any thoughts on changing strings(1) to use -a by default, to avoid
 libbfd parsing, and add a new option to allow previous behaviour for
 people who want it?
 
 About -a, posix says Scan files in their entirety. If -a is not
 specified, it is implementation-defined what portion of each file is
 scanned for strings.
 
 http://pubs.opengroup.org/onlinepubs/9699919799/utilities/strings.html#tag_20_121_04

Unless upstream makes a similar change, I don't think we should do this.

 Index: binutils/strings.c
 ===
 RCS file: /cvs/src/gnu/usr.bin/binutils/binutils/strings.c,v
 retrieving revision 1.8
 diff -u -p -r1.8 strings.c
 --- binutils/strings.c31 Aug 2014 13:40:02 -  1.8
 +++ binutils/strings.c25 Oct 2014 12:01:00 -
 @@ -174,17 +174,21 @@ main (int argc, char **argv)
string_min = -1;
print_addresses = FALSE;
print_filenames = FALSE;
 -  datasection_only = TRUE;
 +  datasection_only = FALSE;
target = NULL;
encoding = 's';
  
 -  while ((optc = getopt_long (argc, argv, afhHn:ot:e:Vv0123456789,
 +  while ((optc = getopt_long (argc, argv, aAfhHn:ot:e:Vv0123456789,
 long_options, (int *) 0)) != EOF)
  {
switch (optc)
   {
   case 'a':
 datasection_only = FALSE;
 +   break;
 +
 + case 'A':
 +   datasection_only = TRUE;
 break;
  
   case 'f':
 Index: binutils/doc/strings.1
 ===
 RCS file: /cvs/src/gnu/usr.bin/binutils/binutils/doc/strings.1,v
 retrieving revision 1.1.1.2
 diff -u -p -r1.1.1.2 strings.1
 --- binutils/doc/strings.12 Nov 2004 20:22:07 -   1.1.1.2
 +++ binutils/doc/strings.125 Oct 2014 12:01:00 -
 @@ -133,7 +133,7 @@
  strings \- print the strings of printable characters in files.
  .SH SYNOPSIS
  .IX Header SYNOPSIS
 -strings [\fB\-afov\fR] [\fB\-\fR\fImin-len\fR]
 +strings [\fB\-Aafov\fR] [\fB\-\fR\fImin-len\fR]
  [\fB\-n\fR \fImin-len\fR] [\fB\-\-bytes=\fR\fImin-len\fR]
  [\fB\-t\fR \fIradix\fR] [\fB\-\-radix=\fR\fIradix\fR]
  [\fB\-e\fR \fIencoding\fR] [\fB\-\-encoding=\fR\fIencoding\fR]
 @@ -163,6 +163,10 @@ files.
  .PD
  Do not scan only the initialized and loaded sections of object files;
  scan the whole files.
 +This is the default in this version.
 +.IP \fB\-A\fR 4
 +.IX Item -A
 +Scan only the initialized and loaded sections of object files.
  .IP \fB\-f\fR 4
  .IX Item -f
  .PD 0

Re: strings, change default to -a?

2014-10-25 Thread Theo de Raadt 
 Date: Sat, 25 Oct 2014 13:18:04 +0100
 From: Stuart Henderson st...@openbsd.org
 
 http://seclists.org/oss-sec/2014/q4/445
 
 Any thoughts on changing strings(1) to use -a by default, to avoid
 libbfd parsing, and add a new option to allow previous behaviour for
 people who want it?
 
 About -a, posix says Scan files in their entirety. If -a is not
 specified, it is implementation-defined what portion of each file is
 scanned for strings.
 
 http://pubs.opengroup.org/onlinepubs/9699919799/utilities/strings.html#tag_20_121_04

Unless upstream makes a similar change, I don't think we should do this.

I agree.  This behavioural change seems too much.  I like my historical
behaviours.