Re: [RFA] new test for separate debug info

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

> Hmm, maybe. That's an easy fix, if needed. Does binutils deal with som?

I don't know.  But I like Daniel's idea; make it "dwarf-2 only".
In the place where you comment that stabs+ does not work, you
can comment that SOM has not even been tried.

Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > > Hmm, maybe. That's an easy fix, if needed. Does binutils deal with som?
 > 
 > I don't know.  But I like Daniel's idea; make it "dwarf-2 only".
 > In the place where you comment that stabs+ does not work, you
 > can comment that SOM has not even been tried.
 > 
 > Michael C

Just curious, but I'd like to see what happens on HPUX first. The test
deals with the case in which strip and objcopy themselves fail as
well. The case for stabs was that objcopy/strip was working w/o errors
or warnings but the result was corrupted. Maybe here they'll just die.

elena

Re: [RFA] new test for separate debug info

[Daniel Jacobowitz]

On Fri, Jan 16, 2004 at 02:55:31PM -0500, Andrew Cagney wrote:
> >> Ya!  Now what would be nice is a similar gdbserver sniff test :-)
> >
> >
> >Er.... I have absolutely no idea what that sentence means :)
> 
> short answer:
> 
> gdb/testsuite/gdb.base/gdbserver.exp

Ah, OK!  I see what you mean now.  Let me think about how to do this.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] new test for separate debug info

[Andrew Cagney]

> > Ya!  Now what would be nice is a similar gdbserver sniff test :-)
> 
> 
> Er.... I have absolutely no idea what that sentence means :)

short answer:

gdb/testsuite/gdb.base/gdbserver.exp

long answer:

There are two ways of testing stuff like separate debug info and gdb-server:

- as part of full and separate dejagnu test run
eg: make check RUNTESTFLAGS=--target_board=gdbserver

- as a small focused functional / unit test
eg: make check RUNTESTFLAGS=gdbserver.exp

Both are correct as both offer distinct advantages and disadvantages 
(mind you I should note that I've witnessed argument disputing this :-). 
  The former provides better coverage, but with no realistic expectation 
that the average GDB developer will ever run it; while the later ensures 
that that specific functionality is always working, but lacks broad 
coverage.

A gdbserver.exp unit test would significantly improve the odds of 
gdbserver working with a standard GDB.  It would also make possible the 
writing of gdbserver specific tests that explore edge cases missed by 
the more standard tests (the attach / detach behavior for instance). 
Also, in support of this, even though separate debug info bit rot and 
broke within months of its initial commit, the breakage wasn't noticed 
for several further months (when the unit test was added).

Andrew

Re: [RFA] new test for separate debug info

[Elena Zannoni]

OK, this is what I checked in.

2004-01-14  Elena Zannoni  <ezannoni@redhat.com>

	* gdb.base/sepdebug.exp: Bail out of the test if we don't know
	what debug info we have. Print a better message if something goes
	wrong while producing the separate debug info file.


Index: sepdebug.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/sepdebug.exp,v
retrieving revision 1.1
diff -u -p -r1.1 sepdebug.exp
--- sepdebug.exp        12 Jan 2004 17:20:41 -0000      1.1
+++ sepdebug.exp        14 Jan 2004 15:05:51 -0000
@@ -72,17 +72,22 @@ if { [test_debug_format "stabs"] } then
     # (see dejagnu/lib/framework.exp)
     unsupported "no separate debug info handling with stabs"
     return -1
+} elseif { [test_debug_format "unknown"] } then {
+    # gdb doesn't know what the debug format is. We are out of luck here.
+    unsupported "unknown debugging format"
+    return -1
 }
 gdb_exit
  
 # Note: the procedure gdb_gnu_strip_debug will produce an executable called
 # ${binfile}, which is just like the executable ($binfile) but without
 # the debuginfo. Instead $binfile has a .gnudebuglink section which contains
-# the name of a idebuginfo only file. This file will be stored in the
+# the name of a debuginfo only file. This file will be stored in the
 # gdb.base/.debug subdirectory.
  
 if [gdb_gnu_strip_debug $binfile] {
-    unsupported "no separate debug info handling"
+    # check that you have a recent version of strip and objcopy installed
+    unsupported "cannot produce separate debug info files"
     return -1
 }
  

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

> My idea was more to incrementally add broken cases to the skip list as
> we find them.

It's not my favorite, but it's okay with me.

  +} elseif { [test_debug_format "unknown"] } then {
  +    # gdb doesn't know what the debug format is. We are out of luck here.
  +    unsupported "unknown debugging format"
  +    return -1

I tested this.  It works for me:

  ...
  Compiled with unknown debugging format.
  Does not include preprocessor macro info.
  (gdb) UNSUPPORTED: gdb.base/sepdebug.exp: unknown debugging format
  testcase /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.exp completed in 1 seconds

   if [gdb_gnu_strip_debug $binfile] {
  -    unsupported "no separate debug info handling"
  +    unsupported "cannot produce separate debug info files (check that you have
  a recent version of strip and objcopy installed)"
       return -1
 }

I would rather that the message be shorter so that it does not blow up
my tables if it appears in a table.  How about:

  if [gdb_gnu_strip_debug $binfile] {
      # check that you have a recent version of strip and objcopy
      unsupported "cannot produce separate debug info files"
  }

Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > I've seen this "unknown debugging format" before, on dwarf-1.
 > I really agree with daniel's idea that the test should be
 > "if format == dwarf-2" rather than "if format != stabs+".
 > 

My idea was more to incrementally add broken cases to the skip list as
we find them. Or one can be in a situation where the test doesn't run
and you don't even know if it really doesn't work. Definitely the
'unknow debugging format' qualifies for skipping.


Index: sepdebug.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/gdb.base/sepdebug.exp,v
retrieving revision 1.1
diff -u -p -r1.1 sepdebug.exp
--- sepdebug.exp        12 Jan 2004 17:20:41 -0000      1.1
+++ sepdebug.exp        13 Jan 2004 23:20:41 -0000
@@ -72,6 +72,10 @@ if { [test_debug_format "stabs"] } then
     # (see dejagnu/lib/framework.exp)
     unsupported "no separate debug info handling with stabs"
     return -1
+} elseif { [test_debug_format "unknown"] } then {
+    # gdb doesn't know what the debug format is. We are out of luck here.
+    unsupported "unknown debugging format"
+    return -1
 }
 gdb_exit
  


 > > Maybe the two unsupported messages should differ a bit more from each
 > > other: "cannot produce separate debug info file" instead?
 > 
 > How about "strip --strip-debug failed", make it really clear.
 > 

Let me think about this, there is also objcopy in the mix. I did try
to have the strip procedure return the error message from strip, so
that it could be displayed, but the message is very long, so I
returned 0 or 1. If you do make check with -v you'll see the real
error.

how is this:


 if [gdb_gnu_strip_debug $binfile] {
-    unsupported "no separate debug info handling"
+    unsupported "cannot produce separate debug info files (check that you have
a recent version of strip and objcopy installed)"
     return -1
 }
  

elena


 > Michael C

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

I've seen this "unknown debugging format" before, on dwarf-1.
I really agree with daniel's idea that the test should be
"if format == dwarf-2" rather than "if format != stabs+".

> Maybe the two unsupported messages should differ a bit more from each
> other: "cannot produce separate debug info file" instead?

How about "strip --strip-debug failed", make it really clear.

Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > Okay, I just ran it on spe191.testdrive.hp.com to see what happens.
 > What happens is:
 > 
 >   (gdb) info source
 >   Current source file is /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.c
 >   Located in /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.c
 >   Contains 150 lines.
 >   Source language is c.
 >   Compiled with unknown debugging format.

aarrrgh

 >   Does not include preprocessor macro info.
 >   (gdb) UNSUPPORTED: gdb.base/sepdebug.exp: no separate debug info handling
 >   testcase /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.exp completed in 0 seconds
 > 
 > This is with HP Ansi C B.11.11.28707.GP.
 > 
 > I'm happy with this outcome, although I am not happy that "info source"
 > prints "compiled with unknown debugging format".  I will investigate and
 > file a PR.
 > 
 > Also, the message does not happen because of "unknown debugging format".
 > It happens because I don't have gnu binutils in my $PATH, and the HP-UX
 > strip barfs on "--strip-debug".

Maybe the two unsupported messages should differ a bit more from each
other: "cannot produce separate debug info file" instead?

 > 
 > I haven't set up a test bed for gcc + binutils on this platform yet,
 > just the native tools.  I did try GNU strip from binutils 2.14 on
 > a hello-world program with "strip --strip-debug hello" and it didn't
 > complain.
 > 

OK, when you do let me know what other marvellous ways the test fails.

elena

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

Okay, I just ran it on spe191.testdrive.hp.com to see what happens.
What happens is:

  (gdb) info source
  Current source file is /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.c
  Located in /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.c
  Contains 150 lines.
  Source language is c.
  Compiled with unknown debugging format.
  Does not include preprocessor macro info.
  (gdb) UNSUPPORTED: gdb.base/sepdebug.exp: no separate debug info handling
  testcase /house/chastain/gdb/s1/gdb/testsuite/gdb.base/sepdebug.exp completed in 0 seconds

This is with HP Ansi C B.11.11.28707.GP.

I'm happy with this outcome, although I am not happy that "info source"
prints "compiled with unknown debugging format".  I will investigate and
file a PR.

Also, the message does not happen because of "unknown debugging format".
It happens because I don't have gnu binutils in my $PATH, and the HP-UX
strip barfs on "--strip-debug".

I haven't set up a test bed for gcc + binutils on this platform yet,
just the native tools.  I did try GNU strip from binutils 2.14 on
a hello-world program with "strip --strip-debug hello" and it didn't
complain.

By the way, HP Ansi C and HP C++ have an objdebug switch:

      +objdebug      This option when used with any of the -g options will
                     cause the debug information to be left in the object
                     files instead of being placed in the a.out.  This will
                     bypass the pxdb step and the links will be faster.
                     Available only for wdb(1) or gdb(1).  +noobjdebug is
                     the default at compile time.  +noobjdebug can be used
                     at link time to continue to place the debug info into
                     the a.out, even if some objects were compiled with
                     +objdebug.
                     NOTE: Object files or archive libraries must not be
                     removed if this new method of debugging is used.

Michael C

Re: [RFA] new test for separate debug info

[Ian Lance Taylor]

Elena Zannoni <ezannoni@redhat.com> writes:

> Daniel Jacobowitz writes:
>  > 
>  > BTW, Elena, should it be testing for the presence of dwarf2 rather than
>  > the absence of stabs?  I suspect you'll be hearing from MichaelC soon
>  > if the test doesn't work on HP-UX.
>  > 
> 
> Hmm, maybe. That's an easy fix, if needed. Does binutils deal with som?

Yes, but native only.  You can't build a cross-assembler which
generates SOM.

Ian

Re: [RFA] new test for separate debug info

[Daniel Jacobowitz]

On Mon, Jan 12, 2004 at 06:10:06PM -0500, Elena Zannoni wrote:
> Daniel Jacobowitz writes:
>  > 
>  > BTW, Elena, should it be testing for the presence of dwarf2 rather than
>  > the absence of stabs?  I suspect you'll be hearing from MichaelC soon
>  > if the test doesn't work on HP-UX.
>  > 
> 
> Hmm, maybe. That's an easy fix, if needed. Does binutils deal with som?

I'm pretty sure it doesn't.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Daniel Jacobowitz writes:
 > 
 > BTW, Elena, should it be testing for the presence of dwarf2 rather than
 > the absence of stabs?  I suspect you'll be hearing from MichaelC soon
 > if the test doesn't work on HP-UX.
 > 

Hmm, maybe. That's an easy fix, if needed. Does binutils deal with som?

Re: [RFA] new test for separate debug info

[Daniel Jacobowitz]

On Mon, Jan 12, 2004 at 05:01:33PM -0500, Andrew Cagney wrote:
> >To pick up an old thread. I am committing this. I have disabled the
> >test for stabs.
> >
> >The next thing to do is test with the new switches that Jakub and Nick
> >have added to binutils.
> >
> >2004-01-12  Elena Zannoni  <ezannoni@redhat.com>
> >
> >	* gdb.base/sepdebug.exp: New file.
> >	* gdb.base/sepdebug.c: New file.
> >	* lib/gdb.exp (separate_debug_filename): New procedure.
> >	(gdb_gnu_strip_debug): New procedure.
> >
> 
> Ya!  Now what would be nice is a similar gdbserver sniff test :-)

Er.... I have absolutely no idea what that sentence means :)

BTW, Elena, should it be testing for the presence of dwarf2 rather than
the absence of stabs?  I suspect you'll be hearing from MichaelC soon
if the test doesn't work on HP-UX.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] new test for separate debug info

[Andrew Cagney]

> To pick up an old thread. I am committing this. I have disabled the
> test for stabs.
> 
> The next thing to do is test with the new switches that Jakub and Nick
> have added to binutils.
> 
> 2004-01-12  Elena Zannoni  <ezannoni@redhat.com>
> 
> 	* gdb.base/sepdebug.exp: New file.
> 	* gdb.base/sepdebug.c: New file.
> 	* lib/gdb.exp (separate_debug_filename): New procedure.
> 	(gdb_gnu_strip_debug): New procedure.
> 

Ya!  Now what would be nice is a similar gdbserver sniff test :-)

Andrew

Re: [RFA] new test for separate debug info

[Elena Zannoni]

To pick up an old thread. I am committing this. I have disabled the
test for stabs.

The next thing to do is test with the new switches that Jakub and Nick
have added to binutils.

2004-01-12  Elena Zannoni  <ezannoni@redhat.com>

	* gdb.base/sepdebug.exp: New file.
	* gdb.base/sepdebug.c: New file.
	* lib/gdb.exp (separate_debug_filename): New procedure.
	(gdb_gnu_strip_debug): New procedure.


====================================

#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
#   2000, 2002, 2003, 2004
#   Free Software Foundation, Inc.

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

# Please email any bugs, comments, and/or additions to this file to:
# bug-gdb@prep.ai.mit.edu

# Based on break.exp, written by Rob Savoye. (rob@cygnus.com)
# Modified to test gdb's handling of separate debug info files.

# This file has two parts. The first is testing that gdb behaves
# normally after reading in an executable and its corresponding
# separate debug file. The second moves the .debug file to a different
# location and tests the "set debug-file-directory" command.


if $tracelevel then {
    strace $tracelevel
}

#
# test running programs
#
set prms_id 0
set bug_id 0

set testfile "sepdebug"
set srcfile ${testfile}.c
set binfile ${objdir}/${subdir}/${testfile}

if  { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug additional_flags=-w}] != "" } {
    gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
}

# FIXME: this is nasty. We need to check for the stabs debug format.
# To do this we must run gdb on the unstripped executable, list 'main'
# (as to have a default source file), use get_debug_format (which does
# 'info source') and then see if the debug info is stabs. If so, we
# bail out.  We cannot do this any other way because get_debug_format
# finds out the debug format using gdb itself, and in case of stabs we
# get an error loading the program if it is already stripped.  An
# alternative would be to find out the debug info from the flags
# passed to dejagnu when the test is run.

gdb_exit
gdb_start
gdb_reinitialize_dir $srcdir/$subdir
gdb_load ${binfile}
gdb_test "list main" "" ""
get_debug_format
if { [test_debug_format "stabs"] } then {
    # the separate debug info feature doesn't work well in binutils with stabs.
    # It produces a corrupted debug info only file, and gdb chokes on it.
    # It is almost impossible to capture the failing message out of gdb,
    # because it happens inside gdb_load. At that point any error message
    # is intercepted by dejagnu itself, and, because of the error threshold,
    # any faulty test result is changed into an UNRESOLVED. 
    # (see dejagnu/lib/framework.exp)
    unsupported "no separate debug info handling with stabs"
    return -1
}
gdb_exit

# Note: the procedure gdb_gnu_strip_debug will produce an executable called
# ${binfile}, which is just like the executable ($binfile) but without
# the debuginfo. Instead $binfile has a .gnudebuglink section which contains
# the name of a idebuginfo only file. This file will be stored in the
# gdb.base/.debug subdirectory.

if [gdb_gnu_strip_debug $binfile] {
    unsupported "no separate debug info handling"
    return -1
}

gdb_exit
gdb_start
gdb_reinitialize_dir $srcdir/$subdir
gdb_load ${binfile}

if [target_info exists gdb_stub] {
    gdb_step_for_stub;
}
#
# test simple breakpoint setting commands
#

# Test deleting all breakpoints when there are none installed,
# GDB should not prompt for confirmation.
# Note that gdb-init.exp provides a "delete_breakpoints" proc
# for general use elsewhere.

send_gdb "delete breakpoints\n"
gdb_expect {
     -re "Delete all breakpoints.*$" {
	    send_gdb "y\n"
	    gdb_expect {
		-re "$gdb_prompt $" {
		    fail "Delete all breakpoints when none (unexpected prompt)"
		}
		timeout	{ fail "Delete all breakpoints when none (timeout after unexpected prompt)" }
	    }
	}
     -re ".*$gdb_prompt $"       { pass "Delete all breakpoints when none" }
    timeout	            { fail "Delete all breakpoints when none (timeout)" }
}

#
# test break at function
#
gdb_test "break main" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function"

#
# test break at quoted function
#
gdb_test "break \"marker2\"" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint quoted function"

#
# test break at function in file
#
gdb_test "break $srcfile:factorial" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function in file"

set bp_location1 [gdb_get_line_number "set breakpoint 1 here"]

#
# test break at line number
#
# Note that the default source file is the last one whose source text
# was printed.  For native debugging, before we've executed the
# program, this is the file containing main, but for remote debugging,
# it's wherever the processor was stopped when we connected to the
# board.  So, to be sure, we do a list command.
#
gdb_test "list main" \
    ".*main \\(argc, argv, envp\\).*" \
    "use `list' to establish default source file"
gdb_test "break $bp_location1" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
    "breakpoint line number"

#
# test duplicate breakpoint
#
gdb_test "break $bp_location1" \
    "Note: breakpoint \[0-9\]+ also set at pc.*Breakpoint \[0-9\]+ at.* file .*$srcfile, line $bp_location1\\." \
    "breakpoint duplicate"

set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]

#
# test break at line number in file
#
gdb_test "break $srcfile:$bp_location2" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
    "breakpoint line number in file"

set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]

#
# Test putting a break at the start of a multi-line if conditional.
# Verify the breakpoint was put at the start of the conditional.
#
gdb_test "break multi_line_if_conditional" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
    "breakpoint at start of multi line if conditional"

gdb_test "break multi_line_while_conditional" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
    "breakpoint at start of multi line while conditional"

set bp_location5 [gdb_get_line_number "set breakpoint 5 here"]
set bp_location6 [gdb_get_line_number "set breakpoint 6 here"]

#
# check to see what breakpoints are set
#
if [target_info exists gdb_stub] {
    set main_line $bp_location5
} else {
    set main_line $bp_location6
}

set bp_location7 [gdb_get_line_number "set breakpoint 7 here"]
set bp_location8 [gdb_get_line_number "set breakpoint 8 here"]
set bp_location9 [gdb_get_line_number "set breakpoint 9 here"]

gdb_test "info break" \
    "Num Type\[ \]+Disp Enb Address\[ \]+What.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$main_line.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in marker2 at .*$srcfile:($bp_location8|$bp_location9).*
\[0-9\]+\[\t \]+breakpoint     keep y.* in factorial at .*$srcfile:$bp_location7.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location2.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in multi_line_if_conditional at .*$srcfile:$bp_location3.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
    "breakpoint info"

# FIXME: The rest of this test doesn't work with anything that can't
# handle arguments.
# Huh? There doesn't *appear* to be anything that passes arguments
# below.
if [istarget "mips-idt-*"] then {
    return
}

#
# run until the breakpoint at main is hit. For non-stubs-using targets.
#
if ![target_info exists use_gdb_stub] {
  if [istarget "*-*-vxworks*"] then {
    send_gdb "run vxmain \"2\"\n"
    set timeout 120
    verbose "Timeout is now $timeout seconds" 2
  } else {
	send_gdb "run\n"
  }
  gdb_expect {
    -re "The program .* has been started already.*y or n. $" {
	send_gdb "y\n"
	exp_continue
    }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $"\
	                    { pass "run until function breakpoint" }
    -re ".*$gdb_prompt $"       { fail "run until function breakpoint" }
    timeout	            { fail "run until function breakpoint (timeout)" }
  }
} else {
    if ![target_info exists gdb_stub] {
	gdb_test continue ".*Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.*\{.*" "stub continue"
    }
}

#
# run until the breakpoint at a line number
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location1.*$bp_location1\[\t \]+printf.*factorial.*" \
			"run until breakpoint set at a line number"

#
# Run until the breakpoint set in a function in a file
#
for {set i 6} {$i >= 1} {incr i -1} {
	gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, factorial \\(value=$i\\) at .*$srcfile:$bp_location7.*$bp_location7\[\t \]+.*if .value > 1. \{.*" \
			"run until file:function($i) breakpoint"
}

#
# Run until the breakpoint set at a quoted function
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile:($bp_location8|$bp_location9).*" \
		"run until quoted breakpoint"
#
# run until the file:function breakpoint at a line number in a file
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location2.*$bp_location2\[\t \]+argc = \\(argc == 12345\\);.*" \
		"run until file:linenum breakpoint"

# Test break at offset +1
set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]

gdb_test "break +1" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
    "breakpoint offset +1"

# Check to see if breakpoint is hit when stepped onto

gdb_test "step" \
    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10.*$bp_location10\[\t \]+return argc;.*breakpoint 10 here.*" \
    "step onto breakpoint"

#
# delete all breakpoints so we can start over, course this can be a test too
#
delete_breakpoints

#
# test temporary breakpoint at function
#

gdb_test "tbreak main" "Breakpoint.*at.* file .*$srcfile, line.*" "Temporary breakpoint function"

#
# test break at function in file
#

gdb_test "tbreak $srcfile:factorial" "Breakpoint.*at.* file .*$srcfile, line.*" \
	"Temporary breakpoint function in file"

#
# test break at line number
#
send_gdb "tbreak $bp_location1\n"
gdb_expect {
    -re "Breakpoint.*at.* file .*$srcfile, line $bp_location1.*$gdb_prompt $" { pass "Temporary breakpoint line number #1" }
	-re ".*$gdb_prompt $"   { pass "Temporary breakpoint line number #1" }
	timeout	    { fail "breakpoint line number #1 (timeout)" }
}

gdb_test "tbreak $bp_location6" "Breakpoint.*at.* file .*$srcfile, line $bp_location6.*" "Temporary breakpoint line number #2"

#
# test break at line number in file
#
send_gdb "tbreak $srcfile:$bp_location2\n"
gdb_expect {
    -re "Breakpoint.*at.* file .*$srcfile, line $bp_location2.*$gdb_prompt $" { pass "Temporary breakpoint line number in file #1" }
	-re ".*$gdb_prompt $"   { pass "Temporary breakpoint line number in file #1" }
	timeout	    { fail "Temporary breakpoint line number in file #1 (timeout)" }
}

set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
gdb_test  "tbreak $srcfile:$bp_location11" "Breakpoint.*at.* file .*$srcfile, line $bp_location11.*" "Temporary breakpoint line number in file #2"

#
# check to see what breakpoints are set (temporary this time)
#
gdb_test "info break" "Num Type.*Disp Enb Address.*What.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$main_line.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in factorial at .*$srcfile:$bp_location7.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location1.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location6.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location2.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location11.*" \
    "Temporary breakpoint info"


#***********

# Verify that catchpoints for fork, vfork and exec don't trigger
# inappropriately.  (There are no calls to those system functions
# in this test program.)
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "catch\n"
gdb_expect {
  -re "Catch requires an event name.*$gdb_prompt $"\
          {pass "catch requires an event name"}
  -re "$gdb_prompt $"\
          {fail "catch requires an event name"}
  timeout {fail "(timeout) catch requires an event name"}
}


set name "set catch fork, never expected to trigger"
send_gdb "catch fork\n"
gdb_expect {
  -re "Catchpoint \[0-9\]* .fork..*$gdb_prompt $"
          {pass $name}
  -re "Catch of fork not yet implemented.*$gdb_prompt $"
	  {pass $name}
  -re "$gdb_prompt $"
          {fail $name}
  timeout {fail "(timeout) $name"}
}


set name "set catch vfork, never expected to trigger"
send_gdb "catch vfork\n"

# If we are on HP-UX 10.20, we expect an error message to be
# printed if we type "catch vfork" at the gdb gdb_prompt.  This is
# because on HP-UX 10.20, we cannot catch vfork events.

if [istarget "hppa*-hp-hpux10.20"] then {
    gdb_expect {
	-re "Catch of vfork events not supported on HP-UX 10.20..*$gdb_prompt $"
		{pass $name}
	-re "$gdb_prompt $"
		{fail $name}
	timeout {fail "(timeout) $name"}
    }
} else {
    gdb_expect {
	-re "Catchpoint \[0-9\]* .vfork..*$gdb_prompt $"
		{pass $name}
	-re "Catch of vfork not yet implemented.*$gdb_prompt $"
		{pass $name}
	-re "$gdb_prompt $"
		{fail $name}
	timeout {fail "(timeout) $name"}
    }
}

set name "set catch exec, never expected to trigger"
send_gdb "catch exec\n"
gdb_expect {
  -re "Catchpoint \[0-9\]* .exec..*$gdb_prompt $"
          {pass $name}
  -re "Catch of exec not yet implemented.*$gdb_prompt $"
	  {pass $name}
  -re "$gdb_prompt $" {fail $name}
  timeout {fail "(timeout) $name"}
}

# Verify that GDB responds gracefully when asked to set a breakpoint
# on a nonexistent source line.
#
send_gdb "break 999\n"
gdb_expect {
  -re "No line 999 in file .*$gdb_prompt $"\
          {pass "break on non-existent source line"}
  -re "$gdb_prompt $"\
          {fail "break on non-existent source line"}
  timeout {fail "(timeout) break on non-existent source line"}
}

# Run to the desired default location. If not positioned here, the
# tests below don't work.
#
gdb_test "until $bp_location1" "main .* at .*:$bp_location1.*" "until bp_location1"


# Verify that GDB allows one to just say "break", which is treated
# as the "default" breakpoint.  Note that GDB gets cute when printing
# the informational message about other breakpoints at the same
# location.  We'll hit that bird with this stone too.
#
send_gdb "break\n"
gdb_expect {
  -re "Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 1st time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 1st time"}
  timeout {fail "(timeout) break on default location, 1st time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoint \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 2nd time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 2nd time"}
  timeout {fail "(timeout) break on default location, 2nd time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoints \[0-9\]* and \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 3rd time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 3rd time"}
  timeout {fail "(timeout) break on default location, 3rd time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoints \[0-9\]*, \[0-9\]* and \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 4th time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 4th time"}
  timeout {fail "(timeout) break on default location, 4th time"}
}

# Verify that a "silent" breakpoint can be set, and that GDB is indeed
# "silent" about its triggering.
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "break $bp_location1\n"
gdb_expect {
  -re "Breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $"\
          {pass "set to-be-silent break bp_location1"}
  -re "$gdb_prompt $"\
          {fail "set to-be-silent break bp_location1"}
  timeout {fail "(timeout) set to-be-silent break bp_location1"}
}

send_gdb "commands $expect_out(1,string)\n"
send_gdb "silent\n"
send_gdb "end\n"
gdb_expect {
  -re ".*$gdb_prompt $"\
          {pass "set silent break bp_location1"}
  timeout {fail "(timeout) set silent break bp_location1"}
}

send_gdb "info break $expect_out(1,string)\n"
gdb_expect {
  -re "\[0-9\]*\[ \t\]*breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*$gdb_prompt $"\
          {pass "info silent break bp_location1"}
  -re "$gdb_prompt $"\
          {fail "info silent break bp_location1"}
  timeout {fail "(timeout) info silent break bp_location1"}
}
send_gdb "continue\n"
gdb_expect {
  -re "Continuing.\r\n$gdb_prompt $"\
          {pass "hit silent break bp_location1"}
  -re "$gdb_prompt $"\
          {fail "hit silent break bp_location1"}
  timeout {fail "(timeout) hit silent break bp_location1"}
}
send_gdb "bt\n"
gdb_expect {
  -re "#0  main .* at .*:$bp_location1.*$gdb_prompt $"\
          {pass "stopped for silent break bp_location1"}
  -re "$gdb_prompt $"\
          {fail "stopped for silent break bp_location1"}
  timeout {fail "(timeout) stopped for silent break bp_location1"}
}

# Verify that GDB can at least parse a breakpoint with the
# "thread" keyword.  (We won't attempt to test here that a
# thread-specific breakpoint really triggers appropriately.
# The gdb.threads subdirectory contains tests for that.)
#
set bp_location12 [gdb_get_line_number "set breakpoint 12 here"]
send_gdb "break $bp_location12 thread 999\n"
gdb_expect {
  -re "Unknown thread 999.*$gdb_prompt $"\
          {pass "thread-specific breakpoint on non-existent thread disallowed"}
  -re "$gdb_prompt $"\
          {fail "thread-specific breakpoint on non-existent thread disallowed"}
  timeout {fail "(timeout) thread-specific breakpoint on non-existent thread disallowed"}
}
send_gdb "break $bp_location12 thread foo\n"
gdb_expect {
  -re "Junk after thread keyword..*$gdb_prompt $"\
          {pass "thread-specific breakpoint on bogus thread ID disallowed"}
  -re "$gdb_prompt $"\
          {fail "thread-specific breakpoint on bogus thread ID disallowed"}
  timeout {fail "(timeout) thread-specific breakpoint on bogus thread ID disallowed"}
}

# Verify that GDB responds gracefully to a breakpoint command with
# trailing garbage.
#
send_gdb "break $bp_location12 foo\n"
gdb_expect {
  -re "Junk at end of arguments..*$gdb_prompt $"\
          {pass "breakpoint with trailing garbage disallowed"}
  -re "$gdb_prompt $"\
          {fail "breakpoint with trailing garbage disallowed"}
  timeout {fail "(timeout) breakpoint with trailing garbage disallowed"}
}

# Verify that GDB responds gracefully to a "clear" command that has
# no matching breakpoint.  (First, get us off the current source line,
# which we know has a breakpoint.)
#
send_gdb "next\n"
gdb_expect {
  -re ".*$gdb_prompt $"\
          {pass "step over breakpoint"}
  timeout {fail "(timeout) step over breakpoint"}
}
send_gdb "clear 81\n"
gdb_expect {
  -re "No breakpoint at 81..*$gdb_prompt $"\
          {pass "clear line has no breakpoint disallowed"}
  -re "$gdb_prompt $"\
          {fail "clear line has no breakpoint disallowed"}
  timeout {fail "(timeout) clear line has no breakpoint disallowed"}
}
send_gdb "clear\n"
gdb_expect {
  -re "No breakpoint at this line..*$gdb_prompt $"\
          {pass "clear current line has no breakpoint disallowed"}
  -re "$gdb_prompt $"\
          {fail "clear current line has no breakpoint disallowed"}
  timeout {fail "(timeout) clear current line has no breakpoint disallowed"}
}

# Verify that we can set and clear multiple breakpoints.
#
# We don't test that it deletes the correct breakpoints.  We do at
# least test that it deletes more than one breakpoint.
#
gdb_test "break marker3" "Breakpoint.*at.*" "break marker3 #1"
gdb_test "break marker3" "Breakpoint.*at.*" "break marker3 #2"
gdb_test "clear marker3" {Deleted breakpoints [0-9]+ [0-9]+.*}

# Verify that a breakpoint can be set via a convenience variable.
#
send_gdb "set \$foo=$bp_location11\n"
gdb_expect {
  -re "$gdb_prompt $"\
          {pass "set convenience variable \$foo to bp_location11"}
  timeout {fail "(timeout) set convenience variable \$foo to bp_location11"}
}
send_gdb "break \$foo\n"
gdb_expect {
  -re "Breakpoint (\[0-9\]*) at .*, line $bp_location11.*$gdb_prompt $"\
          {pass "set breakpoint via convenience variable"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint via convenience variable"}
  timeout {fail "(timeout) set breakpoint via convenience variable"}
}

# Verify that GDB responds gracefully to an attempt to set a
# breakpoint via a convenience variable whose type is not integer.
#
send_gdb "set \$foo=81.5\n"
gdb_expect {
  -re "$gdb_prompt $"\
          {pass "set convenience variable \$foo to 81.5"}
  timeout {fail "(timeout) set convenience variable \$foo to 81.5"}
}
send_gdb "break \$foo\n"
gdb_expect {
  -re "Convenience variables used in line specs must have integer values..*$gdb_prompt $"\
          {pass "set breakpoint via non-integer convenience variable disallowed"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint via non-integer convenience variable disallowed"}
  timeout {fail "(timeout) set breakpoint via non-integer convenience variable disallowed"}
}

# Verify that we can set and trigger a breakpoint in a user-called function.
#
send_gdb "break marker2\n"
gdb_expect {
    -re "Breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*$gdb_prompt $"\
          {pass "set breakpoint on to-be-called function"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint on to-be-called function"}
  timeout {fail "(timeout) set breakpoint on to-be-called function"}
}
send_gdb "print marker2(99)\n"
gdb_expect {
  -re "The program being debugged stopped while in a function called from GDB.\r\nWhen the function .marker2. is done executing, GDB will silently\r\nstop .instead of continuing to evaluate the expression containing\r\nthe function call...*$gdb_prompt $"\
          {pass "hit breakpoint on called function"}
  -re "$gdb_prompt $"\
          {fail "hit breakpoint on called function"}
  timeout {fail "(timeout) hit breakpoint on called function"}
}

# As long as we're stopped (breakpointed) in a called function,
# verify that we can successfully backtrace & such from here.
#
# In this and the following test, the _sr4export check apparently is needed
# for hppa*-*-hpux.
#
send_gdb "bt\n"
gdb_expect {
    -re "#0\[ \t\]*($hex in )?marker2.*:($bp_location8|$bp_location9)\r\n#1.*_sr4export.*$gdb_prompt $"\
            {pass "backtrace while in called function"}
    -re "#0\[ \t\]*($hex in )?marker2.*:($bp_location8|$bp_location9)\r\n#1.*function called from gdb.*$gdb_prompt $"\
	    {pass "backtrace while in called function"}
    -re "$gdb_prompt $"\
            {fail "backtrace while in called function"}
    timeout {fail "(timeout) backtrace while in called function"}
}

# Return from the called function.  For remote targets, it's important to do
# this before runto_main, which otherwise may silently stop on the dummy
# breakpoint inserted by GDB at the program's entry point.
#
send_gdb "finish\n"
gdb_expect {
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.* in _sr4export.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.*function called from gdb.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.*Value returned.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "$gdb_prompt $"\
            {fail "finish from called function"}
    timeout {fail "(timeout) finish from called function"}
}

# Verify that GDB responds gracefully to a "finish" command with
# arguments.
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "finish 123\n"
gdb_expect {
  -re "The \"finish\" command does not take any arguments.\r\n$gdb_prompt $"\
          {pass "finish with arguments disallowed"}
  -re "$gdb_prompt $"\
          {fail "finish with arguments disallowed"}
  timeout {fail "(timeout) finish with arguments disallowed"}
}

# Verify that GDB responds gracefully to a request to "finish" from
# the outermost frame.  On a stub that never exits, this will just
# run to the stubs routine, so we don't get this error...  Thus the 
# second condition.
#

send_gdb "finish\n"
gdb_expect {
  -re "\"finish\" not meaningful in the outermost frame.\r\n$gdb_prompt $"\
          {pass "finish from outermost frame disallowed"}
  -re "Run till exit from.*\r\n$gdb_prompt $" {
     pass "finish from outermost frame disallowed"
  }
  -re "$gdb_prompt $"\
          {fail "finish from outermost frame disallowed"}
  timeout {fail "(timeout) finish from outermost frame disallowed"}
}

# Verify that we can explicitly ask GDB to stop on all shared library
# events, and that it does so.
#
if [istarget "hppa*-*-hpux*"] then {
  if ![runto_main] then { fail "break tests suppressed" }

  send_gdb "set stop-on-solib-events 1\n"
  gdb_expect {
    -re "$gdb_prompt $"\
            {pass "set stop-on-solib-events"}
    timeout {fail "(timeout) set stop-on-solib-events"}
  }

  send_gdb "run\n"
  gdb_expect {
    -re ".*Start it from the beginning.*y or n. $"\
            {send_gdb "y\n"
             gdb_expect {
               -re ".*Stopped due to shared library event.*$gdb_prompt $"\
                       {pass "triggered stop-on-solib-events"}
               -re "$gdb_prompt $"\
                       {fail "triggered stop-on-solib-events"}
               timeout {fail "(timeout) triggered stop-on-solib-events"}
             }
            }
    -re "$gdb_prompt $"\
            {fail "rerun for stop-on-solib-events"}
    timeout {fail "(timeout) rerun for stop-on-solib-events"}
  }

  send_gdb "set stop-on-solib-events 0\n"
  gdb_expect {
    -re "$gdb_prompt $"\
            {pass "reset stop-on-solib-events"}
    timeout {fail "(timeout) reset stop-on-solib-events"}
  }
}

# Hardware breakpoints are unsupported on HP-UX.  Verify that GDB
# gracefully responds to requests to create them.
#
if [istarget "hppa*-*-hpux*"] then {
  if ![runto_main] then { fail "break tests suppressed" }

  send_gdb "hbreak\n"
  gdb_expect {
    -re "No hardware breakpoint support in the target.*$gdb_prompt $"\
            {pass "hw breaks disallowed"}
    -re "$gdb_prompt $"\
            {fail "hw breaks disallowed"}
    timeout {fail "(timeout) hw breaks disallowed"}
  }

  send_gdb "thbreak\n"
  gdb_expect {
    -re "No hardware breakpoint support in the target.*$gdb_prompt $"\
            {pass "temporary hw breaks disallowed"}
    -re "$gdb_prompt $"\
            {fail "temporary hw breaks disallowed"}
    timeout {fail "(timeout) temporary hw breaks disallowed"}
  }
}

#********


#
# Test "next" over recursive function call.
#

proc test_next_with_recursion {} { 
    global gdb_prompt
    global decimal
    global binfile

    if [target_info exists use_gdb_stub] {
	# Reload the program.
	delete_breakpoints
	gdb_load ${binfile};
    } else {
	# FIXME: should be using runto
	gdb_test "kill" "" "kill program" "Kill the program being debugged.*y or n. $" "y"

	delete_breakpoints
    }

    gdb_test "break factorial" "Breakpoint $decimal at .*" "break at factorial"

    # Run until we call factorial with 6

    if [istarget "*-*-vxworks*"] then {
	send_gdb "run vxmain \"6\"\n"
    } else {
	gdb_run_cmd
    }
    gdb_expect {
	-re "Break.* factorial .value=6. .*$gdb_prompt $" {}
	-re ".*$gdb_prompt $" {
	    fail "run to factorial(6)";
	    gdb_suppress_tests;
	}
	timeout { fail "run to factorial(6) (timeout)" ; gdb_suppress_tests }
    }

    # Continue until we call factorial recursively with 5.

    if [gdb_test "continue" \
	"Continuing.*Break.* factorial .value=5. .*" \
	"continue to factorial(5)"] then { gdb_suppress_tests }

    # Do a backtrace just to confirm how many levels deep we are.

    if [gdb_test "backtrace" \
	"#0\[ \t\]+ factorial .value=5..*" \
	"backtrace from factorial(5)"] then { gdb_suppress_tests }

    # Now a "next" should position us at the recursive call, which
    # we will be performing with 4.

    if [gdb_test "next" \
	".* factorial .value - 1.;.*" \
	"next to recursive call"] then { gdb_suppress_tests }

    # Disable the breakpoint at the entry to factorial by deleting them all.
    # The "next" should run until we return to the next line from this
    # recursive call to factorial with 4.
    # Buggy versions of gdb will stop instead at the innermost frame on
    # the line where we are trying to "next" to.

    delete_breakpoints

    if [istarget "mips*tx39-*"] {
	set timeout 60
    }
    # We used to set timeout here for all other targets as well.  This
    # is almost certainly wrong.  The proper timeout depends on the
    # target system in use, and how we communicate with it, so there
    # is no single value appropriate for all targets.  The timeout
    # should be established by the Dejagnu config file(s) for the
    # board, and respected by the test suite.
    #
    # For example, if I'm running GDB over an SSH tunnel talking to a
    # portmaster in California talking to an ancient 68k board running
    # a crummy ROM monitor (a situation I can only wish were
    # hypothetical), then I need a large timeout.  But that's not the
    # kind of knowledge that belongs in this file.

    gdb_test next "\[0-9\]*\[\t \]+return \\(value\\);.*" \
	    "next over recursive call"

    # OK, we should be back in the same stack frame we started from.
    # Do a backtrace just to confirm.

    set result [gdb_test "backtrace" \
	    "#0\[ \t\]+ factorial .value=120.*\r\n#1\[ \t\]+ \[0-9a-fx\]+ in factorial .value=6..*" \
	    "backtrace from factorial(5.1)"]
    if { $result != 0 } { gdb_suppress_tests }

    if [target_info exists gdb,noresults] { gdb_suppress_tests }
  gdb_continue_to_end "recursive next test"
   gdb_stop_suppressing_tests;
}

test_next_with_recursion


#********

# now move the .debug file to a different location so that we can test
# the "set debug-file-directory" command.
  
remote_exec build "mv ${objdir}/${subdir}/.debug/${testfile}.debug ${objdir}/${subdir}"
gdb_exit
gdb_start
gdb_reinitialize_dir $srcdir/$subdir
gdb_test "set debug-file-directory ${objdir}/${subdir}" ".*" "set separate debug location"
gdb_load ${binfile}

if [target_info exists gdb_stub] {
    gdb_step_for_stub;
}

#
# test break at function
#
gdb_test "break main" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function, optimized file"

#
# test break at function
#
gdb_test "break marker4" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint small function, optimized file"

#
# run until the breakpoint at main is hit. For non-stubs-using targets.
#
if ![target_info exists use_gdb_stub] {
  if [istarget "*-*-vxworks*"] then {
    send_gdb "run vxmain \"2\"\n"
    set timeout 120
    verbose "Timeout is now $timeout seconds" 2
  } else {
	send_gdb "run\n"
  }
  gdb_expect {
    -re "The program .* has been started already.*y or n. $" {
	send_gdb "y\n"
	exp_continue
    }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $"\
	                    { pass "run until function breakpoint, optimized file" }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $"\
	                    { pass "run until function breakpoint, optimized file (code motion)" }
    -re ".*$gdb_prompt $"       { fail "run until function breakpoint, optimized file" }
    timeout	            { fail "run until function breakpoint, optimized file (timeout)" }
  }
} else {
    if ![target_info exists gdb_stub] {
	gdb_test continue ".*Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.*\{.*" "stub continue, optimized file"
    }
}

#
# run until the breakpoint at a small function
#

#
# Add a second pass pattern.  The behavior differs here between stabs
# and dwarf for one-line functions.  Stabs preserves two line symbols
# (one before the prologue and one after) with the same line number, 
# but dwarf regards these as duplicates and discards one of them.
# Therefore the address after the prologue (where the breakpoint is)
# has no exactly matching line symbol, and GDB reports the breakpoint
# as if it were in the middle of a line rather than at the beginning.

set bp_location13 [gdb_get_line_number "set breakpoint 13 here"]
set bp_location14 [gdb_get_line_number "set breakpoint 14 here"]
send_gdb "continue\n"
gdb_expect {
    -re "Breakpoint $decimal, marker4 \\(d=177601976\\) at .*$srcfile:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
	pass "run until breakpoint set at small function, optimized file"
    }
    -re "Breakpoint $decimal, $hex in marker4 \\(d=177601976\\) at .*$srcfile:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
	pass "run until breakpoint set at small function, optimized file"
    }
    -re "Breakpoint $decimal, marker4 \\(d=177601976\\) at .*$srcfile:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
        # marker4() is defined at line 46 when compiled with -DPROTOTYPES
	pass "run until breakpoint set at small function, optimized file (line bp_location14)"
    }
    -re ".*$gdb_prompt " {
	fail "run until breakpoint set at small function, optimized file"
    }
    timeout {
	fail "run until breakpoint set at small function, optimized file (timeout)"
    }
}


# Reset the default arguments for VxWorks
if [istarget "*-*-vxworks*"] {
    set timeout 10
    verbose "Timeout is now $timeout seconds" 2
    send_gdb "set args main\n"
    gdb_expect -re ".*$gdb_prompt $" {}
}

=====================================


/* Copyright 1994, 1995, 1999, 2002, 2003, 2004  Free Software Foundation, Inc.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.
 
   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.
 
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

   Please email any bugs, comments, and/or additions to this file to:
   bug-gdb@prep.ai.mit.edu  */

#ifdef vxworks

#  include <stdio.h>

/* VxWorks does not supply atoi.  */
static int
atoi (z)
     char *z;
{
  int i = 0;

  while (*z >= '0' && *z <= '9')
    i = i * 10 + (*z++ - '0');
  return i;
}

/* I don't know of any way to pass an array to VxWorks.  This function
   can be called directly from gdb.  */

vxmain (arg)
char *arg;
{
  char *argv[2];

  argv[0] = "";
  argv[1] = arg;
  main (2, argv, (char **) 0);
}

#else /* ! vxworks */
#  include <stdio.h>
#  include <stdlib.h>
#endif /* ! vxworks */

/*
 * The following functions do nothing useful.  They are included simply
 * as places to try setting breakpoints at.  They are explicitly
 * "one-line functions" to verify that this case works (some versions
 * of gcc have or have had problems with this).
 */

#ifdef PROTOTYPES
int marker1 (void) { return (0); }
int marker2 (int a) { return (1); } /* set breakpoint 8 here */
void marker3 (char *a, char *b) {}
void marker4 (long d) {} /* set breakpoint 14 here */
#else
int marker1 () { return (0); }
int marker2 (a) int a; { return (1); } /* set breakpoint 9 here */
void marker3 (a, b) char *a, *b; {}
void marker4 (d) long d; {}  /* set breakpoint 13 here */
#endif

/*
 *	This simple classical example of recursion is useful for
 *	testing stack backtraces and such.
 */

#ifdef PROTOTYPES
int factorial(int);

int
main (int argc, char **argv, char **envp)
#else
int
main (argc, argv, envp)
int argc;
char *argv[], **envp;
#endif
{
#ifdef usestubs
    set_debug_traps();  /* set breakpoint 5 here */
    breakpoint();
#endif
    if (argc == 12345) {  /* an unlikely value < 2^16, in case uninited */ /* set breakpoint 6 here */
	fprintf (stderr, "usage:  factorial <number>\n");
	return 1;
    }
    printf ("%d\n", factorial (atoi ("6")));  /* set breakpoint 1 here */
    /* set breakpoint 12 here */
    marker1 ();  /* set breakpoint 11 here */
    marker2 (43);
    marker3 ("stack", "trace");
    marker4 (177601976L);
    argc = (argc == 12345); /* This is silly, but we can step off of it */ /* set breakpoint 2 here */
    return argc;  /* set breakpoint 10 here */
}

#ifdef PROTOTYPES
int factorial (int value)
#else
int factorial (value)
int value;
#endif
{
  if (value > 1) {  /* set breakpoint 7 here */
	value *= factorial (value - 1);
    }
    return (value);
}

#ifdef PROTOTYPES
int multi_line_if_conditional (int a, int b, int c)
#else
int multi_line_if_conditional (a, b, c)
  int a, b, c;
#endif
{
  if (a    /* set breakpoint 3 here */
      && b
      && c)
    return 0;
  else
    return 1;
}

#ifdef PROTOTYPES
int multi_line_while_conditional (int a, int b, int c)
#else
int multi_line_while_conditional (a, b, c)
  int a, b, c;
#endif
{
  while (a /* set breakpoint 4 here */
      && b
      && c)
    {
      a--, b--, c--;
    }
  return 0;
}



Index: gdb.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/lib/gdb.exp,v
retrieving revision 1.41
diff -u -p -r1.41 gdb.exp
--- gdb.exp     23 Nov 2003 01:09:19 -0000      1.41
+++ gdb.exp     12 Jan 2004 17:16:13 -0000
@@ -1918,3 +1918,87 @@ proc gdb_skip_bogus_test { msg } {
     return 0;
 }
  
+
+# Note: the procedure gdb_gnu_strip_debug will produce an executable called
+# ${binfile}.dbglnk, which is just like the executable ($binfile) but without
+# the debuginfo. Instead $binfile has a .gnu_debuglink section which contains
+# the name of a idebuginfo only file. This file will be stored in the
+# gdb.base/.debug subdirectory.
+
+# Functions for separate debug info testing
+
+# starting with an executable:
+# foo --> original executable
+
+# at the end of the process we have:
+# foo.stripped --> foo w/o debug info
+# .debug/foo.debug --> foo's debug info
+# foo --> like foo, but with a new .gnu_debuglink section pointing to foo.debug.
+
+# Return the name of the file in which we should stor EXEC's separated
+# debug info. EXEC contains the full path.
+proc separate_debug_filename { exec } {
+
+    # In a .debug subdirectory off the same directory where the testcase
+    # executable is going to be. Something like:
+    # <your-path>/gdb/testsuite/gdb.base/.debug/blah.debug.
+    # This is the default location where gdb expects to findi
+    # the debug info file.
+
+    set exec_dir [file dirname $exec]
+    set exec_file [file tail $exec]
+    set debug_dir [file join $exec_dir ".debug"]
+    set debug_file [file join $debug_dir "${exec_file}.debug"]
+
+    return $debug_file
+}
+
+
+proc gdb_gnu_strip_debug { dest } {
+
+    set debug_file [separate_debug_filename $dest]
+    set strip_to_file_program strip
+    set objcopy_program objcopy
+
+    # Make sure the directory that will hold the separated debug
+    # info actually exists.
+    set debug_dir [file dirname $debug_file]
+    if {! [file isdirectory $debug_dir]} {
+       file mkdir $debug_dir
+    }
+
+    set debug_link [file tail $debug_file]
+    set stripped_file "${dest}.stripped"
+
+    # Get rid of the debug info, and store result in stripped_file
+    # something like gdb/testsuite/gdb.base/blah.stripped.
+    set result [catch "exec $strip_to_file_program --strip-debug ${dest} -o ${stripped_file}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+    # Get rid of everything but the debug info, and store result in debug_file
+    # This will be in the .debug subdirectory, see above.
+    set result [catch "exec $strip_to_file_program --only-keep-debug ${dest} -o ${debug_file}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+    # Link the two previous output files together, adding the .gnu_debuglink
+    # section to the stripped_file, containing a pointer to the debug_file,
+    # save the new file in dest.
+    # This will be the regular executable filename, in the usual location.
+    set result [catch "exec $objcopy_program --add-gnu-debuglink=${debug_file}
${stripped_file} ${dest}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+   return 0
+}
+

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > Elena Zannoni writes:
 > 
 >     # If we have too many warnings or errors,
 >     # the output of the test can't be considered correct.
 >     if { $warning_threshold > 0 && $warncnt >= $warning_threshold
 > 	 || $perror_threshold > 0 && $errcnt >= $perror_threshold } {
 >         verbose "Error/Warning threshold exceeded: \
 >                  $errcnt $warncnt (max. $perror_threshold $warning_threshold)"
 >         set type UNRESOLVED
 >     }
 > 
 > Ick!
 > 
 > > what to do? Maybe just get rid of the unsupported call, and just get
 > > out of the test.
 > 
 > Well, "nothing different" is okay.  I'm not worried about the actual
 > output, as long as we understand why it behaves that way, and stick that
 > understanding into a comment somewhere.
 > 
 > As Daniel alluded to -- what SHOULD happen when an end user tries
 > to do this with stabs+?  Should binutils bonk out with a nice error
 > message?
 > 

I think binutils is a bit confused with separate debug info and stabs.
That would need to be fixed as well. I am really tempted to just bail
out for stabs.

elena


 > Michael C

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

Elena Zannoni writes:

    # If we have too many warnings or errors,
    # the output of the test can't be considered correct.
    if { $warning_threshold > 0 && $warncnt >= $warning_threshold
	 || $perror_threshold > 0 && $errcnt >= $perror_threshold } {
        verbose "Error/Warning threshold exceeded: \
                 $errcnt $warncnt (max. $perror_threshold $warning_threshold)"
        set type UNRESOLVED
    }

Ick!

> what to do? Maybe just get rid of the unsupported call, and just get
> out of the test.

Well, "nothing different" is okay.  I'm not worried about the actual
output, as long as we understand why it behaves that way, and stick that
understanding into a comment somewhere.

As Daniel alluded to -- what SHOULD happen when an end user tries
to do this with stabs+?  Should binutils bonk out with a nice error
message?

Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > eza> What happens is that gdb_load errors out and calls perror, then
 > eza> returns 1.  All I have done is wrap gdb_load in a check for a 1 return
 > eza> value.
 > 
 > Here's the part that's confusing me.  Your gdb_load wrapper
 > says "if [gdb_load ...] { unsupported ...}", but your gdb.sum
 > file says "# of unresolved test cases: 1".  The call to "unsupported"
 > doesn't fit the result of "unresolved"!
 > 

Ah, I didn't notice that. 
Groan, look at dejagnu/lib/framework.exp:

    # If we have too many warnings or errors,
    # the output of the test can't be considered correct.
    if { $warning_threshold > 0 && $warncnt >= $warning_threshold
	 || $perror_threshold > 0 && $errcnt >= $perror_threshold } {
        verbose "Error/Warning threshold exceeded: \
                 $errcnt $warncnt (max. $perror_threshold $warning_threshold)"
        set type UNRESOLVED
    }


Since we got a perror from the previous load, the error threshold is
exceeded and the code above triggers, changing the UNSUPPORTED to
UNRESOLVED.

If I run with -v, I get this bit of output:
Error/Warning threshold exceeded: 1 0 (max. 1 3)

what to do? Maybe just get rid of the unsupported call, and just get
out of the test. But then the failure wouldn't be recorded, there
would just be the error.  Alternatively maybe we could raise temporarily the
error threshold to 2.

global perror_threshold
set perror_threshold [expr $perror_threshold + 1]
if [gdb_load ${binfile}] {
   unsupported "Gdb cannot process a separate debug info file"
   return -1
}
set perror_threshold [expr $perror_threshold - 1]


Not pretty, I know.

elena

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

eza> What happens is that gdb_load errors out and calls perror, then
eza> returns 1.  All I have done is wrap gdb_load in a check for a 1 return
eza> value.

Here's the part that's confusing me.  Your gdb_load wrapper
says "if [gdb_load ...] { unsupported ...}", but your gdb.sum
file says "# of unresolved test cases: 1".  The call to "unsupported"
doesn't fit the result of "unresolved"!

eza> The alternative is to alter the gdb_load (or whatever gets
eza> caled from it) to recognize that error and do something specific, but
eza> I thought that was too ad hoc a behavior for a generic procedure.

Ah, okay.

Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Daniel Jacobowitz writes:
 > On Thu, Nov 13, 2003 at 06:15:17PM -0500, Elena Zannoni wrote:
 > > 
 > > 3rd reply to the same mail...:-)
 > > 
 > > ok, How about this behavior?
 > > 
 > > Running /to/scratch/ezannoni/pristine/src/gdb/testsuite/gdb.base/sepdebug.exp ...
 > > ERROR: couldn't load /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/gdb.base/sepdebug.dbglnk into /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb.
 > 
 > Does it make sense to use this same mechanism with stabs?
 > 

Right now I don't think there is any stabs toolchain/distro that uses it.
And there are apparently problems with the strip/objcopy mechanism with stabs.
For elfutils, I don't know if it even supports stabs at all.

 > i.e. should we be testing it, which implies trying to fix it, or
 > document it as a bad idea and let it go.  I can see either.
 > 

me too.

 > -- 
 > Daniel Jacobowitz
 > MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > eza> woah. I didn't test this. And I guess nobody did until now.
 > eza> RH ships only dwarf2 toolchains.
 > 
 > As an aside, my red hat 8.0 gcc supports both dwarf-2 and stabs+,
 > with dwarf-2 as default.  FSF gcc for i686-pc-linux-gnu has also
 > been this way since 3.1 or 3.2.
 > 

Yes, after rhl7.3 dwarf2 became the default.

 > If I recall correctly, FSF gcc for sparc-sun-solaris is still
 > stabs+ by default, but I wouldn't swear to it.
 > 

Hmm, I should test that too.

 > I'm still waiting for dwarf-1 to die before I take up a
 > stabs+ death watch!
 > 

It'll be a long wait, I am afraid.

elena

 > Michael C

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > Woah, I am confused.
 > 
 > Your source code says "unsupported", but your gdb.sum file says
 > "unresolved", and your gdb.log file says "ERROR".
 > 
 > Is your sandbox a little messed up, or is it me?!
 > 

What happens is that gdb_load errors out and calls perror, then
returns 1.  All I have done is wrap gdb_load in a check for a 1 return
value.  The alternative is to alter the gdb_load (or whatever gets
caled from it) to recognize that error and do something specific, but
I thought that was too ad hoc a behavior for a generic procedure.

elena


 > Michael C
 > 
 > ===
 > 
 >   ok, How about this behavior?
 > 
 >   Running /to/scratch/ezannoni/pristine/src/gdb/testsuite/gdb.base/sepdebug.exp ...
 >   ERROR: couldn't load /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/gdb.base/sepdebug.dbglnk into /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb.
 >    
 > 		  === gdb Summary ===
 >    
 >   # of unresolved testcases       1
 >   /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb version  2003-11-07-cvs -nx
 >    
 > 
 >   I added a wrapper around the gdb_load call:
 > 
 >   if [gdb_load ${binfile}] {
 >       unsupported "Gdb cannot process a separate debug info file"
 >       return -1
 >   }
 > 

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

Woah, I am confused.

Your source code says "unsupported", but your gdb.sum file says
"unresolved", and your gdb.log file says "ERROR".

Is your sandbox a little messed up, or is it me?!

Michael C

===

  ok, How about this behavior?

  Running /to/scratch/ezannoni/pristine/src/gdb/testsuite/gdb.base/sepdebug.exp ...
  ERROR: couldn't load /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/gdb.base/sepdebug.dbglnk into /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb.
   
		  === gdb Summary ===
   
  # of unresolved testcases       1
  /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb version  2003-11-07-cvs -nx
   

  I added a wrapper around the gdb_load call:

  if [gdb_load ${binfile}] {
      unsupported "Gdb cannot process a separate debug info file"
      return -1
  }

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

eza> woah. I didn't test this. And I guess nobody did until now.
eza> RH ships only dwarf2 toolchains.

As an aside, my red hat 8.0 gcc supports both dwarf-2 and stabs+,
with dwarf-2 as default.  FSF gcc for i686-pc-linux-gnu has also
been this way since 3.1 or 3.2.

If I recall correctly, FSF gcc for sparc-sun-solaris is still
stabs+ by default, but I wouldn't swear to it.

I'm still waiting for dwarf-1 to die before I take up a
stabs+ death watch!

Michael C

Re: [RFA] new test for separate debug info

[Daniel Jacobowitz]

On Thu, Nov 13, 2003 at 06:15:17PM -0500, Elena Zannoni wrote:
> 
> 3rd reply to the same mail...:-)
> 
> ok, How about this behavior?
> 
> Running /to/scratch/ezannoni/pristine/src/gdb/testsuite/gdb.base/sepdebug.exp ...
> ERROR: couldn't load /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/gdb.base/sepdebug.dbglnk into /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb.

Does it make sense to use this same mechanism with stabs?

i.e. should we be testing it, which implies trying to fix it, or
document it as a bad idea and let it go.  I can see either.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [RFA] new test for separate debug info

[Elena Zannoni]

3rd reply to the same mail...:-)

ok, How about this behavior?

Running /to/scratch/ezannoni/pristine/src/gdb/testsuite/gdb.base/sepdebug.exp ...
ERROR: couldn't load /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/gdb.base/sepdebug.dbglnk into /to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb.
 
                === gdb Summary ===
 
# of unresolved testcases       1
/to/scratch/ezannoni/pristine/tomago/gdb/testsuite/../../gdb/gdb version  2003-11-07-cvs -nx
 

I added a wrapper around the gdb_load call:

if [gdb_load ${binfile}] {
    unsupported "Gdb cannot process a separate debug info file"
    return -1
}

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:

 > It looks like either binutils or gdb needs some work.

% objdump -g sepdebug.debug
 
sepdebug.debug:     file format elf32-i386
 
sepdebug.debug: .stab: stab entry 0 is corrupt, strx = 0x464c457f, type = 1
sepdebug.debug: .stab: stab entry 2 is corrupt, strx = 0x8048354, type = 52
sepdebug.debug: .stab: stab entry 4 is corrupt, strx = 0x30006, type = 128
debug_name_type: no current file
Last stabs entries before error:
n_type n_desc n_value  string
2      3      00000001
NOMAP  32     00280000
LSYM   0      00000000 char:t(0,2)=r(0,2);0;127;


seems like binutils has some problems too.  Gdb errors out if the test
section is missing, I thought of changing this into a warning, but
then ran into this.

 > 
 > If you could kludge the test to handle "Can't find any code sections
 > in symbol file" then I would be happy with the test script as a test
 > script.
 > 

I'll definitely do this, then, and add a comment.

elena

Re: [RFA] new test for separate debug info

[Elena Zannoni]

Michael Elizabeth Chastain writes:
 > My test bed says:
 > 
 >   binutils HEAD, stabs+
 >     3 ERROR, 7 PASS, 44 FAIL, 1 UNRESOLVED
 >     # not okay

woah. I didn't test this. And I guess nobody did until now. RH ships
only dwarf2 toolchains.


 > 
 > The results are the same for all versions of gcc tested.
 > 
 > Here is an excerpt from gdb.log:
 >  
 >   # gdb HEAD, gcc 3.3.2, binutils HEAD, -gstabs+
 >   (gdb) dir /berman/fsf/_today_/source/gdb/HEAD/src/gdb/testsuite/gdb.base
 >   Source directories searched: /berman/fsf/_today_/source/gdb/HEAD/src/gdb/testsuite/gdb.base:$cdir:$cwd
 >   (gdb) file /berman/fsf/_today_/berman/test/gdb.base/sepdebug
 >   Reading symbols from /berman/fsf/_today_/berman/test/gdb.base/sepdebug...Reading symbols from /berman/fsf/_today_/berman/test/gdb.base/.debug/sepdebug.debug...Can't find any code sections in symbol file
 >   (gdb) ERROR: couldn't load /berman/fsf/_today_/berman/test/gdb.base/sepdebug into gdb.
 >   delete breakpoints
 >   (gdb) UNRESOLVED: gdb.base/sepdebug.exp: Delete all breakpoints when none
 >   From gdb-patches-return-28583-mec.gnu=mindspring.com@sources.redhat.com  Wed Nov 12 17:26:42 2003
 > 
 > It looks like either binutils or gdb needs some work.
 > 

the warning is coming from gdb, dbxread.c. I'll look into this.

 > If you could kludge the test to handle "Can't find any code sections
 > in symbol file" then I would be happy with the test script as a test
 > script.
 > 

OK.

elena

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

My test bed says:

  binutils 2.13.90.0.2-2-rh, dwarf-2
  binutils 2.13.90.0.2-2-rh, stabs+
  binutils 2.14, dwarf-2
  binutils 2.14, stabs+
    1 UNSUPPORTED
    # okay

  binutils HEAD, dwarf-2
    77 PASS
    # okay

  binutils HEAD, stabs+
    3 ERROR, 7 PASS, 44 FAIL, 1 UNRESOLVED
    # not okay

The results are the same for all versions of gcc tested.

Here is an excerpt from gdb.log:
 
  # gdb HEAD, gcc 3.3.2, binutils HEAD, -gstabs+
  (gdb) dir /berman/fsf/_today_/source/gdb/HEAD/src/gdb/testsuite/gdb.base
  Source directories searched: /berman/fsf/_today_/source/gdb/HEAD/src/gdb/testsuite/gdb.base:$cdir:$cwd
  (gdb) file /berman/fsf/_today_/berman/test/gdb.base/sepdebug
  Reading symbols from /berman/fsf/_today_/berman/test/gdb.base/sepdebug...Reading symbols from /berman/fsf/_today_/berman/test/gdb.base/.debug/sepdebug.debug...Can't find any code sections in symbol file
  (gdb) ERROR: couldn't load /berman/fsf/_today_/berman/test/gdb.base/sepdebug into gdb.
  delete breakpoints
  (gdb) UNRESOLVED: gdb.base/sepdebug.exp: Delete all breakpoints when none
  From gdb-patches-return-28583-mec.gnu=mindspring.com@sources.redhat.com  Wed Nov 12 17:26:42 2003

It looks like either binutils or gdb needs some work.

If you could kludge the test to handle "Can't find any code sections
in symbol file" then I would be happy with the test script as a test
script.

I have put up a test result tarball:

  ftp://ftp.shout.net/pub/users/mec/gdb/2003-11-10-sepdebug.tar.gz

The interesting files are:

  native,HEAD,3.3.2,HEAD,stabs+,2/
    gdb.log
    test.tar.gz

I actually run the tests with:

  runtest ...  --target_board unix/gdb:debug_flags=-gdwarf-2\\\\\\ -g2 ...
  runtest ...  --target_board unix/gdb:debug_flags=-gstabs+\\\\\\ -g2 ...

It might be some weird option-processing problem, but I doubt it,
because I run dwarf-2 and stabs+ in the same way.  That is, I don't
just run dwarf-2 by leaving the debug_flags as default.
    
Michael C

Re: [RFA] new test for separate debug info

[Michael Elizabeth Chastain]

ac> FWIW, I think cloning "break.c" is a good idea.  It avoids the problem 
ac> of someone wanting to change/modify that test being landed with the task 
ac> of changing/modifying N other tests.

Me too.  I agree with Andrew's view.

Michael C

Re: [RFA] new test for separate debug info

[Andrew Cagney]

> 	* gdb.base/sepdebug.c: New source file for the above.

FWIW, I think cloning "break.c" is a good idea.  It avoids the problem 
of someone wanting to change/modify that test being landed with the task 
of changing/modifying N other tests.

Andrew

Re: [RFA] new test for separate debug info

[Daniel Jacobowitz]

On Wed, Nov 12, 2003 at 05:13:40PM -0500, Elena Zannoni wrote:
> 
> Ok, I think I addresed all of the comments. Here are the 2 files, in
> their entirety. However, since they are clones of break.c and
> break.exp, I just added, for your viewing pleasure, a diff between
> break.exp and sepdebug.exp.
> 
> 2003-11-12  Elena Zannoni  <ezannoni@redhat.com>
> 
> 	* gdb.base/sepdebug.exp: New test for separate debug info files.
> 	* gdb.base/sepdebug.c: New source file for the above.
> 	* lib/gdb.exp (separate_debug_filename): New procedure.
> 	(gdb_gnu_strip_debug): New procedure.

FWIW, this looks great to me.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

[RFA] new test for separate debug info

[Elena Zannoni]

Ok, I think I addresed all of the comments. Here are the 2 files, in
their entirety. However, since they are clones of break.c and
break.exp, I just added, for your viewing pleasure, a diff between
break.exp and sepdebug.exp.

2003-11-12  Elena Zannoni  <ezannoni@redhat.com>

	* gdb.base/sepdebug.exp: New test for separate debug info files.
	* gdb.base/sepdebug.c: New source file for the above.
	* lib/gdb.exp (separate_debug_filename): New procedure.
	(gdb_gnu_strip_debug): New procedure.

DIffs:

--- break.exp	2003-11-12 10:38:39.000000000 -0500
+++ sepdebug.exp	2003-11-12 15:50:04.000000000 -0500
@@ -19,7 +19,8 @@
 # Please email any bugs, comments, and/or additions to this file to:
 # bug-gdb@prep.ai.mit.edu
 
-# This file was written by Rob Savoye. (rob@cygnus.com)
+# Modified to test gdb's handling of separate debug info files.
+# Based on break.exp, written by Rob Savoye. (rob@cygnus.com)
 
 if $tracelevel then {
     strace $tracelevel
@@ -32,7 +33,7 @@
 set prms_id 0
 set bug_id 0
 
-set testfile "break"
+set testfile "sepdebug"
 set srcfile ${testfile}.c
 set binfile ${objdir}/${subdir}/${testfile}
 
@@ -40,9 +41,18 @@
     gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
 }
 
-if [get_compiler_info ${binfile}] {
+# Note: the procedure gdb_gnu_strip_debug will produce an executable called
+# ${binfile}, which is just like the executable ($binfile) but without
+# the debuginfo. Instead $binfile has a .gnudebuglink section which contains
+# the name of a idebuginfo only file. This file will be stored in the
+# gdb.base/.debug subdirectory.
+
+if [gdb_gnu_strip_debug $binfile] {
+    unsupported "no separate debug info handling"
     return -1
 }
 
 gdb_exit
 gdb_start
@@ -159,12 +169,6 @@
     set main_line $bp_location6
 }
 
-if {$hp_aCC_compiler} {
-    set proto "\\(int\\)"
-} else {
-    set proto ""
-}
-
 set bp_location7 [gdb_get_line_number "set breakpoint 7 here"]
 set bp_location8 [gdb_get_line_number "set breakpoint 8 here"]
 set bp_location9 [gdb_get_line_number "set breakpoint 9 here"]
@@ -173,7 +177,7 @@
     "Num Type\[ \]+Disp Enb Address\[ \]+What.*
 \[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$main_line.*
 \[0-9\]+\[\t \]+breakpoint     keep y.* in marker2 at .*$srcfile:($bp_location8|$bp_location9).*
-\[0-9\]+\[\t \]+breakpoint     keep y.* in factorial$proto at .*$srcfile:$bp_location7.*
+\[0-9\]+\[\t \]+breakpoint     keep y.* in factorial at .*$srcfile:$bp_location7.*
 \[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
 \[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
 \[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location2.*
@@ -302,7 +306,7 @@
 #
 gdb_test "info break" "Num Type.*Disp Enb Address.*What.*\[\r\n\]
 \[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$main_line.*\[\r\n\]
-\[0-9\]+\[\t \]+breakpoint     del.*y.*in factorial$proto at .*$srcfile:$bp_location7.*\[\r\n\]
+\[0-9\]+\[\t \]+breakpoint     del.*y.*in factorial at .*$srcfile:$bp_location7.*\[\r\n\]
 \[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location1.*\[\r\n\]
 \[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location6.*\[\r\n\]
 \[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location2.*\[\r\n\]
@@ -604,7 +608,7 @@
 }
 send_gdb "print marker2(99)\n"
 gdb_expect {
-  -re "The program being debugged stopped while in a function called from GDB.\r\nWhen the function .marker2$proto. is done executing, GDB will silently\r\nstop .instead of continuing to evaluate the expression containing\r\nthe function call...*$gdb_prompt $"\
+  -re "The program being debugged stopped while in a function called from GDB.\r\nWhen the function .marker2. is done executing, GDB will silently\r\nstop .instead of continuing to evaluate the expression containing\r\nthe function call...*$gdb_prompt $"\
           {pass "hit breakpoint on called function"}
   -re "$gdb_prompt $"\
           {fail "hit breakpoint on called function"}
@@ -845,23 +849,15 @@
 
 #********
 
-# build a new file with optimization enabled so that we can try breakpoints
-# on targets with optimized prologues
-
-set binfileo2 ${objdir}/${subdir}/${testfile}o2
-
-if  { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfileo2}" executable {debug additional_flags="-O2" }] != "" } {
-    gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
-}
-
-if [get_compiler_info ${binfileo2}] {
-    return -1
-}
-
+# now move the .debug file to a different location so that we can test
+# the "set debug-file-directory" command.
+  
+remote_exec build "mv ${objdir}/${subdir}/.debug/${testfile}.debug ${objdir}/${subdir}"
 gdb_exit
 gdb_start
 gdb_reinitialize_dir $srcdir/$subdir
-gdb_load ${binfileo2}
+gdb_test "set debug-file-directory ${objdir}/${subdir}" ".*" "set separate debug location"
+gdb_load ${binfile}
 
 if [target_info exists gdb_stub] {
     gdb_step_for_stub;





Index: gdb.exp
===================================================================
RCS file: /cvs/src/src/gdb/testsuite/lib/gdb.exp,v
retrieving revision 1.37
diff -u -p -r1.37 gdb.exp
--- gdb.exp     1 May 2003 01:09:51 -0000       1.37
+++ gdb.exp     12 Nov 2003 21:12:21 -0000
@@ -1840,3 +1840,84 @@ proc gdb_skip_bogus_test { msg } {
     return 0;
 }
  
+# Note: the procedure gdb_gnu_strip_debug will produce an executable called
+# ${binfile}, which is just like the original executable ($binfile) but without
+# the debuginfo. Instead $binfile has a .gnu_debuglink section which contains
+# the name of a debuginfo only file. This file will be stored in the
+# gdb.base/.debug subdirectory.
+
+# starting with an executable:
+# foo --> original executable
+
+# at the end of the process we have:
+# foo.stripped --> foo w/o debug info
+# .debug/foo.debug --> foo's debug info
+# foo --> like foo, but with a new .gnu_debuglink section pointing to foo.debug
.
+
+# Return the name of the file in which we should stor EXEC's separated
+# debug info. EXEC contains the full path.
+proc separate_debug_filename { exec } {
+
+    # In a .debug subdirectory off the same directory where the testcase
+    # executable is going to be. Something like:
+    # <your-path>/gdb/testsuite/gdb.base/.debug/blah.debug.
+    # This is the default location where gdb expects to findi
+    # the debug info file.
+
+    set exec_dir [file dirname $exec]
+    set exec_file [file tail $exec]
+    set debug_dir [file join $exec_dir ".debug"]
+    set debug_file [file join $debug_dir "${exec_file}.debug"]
+
+    return $debug_file
+}
+
+
+proc gdb_gnu_strip_debug { dest } {
+
+    set debug_file [separate_debug_filename $dest]
+    set strip_to_file_program strip
+    set objcopy_program objcopy
+
+    # Make sure the directory that will hold the separated debug
+    # info actually exists.
+    set debug_dir [file dirname $debug_file]
+    if {! [file isdirectory $debug_dir]} {
+       file mkdir $debug_dir
+    }
+
+    set debug_link [file tail $debug_file]
+    set stripped_file "${dest}.stripped"
+
+    # Get rid of the debug info, and store result in stripped_file
+    # something like gdb/testsuite/gdb.base/blah.stripped.
+    set result [catch "exec $strip_to_file_program --strip-debug ${dest} -o ${s
tripped_file}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+    # Get rid of everything but the debug info, and store result in debug_file
+    # This will be in the .debug subdirectory, see above.
+    set result [catch "exec $strip_to_file_program --only-keep-debug ${dest} -o
 ${debug_file}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+    # Link the two previous output files together, adding the .gnu_debuglink
+    # section to the stripped_file, containing a pointer to the debug_file,
+    # save the new file in dest.
+    # This will be the regular executable filename, in the usual location.
+    set result [catch "exec $objcopy_program --add-gnu-debuglink=${debug_file}
${stripped_file} ${dest}" output]
+    verbose "result is $result"
+    verbose "output is $output"
+    if {$result == 1} {
+      return 1
+    }
+
+   return 0
+}
+

sepdebug.exp
-------------

#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
#   2000, 2002, 2003
#   Free Software Foundation, Inc.

# This program is free software; you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation; either version 2 of the License, or
# (at your option) any later version.
# 
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
# 
# You should have received a copy of the GNU General Public License
# along with this program; if not, write to the Free Software
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

# Please email any bugs, comments, and/or additions to this file to:
# bug-gdb@prep.ai.mit.edu

# Modified to test gdb's handling of separate debug info files.
# Based on break.exp, written by Rob Savoye. (rob@cygnus.com)

if $tracelevel then {
    strace $tracelevel
}


#
# test running programs
#
set prms_id 0
set bug_id 0

set testfile "sepdebug"
set srcfile ${testfile}.c
set binfile ${objdir}/${subdir}/${testfile}

if  { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug additional_flags=-w}] != "" } {
    gdb_suppress_entire_file "Testcase compile failed, so all tests in this file will automatically fail."
}

# Note: the procedure gdb_gnu_strip_debug will produce an executable called
# ${binfile}, which is just like the executable ($binfile) but without
# the debuginfo. Instead $binfile has a .gnudebuglink section which contains
# the name of a idebuginfo only file. This file will be stored in the
# gdb.base/.debug subdirectory.

if [gdb_gnu_strip_debug $binfile] {
    unsupported "no separate debug info handling"
    return -1
}
  
gdb_exit
gdb_start
gdb_reinitialize_dir $srcdir/$subdir
gdb_load ${binfile}

if [target_info exists gdb_stub] {
    gdb_step_for_stub;
}
#
# test simple breakpoint setting commands
#

# Test deleting all breakpoints when there are none installed,
# GDB should not prompt for confirmation.
# Note that gdb-init.exp provides a "delete_breakpoints" proc
# for general use elsewhere.

send_gdb "delete breakpoints\n"
gdb_expect {
     -re "Delete all breakpoints.*$" {
	    send_gdb "y\n"
	    gdb_expect {
		-re "$gdb_prompt $" {
		    fail "Delete all breakpoints when none (unexpected prompt)"
		}
		timeout	{ fail "Delete all breakpoints when none (timeout after unexpected prompt)" }
	    }
	}
     -re ".*$gdb_prompt $"       { pass "Delete all breakpoints when none" }
    timeout	            { fail "Delete all breakpoints when none (timeout)" }
}

#
# test break at function
#
gdb_test "break main" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function"

#
# test break at quoted function
#
gdb_test "break \"marker2\"" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint quoted function"

#
# test break at function in file
#
gdb_test "break $srcfile:factorial" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function in file"

#old line 79
set bp_location1 [gdb_get_line_number "set breakpoint 1 here"]

#
# test break at line number
#
# Note that the default source file is the last one whose source text
# was printed.  For native debugging, before we've executed the
# program, this is the file containing main, but for remote debugging,
# it's wherever the processor was stopped when we connected to the
# board.  So, to be sure, we do a list command.
#
gdb_test "list main" \
    ".*main \\(argc, argv, envp\\).*" \
    "use `list' to establish default source file"
gdb_test "break $bp_location1" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
    "breakpoint line number"

#
# test duplicate breakpoint
#
gdb_test "break $bp_location1" \
    "Note: breakpoint \[0-9\]+ also set at pc.*Breakpoint \[0-9\]+ at.* file .*$srcfile, line $bp_location1\\." \
    "breakpoint duplicate"

set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]

#
# test break at line number in file
#
gdb_test "break $srcfile:$bp_location2" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
    "breakpoint line number in file"

set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]

#
# Test putting a break at the start of a multi-line if conditional.
# Verify the breakpoint was put at the start of the conditional.
#
gdb_test "break multi_line_if_conditional" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
    "breakpoint at start of multi line if conditional"

gdb_test "break multi_line_while_conditional" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
    "breakpoint at start of multi line while conditional"

set bp_location5 [gdb_get_line_number "set breakpoint 5 here"]
set bp_location6 [gdb_get_line_number "set breakpoint 6 here"]

#
# check to see what breakpoints are set
#
if [target_info exists gdb_stub] {
    set main_line $bp_location5
} else {
    set main_line $bp_location6
}

set bp_location7 [gdb_get_line_number "set breakpoint 7 here"]
set bp_location8 [gdb_get_line_number "set breakpoint 8 here"]
set bp_location9 [gdb_get_line_number "set breakpoint 9 here"]

gdb_test "info break" \
    "Num Type\[ \]+Disp Enb Address\[ \]+What.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$main_line.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in marker2 at .*$srcfile:($bp_location8|$bp_location9).*
\[0-9\]+\[\t \]+breakpoint     keep y.* in factorial at .*$srcfile:$bp_location7.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location1.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in main at .*$srcfile:$bp_location2.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in multi_line_if_conditional at .*$srcfile:$bp_location3.*
\[0-9\]+\[\t \]+breakpoint     keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
    "breakpoint info"

# FIXME: The rest of this test doesn't work with anything that can't
# handle arguments.
# Huh? There doesn't *appear* to be anything that passes arguments
# below.
if [istarget "mips-idt-*"] then {
    return
}

#
# run until the breakpoint at main is hit. For non-stubs-using targets.
#
if ![target_info exists use_gdb_stub] {
  if [istarget "*-*-vxworks*"] then {
    send_gdb "run vxmain \"2\"\n"
    set timeout 120
    verbose "Timeout is now $timeout seconds" 2
  } else {
	send_gdb "run\n"
  }
  gdb_expect {
    -re "The program .* has been started already.*y or n. $" {
	send_gdb "y\n"
	exp_continue
    }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $"\
	                    { pass "run until function breakpoint" }
    -re ".*$gdb_prompt $"       { fail "run until function breakpoint" }
    timeout	            { fail "run until function breakpoint (timeout)" }
  }
} else {
    if ![target_info exists gdb_stub] {
	gdb_test continue ".*Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.*\{.*" "stub continue"
    }
}

#
# run until the breakpoint at a line number
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location1.*$bp_location1\[\t \]+printf.*factorial.*" \
			"run until breakpoint set at a line number"

#
# Run until the breakpoint set in a function in a file
#
for {set i 6} {$i >= 1} {incr i -1} {
	gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, factorial \\(value=$i\\) at .*$srcfile:$bp_location7.*$bp_location7\[\t \]+.*if .value > 1. \{.*" \
			"run until file:function($i) breakpoint"
}

#
# Run until the breakpoint set at a quoted function
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile:($bp_location8|$bp_location9).*" \
		"run until quoted breakpoint"
#
# run until the file:function breakpoint at a line number in a file
#
gdb_test continue "Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location2.*$bp_location2\[\t \]+argc = \\(argc == 12345\\);.*" \
		"run until file:linenum breakpoint"

# Test break at offset +1
set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]

gdb_test "break +1" \
    "Breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
    "breakpoint offset +1"

# Check to see if breakpoint is hit when stepped onto

gdb_test "step" \
    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10.*$bp_location10\[\t \]+return argc;.*breakpoint 10 here.*" \
    "step onto breakpoint"

#
# delete all breakpoints so we can start over, course this can be a test too
#
delete_breakpoints

#
# test temporary breakpoint at function
#

gdb_test "tbreak main" "Breakpoint.*at.* file .*$srcfile, line.*" "Temporary breakpoint function"

#
# test break at function in file
#

gdb_test "tbreak $srcfile:factorial" "Breakpoint.*at.* file .*$srcfile, line.*" \
	"Temporary breakpoint function in file"

#
# test break at line number
#
send_gdb "tbreak $bp_location1\n"
gdb_expect {
    -re "Breakpoint.*at.* file .*$srcfile, line $bp_location1.*$gdb_prompt $" { pass "Temporary breakpoint line number #1" }
	-re ".*$gdb_prompt $"   { pass "Temporary breakpoint line number #1" }
	timeout	    { fail "breakpoint line number #1 (timeout)" }
}

gdb_test "tbreak $bp_location6" "Breakpoint.*at.* file .*$srcfile, line $bp_location6.*" "Temporary breakpoint line number #2"

#
# test break at line number in file
#
send_gdb "tbreak $srcfile:$bp_location2\n"
gdb_expect {
    -re "Breakpoint.*at.* file .*$srcfile, line $bp_location2.*$gdb_prompt $" { pass "Temporary breakpoint line number in file #1" }
	-re ".*$gdb_prompt $"   { pass "Temporary breakpoint line number in file #1" }
	timeout	    { fail "Temporary breakpoint line number in file #1 (timeout)" }
}

set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
gdb_test  "tbreak $srcfile:$bp_location11" "Breakpoint.*at.* file .*$srcfile, line $bp_location11.*" "Temporary breakpoint line number in file #2"

#
# check to see what breakpoints are set (temporary this time)
#
gdb_test "info break" "Num Type.*Disp Enb Address.*What.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$main_line.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in factorial at .*$srcfile:$bp_location7.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location1.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location6.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location2.*\[\r\n\]
\[0-9\]+\[\t \]+breakpoint     del.*y.*in main at .*$srcfile:$bp_location11.*" \
    "Temporary breakpoint info"


#***********

# Verify that catchpoints for fork, vfork and exec don't trigger
# inappropriately.  (There are no calls to those system functions
# in this test program.)
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "catch\n"
gdb_expect {
  -re "Catch requires an event name.*$gdb_prompt $"\
          {pass "catch requires an event name"}
  -re "$gdb_prompt $"\
          {fail "catch requires an event name"}
  timeout {fail "(timeout) catch requires an event name"}
}


set name "set catch fork, never expected to trigger"
send_gdb "catch fork\n"
gdb_expect {
  -re "Catchpoint \[0-9\]* .fork..*$gdb_prompt $"
          {pass $name}
  -re "Catch of fork not yet implemented.*$gdb_prompt $"
	  {pass $name}
  -re "$gdb_prompt $"
          {fail $name}
  timeout {fail "(timeout) $name"}
}


set name "set catch vfork, never expected to trigger"
send_gdb "catch vfork\n"

# If we are on HP-UX 10.20, we expect an error message to be
# printed if we type "catch vfork" at the gdb gdb_prompt.  This is
# because on HP-UX 10.20, we cannot catch vfork events.

if [istarget "hppa*-hp-hpux10.20"] then {
    gdb_expect {
	-re "Catch of vfork events not supported on HP-UX 10.20..*$gdb_prompt $"
		{pass $name}
	-re "$gdb_prompt $"
		{fail $name}
	timeout {fail "(timeout) $name"}
    }
} else {
    gdb_expect {
	-re "Catchpoint \[0-9\]* .vfork..*$gdb_prompt $"
		{pass $name}
	-re "Catch of vfork not yet implemented.*$gdb_prompt $"
		{pass $name}
	-re "$gdb_prompt $"
		{fail $name}
	timeout {fail "(timeout) $name"}
    }
}

set name "set catch exec, never expected to trigger"
send_gdb "catch exec\n"
gdb_expect {
  -re "Catchpoint \[0-9\]* .exec..*$gdb_prompt $"
          {pass $name}
  -re "Catch of exec not yet implemented.*$gdb_prompt $"
	  {pass $name}
  -re "$gdb_prompt $" {fail $name}
  timeout {fail "(timeout) $name"}
}

# Verify that GDB responds gracefully when asked to set a breakpoint
# on a nonexistent source line.
#
send_gdb "break 999\n"
gdb_expect {
  -re "No line 999 in file .*$gdb_prompt $"\
          {pass "break on non-existent source line"}
  -re "$gdb_prompt $"\
          {fail "break on non-existent source line"}
  timeout {fail "(timeout) break on non-existent source line"}
}

# Run to the desired default location. If not positioned here, the
# tests below don't work.
#
gdb_test "until $bp_location1" "main .* at .*:$bp_location1.*" "until $bp_location1"


# Verify that GDB allows one to just say "break", which is treated
# as the "default" breakpoint.  Note that GDB gets cute when printing
# the informational message about other breakpoints at the same
# location.  We'll hit that bird with this stone too.
#
send_gdb "break\n"
gdb_expect {
  -re "Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 1st time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 1st time"}
  timeout {fail "(timeout) break on default location, 1st time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoint \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 2nd time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 2nd time"}
  timeout {fail "(timeout) break on default location, 2nd time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoints \[0-9\]* and \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 3rd time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 3rd time"}
  timeout {fail "(timeout) break on default location, 3rd time"}
}

send_gdb "break\n"
gdb_expect {
  -re "Note: breakpoints \[0-9\]*, \[0-9\]* and \[0-9\]* also set at .*Breakpoint \[0-9\]*.*$gdb_prompt $"\
          {pass "break on default location, 4th time"}
  -re "$gdb_prompt $"\
          {fail "break on default location, 4th time"}
  timeout {fail "(timeout) break on default location, 4th time"}
}

# Verify that a "silent" breakpoint can be set, and that GDB is indeed
# "silent" about its triggering.
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "break $bp_location1\n"
gdb_expect {
  -re "Breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $"\
          {pass "set to-be-silent break $bp_location1"}
  -re "$gdb_prompt $"\
          {fail "set to-be-silent break $bp_location1"}
  timeout {fail "(timeout) set to-be-silent break $bp_location1"}
}

send_gdb "commands $expect_out(1,string)\n"
send_gdb "silent\n"
send_gdb "end\n"
gdb_expect {
  -re ".*$gdb_prompt $"\
          {pass "set silent break $bp_location1"}
  timeout {fail "(timeout) set silent break $bp_location1"}
}

send_gdb "info break $expect_out(1,string)\n"
gdb_expect {
  -re "\[0-9\]*\[ \t\]*breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*$gdb_prompt $"\
          {pass "info silent break $bp_location1"}
  -re "$gdb_prompt $"\
          {fail "info silent break $bp_location1"}
  timeout {fail "(timeout) info silent break $bp_location1"}
}
send_gdb "continue\n"
gdb_expect {
  -re "Continuing.\r\n$gdb_prompt $"\
          {pass "hit silent break $bp_location1"}
  -re "$gdb_prompt $"\
          {fail "hit silent break $bp_location1"}
  timeout {fail "(timeout) hit silent break $bp_location1"}
}
send_gdb "bt\n"
gdb_expect {
  -re "#0  main .* at .*:$bp_location1.*$gdb_prompt $"\
          {pass "stopped for silent break $bp_location1"}
  -re "$gdb_prompt $"\
          {fail "stopped for silent break $bp_location1"}
  timeout {fail "(timeout) stopped for silent break $bp_location1"}
}

# Verify that GDB can at least parse a breakpoint with the
# "thread" keyword.  (We won't attempt to test here that a
# thread-specific breakpoint really triggers appropriately.
# The gdb.threads subdirectory contains tests for that.)
#
set bp_location12 [gdb_get_line_number "set breakpoint 12 here"]
send_gdb "break $bp_location12 thread 999\n"
gdb_expect {
  -re "Unknown thread 999.*$gdb_prompt $"\
          {pass "thread-specific breakpoint on non-existent thread disallowed"}
  -re "$gdb_prompt $"\
          {fail "thread-specific breakpoint on non-existent thread disallowed"}
  timeout {fail "(timeout) thread-specific breakpoint on non-existent thread disallowed"}
}
send_gdb "break $bp_location12 thread foo\n"
gdb_expect {
  -re "Junk after thread keyword..*$gdb_prompt $"\
          {pass "thread-specific breakpoint on bogus thread ID disallowed"}
  -re "$gdb_prompt $"\
          {fail "thread-specific breakpoint on bogus thread ID disallowed"}
  timeout {fail "(timeout) thread-specific breakpoint on bogus thread ID disallowed"}
}

# Verify that GDB responds gracefully to a breakpoint command with
# trailing garbage.
#
send_gdb "break $bp_location12 foo\n"
gdb_expect {
  -re "Junk at end of arguments..*$gdb_prompt $"\
          {pass "breakpoint with trailing garbage disallowed"}
  -re "$gdb_prompt $"\
          {fail "breakpoint with trailing garbage disallowed"}
  timeout {fail "(timeout) breakpoint with trailing garbage disallowed"}
}

# Verify that GDB responds gracefully to a "clear" command that has
# no matching breakpoint.  (First, get us off the current source line,
# which we know has a breakpoint.)
#
send_gdb "next\n"
gdb_expect {
  -re ".*$gdb_prompt $"\
          {pass "step over breakpoint"}
  timeout {fail "(timeout) step over breakpoint"}
}
send_gdb "clear 81\n"
gdb_expect {
  -re "No breakpoint at 81..*$gdb_prompt $"\
          {pass "clear line has no breakpoint disallowed"}
  -re "$gdb_prompt $"\
          {fail "clear line has no breakpoint disallowed"}
  timeout {fail "(timeout) clear line has no breakpoint disallowed"}
}
send_gdb "clear\n"
gdb_expect {
  -re "No breakpoint at this line..*$gdb_prompt $"\
          {pass "clear current line has no breakpoint disallowed"}
  -re "$gdb_prompt $"\
          {fail "clear current line has no breakpoint disallowed"}
  timeout {fail "(timeout) clear current line has no breakpoint disallowed"}
}

# Verify that we can set and clear multiple breakpoints.
#
# We don't test that it deletes the correct breakpoints.  We do at
# least test that it deletes more than one breakpoint.
#
gdb_test "break marker3" "Breakpoint.*at.*" "break marker3 #1"
gdb_test "break marker3" "Breakpoint.*at.*" "break marker3 #2"
gdb_test "clear marker3" {Deleted breakpoints [0-9]+ [0-9]+.*}

# Verify that a breakpoint can be set via a convenience variable.
#
send_gdb "set \$foo=$bp_location11\n"
gdb_expect {
  -re "$gdb_prompt $"\
          {pass "set convenience variable \$foo to $bp_location11"}
  timeout {fail "(timeout) set convenience variable \$foo to $bp_location11"}
}
send_gdb "break \$foo\n"
gdb_expect {
  -re "Breakpoint (\[0-9\]*) at .*, line $bp_location11.*$gdb_prompt $"\
          {pass "set breakpoint via convenience variable"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint via convenience variable"}
  timeout {fail "(timeout) set breakpoint via convenience variable"}
}

# Verify that GDB responds gracefully to an attempt to set a
# breakpoint via a convenience variable whose type is not integer.
#
send_gdb "set \$foo=81.5\n"
gdb_expect {
  -re "$gdb_prompt $"\
          {pass "set convenience variable \$foo to 81.5"}
  timeout {fail "(timeout) set convenience variable \$foo to 81.5"}
}
send_gdb "break \$foo\n"
gdb_expect {
  -re "Convenience variables used in line specs must have integer values..*$gdb_prompt $"\
          {pass "set breakpoint via non-integer convenience variable disallowed"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint via non-integer convenience variable disallowed"}
  timeout {fail "(timeout) set breakpoint via non-integer convenience variable disallowed"}
}

# Verify that we can set and trigger a breakpoint in a user-called function.
#
send_gdb "break marker2\n"
gdb_expect {
    -re "Breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*$gdb_prompt $"\
          {pass "set breakpoint on to-be-called function"}
  -re "$gdb_prompt $"\
          {fail "set breakpoint on to-be-called function"}
  timeout {fail "(timeout) set breakpoint on to-be-called function"}
}
send_gdb "print marker2(99)\n"
gdb_expect {
  -re "The program being debugged stopped while in a function called from GDB.\r\nWhen the function .marker2. is done executing, GDB will silently\r\nstop .instead of continuing to evaluate the expression containing\r\nthe function call...*$gdb_prompt $"\
          {pass "hit breakpoint on called function"}
  -re "$gdb_prompt $"\
          {fail "hit breakpoint on called function"}
  timeout {fail "(timeout) hit breakpoint on called function"}
}

# As long as we're stopped (breakpointed) in a called function,
# verify that we can successfully backtrace & such from here.
#
# In this and the following test, the _sr4export check apparently is needed
# for hppa*-*-hpux.
#
send_gdb "bt\n"
gdb_expect {
    -re "#0\[ \t\]*($hex in )?marker2.*:($bp_location8|$bp_location9)\r\n#1.*_sr4export.*$gdb_prompt $"\
            {pass "backtrace while in called function"}
    -re "#0\[ \t\]*($hex in )?marker2.*:($bp_location8|$bp_location9)\r\n#1.*function called from gdb.*$gdb_prompt $"\
	    {pass "backtrace while in called function"}
    -re "$gdb_prompt $"\
            {fail "backtrace while in called function"}
    timeout {fail "(timeout) backtrace while in called function"}
}

# Return from the called function.  For remote targets, it's important to do
# this before runto_main, which otherwise may silently stop on the dummy
# breakpoint inserted by GDB at the program's entry point.
#
send_gdb "finish\n"
gdb_expect {
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.* in _sr4export.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.*function called from gdb.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "Run till exit from .*marker2.* at .*($bp_location8|$bp_location9)\r\n.*Value returned.*$gdb_prompt $"\
            {pass "finish from called function"}
    -re "$gdb_prompt $"\
            {fail "finish from called function"}
    timeout {fail "(timeout) finish from called function"}
}

# Verify that GDB responds gracefully to a "finish" command with
# arguments.
#
if ![runto_main] then { fail "break tests suppressed" }

send_gdb "finish 123\n"
gdb_expect {
  -re "The \"finish\" command does not take any arguments.\r\n$gdb_prompt $"\
          {pass "finish with arguments disallowed"}
  -re "$gdb_prompt $"\
          {fail "finish with arguments disallowed"}
  timeout {fail "(timeout) finish with arguments disallowed"}
}

# Verify that GDB responds gracefully to a request to "finish" from
# the outermost frame.  On a stub that never exits, this will just
# run to the stubs routine, so we don't get this error...  Thus the 
# second condition.
#

send_gdb "finish\n"
gdb_expect {
  -re "\"finish\" not meaningful in the outermost frame.\r\n$gdb_prompt $"\
          {pass "finish from outermost frame disallowed"}
  -re "Run till exit from.*\r\n$gdb_prompt $" {
     pass "finish from outermost frame disallowed"
  }
  -re "$gdb_prompt $"\
          {fail "finish from outermost frame disallowed"}
  timeout {fail "(timeout) finish from outermost frame disallowed"}
}

# Verify that we can explicitly ask GDB to stop on all shared library
# events, and that it does so.
#
if [istarget "hppa*-*-hpux*"] then {
  if ![runto_main] then { fail "break tests suppressed" }

  send_gdb "set stop-on-solib-events 1\n"
  gdb_expect {
    -re "$gdb_prompt $"\
            {pass "set stop-on-solib-events"}
    timeout {fail "(timeout) set stop-on-solib-events"}
  }

  send_gdb "run\n"
  gdb_expect {
    -re ".*Start it from the beginning.*y or n. $"\
            {send_gdb "y\n"
             gdb_expect {
               -re ".*Stopped due to shared library event.*$gdb_prompt $"\
                       {pass "triggered stop-on-solib-events"}
               -re "$gdb_prompt $"\
                       {fail "triggered stop-on-solib-events"}
               timeout {fail "(timeout) triggered stop-on-solib-events"}
             }
            }
    -re "$gdb_prompt $"\
            {fail "rerun for stop-on-solib-events"}
    timeout {fail "(timeout) rerun for stop-on-solib-events"}
  }

  send_gdb "set stop-on-solib-events 0\n"
  gdb_expect {
    -re "$gdb_prompt $"\
            {pass "reset stop-on-solib-events"}
    timeout {fail "(timeout) reset stop-on-solib-events"}
  }
}

# Hardware breakpoints are unsupported on HP-UX.  Verify that GDB
# gracefully responds to requests to create them.
#
if [istarget "hppa*-*-hpux*"] then {
  if ![runto_main] then { fail "break tests suppressed" }

  send_gdb "hbreak\n"
  gdb_expect {
    -re "No hardware breakpoint support in the target.*$gdb_prompt $"\
            {pass "hw breaks disallowed"}
    -re "$gdb_prompt $"\
            {fail "hw breaks disallowed"}
    timeout {fail "(timeout) hw breaks disallowed"}
  }

  send_gdb "thbreak\n"
  gdb_expect {
    -re "No hardware breakpoint support in the target.*$gdb_prompt $"\
            {pass "temporary hw breaks disallowed"}
    -re "$gdb_prompt $"\
            {fail "temporary hw breaks disallowed"}
    timeout {fail "(timeout) temporary hw breaks disallowed"}
  }
}

#********


#
# Test "next" over recursive function call.
#

proc test_next_with_recursion {} { 
    global gdb_prompt
    global decimal
    global binfile

    if [target_info exists use_gdb_stub] {
	# Reload the program.
	delete_breakpoints
	gdb_load ${binfile};
    } else {
	# FIXME: should be using runto
	gdb_test "kill" "" "kill program" "Kill the program being debugged.*y or n. $" "y"

	delete_breakpoints
    }

    gdb_test "break factorial" "Breakpoint $decimal at .*" "break at factorial"

    # Run until we call factorial with 6

    if [istarget "*-*-vxworks*"] then {
	send_gdb "run vxmain \"6\"\n"
    } else {
	gdb_run_cmd
    }
    gdb_expect {
	-re "Break.* factorial .value=6. .*$gdb_prompt $" {}
	-re ".*$gdb_prompt $" {
	    fail "run to factorial(6)";
	    gdb_suppress_tests;
	}
	timeout { fail "run to factorial(6) (timeout)" ; gdb_suppress_tests }
    }

    # Continue until we call factorial recursively with 5.

    if [gdb_test "continue" \
	"Continuing.*Break.* factorial .value=5. .*" \
	"continue to factorial(5)"] then { gdb_suppress_tests }

    # Do a backtrace just to confirm how many levels deep we are.

    if [gdb_test "backtrace" \
	"#0\[ \t\]+ factorial .value=5..*" \
	"backtrace from factorial(5)"] then { gdb_suppress_tests }

    # Now a "next" should position us at the recursive call, which
    # we will be performing with 4.

    if [gdb_test "next" \
	".* factorial .value - 1.;.*" \
	"next to recursive call"] then { gdb_suppress_tests }

    # Disable the breakpoint at the entry to factorial by deleting them all.
    # The "next" should run until we return to the next line from this
    # recursive call to factorial with 4.
    # Buggy versions of gdb will stop instead at the innermost frame on
    # the line where we are trying to "next" to.

    delete_breakpoints

    if [istarget "mips*tx39-*"] {
	set timeout 60
    }
    # We used to set timeout here for all other targets as well.  This
    # is almost certainly wrong.  The proper timeout depends on the
    # target system in use, and how we communicate with it, so there
    # is no single value appropriate for all targets.  The timeout
    # should be established by the Dejagnu config file(s) for the
    # board, and respected by the test suite.
    #
    # For example, if I'm running GDB over an SSH tunnel talking to a
    # portmaster in California talking to an ancient 68k board running
    # a crummy ROM monitor (a situation I can only wish were
    # hypothetical), then I need a large timeout.  But that's not the
    # kind of knowledge that belongs in this file.

    gdb_test next "\[0-9\]*\[\t \]+return \\(value\\);.*" \
	    "next over recursive call"

    # OK, we should be back in the same stack frame we started from.
    # Do a backtrace just to confirm.

    set result [gdb_test "backtrace" \
	    "#0\[ \t\]+ factorial .value=120.*\r\n#1\[ \t\]+ \[0-9a-fx\]+ in factorial .value=6..*" \
	    "backtrace from factorial(5.1)"]
    if { $result != 0 } { gdb_suppress_tests }

    if [target_info exists gdb,noresults] { gdb_suppress_tests }
  gdb_continue_to_end "recursive next test"
   gdb_stop_suppressing_tests;
}

test_next_with_recursion


#********

# now move the .debug file to a different location so that we can test
# the "set debug-file-directory" command.
  
remote_exec build "mv ${objdir}/${subdir}/.debug/${testfile}.debug ${objdir}/${subdir}"
gdb_exit
gdb_start
gdb_reinitialize_dir $srcdir/$subdir
gdb_test "set debug-file-directory ${objdir}/${subdir}" ".*" "set separate debug location"
gdb_load ${binfile}

if [target_info exists gdb_stub] {
    gdb_step_for_stub;
}

#
# test break at function
#
gdb_test "break main" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint function, optimized file"

#
# test break at function
#
gdb_test "break marker4" \
    "Breakpoint.*at.* file .*$srcfile, line.*" \
    "breakpoint small function, optimized file"

#
# run until the breakpoint at main is hit. For non-stubs-using targets.
#
if ![target_info exists use_gdb_stub] {
  if [istarget "*-*-vxworks*"] then {
    send_gdb "run vxmain \"2\"\n"
    set timeout 120
    verbose "Timeout is now $timeout seconds" 2
  } else {
	send_gdb "run\n"
  }
  gdb_expect {
    -re "The program .* has been started already.*y or n. $" {
	send_gdb "y\n"
	exp_continue
    }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $"\
	                    { pass "run until function breakpoint, optimized file" }
    -re "Starting program.*Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $"\
	                    { pass "run until function breakpoint, optimized file (code motion)" }
    -re ".*$gdb_prompt $"       { fail "run until function breakpoint, optimized file" }
    timeout	            { fail "run until function breakpoint, optimized file (timeout)" }
  }
} else {
    if ![target_info exists gdb_stub] {
	gdb_test continue ".*Continuing\\..*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.*\{.*" "stub continue, optimized file"
    }
}

#
# run until the breakpoint at a small function
#

#
# Add a second pass pattern.  The behavior differs here between stabs
# and dwarf for one-line functions.  Stabs preserves two line symbols
# (one before the prologue and one after) with the same line number, 
# but dwarf regards these as duplicates and discards one of them.
# Therefore the address after the prologue (where the breakpoint is)
# has no exactly matching line symbol, and GDB reports the breakpoint
# as if it were in the middle of a line rather than at the beginning.

set bp_location13 [gdb_get_line_number "set breakpoint 13 here"]
set bp_location14 [gdb_get_line_number "set breakpoint 14 here"]
send_gdb "continue\n"
gdb_expect {
    -re "Breakpoint $decimal, marker4 \\(d=177601976\\) at .*$srcfile:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
	pass "run until breakpoint set at small function, optimized file"
    }
    -re "Breakpoint $decimal, $hex in marker4 \\(d=177601976\\) at .*$srcfile:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
	pass "run until breakpoint set at small function, optimized file"
    }
    -re "Breakpoint $decimal, marker4 \\(d=177601976\\) at .*$srcfile:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
        # marker4() is defined at line 46 when compiled with -DPROTOTYPES
	pass "run until breakpoint set at small function, optimized file (line $bp_location14)"
    }
    -re ".*$gdb_prompt " {
	fail "run until breakpoint set at small function, optimized file"
    }
    timeout {
	fail "run until breakpoint set at small function, optimized file (timeout)"
    }
}


# Reset the default arguments for VxWorks
if [istarget "*-*-vxworks*"] {
    set timeout 10
    verbose "Timeout is now $timeout seconds" 2
    send_gdb "set args main\n"
    gdb_expect -re ".*$gdb_prompt $" {}
}


sepdebug.c
-----------------------
/* Copyright 1994, 1995, 1999, 2002, 2003  Free Software Foundation, Inc.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2 of the License, or
   (at your option) any later version.
 
   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.
 
   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software
   Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  

   Please email any bugs, comments, and/or additions to this file to:
   bug-gdb@prep.ai.mit.edu  */

#ifdef vxworks

#  include <stdio.h>

/* VxWorks does not supply atoi.  */
static int
atoi (z)
     char *z;
{
  int i = 0;

  while (*z >= '0' && *z <= '9')
    i = i * 10 + (*z++ - '0');
  return i;
}

/* I don't know of any way to pass an array to VxWorks.  This function
   can be called directly from gdb.  */

vxmain (arg)
char *arg;
{
  char *argv[2];

  argv[0] = "";
  argv[1] = arg;
  main (2, argv, (char **) 0);
}

#else /* ! vxworks */
#  include <stdio.h>
#  include <stdlib.h>
#endif /* ! vxworks */

/*
 * The following functions do nothing useful.  They are included simply
 * as places to try setting breakpoints at.  They are explicitly
 * "one-line functions" to verify that this case works (some versions
 * of gcc have or have had problems with this).
 */

#ifdef PROTOTYPES
int marker1 (void) { return (0); }
int marker2 (int a) { return (1); } /* set breakpoint 8 here */
void marker3 (char *a, char *b) {}
void marker4 (long d) {} /* set breakpoint 14 here */
#else
int marker1 () { return (0); }
int marker2 (a) int a; { return (1); } /* set breakpoint 9 here */
void marker3 (a, b) char *a, *b; {}
void marker4 (d) long d; {}  /* set breakpoint 13 here */
#endif

/*
 *	This simple classical example of recursion is useful for
 *	testing stack backtraces and such.
 */

#ifdef PROTOTYPES
int factorial(int);

int
main (int argc, char **argv, char **envp)
#else
int
main (argc, argv, envp)
int argc;
char *argv[], **envp;
#endif
{
#ifdef usestubs
    set_debug_traps();  /* set breakpoint 5 here */
    breakpoint();
#endif
    if (argc == 12345) {  /* an unlikely value < 2^16, in case uninited */ /* set breakpoint 6 here */
	fprintf (stderr, "usage:  factorial <number>\n");
	return 1;
    }
    printf ("%d\n", factorial (atoi ("6")));  /* set breakpoint 1 here */
    /* set breakpoint 12 here */
    marker1 ();  /* set breakpoint 11 here */
    marker2 (43);
    marker3 ("stack", "trace");
    marker4 (177601976L);
    argc = (argc == 12345); /* This is silly, but we can step off of it */ /* set breakpoint 2 here */
    return argc;  /* set breakpoint 10 here */
}

#ifdef PROTOTYPES
int factorial (int value)
#else
int factorial (value)
int value;
#endif
{
  if (value > 1) {  /* set breakpoint 7 here */
	value *= factorial (value - 1);
    }
    return (value);
}

#ifdef PROTOTYPES
int multi_line_if_conditional (int a, int b, int c)
#else
int multi_line_if_conditional (a, b, c)
  int a, b, c;
#endif
{
  if (a    /* set breakpoint 3 here */
      && b
      && c)
    return 0;
  else
    return 1;
}

#ifdef PROTOTYPES
int multi_line_while_conditional (int a, int b, int c)
#else
int multi_line_while_conditional (a, b, c)
  int a, b, c;
#endif
{
  while (a /* set breakpoint 4 here */
      && b
      && c)
    {
      a--, b--, c--;
    }
  return 0;
}