[PATCH] testsuite: Add (extensive) hardware breakpoint testing

[PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

Hi,

 This change adds a test case for hardware breakpoint testing.  This was 
based on gdb.base/break.exp and originally called gdb.base/hbreak.exp, but 
I can see that since its inception a small smoke test already took that 
name, so let's call this one hbreak2.exp unless you have a better idea.

 There's little to add here, the contents should be self explanatory.  As 
hardware breakpoint support is not granted, some checks are performed at 
the beginning to verify this feature is supported at all to prevent 
unjustified failures.  This seems to work well enough for me, but 
obviously I could only test a subset of targets, so something may still 
pop up as people try it.

 Also, as the number of hardware breakpoints in a given processor is 
likely limited (though some simulators may and do actually provide an 
infinite number) and may be implementation specific (e.g. x86 defines a 
fixed number of four such breakpoints, but MIPS can have a varying 
number), this test refrains from using more than one such breakpoint at a 
time.  This may cause potential problems with multiple such breakpoints 
installed at a time to escape unnoticed, but let's add one item at a time.  
The test can be further improved later on or another one added.

 I have tested this change with what I have available, trying to weigh the 
amount of effort against the expected benefits.  So I focused on making 
sure the lack of support is correctly detected and the tests pass on 
targets that do support this kind of breakpoints.

I covered the former with the following targets:

* mips-linux-gnu, remotely via gdbserver -- this passes "hardware 
  breakpoint support", but trips on "hardware breakpoint insertion":

 "Warning:
  Cannot insert hardware breakpoint 2.
  Could not insert hardware breakpoints:
  You may have requested too many hardware breakpoints/watchpoints."

* mips-linux-gnu, natively -- this trips on "hardware breakpoints": 

  "No hardware breakpoint support in the target."

* powerpc64-linux-gnu, natively -- this trips on "hardware breakpoints" 
  (presumably because this box does not implement them):

 "Hardware breakpoints used exceeds limit."

  (is that a typo there in this message?).

I covered the latter with:

* mips-sde-elf, remotely using a simulator -- this passes all the tests:

  # of expected passes            71

* i686-linux-gnu, natively -- this should work, but it does not:

  # of expected passes            16
  # of unexpected failures        55

  and the transcript shows:

  (gdb) break main
  Breakpoint 1 at 0x804849c: file .../gdb/testsuite/gdb.base/break.c, line 93.
  (gdb) run
  Starting program: .../gdb/testsuite/gdb.base/break

  Breakpoint 1, main (argc=1, argv=0xbffff134, envp=0xbffff13c) at .../gdb/testsuite/gdb.base/break.c:93
  93          if (argc == 12345) {  /* an unlikely value < 2^16, in case uninited*/ /* set breakpoint 6 here */
  (gdb) delete breakpoints
  Delete all breakpoints? (y or n) y
  (gdb) info breakpoints
  No breakpoints or watchpoints.
  (gdb) hbreak main
  Hardware assisted breakpoint 2 at 0x804849c: file .../gdb/testsuite/gdb.base/break.c, line 93.
  (gdb) PASS: gdb.base/hbreak2.exp: hardware breakpoint support
  run
  The program being debugged has been started already.
  Start it from the beginning? (y or n) y
  Starting program: .../gdb/testsuite/gdb.base/break
  720
  [Inferior 1 (process 16835) exited normally]
  (gdb) FAIL: gdb.base/hbreak2.exp: hardware breakpoint insertion (the program exited)

  that is the breakpoint set has been missed -- i.e. not a test case 
  problem.

* i686-mingw, natively -- likewise:

  # of expected passes            8
  # of unexpected failures        63

  and the transcript:

  (gdb) break main
  Breakpoint 1 at 0x4013e2: file break.c, line 93.
  (gdb) run
  Starting program: .../gdb/testsuite/gdb.d/break.exe
  [New Thread 5452.0x197c]

  Breakpoint 1, main (argc=1, argv=0x5a10d8, envp=0x5a1688) at break.c:93
  93          if (argc == 12345) {  /* an unlikely value < 2^16, in case uninited*/ /* set breakpoint 6 here */
  (gdb) delete breakpoints
  Delete all breakpoints? (y or n) [answered Y; input not from terminal]
  (gdb) info breakpoints
  No breakpoints or watchpoints.
  (gdb) hbreak main
  Hardware assisted breakpoint 2 at 0x4013e2: file break.c, line 93.
  (gdb) PASS: gdb.base/hbreak2.exp: hardware breakpoint support
  run
  The program being debugged has been started already.
  Start it from the beginning? (y or n) [answered Y; input not from terminal]
  Starting program: .../gdb/testsuite/gdb.d/break.exe
  [New Thread 5000.0xef8]
  720
  [Inferior 1 (process 5000) exited normally]
  (gdb) FAIL: gdb.base/hbreak2.exp: hardware breakpoint insertion (the program exited)

  i.e. the same reason.

 The consistency of failures between i686-linux-gnu and i686-mingw makes 
me fairly sure that's a bug in x86 support in GDB of some sort rather than 
a problem with my setup -- is that a known bug?  Anyway, this (together 
with the PowerPC failure quoted above) has left me out of native targets 
to test, but the transcripts of the test runs make me confident the case 
handles both remote and native targets (that use "target ...; continue" 
and "run" respectively) correctly and will do the right thing once we have 
a working native target to test.

 Questions or comments?  Otherwise OK to apply?

2011-11-11  Maciej W. Rozycki  <macro@mips.com>
	    Maciej W. Rozycki  <macro@codesourcery.com>
	    Daniel Jacobowitz  <dan@codesourcery.com>

	* gdb.base/hbreak2.exp: New test for "hbreak" and "thbreak".

  Maciej

gdb-test-hbreak.diff
Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2011-11-10 13:23:14.525919195 +0000
@@ -0,0 +1,629 @@
+#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
+#   2000, 2002, 2003, 2007, 2008, 2009, 2010, 2011
+#   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 3 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, see <http://www.gnu.org/licenses/>.
+
+# Based on break.exp by Rob Savoye. (rob@cygnus.com)
+# Only one hardware breakpoint is set at a time as targets may limit
+# the number available.
+
+
+if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
+    return -1
+}
+set srcfile break.c
+set srcfile1 break1.c
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test whether the target supports hardware breakpoints at all.
+#
+gdb_test_multiple "hbreak main" "hardware breakpoint support" {
+    -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
+	pass "hardware breakpoint support"
+    }
+}
+gdb_run_cmd
+gdb_test_multiple "" "hardware breakpoint insertion" {
+    -re "Warning:\[\r\n\]+Cannot insert hardware breakpoint \[0-9\]+\.\[\r\n\]+Could not insert hardware breakpoints:\[\r\n\]+You may have requested too many hardware breakpoints/watchpoints\.\[\r\n\]+.*$gdb_prompt $" {
+	unsupported "hardware breakpoint insertion"
+	return
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:.*\[\r\n\]+.*\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "hardware breakpoint insertion"
+    }
+}
+delete_breakpoints
+
+#
+# Test simple hardware breakpoint setting commands.
+#
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file"
+delete_breakpoints
+
+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 "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number"
+delete_breakpoints
+
+set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]
+
+#
+# Test break at line number in file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file"
+delete_breakpoints
+
+set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
+set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]
+
+#
+# Test putting a break at the start of a multi-line if conditional.
+# Verify the breakpoint was put at the start of the conditional.
+#
+gdb_test "hbreak multi_line_if_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
+    "hardware breakpoint at start of multi line if conditional"
+delete_breakpoints
+
+gdb_test "hbreak multi_line_while_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
+    "hardware 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 $bp_location6
+}
+
+if {$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" $srcfile1]
+set bp_location9 [gdb_get_line_number "set breakpoint 9 here" $srcfile1]
+
+gdb_test "info break" \
+    "Num     Type\[ \]+Disp Enb Address\[ \]+What.*
+\[0-9\]+\[\t \]+hw breakpoint  keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
+    "hardware breakpoint info"
+delete_breakpoints
+
+# 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.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function (2)"
+gdb_run_cmd
+gdb_test "" \
+    "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*" \
+    "run until function breakpoint"
+delete_breakpoints
+
+#
+# Run until the breakpoint at a line number.
+#
+gdb_test "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number (2)"
+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"
+delete_breakpoints
+
+#
+# Run until the breakpoint set in a function in a file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file (2)"
+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($i) breakpoint"
+}
+delete_breakpoints
+
+#
+# Run until the breakpoint set at a quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function (2)"
+gdb_test "continue" \
+    "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*" \
+    "run until quoted breakpoint"
+delete_breakpoints
+#
+# Run until the file:function breakpoint at a line number in a file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file (2)"
+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"
+delete_breakpoints
+
+# Test break at offset +1.
+set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]
+
+gdb_test "hbreak +1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
+    "hardware 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 hardware breakpoint"
+delete_breakpoints
+
+# Check to see if breakpoint can be set on ending brace of function.
+set bp_location10a [gdb_get_line_number "set breakpoint 10a here"]
+
+gdb_test "hbreak $bp_location10a" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10a\\." \
+    "setting hardware breakpoint at }"
+
+gdb_test "continue" \
+    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10a.*$bp_location10a\[\t \]+}.*breakpoint 10a here.*" \
+    "continue to hardware breakpoint at }"
+
+#
+# Delete all breakpoints so we can start over, course this can be a test too.
+#
+delete_breakpoints
+
+#
+# Test temporary breakpoint at function.
+#
+
+gdb_test "thbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+
+gdb_test "thbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function in file"
+delete_breakpoints
+
+#
+# Test break at line number.
+#
+gdb_test "thbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1.*" \
+    "temporary hardware breakpoint line number #1"
+delete_breakpoints
+
+gdb_test "thbreak $bp_location6" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location6.*" \
+    "temporary hardware breakpoint line number #2"
+delete_breakpoints
+
+#
+# Test break at line number in file.
+#
+gdb_test "thbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2.*" \
+    "temporary hardware breakpoint line number in file #1"
+delete_breakpoints
+
+set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
+gdb_test "thbreak $srcfile:$bp_location11" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location11.*" \
+    "temporary hardware 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.*\[\r\n\]
+\[0-9\]+\[\t \]+hw breakpoint  del.*y.*in main at .*$srcfile:$bp_location11.*" \
+    "temporary hardware breakpoint info"
+
+
+#***********
+
+if ![runto_main] then { fail "break tests suppressed" }
+
+# Verify that GDB responds gracefully when asked to set a breakpoint
+# on a nonexistent source line.
+#
+gdb_test "hbreak 999" \
+    "No line 999 in file .*" \
+    "hardware 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 "hbreak", which is treated
+# as the "default" breakpoint.
+#
+gdb_test "hbreak" "Hardware assisted breakpoint \[0-9\]*.*" \
+    "hardware break on default location"
+
+# 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" }
+
+gdb_test_multiple "hbreak $bp_location1" \
+    "set to-be-silent hardware break bp_location1" {
+	-re "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $" {
+	    pass "set to-be-silent hardware break bp_location1"
+	}
+    }
+
+send_gdb "commands $expect_out(1,string)\n"
+send_gdb "silent\n"
+send_gdb "end\n"
+gdb_test "" "set silent hardware break bp_location1"
+
+gdb_test "info break $expect_out(1,string)" \
+    "\[0-9\]*\[ \t\]*hw breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*" \
+    "info silent hardware break bp_location1"
+
+gdb_test "continue" "Continuing." \
+    "hit silent hardware break bp_location1"
+
+gdb_test "bt" "#0  main .* at .*:$bp_location1.*" \
+    "stopped for silent hardware 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"]
+gdb_test "hbreak $bp_location12 thread 999" "Unknown thread 999.*" \
+    "thread-specific hardware breakpoint on non-existent thread disallowed"
+
+gdb_test "hbreak $bp_location12 thread foo" \
+    "Junk after thread keyword.*" \
+    "thread-specific hardware breakpoint on bogus thread ID disallowed"
+
+# Verify that GDB responds gracefully to a breakpoint command with
+# trailing garbage.
+#
+gdb_test "hbreak $bp_location12 foo" \
+    "Junk at end of arguments.*" \
+    "hardware breakpoint with trailing garbage disallowed"
+
+# Verify that GDB responds gracefully to a "clear" command that has
+# no matching breakpoint.  (First, get us off the current source line,
+# which we know has a breakpoint.)
+#
+gdb_test "next" "marker1.*" "step over hardware breakpoint"
+
+gdb_test "clear 81" "No breakpoint at 81.*" \
+    "clear line has no breakpoint disallowed"
+
+gdb_test "clear" "No breakpoint at this line.*" \
+    "clear current line has no breakpoint disallowed"
+delete_breakpoints
+
+# Verify that a breakpoint can be set via a convenience variable.
+#
+gdb_test_no_output "set \$foo=$bp_location11" \
+    "set convenience variable \$foo to bp_location11"
+
+gdb_test "hbreak \$foo" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location11.*" \
+    "set hardware breakpoint via convenience variable"
+delete_breakpoints
+
+# Verify that GDB responds gracefully to an attempt to set a
+# breakpoint via a convenience variable whose type is not integer.
+#
+gdb_test_no_output "set \$foo=81.5" \
+    "set convenience variable \$foo to 81.5"
+
+gdb_test "hbreak \$foo" \
+    "Convenience variables used in line specs must have integer values.*" \
+    "set hardware breakpoint via non-integer convenience variable disallowed"
+
+# Verify that we can set and trigger a breakpoint in a user-called function.
+#
+gdb_test "hbreak marker2" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*" \
+    "set hardware breakpoint on to-be-called function"
+
+gdb_test "print marker2(99)" \
+    "The program being debugged stopped while in a function called from GDB.\r\nEvaluation of the expression containing the function\r\n.marker2$proto. will be abandoned.\r\nWhen the function is done executing, GDB will silently stop.*" \
+    "hit hardware 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.
+#
+gdb_test_multiple "bt" "backtrace while in called function" {
+    -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 $" {
+	pass "backtrace while in called function"
+    }
+}
+
+# Return from the called function.  For remote targets, it's important to do
+# this before runto_main, which otherwise may silently stop on the dummy
+# breakpoint inserted by GDB at the program's entry point.
+#
+gdb_test_multiple "finish" "finish from called function" {
+    -re "Run till exit from .*marker2.* at .*($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"
+    }
+}
+
+#********
+
+
+#
+# Test "next" over recursive function call.
+#
+
+proc test_next_with_recursion {} {
+    global gdb_prompt
+    global decimal
+    global binfile
+
+    delete_breakpoints
+
+    # Can't set a hardware breakpoint without a live target, so do it now
+    # before it's killed below.
+    gdb_test "hbreak factorial" \
+	"Hardware assisted breakpoint $decimal at .*" \
+	"hardware break at factorial"
+
+    gdb_test "kill" "" "kill program" \
+	"Kill the program being debugged.*y or n. $" "y"
+
+    # Run until we call factorial with 6
+
+    gdb_run_cmd
+    if [gdb_test "" \
+	"Break.* factorial .value=6. .*" \
+	"run to factorial(6)"] then { 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
+
+
+#********
+
+# Build a new file with optimization enabled so that we can try breakpoints
+# on targets with optimized prologues.
+
+if { [prepare_for_testing hbreak2.exp "breako2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
+    return -1
+}
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*, line.*" \
+    "hardware breakpoint function, optimized file"
+
+#
+# Run until the breakpoint at main is hit.  For non-stubs-using targets.
+#
+gdb_run_cmd
+gdb_test_multiple "" "run until hardware function breakpoint, optimized file" {
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file"
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file (code motion)"
+    }
+}
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak marker4" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint small function, 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" $srcfile1]
+set bp_location14 [gdb_get_line_number "set breakpoint 14 here" $srcfile1]
+
+gdb_test_multiple "continue" \
+    "run until hardware breakpoint set at small function, optimized file" {
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, $hex in marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
+	    # marker4() is defined at line 46 when compiled with -DPROTOTYPES
+	    pass "run until hardware breakpoint set at small function, optimized file (line bp_location14)"
+	}
+	-re "Breakpoint $decimal, factorial \\(.*\\) .*\{\r\n$gdb_prompt" {
+	    # GCC 4.3 emits bad line number information - see gcc/36748.
+	    if { [test_compiler_info "gcc-4-3-*"] } {
+		setup_xfail *-*-*
+	    }
+	    fail "run until hardware breakpoint set at small function, optimized file"
+	}
+}
+
+
+# Reset the default arguments for VxWorks
+if [istarget "*-*-vxworks*"] {
+    set timeout 10
+    verbose "Timeout is now $timeout seconds" 2
+    gdb_test_no_output "set args main"
+}

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Doug Evans]

Hi.  My two bit review.

On Fri, Nov 11, 2011 at 4:12 AM, Maciej W. Rozycki
<macro@codesourcery.com> wrote:
>[...]
>  The consistency of failures between i686-linux-gnu and i686-mingw makes
> me fairly sure that's a bug in x86 support in GDB of some sort rather than
> a problem with my setup -- is that a known bug?

OOC, have you tried amd64-linux?
{i386,amd64}-linux are important enough targets that I think this
should be fixed for 7.4.
Joel?

>  Questions or comments?  Otherwise OK to apply?

One high level comment:
There's a lot of basic linespec testing that would be nice to make table driven.
E.g., {hbreak,thbreak} x {function, file:lineno, ...}, followed by
delete_breakpoints.
[I might not do this when testing running to the breakpoints, dunno if
that is easily made data-driven, though that would be nice too.]
If we're going to do (semi-)exhaustive linespec testing, I'd rather
have a table than a lot of repetitive code.

The same applies to break,tbreak.
They too could be table driven.

OTOH, maybe it's better to do simple linespec testing separately.
There I would move {break,tbreak,hbreak,thbreak} x {function,
file:lineno, ...} into its own file,
have a table to drive it, and do a lot of testing with less code.
It'd be more easily extensible too.

I wouldn't make this a requirement, just mentioning it.

>
> 2011-11-11  Maciej W. Rozycki  <macro@mips.com>
>            Maciej W. Rozycki  <macro@codesourcery.com>
>            Daniel Jacobowitz  <dan@codesourcery.com>
>
>        * gdb.base/hbreak2.exp: New test for "hbreak" and "thbreak".
>
>  Maciej
>
> gdb-test-hbreak.diff
> Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
> ===================================================================
> --- /dev/null   1970-01-01 00:00:00.000000000 +0000
> +++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp      2011-11-10 13:23:14.525919195 +0000
> @@ -0,0 +1,629 @@
> +#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
> +#   2000, 2002, 2003, 2007, 2008, 2009, 2010, 2011
> +#   Free Software Foundation, Inc.

This is a new file.  I'm not sure what the rule is, but I think this
should just say 2011.

> +
> +# 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 3 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, see <http://www.gnu.org/licenses/>.
> +
> +# Based on break.exp by Rob Savoye. (rob@cygnus.com)
> +# Only one hardware breakpoint is set at a time as targets may limit
> +# the number available.
> +
> +
> +if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
> +    return -1
> +}
> +set srcfile break.c
> +set srcfile1 break1.c
> +
> +if ![runto_main] then { fail "break tests suppressed" }
> +delete_breakpoints
> +
> +#
> +# Test whether the target supports hardware breakpoints at all.
> +#
> +gdb_test_multiple "hbreak main" "hardware breakpoint support" {
> +    -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
> +       unsupported "hardware breakpoints"
> +       return
> +    }
> +    -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
> +       unsupported "hardware breakpoints"
> +       return
> +    }
> +    -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
> +       pass "hardware breakpoint support"
> +    }
> +}
> +gdb_run_cmd
> +gdb_test_multiple "" "hardware breakpoint insertion" {
> +    -re "Warning:\[\r\n\]+Cannot insert hardware breakpoint \[0-9\]+\.\[\r\n\]+Could not insert hardware breakpoints:\[\r\n\]+You may have requested too many hardware breakpoints/watchpoints\.\[\r\n\]+.*$gdb_prompt $" {
> +       unsupported "hardware breakpoint insertion"
> +       return
> +    }
> +    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:.*\[\r\n\]+.*\[\t \]+if .argc.* \{.*$gdb_prompt $" {
> +       pass "hardware breakpoint insertion"
> +    }
> +}
> +delete_breakpoints
> +
> +#
> +# Test simple hardware breakpoint setting commands.
> +#
> +
> +#
> +# Test break at function.
> +#
> +gdb_test "hbreak main" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "hardware breakpoint function"
> +delete_breakpoints
> +
> +#
> +# Test break at quoted function.
> +#
> +gdb_test "hbreak \"marker2\"" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
> +    "hardware breakpoint quoted function"
> +delete_breakpoints
> +
> +#
> +# Test break at function in file.
> +#
> +gdb_test "hbreak $srcfile:factorial" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "hardware breakpoint function in file"
> +delete_breakpoints
> +
> +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 "hbreak $bp_location1" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
> +    "hardware breakpoint line number"
> +delete_breakpoints
> +
> +set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]
> +
> +#
> +# Test break at line number in file.
> +#
> +gdb_test "hbreak $srcfile:$bp_location2" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
> +    "hardware breakpoint line number in file"
> +delete_breakpoints
> +
> +set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
> +set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]
> +
> +#
> +# Test putting a break at the start of a multi-line if conditional.
> +# Verify the breakpoint was put at the start of the conditional.
> +#
> +gdb_test "hbreak multi_line_if_conditional" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
> +    "hardware breakpoint at start of multi line if conditional"
> +delete_breakpoints
> +
> +gdb_test "hbreak multi_line_while_conditional" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
> +    "hardware 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 $bp_location6
> +}
> +
> +if {$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" $srcfile1]
> +set bp_location9 [gdb_get_line_number "set breakpoint 9 here" $srcfile1]
> +
> +gdb_test "info break" \
> +    "Num     Type\[ \]+Disp Enb Address\[ \]+What.*
> +\[0-9\]+\[\t \]+hw breakpoint  keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
> +    "hardware breakpoint info"
> +delete_breakpoints
> +
> +# 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.

I think I'd just remove the FIXME and Huh? comments.
And then ideally remove the mips-idt-* test.

> +if [istarget "mips-idt-*"] then {
> +    return
> +}
> +
> +#
> +# Run until the breakpoint at main is hit.  For non-stubs-using targets.
> +#
> +gdb_test "hbreak main" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "hardware breakpoint function (2)"
> +gdb_run_cmd
> +gdb_test "" \
> +    "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*" \
> +    "run until function breakpoint"
> +delete_breakpoints
> +
> +#
> +# Run until the breakpoint at a line number.
> +#
> +gdb_test "hbreak $bp_location1" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
> +    "hardware breakpoint line number (2)"
> +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"
> +delete_breakpoints
> +
> +#
> +# Run until the breakpoint set in a function in a file.
> +#
> +gdb_test "hbreak $srcfile:factorial" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "hardware breakpoint function in file (2)"
> +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($i) breakpoint"
> +}
> +delete_breakpoints
> +
> +#
> +# Run until the breakpoint set at a quoted function.
> +#
> +gdb_test "hbreak \"marker2\"" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
> +    "hardware breakpoint quoted function (2)"
> +gdb_test "continue" \
> +    "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*" \
> +    "run until quoted breakpoint"
> +delete_breakpoints
> +#
> +# Run until the file:function breakpoint at a line number in a file.
> +#
> +gdb_test "hbreak $srcfile:$bp_location2" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
> +    "hardware breakpoint line number in file (2)"
> +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"
> +delete_breakpoints
> +
> +# Test break at offset +1.
> +set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]
> +
> +gdb_test "hbreak +1" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
> +    "hardware 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 hardware breakpoint"
> +delete_breakpoints
> +
> +# Check to see if breakpoint can be set on ending brace of function.
> +set bp_location10a [gdb_get_line_number "set breakpoint 10a here"]
> +
> +gdb_test "hbreak $bp_location10a" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10a\\." \
> +    "setting hardware breakpoint at }"
> +
> +gdb_test "continue" \
> +    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10a.*$bp_location10a\[\t \]+}.*breakpoint 10a here.*" \
> +    "continue to hardware breakpoint at }"
> +
> +#
> +# Delete all breakpoints so we can start over, course this can be a test too.
> +#
> +delete_breakpoints
> +
> +#
> +# Test temporary breakpoint at function.
> +#
> +
> +gdb_test "thbreak main" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "temporary hardware breakpoint function"
> +delete_breakpoints
> +
> +#
> +# Test break at function in file.
> +#
> +
> +gdb_test "thbreak $srcfile:factorial" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
> +    "temporary hardware breakpoint function in file"
> +delete_breakpoints
> +
> +#
> +# Test break at line number.
> +#
> +gdb_test "thbreak $bp_location1" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1.*" \
> +    "temporary hardware breakpoint line number #1"
> +delete_breakpoints
> +
> +gdb_test "thbreak $bp_location6" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location6.*" \
> +    "temporary hardware breakpoint line number #2"
> +delete_breakpoints
> +
> +#
> +# Test break at line number in file.
> +#
> +gdb_test "thbreak $srcfile:$bp_location2" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2.*" \
> +    "temporary hardware breakpoint line number in file #1"
> +delete_breakpoints
> +
> +set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
> +gdb_test "thbreak $srcfile:$bp_location11" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location11.*" \
> +    "temporary hardware 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.*\[\r\n\]
> +\[0-9\]+\[\t \]+hw breakpoint  del.*y.*in main at .*$srcfile:$bp_location11.*" \
> +    "temporary hardware breakpoint info"
> +
> +
> +#***********
> +
> +if ![runto_main] then { fail "break tests suppressed" }
> +
> +# Verify that GDB responds gracefully when asked to set a breakpoint
> +# on a nonexistent source line.
> +#
> +gdb_test "hbreak 999" \
> +    "No line 999 in file .*" \
> +    "hardware 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 "hbreak", which is treated
> +# as the "default" breakpoint.
> +#
> +gdb_test "hbreak" "Hardware assisted breakpoint \[0-9\]*.*" \
> +    "hardware break on default location"
> +
> +# 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" }
> +
> +gdb_test_multiple "hbreak $bp_location1" \
> +    "set to-be-silent hardware break bp_location1" {
> +       -re "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $" {
> +           pass "set to-be-silent hardware break bp_location1"
> +       }
> +    }
> +
> +send_gdb "commands $expect_out(1,string)\n"
> +send_gdb "silent\n"
> +send_gdb "end\n"
> +gdb_test "" "set silent hardware break bp_location1"
> +
> +gdb_test "info break $expect_out(1,string)" \
> +    "\[0-9\]*\[ \t\]*hw breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*" \
> +    "info silent hardware break bp_location1"
> +
> +gdb_test "continue" "Continuing." \
> +    "hit silent hardware break bp_location1"
> +
> +gdb_test "bt" "#0  main .* at .*:$bp_location1.*" \
> +    "stopped for silent hardware 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"]
> +gdb_test "hbreak $bp_location12 thread 999" "Unknown thread 999.*" \
> +    "thread-specific hardware breakpoint on non-existent thread disallowed"
> +
> +gdb_test "hbreak $bp_location12 thread foo" \
> +    "Junk after thread keyword.*" \
> +    "thread-specific hardware breakpoint on bogus thread ID disallowed"
> +
> +# Verify that GDB responds gracefully to a breakpoint command with
> +# trailing garbage.
> +#
> +gdb_test "hbreak $bp_location12 foo" \
> +    "Junk at end of arguments.*" \
> +    "hardware breakpoint with trailing garbage disallowed"
> +
> +# Verify that GDB responds gracefully to a "clear" command that has
> +# no matching breakpoint.  (First, get us off the current source line,
> +# which we know has a breakpoint.)
> +#
> +gdb_test "next" "marker1.*" "step over hardware breakpoint"
> +
> +gdb_test "clear 81" "No breakpoint at 81.*" \
> +    "clear line has no breakpoint disallowed"
> +
> +gdb_test "clear" "No breakpoint at this line.*" \
> +    "clear current line has no breakpoint disallowed"
> +delete_breakpoints
> +
> +# Verify that a breakpoint can be set via a convenience variable.
> +#
> +gdb_test_no_output "set \$foo=$bp_location11" \
> +    "set convenience variable \$foo to bp_location11"
> +
> +gdb_test "hbreak \$foo" \
> +    "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location11.*" \
> +    "set hardware breakpoint via convenience variable"
> +delete_breakpoints
> +
> +# Verify that GDB responds gracefully to an attempt to set a
> +# breakpoint via a convenience variable whose type is not integer.
> +#
> +gdb_test_no_output "set \$foo=81.5" \
> +    "set convenience variable \$foo to 81.5"
> +
> +gdb_test "hbreak \$foo" \
> +    "Convenience variables used in line specs must have integer values.*" \
> +    "set hardware breakpoint via non-integer convenience variable disallowed"
> +
> +# Verify that we can set and trigger a breakpoint in a user-called function.
> +#
> +gdb_test "hbreak marker2" \
> +    "Hardware assisted breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*" \
> +    "set hardware breakpoint on to-be-called function"
> +
> +gdb_test "print marker2(99)" \
> +    "The program being debugged stopped while in a function called from GDB.\r\nEvaluation of the expression containing the function\r\n.marker2$proto. will be abandoned.\r\nWhen the function is done executing, GDB will silently stop.*" \
> +    "hit hardware 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.
> +#
> +gdb_test_multiple "bt" "backtrace while in called function" {
> +    -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 $" {
> +       pass "backtrace while in called function"
> +    }
> +}
> +
> +# Return from the called function.  For remote targets, it's important to do
> +# this before runto_main, which otherwise may silently stop on the dummy
> +# breakpoint inserted by GDB at the program's entry point.
> +#
> +gdb_test_multiple "finish" "finish from called function" {
> +    -re "Run till exit from .*marker2.* at .*($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"
> +    }
> +}
> +
> +#********
> +
> +
> +#
> +# Test "next" over recursive function call.
> +#
> +
> +proc test_next_with_recursion {} {
> +    global gdb_prompt
> +    global decimal
> +    global binfile
> +
> +    delete_breakpoints
> +
> +    # Can't set a hardware breakpoint without a live target, so do it now
> +    # before it's killed below.
> +    gdb_test "hbreak factorial" \
> +       "Hardware assisted breakpoint $decimal at .*" \
> +       "hardware break at factorial"
> +
> +    gdb_test "kill" "" "kill program" \
> +       "Kill the program being debugged.*y or n. $" "y"
> +
> +    # Run until we call factorial with 6
> +
> +    gdb_run_cmd
> +    if [gdb_test "" \
> +       "Break.* factorial .value=6. .*" \
> +       "run to factorial(6)"] then { 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

I would do:

  set oldtimeout $timeout
  if [istarget ...] {
    set timeout 60
  }

and then restore timeout at the end of the function.

> +    }
> +    # 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
> +
> +
> +#********
> +
> +# Build a new file with optimization enabled so that we can try breakpoints
> +# on targets with optimized prologues.
> +
> +if { [prepare_for_testing hbreak2.exp "breako2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
> +    return -1
> +}
> +
> +if ![runto_main] then { fail "break tests suppressed" }
> +delete_breakpoints
> +
> +#
> +# Test break at function.
> +#
> +gdb_test "hbreak main" \
> +    "Hardware assisted breakpoint.*at.* file .*, line.*" \
> +    "hardware breakpoint function, optimized file"
> +
> +#
> +# Run until the breakpoint at main is hit.  For non-stubs-using targets.
> +#
> +gdb_run_cmd
> +gdb_test_multiple "" "run until hardware function breakpoint, optimized file" {
> +    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $" {
> +       pass "run until hardware function breakpoint, optimized file"
> +    }
> +    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $" {
> +       pass "run until hardware function breakpoint, optimized file (code motion)"
> +    }
> +}
> +delete_breakpoints
> +
> +#
> +# Test break at function.
> +#
> +gdb_test "hbreak marker4" \
> +    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
> +    "hardware breakpoint small function, 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" $srcfile1]
> +set bp_location14 [gdb_get_line_number "set breakpoint 14 here" $srcfile1]
> +
> +gdb_test_multiple "continue" \
> +    "run until hardware breakpoint set at small function, optimized file" {
> +       -re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
> +           pass "run until hardware breakpoint set at small function, optimized file"
> +       }
> +       -re "Breakpoint $decimal, $hex in marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
> +           pass "run until hardware breakpoint set at small function, optimized file"
> +       }
> +       -re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
> +           # marker4() is defined at line 46 when compiled with -DPROTOTYPES
> +           pass "run until hardware breakpoint set at small function, optimized file (line bp_location14)"
> +       }
> +       -re "Breakpoint $decimal, factorial \\(.*\\) .*\{\r\n$gdb_prompt" {
> +           # GCC 4.3 emits bad line number information - see gcc/36748.
> +           if { [test_compiler_info "gcc-4-3-*"] } {
> +               setup_xfail *-*-*
> +           }
> +           fail "run until hardware breakpoint set at small function, optimized file"
> +       }
> +}
> +
> +
> +# Reset the default arguments for VxWorks
> +if [istarget "*-*-vxworks*"] {
> +    set timeout 10
> +    verbose "Timeout is now $timeout seconds" 2

Timeout restoration is no longer needed at the end of a file.
I would delete these lines.

> +    gdb_test_no_output "set args main"
> +}
>

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Joel Brobecker]

> >  The consistency of failures between i686-linux-gnu and i686-mingw makes
> > me fairly sure that's a bug in x86 support in GDB of some sort rather than
> > a problem with my setup -- is that a known bug?
> 
> OOC, have you tried amd64-linux?
> {i386,amd64}-linux are important enough targets that I think this
> should be fixed for 7.4. Joel?

I think we should look at the context as well before making a decision:
  - Is that a regression? If it's been like that in previous versions,
    then maybe it's OK for it to fail for another version...
  - Are hardware breakpoints used much? Probably by the people who
    debug programs in ROM(/flash?).

We should probably try to investigate the problem quickly and determine
the extent of the problem. If we think we should fix that for 7.4, and
someone is willing to take responsibility for it, then let's make it
a high priority item on our list, and not release 7.4 without the fix.

-- 
Joel

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Fri, 11 Nov 2011, Joel Brobecker wrote:

> > >  The consistency of failures between i686-linux-gnu and i686-mingw makes
> > > me fairly sure that's a bug in x86 support in GDB of some sort rather than
> > > a problem with my setup -- is that a known bug?
> > 
> > OOC, have you tried amd64-linux?

 I have now, and the result is the same -- breakpoints silently missed.

> > {i386,amd64}-linux are important enough targets that I think this
> > should be fixed for 7.4. Joel?
> 
> I think we should look at the context as well before making a decision:
>   - Is that a regression? If it's been like that in previous versions,
>     then maybe it's OK for it to fail for another version...
>   - Are hardware breakpoints used much? Probably by the people who
>     debug programs in ROM(/flash?).

 Plus self-modifying code -- although I agree that hardware execution 
breakpoints are less needed with hosted systems than on bare-iron targets.  

> We should probably try to investigate the problem quickly and determine
> the extent of the problem. If we think we should fix that for 7.4, and
> someone is willing to take responsibility for it, then let's make it
> a high priority item on our list, and not release 7.4 without the fix.

 Well, OOC I looked at the relevant source and the bug (and consequently 
the fix) is quite obvious -- the debug registers are only actually pushed 
down to hardware on data breakpoints and not execution breakpoints.

 I'm getting:

# of expected passes            69
# of unexpected failures        2

now on x86_64-linux-gnu.  The two failures are some matching problems with 
my hbreak2.exp -- I'll investigate and fix that up.

 OK to apply?

2011-11-11  Maciej W. Rozycki  <macro@codesourcery.com>

	gdb/
	* i386-nat.c (i386_insert_hw_breakpoint): Call 
	i386_update_inferior_debug_regs.
	(i386_remove_hw_breakpoint): Likewise.

  Maciej

gdb-i386-hbreak.diff
Index: gdb-fsf-trunk-quilt/gdb/i386-nat.c
===================================================================
--- gdb-fsf-trunk-quilt.orig/gdb/i386-nat.c	2011-11-07 13:42:30.000000000 +0000
+++ gdb-fsf-trunk-quilt/gdb/i386-nat.c	2011-11-11 18:47:40.555635587 +0000
@@ -684,9 +684,15 @@ i386_insert_hw_breakpoint (struct gdbarc
 {
   unsigned len_rw = i386_length_and_rw_bits (1, hw_execute);
   CORE_ADDR addr = bp_tgt->placed_address;
-  int retval = i386_insert_aligned_watchpoint (&dr_mirror,
+  /* Work on a local copy of the debug registers, and on success,
+     commit the change back to the inferior.  */
+  struct i386_debug_reg_state local_state = dr_mirror;
+  int retval = i386_insert_aligned_watchpoint (&local_state,
 					       addr, len_rw) ? EBUSY : 0;
 
+  if (retval == 0)
+    i386_update_inferior_debug_regs (&local_state);
+
   if (maint_show_dr)
     i386_show_dr (&dr_mirror, "insert_hwbp", addr, 1, hw_execute);
 
@@ -702,9 +708,15 @@ i386_remove_hw_breakpoint (struct gdbarc
 {
   unsigned len_rw = i386_length_and_rw_bits (1, hw_execute);
   CORE_ADDR addr = bp_tgt->placed_address;
-  int retval = i386_remove_aligned_watchpoint (&dr_mirror,
+  /* Work on a local copy of the debug registers, and on success,
+     commit the change back to the inferior.  */
+  struct i386_debug_reg_state local_state = dr_mirror;
+  int retval = i386_remove_aligned_watchpoint (&local_state,
 					       addr, len_rw);
 
+  if (retval == 0)
+    i386_update_inferior_debug_regs (&local_state);
+
   if (maint_show_dr)
     i386_show_dr (&dr_mirror, "remove_hwbp", addr, 1, hw_execute);
 

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Pedro Alves]

On Friday 11 November 2011 18:59:10, Maciej W. Rozycki wrote:
>  Well, OOC I looked at the relevant source and the bug (and consequently 
> the fix) is quite obvious -- the debug registers are only actually pushed 
> down to hardware on data breakpoints and not execution breakpoints.

Ugh, my fault...

> 2011-11-11  Maciej W. Rozycki  <macro@codesourcery.com>
> 
> 	gdb/
> 	* i386-nat.c (i386_insert_hw_breakpoint): Call 
> 	i386_update_inferior_debug_regs.
> 	(i386_remove_hw_breakpoint): Likewise.

Okay.  Thanks!

-- 
Pedro Alves

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Sat, 12 Nov 2011, Pedro Alves wrote:

> >  Well, OOC I looked at the relevant source and the bug (and consequently 
> > the fix) is quite obvious -- the debug registers are only actually pushed 
> > down to hardware on data breakpoints and not execution breakpoints.
> 
> Ugh, my fault...

 Well, that happens sometimes and it just shows we do really need this 
test case. :)

 NB I have deliberately omitted the skip_hw_breakpoint_tests check here as 
I think we should test the failure modes too.  If a target does not 
support hardware breakpoints, then it should handle any requests to use 
one gracefully, which this test case will record as unsupported.  If any 
targets still score failures here, then they need to be investigated.

> > 2011-11-11  Maciej W. Rozycki  <macro@codesourcery.com>
> > 
> > 	gdb/
> > 	* i386-nat.c (i386_insert_hw_breakpoint): Call 
> > 	i386_update_inferior_debug_regs.
> > 	(i386_remove_hw_breakpoint): Likewise.
> 
> Okay.  Thanks!

 Committed now, thanks for the review.

  Maciej

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Fri, 11 Nov 2011, Doug Evans wrote:

> {i386,amd64}-linux are important enough targets that I think this
> should be fixed for 7.4.

 OK, so we've got this part sorted out -- and I verified the two failures 
observed are also present with break.exp, so they're a generic 
x86_64-linux-gnu problem and nothing wrong with this test.

> One high level comment:
> There's a lot of basic linespec testing that would be nice to make table driven.
> E.g., {hbreak,thbreak} x {function, file:lineno, ...}, followed by
> delete_breakpoints.
> [I might not do this when testing running to the breakpoints, dunno if
> that is easily made data-driven, though that would be nice too.]
> If we're going to do (semi-)exhaustive linespec testing, I'd rather
> have a table than a lot of repetitive code.
> 
> The same applies to break,tbreak.
> They too could be table driven.
> 
> OTOH, maybe it's better to do simple linespec testing separately.
> There I would move {break,tbreak,hbreak,thbreak} x {function,
> file:lineno, ...} into its own file,
> have a table to drive it, and do a lot of testing with less code.
> It'd be more easily extensible too.
> 
> I wouldn't make this a requirement, just mentioning it.

 I agree lots of older test cases ask for cleaning up -- I refrained from 
diverging from break.exp too much, although I did some obvious tidy-ups, 
like removing the remaining send_gdb/gdb_expect pieces in favour to 
gdb_test/gdb_test_multiple.  Interestingly enough someone did that to 
break.exp too at one point, but didn't catch all the places.  I think 
it'll make sense to complete it using hbreak2.exp as a reference for a 
change.  I can't commit to doing it now, but I'll try as my time permits.

> > @@ -0,0 +1,629 @@
> > +#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
> > +#   2000, 2002, 2003, 2007, 2008, 2009, 2010, 2011
> > +#   Free Software Foundation, Inc.
> 
> This is a new file.  I'm not sure what the rule is, but I think this
> should just say 2011.

 I wasn't sure -- the contents of the file certainly are not new, as this 
is break.exp taken as a whole and then edited.  Not even quite heavily -- 
you can run a diff or compare the files side by side and they are still 
pretty close to each other.  I have removed the older years now; they can 
be still added back if needed.

> > +# 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.
> 
> I think I'd just remove the FIXME and Huh? comments.
> And then ideally remove the mips-idt-* test.

 Again, this was taken from break.exp verbatim.  I have removed it now as 
it seems to make little sense indeed, but in this case I think the former 
should be updated accordingly.

> > +    if [istarget "mips*tx39-*"] {
> > +       set timeout 60
> 
> I would do:
> 
>   set oldtimeout $timeout
>   if [istarget ...] {
>     set timeout 60
>   }
> 
> and then restore timeout at the end of the function.

 Hmm, the TX39 is an R3000 clone and probably does not have hardware 
breakpoints in the first place. ;)  So this is another break.exp 
compatibility place -- I'm leaving it in place for this reason.

 As to your observation -- does it really matter?  The timeout variable is 
local to this function (accessed with upvar from gdb_test), so it's wiped 
away at the point the procedure returns.  I'm therefore leaving it as it 
is unless you have further notes.

> > +# Reset the default arguments for VxWorks
> > +if [istarget "*-*-vxworks*"] {
> > +    set timeout 10
> > +    verbose "Timeout is now $timeout seconds" 2
> 
> Timeout restoration is no longer needed at the end of a file.
> I would delete these lines.
> 
> > +    gdb_test_no_output "set args main"
> > +}

 Hmm, this actually looks like a blind copy&paste from gdb.base/a2-run.exp 
with the timeout bit added later on.  I have therefore removed the whole 
conditional -- it's not like we change the arguments anywhere here.

 Here's an updated version -- any further comments?

2011-11-11  Maciej W. Rozycki  <macro@mips.com>
            Maciej W. Rozycki  <macro@codesourcery.com>
            Daniel Jacobowitz  <dan@codesourcery.com>

	* gdb.base/hbreak2.exp: New test for "hbreak" and "thbreak".

  Maciej

gdb-test-hbreak.diff
Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2011-11-11 19:47:24.495606633 +0000
@@ -0,0 +1,612 @@
+#   Copyright 2011 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 3 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, see <http://www.gnu.org/licenses/>.
+
+# Based on break.exp by Rob Savoye. (rob@cygnus.com)
+# Hardware breakpoint support by Maciej W. Rozycki and Daniel Jacobowitz.
+# Only one hardware breakpoint is set at a time as targets may limit
+# the number available.
+
+
+if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
+    return -1
+}
+set srcfile break.c
+set srcfile1 break1.c
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test whether the target supports hardware breakpoints at all.
+#
+gdb_test_multiple "hbreak main" "hardware breakpoint support" {
+    -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
+	pass "hardware breakpoint support"
+    }
+}
+gdb_run_cmd
+gdb_test_multiple "" "hardware breakpoint insertion" {
+    -re "Warning:\[\r\n\]+Cannot insert hardware breakpoint \[0-9\]+\.\[\r\n\]+Could not insert hardware breakpoints:\[\r\n\]+You may have requested too many hardware breakpoints/watchpoints\.\[\r\n\]+.*$gdb_prompt $" {
+	unsupported "hardware breakpoint insertion"
+	return
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:.*\[\r\n\]+.*\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "hardware breakpoint insertion"
+    }
+}
+delete_breakpoints
+
+#
+# Test simple hardware breakpoint setting commands.
+#
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file"
+delete_breakpoints
+
+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 "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number"
+delete_breakpoints
+
+set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]
+
+#
+# Test break at line number in file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file"
+delete_breakpoints
+
+set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
+set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]
+
+#
+# Test putting a break at the start of a multi-line if conditional.
+# Verify the breakpoint was put at the start of the conditional.
+#
+gdb_test "hbreak multi_line_if_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
+    "hardware breakpoint at start of multi line if conditional"
+delete_breakpoints
+
+gdb_test "hbreak multi_line_while_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
+    "hardware 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 $bp_location6
+}
+
+if {$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" $srcfile1]
+set bp_location9 [gdb_get_line_number "set breakpoint 9 here" $srcfile1]
+
+gdb_test "info break" \
+    "Num     Type\[ \]+Disp Enb Address\[ \]+What.*
+\[0-9\]+\[\t \]+hw breakpoint  keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
+    "hardware breakpoint info"
+delete_breakpoints
+
+#
+# Run until the breakpoint at main is hit.  For non-stubs-using targets.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function (2)"
+gdb_run_cmd
+gdb_test "" \
+    "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*" \
+    "run until function breakpoint"
+delete_breakpoints
+
+#
+# Run until the breakpoint at a line number.
+#
+gdb_test "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number (2)"
+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"
+delete_breakpoints
+
+#
+# Run until the breakpoint set in a function in a file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file (2)"
+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($i) breakpoint"
+}
+delete_breakpoints
+
+#
+# Run until the breakpoint set at a quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function (2)"
+gdb_test "continue" \
+    "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*" \
+    "run until quoted breakpoint"
+delete_breakpoints
+#
+# Run until the file:function breakpoint at a line number in a file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file (2)"
+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"
+delete_breakpoints
+
+# Test break at offset +1.
+set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]
+
+gdb_test "hbreak +1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
+    "hardware 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 hardware breakpoint"
+delete_breakpoints
+
+# Check to see if breakpoint can be set on ending brace of function.
+set bp_location10a [gdb_get_line_number "set breakpoint 10a here"]
+
+gdb_test "hbreak $bp_location10a" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10a\\." \
+    "setting hardware breakpoint at }"
+
+gdb_test "continue" \
+    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10a.*$bp_location10a\[\t \]+}.*breakpoint 10a here.*" \
+    "continue to hardware breakpoint at }"
+
+#
+# Delete all breakpoints so we can start over, course this can be a test too.
+#
+delete_breakpoints
+
+#
+# Test temporary breakpoint at function.
+#
+
+gdb_test "thbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+
+gdb_test "thbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function in file"
+delete_breakpoints
+
+#
+# Test break at line number.
+#
+gdb_test "thbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1.*" \
+    "temporary hardware breakpoint line number #1"
+delete_breakpoints
+
+gdb_test "thbreak $bp_location6" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location6.*" \
+    "temporary hardware breakpoint line number #2"
+delete_breakpoints
+
+#
+# Test break at line number in file.
+#
+gdb_test "thbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2.*" \
+    "temporary hardware breakpoint line number in file #1"
+delete_breakpoints
+
+set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
+gdb_test "thbreak $srcfile:$bp_location11" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location11.*" \
+    "temporary hardware 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.*\[\r\n\]
+\[0-9\]+\[\t \]+hw breakpoint  del.*y.*in main at .*$srcfile:$bp_location11.*" \
+    "temporary hardware breakpoint info"
+
+
+#***********
+
+if ![runto_main] then { fail "break tests suppressed" }
+
+# Verify that GDB responds gracefully when asked to set a breakpoint
+# on a nonexistent source line.
+#
+gdb_test "hbreak 999" \
+    "No line 999 in file .*" \
+    "hardware 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 "hbreak", which is treated
+# as the "default" breakpoint.
+#
+gdb_test "hbreak" "Hardware assisted breakpoint \[0-9\]*.*" \
+    "hardware break on default location"
+
+# 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" }
+
+gdb_test_multiple "hbreak $bp_location1" \
+    "set to-be-silent hardware break bp_location1" {
+	-re "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $" {
+	    pass "set to-be-silent hardware break bp_location1"
+	}
+    }
+
+send_gdb "commands $expect_out(1,string)\n"
+send_gdb "silent\n"
+send_gdb "end\n"
+gdb_test "" "set silent hardware break bp_location1"
+
+gdb_test "info break $expect_out(1,string)" \
+    "\[0-9\]*\[ \t\]*hw breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*" \
+    "info silent hardware break bp_location1"
+
+gdb_test "continue" "Continuing." \
+    "hit silent hardware break bp_location1"
+
+gdb_test "bt" "#0  main .* at .*:$bp_location1.*" \
+    "stopped for silent hardware 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"]
+gdb_test "hbreak $bp_location12 thread 999" "Unknown thread 999.*" \
+    "thread-specific hardware breakpoint on non-existent thread disallowed"
+
+gdb_test "hbreak $bp_location12 thread foo" \
+    "Junk after thread keyword.*" \
+    "thread-specific hardware breakpoint on bogus thread ID disallowed"
+
+# Verify that GDB responds gracefully to a breakpoint command with
+# trailing garbage.
+#
+gdb_test "hbreak $bp_location12 foo" \
+    "Junk at end of arguments.*" \
+    "hardware breakpoint with trailing garbage disallowed"
+
+# Verify that GDB responds gracefully to a "clear" command that has
+# no matching breakpoint.  (First, get us off the current source line,
+# which we know has a breakpoint.)
+#
+gdb_test "next" "marker1.*" "step over hardware breakpoint"
+
+gdb_test "clear 81" "No breakpoint at 81.*" \
+    "clear line has no breakpoint disallowed"
+
+gdb_test "clear" "No breakpoint at this line.*" \
+    "clear current line has no breakpoint disallowed"
+delete_breakpoints
+
+# Verify that a breakpoint can be set via a convenience variable.
+#
+gdb_test_no_output "set \$foo=$bp_location11" \
+    "set convenience variable \$foo to bp_location11"
+
+gdb_test "hbreak \$foo" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location11.*" \
+    "set hardware breakpoint via convenience variable"
+delete_breakpoints
+
+# Verify that GDB responds gracefully to an attempt to set a
+# breakpoint via a convenience variable whose type is not integer.
+#
+gdb_test_no_output "set \$foo=81.5" \
+    "set convenience variable \$foo to 81.5"
+
+gdb_test "hbreak \$foo" \
+    "Convenience variables used in line specs must have integer values.*" \
+    "set hardware breakpoint via non-integer convenience variable disallowed"
+
+# Verify that we can set and trigger a breakpoint in a user-called function.
+#
+gdb_test "hbreak marker2" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*" \
+    "set hardware breakpoint on to-be-called function"
+
+gdb_test "print marker2(99)" \
+    "The program being debugged stopped while in a function called from GDB.\r\nEvaluation of the expression containing the function\r\n.marker2$proto. will be abandoned.\r\nWhen the function is done executing, GDB will silently stop.*" \
+    "hit hardware 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.
+#
+gdb_test_multiple "bt" "backtrace while in called function" {
+    -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 $" {
+	pass "backtrace while in called function"
+    }
+}
+
+# Return from the called function.  For remote targets, it's important to do
+# this before runto_main, which otherwise may silently stop on the dummy
+# breakpoint inserted by GDB at the program's entry point.
+#
+gdb_test_multiple "finish" "finish from called function" {
+    -re "Run till exit from .*marker2.* at .*($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"
+    }
+}
+
+#********
+
+
+#
+# Test "next" over recursive function call.
+#
+
+proc test_next_with_recursion {} {
+    global gdb_prompt
+    global decimal
+    global binfile
+
+    delete_breakpoints
+
+    # Can't set a hardware breakpoint without a live target, so do it now
+    # before it's killed below.
+    gdb_test "hbreak factorial" \
+	"Hardware assisted breakpoint $decimal at .*" \
+	"hardware break at factorial"
+
+    gdb_test "kill" "" "kill program" \
+	"Kill the program being debugged.*y or n. $" "y"
+
+    # Run until we call factorial with 6
+
+    gdb_run_cmd
+    if [gdb_test "" \
+	"Break.* factorial .value=6. .*" \
+	"run to factorial(6)"] then { 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
+
+
+#********
+
+# Build a new file with optimization enabled so that we can try breakpoints
+# on targets with optimized prologues.
+
+if { [prepare_for_testing hbreak2.exp "breako2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
+    return -1
+}
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*, line.*" \
+    "hardware breakpoint function, optimized file"
+
+#
+# Run until the breakpoint at main is hit.  For non-stubs-using targets.
+#
+gdb_run_cmd
+gdb_test_multiple "" "run until hardware function breakpoint, optimized file" {
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file"
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file (code motion)"
+    }
+}
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak marker4" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint small function, 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" $srcfile1]
+set bp_location14 [gdb_get_line_number "set breakpoint 14 here" $srcfile1]
+
+gdb_test_multiple "continue" \
+    "run until hardware breakpoint set at small function, optimized file" {
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, $hex in marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
+	    # marker4() is defined at line 46 when compiled with -DPROTOTYPES
+	    pass "run until hardware breakpoint set at small function, optimized file (line bp_location14)"
+	}
+	-re "Breakpoint $decimal, factorial \\(.*\\) .*\{\r\n$gdb_prompt" {
+	    # GCC 4.3 emits bad line number information - see gcc/36748.
+	    if { [test_compiler_info "gcc-4-3-*"] } {
+		setup_xfail *-*-*
+	    }
+	    fail "run until hardware breakpoint set at small function, optimized file"
+	}
+}

Re: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Fri, 11 Nov 2011, Maciej W. Rozycki wrote:

> On Fri, 11 Nov 2011, Doug Evans wrote:
> 
> > {i386,amd64}-linux are important enough targets that I think this
> > should be fixed for 7.4.
> 
>  OK, so we've got this part sorted out -- and I verified the two failures 
> observed are also present with break.exp, so they're a generic 
> x86_64-linux-gnu problem and nothing wrong with this test.
> 
> > One high level comment:
> > There's a lot of basic linespec testing that would be nice to make table driven.
> > E.g., {hbreak,thbreak} x {function, file:lineno, ...}, followed by
> > delete_breakpoints.
> > [I might not do this when testing running to the breakpoints, dunno if
> > that is easily made data-driven, though that would be nice too.]
> > If we're going to do (semi-)exhaustive linespec testing, I'd rather
> > have a table than a lot of repetitive code.
> > 
> > The same applies to break,tbreak.
> > They too could be table driven.
> > 
> > OTOH, maybe it's better to do simple linespec testing separately.
> > There I would move {break,tbreak,hbreak,thbreak} x {function,
> > file:lineno, ...} into its own file,
> > have a table to drive it, and do a lot of testing with less code.
> > It'd be more easily extensible too.
> > 
> > I wouldn't make this a requirement, just mentioning it.
> 
>  I agree lots of older test cases ask for cleaning up -- I refrained from 
> diverging from break.exp too much, although I did some obvious tidy-ups, 
> like removing the remaining send_gdb/gdb_expect pieces in favour to 
> gdb_test/gdb_test_multiple.  Interestingly enough someone did that to 
> break.exp too at one point, but didn't catch all the places.  I think 
> it'll make sense to complete it using hbreak2.exp as a reference for a 
> change.  I can't commit to doing it now, but I'll try as my time permits.
> 
> > > @@ -0,0 +1,629 @@
> > > +#   Copyright 1988, 1990, 1991, 1992, 1994, 1995, 1996, 1997, 1998, 1999,
> > > +#   2000, 2002, 2003, 2007, 2008, 2009, 2010, 2011
> > > +#   Free Software Foundation, Inc.
> > 
> > This is a new file.  I'm not sure what the rule is, but I think this
> > should just say 2011.
> 
>  I wasn't sure -- the contents of the file certainly are not new, as this 
> is break.exp taken as a whole and then edited.  Not even quite heavily -- 
> you can run a diff or compare the files side by side and they are still 
> pretty close to each other.  I have removed the older years now; they can 
> be still added back if needed.
> 
> > > +# 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.
> > 
> > I think I'd just remove the FIXME and Huh? comments.
> > And then ideally remove the mips-idt-* test.
> 
>  Again, this was taken from break.exp verbatim.  I have removed it now as 
> it seems to make little sense indeed, but in this case I think the former 
> should be updated accordingly.
> 
> > > +    if [istarget "mips*tx39-*"] {
> > > +       set timeout 60
> > 
> > I would do:
> > 
> >   set oldtimeout $timeout
> >   if [istarget ...] {
> >     set timeout 60
> >   }
> > 
> > and then restore timeout at the end of the function.
> 
>  Hmm, the TX39 is an R3000 clone and probably does not have hardware 
> breakpoints in the first place. ;)  So this is another break.exp 
> compatibility place -- I'm leaving it in place for this reason.
> 
>  As to your observation -- does it really matter?  The timeout variable is 
> local to this function (accessed with upvar from gdb_test), so it's wiped 
> away at the point the procedure returns.  I'm therefore leaving it as it 
> is unless you have further notes.
> 
> > > +# Reset the default arguments for VxWorks
> > > +if [istarget "*-*-vxworks*"] {
> > > +    set timeout 10
> > > +    verbose "Timeout is now $timeout seconds" 2
> > 
> > Timeout restoration is no longer needed at the end of a file.
> > I would delete these lines.
> > 
> > > +    gdb_test_no_output "set args main"
> > > +}
> 
>  Hmm, this actually looks like a blind copy&paste from gdb.base/a2-run.exp 
> with the timeout bit added later on.  I have therefore removed the whole 
> conditional -- it's not like we change the arguments anywhere here.
> 
>  Here's an updated version -- any further comments?

 The update below, taken from break.exp, is required to fix:

(gdb) hbreak 999
No line 999 in the current file.
Make hw breakpoint pending on future shared library load? (y or [n]) n
(gdb) FAIL: gdb.base/hbreak2.exp: hardware break on non-existent source line (got interactive prompt)

on some platforms.  I have integrated it into my patch now.

 Any progress with this test case otherwise?

  Maciej

Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- gdb-fsf-trunk-quilt.orig/gdb/testsuite/gdb.base/hbreak2.exp	2011-12-08 15:28:05.585562274 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2011-12-08 15:04:31.305404817 +0000
@@ -304,6 +304,7 @@ if ![runto_main] then { fail "break test
 # Verify that GDB responds gracefully when asked to set a breakpoint
 # on a nonexistent source line.
 #
+gdb_test_no_output "set breakpoint pending off"
 gdb_test "hbreak 999" \
     "No line 999 in file .*" \
     "hardware break on non-existent source line"

Ping: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

Hi,

 Where are we with this proposal of mine:

http://sourceware.org/ml/gdb-patches/2011-11/msg00316.html

?  Here's another update to add changes corresponding to recent 
adjustments to break.exp and to accomodate the year roll-over:

Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- gdb-fsf-trunk-quilt.orig/gdb/testsuite/gdb.base/hbreak2.exp	2012-02-25 00:52:04.975561097 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2012-02-25 00:31:13.935653572 +0000
@@ -1,4 +1,4 @@
-#   Copyright 2011 Free Software Foundation, Inc.
+#   Copyright 2012 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
@@ -132,14 +132,7 @@ gdb_test "hbreak multi_line_while_condit
 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 main_line $bp_location6
 
 if {$hp_aCC_compiler} {
     set proto "\\(int\\)"
@@ -306,7 +299,7 @@ if ![runto_main] then { fail "break test
 #
 gdb_test_no_output "set breakpoint pending off"
 gdb_test "hbreak 999" \
-    "No line 999 in file .*" \
+    "No line 999 in the current file." \
     "hardware break on non-existent source line"
 
 # Run to the desired default location.  If not positioned here, the
@@ -334,10 +327,7 @@ gdb_test_multiple "hbreak $bp_location1"
 	}
     }
 
-send_gdb "commands $expect_out(1,string)\n"
-send_gdb "silent\n"
-send_gdb "end\n"
-gdb_test "" "set silent hardware break bp_location1"
+gdb_test "commands $expect_out(1,string)\nsilent\nend" ">end" "set silent break bp_location1"
 
 gdb_test "info break $expect_out(1,string)" \
     "\[0-9\]*\[ \t\]*hw breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*" \

It looks to me bp_location5 is not used in these two tests anymore, so 
perhaps the command to set it can be removed now.

 A regenerated current version of the original patch follows.

  Maciej

2012-02-27  Maciej W. Rozycki  <macro@mips.com>
            Maciej W. Rozycki  <macro@codesourcery.com>
            Daniel Jacobowitz  <dan@codesourcery.com>

	* gdb.base/hbreak2.exp: New test for "hbreak" and "thbreak".

gdb-test-hbreak.diff
Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- /dev/null	1970-01-01 00:00:00.000000000 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2012-02-25 00:31:13.935653572 +0000
@@ -0,0 +1,603 @@
+#   Copyright 2012 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 3 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, see <http://www.gnu.org/licenses/>.
+
+# Based on break.exp by Rob Savoye. (rob@cygnus.com)
+# Hardware breakpoint support by Maciej W. Rozycki and Daniel Jacobowitz.
+# Only one hardware breakpoint is set at a time as targets may limit
+# the number available.
+
+
+if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
+    return -1
+}
+set srcfile break.c
+set srcfile1 break1.c
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test whether the target supports hardware breakpoints at all.
+#
+gdb_test_multiple "hbreak main" "hardware breakpoint support" {
+    -re "No hardware breakpoint support in the target.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware breakpoints used exceeds limit.*$gdb_prompt $" {
+	unsupported "hardware breakpoints"
+	return
+    }
+    -re "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*$gdb_prompt $" {
+	pass "hardware breakpoint support"
+    }
+}
+gdb_run_cmd
+gdb_test_multiple "" "hardware breakpoint insertion" {
+    -re "Warning:\[\r\n\]+Cannot insert hardware breakpoint \[0-9\]+\.\[\r\n\]+Could not insert hardware breakpoints:\[\r\n\]+You may have requested too many hardware breakpoints/watchpoints\.\[\r\n\]+.*$gdb_prompt $" {
+	unsupported "hardware breakpoint insertion"
+	return
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:.*\[\r\n\]+.*\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "hardware breakpoint insertion"
+    }
+}
+delete_breakpoints
+
+#
+# Test simple hardware breakpoint setting commands.
+#
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file"
+delete_breakpoints
+
+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 "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number"
+delete_breakpoints
+
+set bp_location2 [gdb_get_line_number "set breakpoint 2 here"]
+
+#
+# Test break at line number in file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file"
+delete_breakpoints
+
+set bp_location3 [gdb_get_line_number "set breakpoint 3 here"]
+set bp_location4 [gdb_get_line_number "set breakpoint 4 here"]
+
+#
+# Test putting a break at the start of a multi-line if conditional.
+# Verify the breakpoint was put at the start of the conditional.
+#
+gdb_test "hbreak multi_line_if_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location3\\." \
+    "hardware breakpoint at start of multi line if conditional"
+delete_breakpoints
+
+gdb_test "hbreak multi_line_while_conditional" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location4\\." \
+    "hardware 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"]
+
+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" $srcfile1]
+set bp_location9 [gdb_get_line_number "set breakpoint 9 here" $srcfile1]
+
+gdb_test "info break" \
+    "Num     Type\[ \]+Disp Enb Address\[ \]+What.*
+\[0-9\]+\[\t \]+hw breakpoint  keep y.* in multi_line_while_conditional at .*$srcfile:$bp_location4" \
+    "hardware breakpoint info"
+delete_breakpoints
+
+#
+# Run until the breakpoint at main is hit.  For non-stubs-using targets.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function (2)"
+gdb_run_cmd
+gdb_test "" \
+    "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*" \
+    "run until function breakpoint"
+delete_breakpoints
+
+#
+# Run until the breakpoint at a line number.
+#
+gdb_test "hbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1\\." \
+    "hardware breakpoint line number (2)"
+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"
+delete_breakpoints
+
+#
+# Run until the breakpoint set in a function in a file.
+#
+gdb_test "hbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "hardware breakpoint function in file (2)"
+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($i) breakpoint"
+}
+delete_breakpoints
+
+#
+# Run until the breakpoint set at a quoted function.
+#
+gdb_test "hbreak \"marker2\"" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint quoted function (2)"
+gdb_test "continue" \
+    "Continuing\\..*Breakpoint \[0-9\]+, (0x\[0-9a-f\]+ in )?marker2 \\(a=43\\) at .*$srcfile1:($bp_location8|$bp_location9).*" \
+    "run until quoted breakpoint"
+delete_breakpoints
+#
+# Run until the file:function breakpoint at a line number in a file.
+#
+gdb_test "hbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2\\." \
+    "hardware breakpoint line number in file (2)"
+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"
+delete_breakpoints
+
+# Test break at offset +1.
+set bp_location10 [gdb_get_line_number "set breakpoint 10 here"]
+
+gdb_test "hbreak +1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10\\." \
+    "hardware 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 hardware breakpoint"
+delete_breakpoints
+
+# Check to see if breakpoint can be set on ending brace of function.
+set bp_location10a [gdb_get_line_number "set breakpoint 10a here"]
+
+gdb_test "hbreak $bp_location10a" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location10a\\." \
+    "setting hardware breakpoint at }"
+
+gdb_test "continue" \
+    ".*Breakpoint \[0-9\]+, main \\(argc=.*, argv=.*, envp=.*\\) at .*$srcfile:$bp_location10a.*$bp_location10a\[\t \]+}.*breakpoint 10a here.*" \
+    "continue to hardware breakpoint at }"
+
+#
+# Delete all breakpoints so we can start over, course this can be a test too.
+#
+delete_breakpoints
+
+#
+# Test temporary breakpoint at function.
+#
+
+gdb_test "thbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function"
+delete_breakpoints
+
+#
+# Test break at function in file.
+#
+
+gdb_test "thbreak $srcfile:factorial" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line.*" \
+    "temporary hardware breakpoint function in file"
+delete_breakpoints
+
+#
+# Test break at line number.
+#
+gdb_test "thbreak $bp_location1" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location1.*" \
+    "temporary hardware breakpoint line number #1"
+delete_breakpoints
+
+gdb_test "thbreak $bp_location6" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location6.*" \
+    "temporary hardware breakpoint line number #2"
+delete_breakpoints
+
+#
+# Test break at line number in file.
+#
+gdb_test "thbreak $srcfile:$bp_location2" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location2.*" \
+    "temporary hardware breakpoint line number in file #1"
+delete_breakpoints
+
+set bp_location11 [gdb_get_line_number "set breakpoint 11 here"]
+gdb_test "thbreak $srcfile:$bp_location11" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile, line $bp_location11.*" \
+    "temporary hardware 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.*\[\r\n\]
+\[0-9\]+\[\t \]+hw breakpoint  del.*y.*in main at .*$srcfile:$bp_location11.*" \
+    "temporary hardware breakpoint info"
+
+
+#***********
+
+if ![runto_main] then { fail "break tests suppressed" }
+
+# Verify that GDB responds gracefully when asked to set a breakpoint
+# on a nonexistent source line.
+#
+gdb_test_no_output "set breakpoint pending off"
+gdb_test "hbreak 999" \
+    "No line 999 in the current file." \
+    "hardware 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 "hbreak", which is treated
+# as the "default" breakpoint.
+#
+gdb_test "hbreak" "Hardware assisted breakpoint \[0-9\]*.*" \
+    "hardware break on default location"
+
+# 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" }
+
+gdb_test_multiple "hbreak $bp_location1" \
+    "set to-be-silent hardware break bp_location1" {
+	-re "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location1.*$gdb_prompt $" {
+	    pass "set to-be-silent hardware break bp_location1"
+	}
+    }
+
+gdb_test "commands $expect_out(1,string)\nsilent\nend" ">end" "set silent break bp_location1"
+
+gdb_test "info break $expect_out(1,string)" \
+    "\[0-9\]*\[ \t\]*hw breakpoint.*:$bp_location1\r\n\[ \t\]*silent.*" \
+    "info silent hardware break bp_location1"
+
+gdb_test "continue" "Continuing." \
+    "hit silent hardware break bp_location1"
+
+gdb_test "bt" "#0  main .* at .*:$bp_location1.*" \
+    "stopped for silent hardware 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"]
+gdb_test "hbreak $bp_location12 thread 999" "Unknown thread 999.*" \
+    "thread-specific hardware breakpoint on non-existent thread disallowed"
+
+gdb_test "hbreak $bp_location12 thread foo" \
+    "Junk after thread keyword.*" \
+    "thread-specific hardware breakpoint on bogus thread ID disallowed"
+
+# Verify that GDB responds gracefully to a breakpoint command with
+# trailing garbage.
+#
+gdb_test "hbreak $bp_location12 foo" \
+    "Junk at end of arguments.*" \
+    "hardware breakpoint with trailing garbage disallowed"
+
+# Verify that GDB responds gracefully to a "clear" command that has
+# no matching breakpoint.  (First, get us off the current source line,
+# which we know has a breakpoint.)
+#
+gdb_test "next" "marker1.*" "step over hardware breakpoint"
+
+gdb_test "clear 81" "No breakpoint at 81.*" \
+    "clear line has no breakpoint disallowed"
+
+gdb_test "clear" "No breakpoint at this line.*" \
+    "clear current line has no breakpoint disallowed"
+delete_breakpoints
+
+# Verify that a breakpoint can be set via a convenience variable.
+#
+gdb_test_no_output "set \$foo=$bp_location11" \
+    "set convenience variable \$foo to bp_location11"
+
+gdb_test "hbreak \$foo" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line $bp_location11.*" \
+    "set hardware breakpoint via convenience variable"
+delete_breakpoints
+
+# Verify that GDB responds gracefully to an attempt to set a
+# breakpoint via a convenience variable whose type is not integer.
+#
+gdb_test_no_output "set \$foo=81.5" \
+    "set convenience variable \$foo to 81.5"
+
+gdb_test "hbreak \$foo" \
+    "Convenience variables used in line specs must have integer values.*" \
+    "set hardware breakpoint via non-integer convenience variable disallowed"
+
+# Verify that we can set and trigger a breakpoint in a user-called function.
+#
+gdb_test "hbreak marker2" \
+    "Hardware assisted breakpoint (\[0-9\]*) at .*, line ($bp_location8|$bp_location9).*" \
+    "set hardware breakpoint on to-be-called function"
+
+gdb_test "print marker2(99)" \
+    "The program being debugged stopped while in a function called from GDB.\r\nEvaluation of the expression containing the function\r\n.marker2$proto. will be abandoned.\r\nWhen the function is done executing, GDB will silently stop.*" \
+    "hit hardware 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.
+#
+gdb_test_multiple "bt" "backtrace while in called function" {
+    -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 $" {
+	pass "backtrace while in called function"
+    }
+}
+
+# Return from the called function.  For remote targets, it's important to do
+# this before runto_main, which otherwise may silently stop on the dummy
+# breakpoint inserted by GDB at the program's entry point.
+#
+gdb_test_multiple "finish" "finish from called function" {
+    -re "Run till exit from .*marker2.* at .*($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"
+    }
+}
+
+#********
+
+
+#
+# Test "next" over recursive function call.
+#
+
+proc test_next_with_recursion {} {
+    global gdb_prompt
+    global decimal
+    global binfile
+
+    delete_breakpoints
+
+    # Can't set a hardware breakpoint without a live target, so do it now
+    # before it's killed below.
+    gdb_test "hbreak factorial" \
+	"Hardware assisted breakpoint $decimal at .*" \
+	"hardware break at factorial"
+
+    gdb_test "kill" "" "kill program" \
+	"Kill the program being debugged.*y or n. $" "y"
+
+    # Run until we call factorial with 6
+
+    gdb_run_cmd
+    if [gdb_test "" \
+	"Break.* factorial .value=6. .*" \
+	"run to factorial(6)"] then { 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
+
+
+#********
+
+# Build a new file with optimization enabled so that we can try breakpoints
+# on targets with optimized prologues.
+
+if { [prepare_for_testing hbreak2.exp "breako2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
+    return -1
+}
+
+if ![runto_main] then { fail "break tests suppressed" }
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak main" \
+    "Hardware assisted breakpoint.*at.* file .*, line.*" \
+    "hardware breakpoint function, optimized file"
+
+#
+# Run until the breakpoint at main is hit.  For non-stubs-using targets.
+#
+gdb_run_cmd
+gdb_test_multiple "" "run until hardware function breakpoint, optimized file" {
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$srcfile:$bp_location6.*$bp_location6\[\t \]+if .argc.* \{.*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file"
+    }
+    -re "Breakpoint \[0-9\]+,.*main .*argc.*argv.* at .*$gdb_prompt $" {
+	pass "run until hardware function breakpoint, optimized file (code motion)"
+    }
+}
+delete_breakpoints
+
+#
+# Test break at function.
+#
+gdb_test "hbreak marker4" \
+    "Hardware assisted breakpoint.*at.* file .*$srcfile1, line.*" \
+    "hardware breakpoint small function, 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" $srcfile1]
+set bp_location14 [gdb_get_line_number "set breakpoint 14 here" $srcfile1]
+
+gdb_test_multiple "continue" \
+    "run until hardware breakpoint set at small function, optimized file" {
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, $hex in marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location13\[\r\n\]+$bp_location13\[\t \]+void marker4.*" {
+	    pass "run until hardware breakpoint set at small function, optimized file"
+	}
+	-re "Breakpoint $decimal, marker4 \\(d=(d@entry=)?177601976\\) at .*$srcfile1:$bp_location14\[\r\n\]+$bp_location14\[\t \]+void marker4.*" {
+	    # marker4() is defined at line 46 when compiled with -DPROTOTYPES
+	    pass "run until hardware breakpoint set at small function, optimized file (line bp_location14)"
+	}
+	-re "Breakpoint $decimal, factorial \\(.*\\) .*\{\r\n$gdb_prompt" {
+	    # GCC 4.3 emits bad line number information - see gcc/36748.
+	    if { [test_compiler_info "gcc-4-3-*"] } {
+		setup_xfail *-*-*
+	    }
+	    fail "run until hardware breakpoint set at small function, optimized file"
+	}
+}

Re: Ping: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Pedro Alves]

On 02/27/2012 09:04 PM, Maciej W. Rozycki wrote:
> Hi,
> 
>  Where are we with this proposal of mine:
> 
> http://sourceware.org/ml/gdb-patches/2011-11/msg00316.html
> 
> ?  Here's another update to add changes corresponding to recent 
> adjustments to break.exp and to accomodate the year roll-over:

I read the thread from the start, and I diffed the new proposed test against break.exp,
and I think we should go ahead and put this in.  Only a couple remarks below.

> +#   Copyright 2012 Free Software Foundation, Inc.

I think dropping the original copyright years was a mistake.

> +if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {

s/"break"/"hbreak2", so that running this test doesn't overwrite break.exp's binary.
More instances of this.

-- 
Pedro Alves

Re: Ping: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Tue, 28 Feb 2012, Pedro Alves wrote:

> I read the thread from the start, and I diffed the new proposed test against break.exp,
> and I think we should go ahead and put this in.  Only a couple remarks below.
> 
> > +#   Copyright 2012 Free Software Foundation, Inc.
> 
> I think dropping the original copyright years was a mistake.

 TBH I've never been particularly convinced to do that since this is 
derived work (and clearly not in a trivial amount).  Thanks for backing me 
up.

> > +if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
> 
> s/"break"/"hbreak2", so that running this test doesn't overwrite break.exp's binary.
> More instances of this.

 Good point, apparently there is just one other place requiring this 
adjustment.

 Thanks for your review, here's the resulting update.  Any further 
thoughts, anyone?

  Maciej

Index: gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp
===================================================================
--- gdb-fsf-trunk-quilt.orig/gdb/testsuite/gdb.base/hbreak2.exp	2012-02-28 18:50:20.455581421 +0000
+++ gdb-fsf-trunk-quilt/gdb/testsuite/gdb.base/hbreak2.exp	2012-02-28 18:49:31.835596342 +0000
@@ -1,4 +1,5 @@
-#   Copyright 2012 Free Software Foundation, Inc.
+#   Copyright 1988, 1990-1992, 1994-2000, 2002-2003, 2007-2012 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
@@ -19,7 +20,7 @@
 # the number available.
 
 
-if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
+if { [prepare_for_testing hbreak2.exp "hbreak2" {break.c break1.c} {debug nowarnings}] } {
     return -1
 }
 set srcfile break.c
@@ -530,7 +531,7 @@ test_next_with_recursion
 # Build a new file with optimization enabled so that we can try breakpoints
 # on targets with optimized prologues.
 
-if { [prepare_for_testing hbreak2.exp "breako2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
+if { [prepare_for_testing hbreak2.exp "hbreak2o2" {break.c break1.c} {debug nowarnings optimize=-O2}] } {
     return -1
 }
 

Re: Ping: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Pedro Alves]

On 02/28/2012 07:42 PM, Maciej W. Rozycki wrote:
> On Tue, 28 Feb 2012, Pedro Alves wrote:

>>> +if { [prepare_for_testing hbreak2.exp "break" {break.c break1.c} {debug nowarnings}] } {
>>
>> s/"break"/"hbreak2", so that running this test doesn't overwrite break.exp's binary.
>> More instances of this.
> 
>  Good point, apparently there is just one other place requiring this 
> adjustment.

I've added this to the testcase cookbook at <http://sourceware.org/gdb/wiki/GDBTestcaseCookbook>.

> 
>  Thanks for your review, here's the resulting update.  Any further 
> thoughts, anyone?

Not from me.  Thanks for doing this.

-- 
Pedro Alves

Re: Ping: [PATCH] testsuite: Add (extensive) hardware breakpoint testing

[Maciej W. Rozycki]

On Thu, 1 Mar 2012, Pedro Alves wrote:

> >  Thanks for your review, here's the resulting update.  Any further 
> > thoughts, anyone?
> 
> Not from me.  Thanks for doing this.

 I have committed this change now.  Thanks, everyone.

  Maciej