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

Doug Evans dje@google.com
Fri Nov 11 17:08:00 GMT 2011


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"
> +}
>



More information about the Gdb-patches mailing list