Re: [patch v9 22/23] testsuite, gdb.btrace: add btrace tests

[Jan Kratochvil <jan.kratochvil@redhat.com>]

On Mon, 04 Mar 2013 18:06:09 +0100, Markus Metzger wrote:
> From: Christian Himpel <christian.himpel@intel.com>
> 
> Recursive function calls are listed as only one function call.  This is
> marked as an XFAIL.
> 
> 2013-03-04  Christian Himpel  <christian.himpel@intel.com>
> 
> gdb/testsuite
> 	* Makefile.in: Add btrace testsuite.
> 	* configure: Regenerated.
> 	* configure.ac: Add btrace testsuite.
> 	* gdb.btrace/Makefile.in: New file.
> 	* gdb.btrace/enable.c: New file.
> 	* gdb.btrace/enable.exp: New file.
> 	* gdb.btrace/function_call_history.c: New file.
> 	* gdb.btrace/function_call_history.exp: New file.
> 	* gdb.btrace/instruction_history.c: New file.
> 	* gdb.btrace/instruction_history.exp: New file.
> 	* gdb.btrace/x86-instruction_history.S: New file.
> 	* lib/gdb.exp: Add btrace skip proc.
[...]
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/enable.exp
> @@ -0,0 +1,83 @@
> +# This testcase is part of GDB, the GNU debugger.
> +#
> +# Copyright 2013 Free Software Foundation, Inc.
> +#
> +# Contributed by Intel Corp. <christian.himpel@intel.com>
> +#
> +# 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/>.
> +
> +# check for btrace support
> +if { [skip_btrace_tests] } { return -1 }
> +
> +# start fresh - without an executable
> +gdb_exit
> +gdb_start
> +
> +# record cannot be stopped, if it was never active
> +gdb_test "record stop" "No record target is currently active\..*" "record stop without target"

\ -> \\
First processing is by TCL, second processing is by regex.
This is regex backslash.  This way it acts the same as:
	gdb_test "record stop" "No record target is currently active..*" "record stop without target"

One can also use {...} instead of "..." which avoids the Tcl processing, it
would be then:
	gdb_test "record stop" {No record target is currently active\.} "record stop without target"

And that trailing .* is not needed there, gdb_test automatically matches
trailing "\r\n$gdb_prompt $".


> +
> +# btrace cannot be enabled without a running inferior
> +gdb_test "record btrace" "The program is not being run\." "record btrace without running program"

Again \ -> \\ .




> +
> +# no function and no instruction history without btrace enabled
> +gdb_test "record function-call-history" "No record target is currently active\..*" "record function-call-history without target"

\..* -> \\.

> +gdb_test "record instruction-history" "No record target is currently active\..*" "record instruction-history without target"

\..* -> \\.

> +gdb_test "info record" "No record target is currently active\..*" "info record without target"

\..* -> \\.

> +
> +# start inferior
> +standard_testfile

standard_testfile should be at the top of file; it does not start/contol GDB
any way.


> +if [prepare_for_testing $testfile.exp $testfile {} {debug}] {
> +    return -1
> +}

> +runto_main

if ![runto_main] {
    return -1
}


> +
> +# enable btrace
> +gdb_test_no_output "record btrace" "record btrace"
> +gdb_test "record function-call-history" "No trace\." "record function-call-history without trace"

\ -> \\

> +gdb_test "record instruction-history" "No trace\." "record instruction-history without trace"

\ -> \\

> +
> +# btrace cannot be enabled twice
> +gdb_test "record btrace" "The process is already being recorded\." "record btrace the second time"

\ -> \\

> +
> +# full record cannot be activated as long as btrace is active
> +gdb_test "record full" "Process record target already running\.  Use \"record stop\" to stop record target first\." "record full cannot be enabled"

\. -> \\.
twice


> +
> +# no trace recorded yet
> +gdb_test "info record" "Active record target: record-btrace\r\nRecorded 0 instructions in 0 functions for thread 1.*\." "info record without trace"

\. -> \\.


Many times below again and in the other *.exp files.


> +
> +# stop btrace record
> +gdb_test "record stop" "Process record is stopped and all execution logs are deleted\." "record stop"
> +gdb_test "record stop" "No record target is currently active\..*" "record stop the second time"
> +
> +# enable btrace again
> +gdb_test_no_output "record btrace" "record btrace re-enable"
> +gdb_test "record btrace" "The process is already being recorded\." "record btrace re-enable twice"
> +


> +# rerun
> +send_gdb "run\n"
> +gdb_expect {
> +    -re "The program .* has been started already.*y or n. $" {
> +        send_gdb "y\n"
> +        exp_continue
> +    }
> +    -re "Starting program.*$gdb_prompt $" {
> +        pass "rerun to main"
> +    }
> +    -re "$gdb_prompt $" {
> +        fail "rerun to main"
> +    }
> +    timeout {
> +        fail "(timeout) rerun to main"
> +    }
> +}

In general one should never use gdb_expect.  And send_gdb is used only
exceptionally.  For complicated cases there is gdb_test_multiple.  But in this
case why there isn't just another runto_main?

if ![runto_main] {
    return -1
}


> diff --git a/gdb/testsuite/gdb.btrace/function_call_history.c b/gdb/testsuite/gdb.btrace/function_call_history.c
> new file mode 100644
> index 0000000..a76b7c3
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/function_call_history.c
> @@ -0,0 +1,45 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> +   Copyright 2013 Free Software Foundation, Inc.
> +
> +   Contributed by Intel Corp. <christian.himpel@intel.com>
> +
> +   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/>.  */
> +
> +int
> +inc (int i)
> +{
> +  return i+1;
> +}
> +
> +int
> +fib (int n)
> +{
> +  if (n <= 1)
> +    return n;
> +
> +  return fib(n-2) + fib(n-1);
> +}
> +
> +int
> +main (void)
> +{
> +  int i, j;
> +
> +  for (i = 0; i < 10; i++)
> +    j += inc(i);
> +
> +  j += fib(3); /* bp.1 */
> +  return j; /* bp.2 */
> +}
> diff --git a/gdb/testsuite/gdb.btrace/function_call_history.exp b/gdb/testsuite/gdb.btrace/function_call_history.exp
> new file mode 100644
> index 0000000..711dd20
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/function_call_history.exp
> @@ -0,0 +1,211 @@
> +# This testcase is part of GDB, the GNU debugger.
> +#
> +# Copyright 2013 Free Software Foundation, Inc.
> +#
> +# Contributed by Intel Corp. <christian.himpel@intel.com>
> +#
> +# 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/>.
> +
> +# check for btrace support
> +if { [skip_btrace_tests] } { return -1 }
> +
> +# start inferior
> +standard_testfile
> +if [prepare_for_testing function_call_history.exp $testfile {} {debug}] {
> +    return -1
> +}


> +runto_main

if ![runto_main] {
    return -1
}


> +
> +# start btrace
> +gdb_test_no_output "record btrace"
> +
> +# set bp after increment loop and continue
> +set bp_location [gdb_get_line_number "bp.1" $testfile.c]
> +gdb_breakpoint $bp_location
> +gdb_continue_to_breakpoint "cont to $bp_location" ".*$testfile.c:$bp_location.*"
> +
> +# show function call history with unlimited size, we expect to see all 21 entries
> +gdb_test_no_output "set record function-call-history-size 0"
> +gdb_test "record function-call-history" "0	 main\r
> +1	 inc\r
> +2	 main\r
> +3	 inc\r
> +4	 main\r
> +5	 inc\r
> +6	 main\r
> +7	 inc\r
> +8	 main\r
> +9	 inc\r
> +10	 main\r
> +11	 inc\r
> +12	 main\r
> +13	 inc\r
> +14	 main\r
> +15	 inc\r
> +16	 main\r
> +17	 inc\r
> +18	 main\r
> +19	 inc\r
> +20	 main\r" "record function-call-history - with size unlimited"
> +
> +# show function call history with size of 21, we expect to see all 21 entries
> +gdb_test_no_output "set record function-call-history-size 21"
> +# show function call history
> +gdb_test "record function-call-history 0" "0	 main\r
> +1	 inc\r
> +2	 main\r
> +3	 inc\r
> +4	 main\r
> +5	 inc\r
> +6	 main\r
> +7	 inc\r
> +8	 main\r
> +9	 inc\r
> +10	 main\r
> +11	 inc\r
> +12	 main\r
> +13	 inc\r
> +14	 main\r
> +15	 inc\r
> +16	 main\r
> +17	 inc\r
> +18	 main\r
> +19	 inc\r
> +20	 main\r" "record function-call-history - show all 21 entries"
> +
> +# show first 15 entries
> +gdb_test_no_output "set record function-call-history-size 15"
> +gdb_test "record function-call-history 0" "0	 main\r
> +1	 inc\r
> +2	 main\r
> +3	 inc\r
> +4	 main\r
> +5	 inc\r
> +6	 main\r
> +7	 inc\r
> +8	 main\r
> +9	 inc\r
> +10	 main\r
> +11	 inc\r
> +12	 main\r
> +13	 inc\r
> +14	 main\r" "record function-call-history - show first 15 entries"
> +
> +# show last 6 entries
> +gdb_test "record function-call-history +" "15	 inc\r
> +16	 main\r
> +17	 inc\r
> +18	 main\r
> +19	 inc\r
> +20	 main\r" "record function-call-history - show last 6 entries"
> +
> +# moving further should not work
> +gdb_test "record function-call-history +" "At the end of the branch trace record\." "record function-call-history - at the end (1)"
> +
> +# make sure we cannot move any further a second time
> +gdb_test "record function-call-history +" "At the end of the branch trace record\." "record function-call-history - at the end (2)"
> +
> +# moving back showing the latest 15 function calls
> +gdb_test "record function-call-history -" "6	 main\r
> +7	 inc\r
> +8	 main\r
> +9	 inc\r
> +10	 main\r
> +11	 inc\r
> +12	 main\r
> +13	 inc\r
> +14	 main\r
> +15	 inc\r
> +16	 main\r
> +17	 inc\r
> +18	 main\r
> +19	 inc\r
> +20	 main\r" "record function-call-history - show last 15 entries"
> +
> +# moving further back shows the 6 first function calls
> +gdb_test "record function-call-history -" "0	 main\r
> +1	 inc\r
> +2	 main\r
> +3	 inc\r
> +4	 main\r
> +5	 inc\r" "record function-call-history - show first 6 entries"
> +
> +# moving further back shouldn't work
> +gdb_test "record function-call-history -" "At the start of the branch trace record\." "record function-call-history - at the start (1)"
> +
> +# make sure we cannot move any further back
> +gdb_test "record function-call-history -" "At the start of the branch trace record\." "record function-call-history - at the start (2)"
> +
> +# moving forward again, but this time with file and line number, expected to see the first 15 entries
> +gdb_test "record function-call-history /l +" "
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r" "record function-call-history /l - show first 15 entries"
> +
> +# moving forward and expect to see the latest 6 entries
> +gdb_test "record function-call-history /l +" "
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-41	 main\r
> +.*$srcfile:22-24	 inc\r
> +.*$srcfile:40-43	 main\r" "record function-call-history /l - show last 6 entries"
> +
> +# moving further forward shouldn't work
> +gdb_test "record function-call-history /l +" "At the end of the branch trace record\." "record function-call-history /l - at the end (1)"
> +gdb_test "record function-call-history /l" "At the end of the branch trace record\." "record function-call-history /l - at the end (2)"
> +
> +set expected_range "3	 inc\r
> +4	 main\r
> +5	 inc\r
> +6	 main\r
> +7	 inc\r
> +8	 main\r
> +9	 inc\r"
> +
> +# show functions in instruction range
> +gdb_test "record function-call-history 3,10" $expected_range "absolute instruction range"
> +gdb_test "record function-call-history 3,+7" $expected_range "relative positive instruction range"
> +gdb_test "record function-call-history 10,-7" $expected_range "relative negative instruction range"
> +
> +# set bp after fib recursion and continue
> +set bp_location [gdb_get_line_number "bp.2" $testfile.c]
> +gdb_breakpoint $bp_location
> +gdb_continue_to_breakpoint "cont to $bp_location" ".*$testfile.c:$bp_location.*"
> +
> +# at this point we expect to have main, fib, ..., fib, main, where fib occurs 8 times,
> +# so we limit the output to only show the latest 10 function calls
> +gdb_test_no_output "set record function-call-history-size 10"
> +set message "show recursive function call history"
> +gdb_test_multiple "record function-call-history" $message {
> +    -re "13	 main\r\n14	 fib\r\n15	 fib\r\n16	 fib\r\n17	 fib\r\n18	 fib\r\n19	 fib\r\n20	 fib\r\n21	 fib\r\n22	 main\r\n$gdb_prompt " {
> +        pass $message
> +    }
> +    -re "13	 inc\r\n14	 main\r\n15	 inc\r\n16	 main\r\n17	 inc\r\n18	 main\r\n19	 inc\r\n20	 main\r\n21	 fib\r\n22	 main\r\n$gdb_prompt " {
> +        # recursive function calls appear only as 1 call
> +        xfail $message

kfail, as already discussed before.


A nitpick but there can (and therefore should) be trailing $ match so that no
more accidental input is accepted there, therefore in both cases:
	main\r\n$gdb_prompt " {
->
	main\r\n$gdb_prompt $" {



> +    }
> +}
> diff --git a/gdb/testsuite/gdb.btrace/instruction_history.c b/gdb/testsuite/gdb.btrace/instruction_history.c
> new file mode 100644
> index 0000000..2dc3c6b
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/instruction_history.c
> @@ -0,0 +1,26 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> +   Copyright 2013 Free Software Foundation, Inc.
> +
> +   Contributed by Intel Corp. <christian.himpel@intel.com>
> +
> +   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/>.  */
> +
> +int
> +main (void)
> +{
> +  // loop is declared in x86-instruction_history.S

Rather /* comments */ ; although one will probably never use such new feature
with old compilers not supporting // for .c .


> +  loop ();
> +  return 0;
> +}
> diff --git a/gdb/testsuite/gdb.btrace/instruction_history.exp b/gdb/testsuite/gdb.btrace/instruction_history.exp
> new file mode 100644
> index 0000000..92288eb
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/instruction_history.exp
> @@ -0,0 +1,190 @@
> +# This testcase is part of GDB, the GNU debugger.
> +#
> +# Copyright 2013 Free Software Foundation, Inc.
> +#
> +# Contributed by Intel Corp. <christian.himpel@intel.com>
> +#
> +# 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/>.
> +
> +# check for btrace support
> +if { [skip_btrace_tests] } { return -1 }
> +
> +# compile and run to main
> +standard_testfile
> +if [prepare_for_testing instruction_history.exp $testfile "$testfile.c x86-$testfile.S" {debug}] {

standard_testfile has parameters for this purpose:
	standard_testfile .c x86-instruction_history.S
	if [prepare_for_testing $testfile.exp $testfile "$srcfile $srcfile2" {debug}] {

And please rename x86-instruction_history.S to instruction_history-x86.S so
that completion works for all the files belonging to a testcase.


> +    return -1
> +}
> +

> +runto_main

if ![runto_main] {
    return -1
}


> +
> +# set bp before loop and continue
> +set bp_location [gdb_get_line_number "bp.1" x86-$testfile.S]
> +gdb_breakpoint x86-$testfile.S:$bp_location
> +gdb_continue_to_breakpoint "cont to $bp_location" ".*x86-$testfile.S:$bp_location.*"
> +
> +# start btrace
> +gdb_test_no_output "record btrace"
> +
> +# set bp after loop and continue
> +set bp_location [gdb_get_line_number "bp.2" x86-$testfile.S]
> +gdb_breakpoint x86-$testfile.S:$bp_location
> +gdb_continue_to_breakpoint "cont to $bp_location" ".*x86-$testfile.S:$bp_location.*"
> +
> +# The following test cases test if "browsing" through the
> +# instruction history works as expected. So for the tests
> +# it is necessary to count the number of lines that are
> +# shown by the "record instruction-history" command.
> +
> +set testname "determine number of recorded instructions"
> +gdb_test_multiple "info record" $testname {
> +    -re "Active record target: record-btrace\r\nRecorded \(\[0-9\]*\) instructions in \(\[0-9\]*\) functions for thread 1 .*\.\r\n$gdb_prompt " {

Again trailing $gdb_prompt $" like in the other file.


> +        set traced $expect_out(1,string)
> +        set traced_functions $expect_out(2,string)
> +        pass $testname
> +    }
> +}
> +
> +# we have exactly 7 instructions here
> +set message "exactly 7 instructions"
> +if { $traced != 7 } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +# test that we see the expected instructions
> +gdb_test "record instruction-history 1,6" "1\t    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r

It is OK as is but FYI one can do the escaping automatically at runtime for any
string using 'string_to_regexp' function.


> +2\t    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tdec    %eax\r
> +3\t    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tjmp    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +4\t    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tcmp    \\\$0x0,%eax\r
> +5\t    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r"
> +
> +gdb_test "record instruction-history /f 1,+5" "1\t    0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +2\t    0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tdec    %eax\r
> +3\t    0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tjmp    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +4\t    0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tcmp    \\\$0x0,%eax\r
> +5\t    0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r"
> +
> +gdb_test "record instruction-history /p 6,-5" "1\t 0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +2\t 0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tdec    %eax\r
> +3\t 0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tjmp    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +4\t 0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tcmp    \\\$0x0,%eax\r
> +5\t 0x\[0-9a-f\]+ <loop\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r"
> +
> +gdb_test "record instruction-history /pf 1,6" "1\t 0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +2\t 0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tdec    %eax\r
> +3\t 0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tjmp    0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r
> +4\t 0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tcmp    \\\$0x0,%eax\r
> +5\t 0x\[0-9a-f\]+ <\\+\[0-9\]+>:\tje     0x\[0-9a-f\]+ <loop\\+\[0-9\]+>\r"
> +
> +# the following tests are checking the iterators
> +# to avoid lots of regexps, we just check the number of lines that
> +# were printed during command execution.
> +
> +# test_lines_output returns the output lines from command as a list.
> +proc test_lines_output { command message } {
> +    global gdb_prompt
> +    set message "test_lines_output: $message"
> +    gdb_test_multiple $command $message {
> +        -re "\n\(.*\)\r\n$gdb_prompt " {

trailing $ again.


> +            return [split [string trim $expect_out(1,string)] "\n"]
> +        }
> +    }
> +}
> +
> +# test_lines_length returns the number of lines from command.
> +proc test_lines_length { command message } {
> +    return [llength [test_lines_output $command $message]]
> +}
> +
> +# show instruction history with unlimited size, we expect to see
> +# all $traced instructions
> +gdb_test_no_output "set record instruction-history-size 0"
> +set message "record instruction-history - unlimited"
> +set lines [test_lines_length "record instruction-history 0" $message]
> +if { $traced != $lines } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +gdb_test_no_output "set record instruction-history-size $traced"
> +set message "record instruction-history - traced"
> +set lines [test_lines_length "record instruction-history 0" $message]
> +if { $traced != $lines } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +# test that the iterator works
> +set history_size 3
> +gdb_test_no_output "set record instruction-history-size $history_size"
> +set message "browse history forward start"
> +set lines [test_lines_length "record instruction-history 0" $message]
> +if { $lines != $history_size } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +set message "browse history forward middle"
> +set lines [test_lines_length "record instruction-history +" $message]
> +if { $lines != $history_size } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +set message "browse history forward last"
> +set lines [test_lines_length "record instruction-history +" $message]
> +if { $lines != 1 } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +gdb_test "record instruction-history" "At the end of the branch trace record\." "browse history forward beyond 1"
> +
> +# make sure we cannot move further
> +gdb_test "record instruction-history" "At the end of the branch trace record\." "browse history forward beyond 2"
> +
> +set message "browse history backward last"
> +set lines [test_lines_length "record instruction-history -" $message]
> +if { $lines != $history_size } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +set message "browse history backward middle"
> +set lines [test_lines_length "record instruction-history -" $message]
> +if { $lines != $history_size } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +set message "browse history backward first"
> +set lines [test_lines_length "record instruction-history -" $message]
> +if { $lines != 1 } {
> +    fail $message
> +} else {
> +    pass $message
> +}
> +
> +gdb_test "record instruction-history -" "At the start of the branch trace record\." "browse history backward beyond 1"
> +
> +# make sure we cannot move further back
> +gdb_test "record instruction-history -" "At the start of the branch trace record\." "browse history backward beyond 2"
> diff --git a/gdb/testsuite/gdb.btrace/x86-instruction_history.S b/gdb/testsuite/gdb.btrace/x86-instruction_history.S
> new file mode 100644
> index 0000000..5c74b46
> --- /dev/null
> +++ b/gdb/testsuite/gdb.btrace/x86-instruction_history.S
> @@ -0,0 +1,32 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> +   Copyright 2013 Free Software Foundation, Inc.
> +
> +   Contributed by Intel Corp. <christian.himpel@intel.com>
> +
> +   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/>.  */
> +
> +	.text
> +	.globl loop
> +	.type  loop, @function
> +
> +loop:
> +	movl $0x2, %eax /* bp.1 */
> +.L1:
> +	cmpl $0, %eax
> +	je .L2
> +	decl %eax
> +	jmp .L1
> +.L2:
> +	ret /* bp.2 */
> diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
> index 8b16b38..1393455 100644
> --- a/gdb/testsuite/lib/gdb.exp
> +++ b/gdb/testsuite/lib/gdb.exp
> @@ -2097,6 +2097,75 @@ proc skip_vsx_tests {} {
>      return $skip_vsx_tests_saved
>  }
>  
> +# Run a test on the target to see if it supports btrace hardware.  Return 0 if so,
> +# 1 if it does not.  Based on 'check_vmx_hw_available' from the GCC testsuite.
> +
> +proc skip_btrace_tests {} {
> +    global skip_btrace_tests_saved
> +    global srcdir subdir gdb_prompt inferior_exited_re
> +
> +    # Use the cached value, if it exists.
> +    set me "skip_btrace_tests"
> +    if [info exists skip_btrace_tests_saved] {
> +        verbose "$me:  returning saved $skip_btrace_tests_saved" 2
> +        return $skip_btrace_tests_saved
> +    }
> +
> +    if { ![istarget "i?86-*-*"] && ![istarget "x86_64-*-*"] } {
> +        verbose "$me:  target does not support btrace, returning 1" 2
> +        return [set skip_btrace_tests_saved 1]
> +    }
> +
> +    # Set up, compile, and execute a test program.
> +    # Include the current process ID in the file names to prevent conflicts
> +    # with invocations for multiple testsuites.
> +    set src btrace[pid].c
> +    set exe btrace[pid].x

Use:
	set src [standard_output_file btrace[pid].c]
	set exe [standard_output_file btrace[pid].x]


> +
> +    set f [open $src "w"]
> +    puts $f "int main() { return 0; }"

For C there should be:
       puts $f "int main(void) { return 0; }"


> +    close $f
> +
> +    verbose "$me:  compiling testfile $src" 2
> +    set compile_flags {debug nowarnings quiet}
> +    set lines [gdb_compile $src $exe executable $compile_flags]
> +    file delete $src
> +
> +    if ![string match "" $lines] then {
> +        verbose "$me:  testfile compilation failed, returning 1" 2
> +        return [set skip_btrace_tests_saved 1]
> +    }
> +
> +    # No error message, compilation succeeded so now run it via gdb.
> +
> +    gdb_exit
> +    gdb_start
> +    gdb_reinitialize_dir $srcdir/$subdir
> +    gdb_load "$exe"

This can be replaced by:
	clean_restart btrace[pid].x

Unfortunately one cannot use $exe as that one has standard_output_file already
applied which must not be done for clean_restart.


> +    runto_main

if ![runto_main] {
    return [...]
}


> +    # In case of an unexpected output, we return 2 as a fail value.
> +    set skip_btrace_tests_saved 2
> +    gdb_test_multiple "record btrace" "check btrace support" {
> +        -re "You can't do that when your target is.*\r\n$gdb_prompt $" {
> +            set skip_btrace_tests_saved 1
> +        }
> +        -re "Target does not support branch tracing.*\r\n$gdb_prompt $" {
> +            set skip_btrace_tests_saved 1
> +        }
> +        -re "Could not enable branch tracing.*\r\n$gdb_prompt $" {
> +            set skip_btrace_tests_saved 1
> +        }
> +        -re "$gdb_prompt $" {

To test no output has been produced (instead of any output has been produced):
        -re "^record btrace\r\n$gdb_prompt $" {


> +            set skip_btrace_tests_saved 0
> +        }
> +    }
> +    gdb_exit
> +    remote_file build delete $exe
> +
> +    verbose "$me:  returning $skip_btrace_tests_saved" 2
> +    return $skip_btrace_tests_saved
> +}
> +
>  # Skip all the tests in the file if you are not on an hppa running
>  # hpux target.
>  
> -- 
> 1.7.1

If you follow the advices there shold not be an issue but it would be probably
worth another review.


Thanks,
Jan