Fix MI dprintf-insert not printing when a location is pending.
Commit Message
This patch fixes the "Format string required" error when trying to print
a dprintf on a pending location when set via the MI interface even if the
format string is entered correctly.
This patch also adds a test case to check that issue called
mi-dprintf-pending.exp.
gdb/ChangeLog:
PR breakpoints/16465
* breakpoint.c (create_breakpoint): Fix missing extra_string.
gdb/testsuite/ChangeLog:
PR breakpoints/16465
* gdb.mi/mi-dprintf-pending.c: New file.
* gdb.mi/mi-dprintf-pending.exp: New test.
* gdb.mi/mi-dprintf-pendshr.c: New file.
---
gdb/breakpoint.c | 13 ++++-
gdb/testsuite/gdb.mi/mi-dprintf-pending.c | 24 +++++++++
gdb/testsuite/gdb.mi/mi-dprintf-pending.exp | 77 +++++++++++++++++++++++++++++
gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c | 27 ++++++++++
4 files changed, 139 insertions(+), 2 deletions(-)
create mode 100644 gdb/testsuite/gdb.mi/mi-dprintf-pending.c
create mode 100644 gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
create mode 100644 gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
Comments
On 03/26/2015 09:47 AM, Antoine Tremblay wrote:
> gdb/ChangeLog:
>
> PR breakpoints/16465
> * breakpoint.c (create_breakpoint): Fix missing extra_string.
This is a real nit, so please don't go making any changes here unless a
maintainer requests it, but this changelog entry doesn't really explain
the change you've made. ["Save `extra_string' for pending breakpoints."
is much more descriptive/helpful.]
> @@ -9783,10 +9786,16 @@ create_breakpoint (struct gdbarch *gdbarch,
> cond_string = xstrdup (cond_string);
> make_cleanup (xfree, cond_string);
> }
> + /* Create a private copy of any extra string. */
> + if (extra_string)
We explicitly test against NULL for pointers. [I know there are quite a
few violations of this in this function. All are awaiting an
easy/obvious separate cleanup. :-)]
> + {
> + extra_string = xstrdup (extra_string);
> + make_cleanup (xfree, extra_string);
> + }
> b->cond_string = cond_string;
> + b->extra_string = extra_string;
> b->thread = thread;
> }
> - b->extra_string = NULL;
> b->ignore_count = ignore_count;
> b->disposition = tempflag ? disp_del : disp_donttouch;
> b->condition_not_parsed = 1;
FWIW, I have pretty much an identical change in my locations API
refactor, where I ran across this problem (and more) during testing.
> diff --git a/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
> new file mode 100644
> index 0000000..6832f1d
> --- /dev/null
> +++ b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
> @@ -0,0 +1,77 @@
> +if {[skip_shlib_tests]} {
> +if [get_compiler_info] {
> +if { [gdb_compile_shlib $libsrc1 $lib_sl1 $lib_opts] != "" } {
> +if { [gdb_compile $srcdir/$subdir/$srcfile $binfile executable $exec_opts] != ""} {
This is a /big/ nitpick, but it's something that consistently irritates
me: compare the coding style of the four statements above. While there
is no "rule" governing which is most correct/desired, I always use the
first. I would ask you to choose one style and be consistent, but I am
not asking you to make any changes right now. Just please keep this in
mind in future patches.
> +# Set pending dprintf via MI.
> +mi_gdb_test "-dprintf-insert -f pendfunc1 \"hello\"" \
> + ".*\\^done,bkpt={number=\"1\",type=\"dprintf\",disp=\"keep\",enabled=\"y\",addr=\"<PENDING>\",pending=\"pendfunc1\",times=\"0\",original-location=\"pendfunc1\"}" \
> + "mi set dprintf"
> +
> +mi_gdb_test "-break-insert $bp_location1" ".*" "mi insert breakpoint bp_location1"
Is it possible to use mi_make_breakpoint for these tests?
> +
> +mi_run_cmd
> +
> +set msg "mi dprintf"
> +gdb_expect {
> + -re ".*~\"hello\"" {
> + pass $msg
> + }
> + -re ".*$mi_gdb_prompt$" {
> + fail $msg
> + }
> + timeout {
> + fail $msg
> + }
> +}
This a pretty common test suite idiom, I think. Can mi_gdb_test be used
instead of gdb_expect?
> +mi_expect_stop ".*" ".*" ".*" ".*" ".*" "" "$msg stop"
> diff --git a/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c b/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
> new file mode 100644
> index 0000000..fe49a8d
> --- /dev/null
> +++ b/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
> @@ -0,0 +1,27 @@
> +/* This testcase is part of GDB, the GNU debugger.
> +
> + Copyright 2015 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/>. */
> +
> +void
> +pendfunc1 ()
> +{
> +}
> +
> +void
> +pendfunc ()
> +{
> + pendfunc1();
> +}
>
IIRC, we are now requiring test case conformance to the coding standard.
[At least, that's what I've been told in the past.] So, "(void)" in the
function decls and spaces between function names and '('. [I don't think
we're requiring function comments for trivial stuff like this, though.]
Keith
On 03/26/2015 03:10 PM, Keith Seitz wrote:
> On 03/26/2015 09:47 AM, Antoine Tremblay wrote:
>> gdb/ChangeLog:
>>
>> PR breakpoints/16465
>> * breakpoint.c (create_breakpoint): Fix missing extra_string.
>
> This is a real nit, so please don't go making any changes here unless a
> maintainer requests it, but this changelog entry doesn't really explain
> the change you've made. ["Save `extra_string' for pending breakpoints."
> is much more descriptive/helpful.]
>
Indeed why not :)
>> @@ -9783,10 +9786,16 @@ create_breakpoint (struct gdbarch *gdbarch,
>> cond_string = xstrdup (cond_string);
>> make_cleanup (xfree, cond_string);
>> }
>> + /* Create a private copy of any extra string. */
>> + if (extra_string)
>
> We explicitly test against NULL for pointers. [I know there are quite a
> few violations of this in this function. All are awaiting an
> easy/obvious separate cleanup. :-)]
>
Yes, it will look weird to just change the one I added though...
That's what I did anyway so that the others can be part of a cleanup patch.
>> + {
>> + extra_string = xstrdup (extra_string);
>> + make_cleanup (xfree, extra_string);
>> + }
>> b->cond_string = cond_string;
>> + b->extra_string = extra_string;
>> b->thread = thread;
>> }
>> - b->extra_string = NULL;
>> b->ignore_count = ignore_count;
>> b->disposition = tempflag ? disp_del : disp_donttouch;
>> b->condition_not_parsed = 1;
>
> FWIW, I have pretty much an identical change in my locations API
> refactor, where I ran across this problem (and more) during testing.
:) Glad it can remove unrelated stuff from your patch
>
>> diff --git a/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
>> b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
>> new file mode 100644
>> index 0000000..6832f1d
>> --- /dev/null
>> +++ b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp
>> @@ -0,0 +1,77 @@
>> +if {[skip_shlib_tests]} {
>> +if [get_compiler_info] {
>> +if { [gdb_compile_shlib $libsrc1 $lib_sl1 $lib_opts] != "" } {
>> +if { [gdb_compile $srcdir/$subdir/$srcfile $binfile executable
>> $exec_opts] != ""} {
>
> This is a /big/ nitpick, but it's something that consistently irritates
> me: compare the coding style of the four statements above. While there
> is no "rule" governing which is most correct/desired, I always use the
> first. I would ask you to choose one style and be consistent, but I am
> not asking you to make any changes right now. Just please keep this in
> mind in future patches.
>
I really had not noticed that, thanks for making me notice! I must
confess I do a lot of copy & paste when writing a test. The if {[func]}
seems indeed the one used in all the tcl docs so I'll use that.
>> +# Set pending dprintf via MI.
>> +mi_gdb_test "-dprintf-insert -f pendfunc1 \"hello\"" \
>> +
>> ".*\\^done,bkpt={number=\"1\",type=\"dprintf\",disp=\"keep\",enabled=\"y\",addr=\"<PENDING>\",pending=\"pendfunc1\",times=\"0\",original-location=\"pendfunc1\"}"
>> \
>> + "mi set dprintf"
>> +
>> +mi_gdb_test "-break-insert $bp_location1" ".*" "mi insert breakpoint
>> bp_location1"
>
> Is it possible to use mi_make_breakpoint for these tests?
Unfortunately for the dprintf one mi_make_breakpoint doesn't support
pending breakpoints, it creates something like :
bkpt={number="2",type=".*",disp=".*",enabled=".*",addr=".*",func=".*",
file=".*/myfile.c",fullname=".*",line="3",thread-groups=\[.*\],
times="0".*original-location=".*"}
But with pending funcs it should be pending= ...
It could be the subject of another patch to add that support.
I used mi_create_breakpoint for the other breakpoint now
>
>> +
>> +mi_run_cmd
>> +
>> +set msg "mi dprintf"
>> +gdb_expect {
>> + -re ".*~\"hello\"" {
>> + pass $msg
>> + }
>> + -re ".*$mi_gdb_prompt$" {
>> + fail $msg
>> + }
>> + timeout {
>> + fail $msg
>> + }
>> +}
>
> This a pretty common test suite idiom, I think. Can mi_gdb_test be used
> instead of gdb_expect?
That I can't since mi_gdb_test requires a command and in this case I'm
just doing expect on that comes after mi_run_cmd, there's no command
associated with it..
>
>> +mi_expect_stop ".*" ".*" ".*" ".*" ".*" "" "$msg stop"
>> diff --git a/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
>> b/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
>> new file mode 100644
>> index 0000000..fe49a8d
>> --- /dev/null
>> +++ b/gdb/testsuite/gdb.mi/mi-dprintf-pendshr.c
>> @@ -0,0 +1,27 @@
>> +/* This testcase is part of GDB, the GNU debugger.
>> +
>> + Copyright 2015 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/>. */
>> +
>> +void
>> +pendfunc1 ()
>> +{
>> +}
>> +
>> +void
>> +pendfunc ()
>> +{
>> + pendfunc1();
>> +}
>>
>
> IIRC, we are now requiring test case conformance to the coding standard.
> [At least, that's what I've been told in the past.] So, "(void)" in the
> function decls and spaces between function names and '('. [I don't think
> we're requiring function comments for trivial stuff like this, though.]
>
Indeed that's really my old habits dying hard... fixed sorry about that.
Thanks for the review ! , Patch v2 is coming up in a minute as a
separate email...
Antoine
On 03/27/2015 05:34 AM, Antoine Tremblay wrote:
> On 03/26/2015 03:10 PM, Keith Seitz wrote:
>> On 03/26/2015 09:47 AM, Antoine Tremblay wrote:
>>> +# Set pending dprintf via MI.
>>> +mi_gdb_test "-dprintf-insert -f pendfunc1 \"hello\"" \
>>> +
>>> ".*\\^done,bkpt={number=\"1\",type=\"dprintf\",disp=\"keep\",enabled=\"y\",addr=\"<PENDING>\",pending=\"pendfunc1\",times=\"0\",original-location=\"pendfunc1\"}"
>>>
>>> \
>>> + "mi set dprintf"
>>> +
>>> +mi_gdb_test "-break-insert $bp_location1" ".*" "mi insert breakpoint
>>> bp_location1"
>>
>> Is it possible to use mi_make_breakpoint for these tests?
>
> Unfortunately for the dprintf one mi_make_breakpoint doesn't support
> pending breakpoints, it creates something like :
> bkpt={number="2",type=".*",disp=".*",enabled=".*",addr=".*",func=".*",
> file=".*/myfile.c",fullname=".*",line="3",thread-groups=\[.*\],
> times="0".*original-location=".*"}
>
> But with pending funcs it should be pending= ...
Ah, I see. Thank you for pointing that out. I *thought* that I did that
already!
> It could be the subject of another patch to add that support.
>
Indeed.
> I used mi_create_breakpoint for the other breakpoint now
>
Excellent.
>>> +mi_run_cmd
>>> +
>>> +set msg "mi dprintf"
>>> +gdb_expect {
>>> + -re ".*~\"hello\"" {
>>> + pass $msg
>>> + }
>>> + -re ".*$mi_gdb_prompt$" {
>>> + fail $msg
>>> + }
>>> + timeout {
>>> + fail $msg
>>> + }
>>> +}
>>
>> This a pretty common test suite idiom, I think. Can mi_gdb_test be used
>> instead of gdb_expect?
>
> That I can't since mi_gdb_test requires a command and in this case I'm
> just doing expect on that comes after mi_run_cmd, there's no command
> associated with it..
While not common in gdb.mi (with mi_gdb_test), the rest of gdb.* uses
the idiom:
gdb_test "" ...
all the time.
Looking at the description of mi_gdb_test:
# mi_gdb_test COMMAND PATTERN MESSAGE [IPATTERN] -- send a command to gdb;
# test the result.
#
# COMMAND is the command to execute, send to GDB with send_gdb. If
# this is the null string no command is sent.
[Very quickly] Browsing through the code of that function, it seems like
it should work. If it does not, would you please file a bug? It
*definitely* should work.
Keith
@@ -9774,7 +9774,10 @@ create_breakpoint (struct gdbarch *gdbarch,
b->addr_string = copy_arg;
if (parse_arg)
- b->cond_string = NULL;
+ {
+ b->cond_string = NULL;
+ b->extra_string = NULL;
+ }
else
{
/* Create a private copy of condition string. */
@@ -9783,10 +9786,16 @@ create_breakpoint (struct gdbarch *gdbarch,
cond_string = xstrdup (cond_string);
make_cleanup (xfree, cond_string);
}
+ /* Create a private copy of any extra string. */
+ if (extra_string)
+ {
+ extra_string = xstrdup (extra_string);
+ make_cleanup (xfree, extra_string);
+ }
b->cond_string = cond_string;
+ b->extra_string = extra_string;
b->thread = thread;
}
- b->extra_string = NULL;
b->ignore_count = ignore_count;
b->disposition = tempflag ? disp_del : disp_donttouch;
b->condition_not_parsed = 1;
new file mode 100644
@@ -0,0 +1,24 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2015 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/>. */
+
+extern void pendfunc ();
+
+int main()
+{
+ pendfunc ();
+ return 0; /* set breakpoint 1 here */
+}
new file mode 100644
@@ -0,0 +1,77 @@
+# Copyright 2015 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/>.
+
+# This test checks if dprintf prints correctly when it's pending.
+# See PR breakpoints/16465.
+
+load_lib mi-support.exp
+
+if {[skip_shlib_tests]} {
+ return 0
+}
+
+standard_testfile mi-dprintf-pending.c
+
+set libfile1 "mi-dprintf-pendshr"
+set libsrc1 $srcdir/$subdir/$libfile1.c
+set lib_sl1 [standard_output_file $libfile1.sl]
+set lib_opts debug
+set exec_opts [list debug shlib=$lib_sl1 shlib_load]
+
+if [get_compiler_info] {
+ return -1
+}
+
+if { [gdb_compile_shlib $libsrc1 $lib_sl1 $lib_opts] != "" } {
+ untested "Could not compile $libsrc1"
+ return -1
+}
+
+if { [gdb_compile $srcdir/$subdir/$srcfile $binfile executable $exec_opts] != ""} {
+ untested "Could not compile $srcdir/$subdir/$srcfile"
+ return -1
+}
+
+# Start with a fresh gdb.
+gdb_exit
+mi_gdb_start
+mi_gdb_reinitialize_dir $srcdir/$subdir
+mi_gdb_load ${binfile}
+mi_load_shlibs $lib_sl1
+
+set bp_location1 [gdb_get_line_number "set breakpoint 1 here"]
+
+# Set pending dprintf via MI.
+mi_gdb_test "-dprintf-insert -f pendfunc1 \"hello\"" \
+ ".*\\^done,bkpt={number=\"1\",type=\"dprintf\",disp=\"keep\",enabled=\"y\",addr=\"<PENDING>\",pending=\"pendfunc1\",times=\"0\",original-location=\"pendfunc1\"}" \
+ "mi set dprintf"
+
+mi_gdb_test "-break-insert $bp_location1" ".*" "mi insert breakpoint bp_location1"
+
+mi_run_cmd
+
+set msg "mi dprintf"
+gdb_expect {
+ -re ".*~\"hello\"" {
+ pass $msg
+ }
+ -re ".*$mi_gdb_prompt$" {
+ fail $msg
+ }
+ timeout {
+ fail $msg
+ }
+}
+mi_expect_stop ".*" ".*" ".*" ".*" ".*" "" "$msg stop"
new file mode 100644
@@ -0,0 +1,27 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2015 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/>. */
+
+void
+pendfunc1 ()
+{
+}
+
+void
+pendfunc ()
+{
+ pendfunc1();
+}