Re: [PATCH] Fix MI dprintf-insert not printing when a location is pending.

[Antoine Tremblay <antoine.tremblay@ericsson.com>]

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