Re: [PATCH] gdb/mi: add no-history stop reason

[Bruno Larsen <blarsen@redhat.com>]

On 03/01/2023 13:04, Andrew Burgess wrote:
> Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:
>
>> When executing in reverse and runs out of recorded history, GDB prints
>> a warning to the user, but does not add a reason in the stopped record,
>> for example:
>>
>> *stopped,frame={addr="0x000000000040113e",func="main",args=[],file="/home/blarsen/Documents/fsf_build/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.reverse/solib-reverse.c",fullname="/home/blarsen/Documents/binutils-gdb/gdb/testsuite/gdb.reverse/solib-reverse.c",line="27",arch="i386:x86-64"},thread-id="1",stopped-threads="all",core="1"
>>
>> This problem was reported as record/29260.
>>
>> This commit adds the reason no-history to the record, making it easier
>> for interfaces using the mi interpreter to report the result.
>>
>> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29260
>> ---
>>   gdb/NEWS            | 5 +++++
>>   gdb/doc/gdb.texinfo | 2 ++
>>   gdb/infrun.c        | 5 ++++-
>>   3 files changed, 11 insertions(+), 1 deletion(-)
> I think this should have a test added to gdb.reverse/.
There is already a test called gdb.mi/mi-reverse that handles executing 
reverse instructions in mi mode, including running till the end of the 
history.  I forgot to change it, but it will cover this change.
>
>> diff --git a/gdb/NEWS b/gdb/NEWS
>> index e61f06081de..3673bed8f46 100644
>> --- a/gdb/NEWS
>> +++ b/gdb/NEWS
>> @@ -5,6 +5,11 @@
>>   
>>   * MI version 1 has been removed.
>>   
>> +* MI changes
>> +
>> +** mi now reports 'no_history' as a stop reason when hitting the end of the
> s/no_history/no-history/
>
>> +   reverse execution history.
>> +
>>   *** Changes in GDB 13
>>   
>>   * MI version 1 is deprecated, and will be removed in GDB 14.
>> diff --git a/gdb/doc/gdb.texinfo b/gdb/doc/gdb.texinfo
>> index a72b2b9eb26..8982151a7e3 100644
>> --- a/gdb/doc/gdb.texinfo
>> +++ b/gdb/doc/gdb.texinfo
>> @@ -30851,6 +30851,8 @@ The inferior returned from a system call.  This is reported when
>>   @item exec
>>   The inferior called @code{exec}.  This is reported when @code{catch exec}
>>   (@pxref{Set Catchpoints}) has been used.
>> +@item no-history
>> +There isn't enough history recorded to continue reverse execution.
>>   @end table
>>   
>>   The @var{id} field identifies the global thread ID of the thread
>> diff --git a/gdb/infrun.c b/gdb/infrun.c
>> index d5f97e33625..601d1a54701 100644
>> --- a/gdb/infrun.c
>> +++ b/gdb/infrun.c
>> @@ -8406,7 +8406,10 @@ print_signal_received_reason (struct ui_out *uiout, enum gdb_signal siggnal)
>>   void
>>   print_no_history_reason (struct ui_out *uiout)
>>   {
>> -  uiout->text ("\nNo more reverse-execution history.\n");
>> +  if (uiout->is_mi_like_p ())
>> +    uiout->field_string ("reason", "no-history");
> If we look at how similar cases are handled in e.g. breakpoint.c, then
> we see that GDB looks up all these stop reason strings by calling
> async_reason_lookup.
>
> So, I think we should add a new reason to 'enum async_reply_reason' in
> mi/mi-common.h.  The 'no-history' string should be added to
> 'async_reason_string_lookup' in mi-common.c, and then you should call
> async_reason_lookup here.
I did see other places doing it, but I thought this was used for more 
complex stuff. Thanks for explaining!
>
> Thanks,
> Andrew
>> +  else
>> +    uiout->text ("\nNo more reverse-execution history.\n");
>>   }
>>   
>>   /* Print current location without a level number, if we have changed
>> -- 
>> 2.39.0


-- 
Cheers,
Bruno