Re: [PATCH 1/2] [gdb/testsuite] Disable inferior output in gdb.base/foll-vfork.exp

[Tom de Vries <tdevries@suse.de>]

On 11/16/21 4:01 PM, Simon Marchi wrote:
> On 2021-11-15 5:11 p.m., Tom de Vries via Gdb-patches wrote:
>> On 11/15/21 5:39 PM, Simon Marchi wrote:
>>> On 2021-11-15 6:21 a.m., Tom de Vries via Gdb-patches wrote:
>>>> Test-case gdb.base/foll-vfork.exp has inferior output that is not needed, but
>>>> which makes the regexp matching more difficult (see commit 1f28b70def1
>>>> "[gdb/testsuite] Fix regexp in gdb.base/foll-vfork.exp").
>>>>
>>>> Disable the inferior output using '#if DEBUG'.
>>>>
>>>> Tested on x86_64-linux.
>>>> ---
>>>>  gdb/testsuite/gdb.base/foll-vfork-exit.c | 14 ++++++++++++--
>>>>  gdb/testsuite/gdb.base/foll-vfork.c      |  9 ++++++++-
>>>>  gdb/testsuite/gdb.base/foll-vfork.exp    | 17 +++++++++++------
>>>>  gdb/testsuite/gdb.base/vforked-prog.c    |  7 ++++++-
>>>>  4 files changed, 37 insertions(+), 10 deletions(-)
>>>>
>>>> diff --git a/gdb/testsuite/gdb.base/foll-vfork-exit.c b/gdb/testsuite/gdb.base/foll-vfork-exit.c
>>>> index 6c263cdc057..15e272afe82 100644
>>>> --- a/gdb/testsuite/gdb.base/foll-vfork-exit.c
>>>> +++ b/gdb/testsuite/gdb.base/foll-vfork-exit.c
>>>> @@ -29,12 +29,22 @@ main ()
>>>>    pid = vfork (); /* VFORK */
>>>>    if (pid == 0)
>>>>      {
>>>> -      printf ("I'm the child!\n");
>>>> +      const char *s = "I'm the child!";
>>>> +#if DEBUG
>>>> +      printf ("%s\n", s);
>>>> +#else
>>>> +    const char *volatile v = s;
>>>> +#endif
>>>
>>> In my original comment, I meant that we could have simply deleted the
>>> printfs, instead of deleting the regexp $ anchor, as done in
>>> 1f28b70def1.  I don't think these printfs provide any value to debug the
>>> test case at all.
>>>
>>> Since 1f28b70def1 is already merged, the printfs can just stay as-is
>>> (since the regexp accepts them).  Or we can remove them and put back the
>>> anchor.  I slightly prefer the latter, because it makes the test case
>>> check slightly more restrictive (if there is some spurious output after
>>> the prompt, the test will fail).  But it won't make a big difference in
>>> the grand scheme of things.
>>
>> So, is this what you meant?
> 
> Yeah, pretty much.  Could the -re entries use -wrap?
> 

Sure.

> You still assign the strings to volatile variables to make the
> gdb_get_line_number calls work, I suppose?
> 

Indeed.

I'll commit this variant, updated to use -wrap.

Thanks,
- Tom