Re: [PATCH 01/11] change gdb.base/skip.exp to use finish instead of step

[Bruno Larsen <blarsen@redhat.com>]

On 3/2/22 13:39, Andrew Burgess wrote:
> Pedro Alves <pedro@palves.net> writes:
> 
>> On 2022-01-26 19:50, Bruno Larsen via Gdb-patches wrote:
>>> skip.exp was making use of a fixed amount of step commands to exit
>>> some functions.  This caused some problems when testing GDB with clang,
>>> as it doesn't add epilogue information for functions.  Since the step
>>> commands weren't testing important features, they were changed to finish
>>> commands.
>>
>> I'm not a fan of _completely_ avoiding stepping out of functions, due to some
>> compilers misbehaving.  I mean, if we're going to do this throughout
>> the testsuite, then we should at least have one testcase that is explicitly
>> about stepping out of a function with step, one that exhibits the problem,
>> and we can concentrate the xfails there (and we should press the compiler
>> folks to fix it).  I'm not sure we have one.  If we do, you should mention it
>> explicitly in the commit log.  Afterall, users will be doing this too.
> 
> Agreed, and there's probably scope for a proc in lib/gdb.exp that does
> something like "step until we reach a source line matching REGEXP",
> which could be used instead of finish in some cases.
> 
> Thanks,
> Andrew
> 

Interesting suggestion. I'm not sure "step until we reach a source matching REGEXP" is the best approach, since anything off in that regexp means that we might be single stepping until the program exits - a bit too time-consuming for my liking. I think that a proc that single steps until it exits the current frame. Does this sound reasonable?

-- 
Cheers!
Bruno Larsen