Re: [PATCH v2] Slightly tweak and clarify target_resume's interface

[Pedro Alves <pedro@palves.net>]

On 2022-04-28 21:38, Simon Marchi wrote:

>> I also thought of making iterate_over_lwps itself not iterate if the
>> ptid is an lwp, but this seems like the only spot where that would make
>> a difference so I didn't bother.  Plus, we'd call the callback only
>> to have it do nothing (the second parameter is an "expect this lwp" parameter).
> 
> What do you mean not iterate?  We still want iterate_over_lwps to call
> the callback on that specific LWP, don't we?  And since lwps are stored
> in a linked list, it has to iterate.  I suppose it could break early
> once it has found the lwp though.

Yeah, I was thinking of something like done in
0618ae41497998792701b72f34ea4859cf95aafc ("gdb: optimize all_matching_threads_iterator").
To avoid iterating to find the lwp, we'd add a hash map like done for threads, with
inferior::ptid_thread_map.

> 
>> Here's the updated patch.  Retested on native and native-extended-gdbserver x86-64 GNU/Linux.
>>
>> WDYT?
> 
> LGTM, thanks!

I'll check it in, then, thanks.