Re: [PATCH] gdb: don't pass TARGET_WNOHANG to targets that can't async (PR 26642)

[Pedro Alves <pedro@palves.net>]

On 10/13/20 4:06 PM, Simon Marchi wrote:
> On 2020-10-12 10:48 a.m., Pedro Alves wrote:

>> I'd rather go without this hunk.  Would it still fix things?
> 
> Yes, since infrun won't ask linux-nat to TARGET_WNOHANG when
> target_async_permitted is false.  I'll remove it, since it can never
> happen with the assert in target_wait.  If anything, we could put an
> assert here as a double-check, if you think it could help.

I don't think it could help.

> I'll send a v2 shortly.
> 
> I was wondering, do you see a better way to fix this than what I did in
> do_target_wait_1, forcing TARGET_WNOHANG to off?  It doesn't feel right
> to have the caller ask for something and just do something else.  But I
> couldn't think of anything better (that's also suitable for the release
> branch, at least).

I don't -- you can't do it at the caller because there you don't know
which inferior/target will be called.

We could just change the semantics of the do_target_wait slightly, like:

 static bool
 do_target_wait (ptid_t wait_ptid, execution_control_state *ecs,
-		target_wait_flags options)
+		bool dont_block_if_possible)
 {

That doesn't seem that much better to me, though.