Re: [PATCH] tools/memory-model: Weaken ctrl dependency definition in explanation.txt

From: Alan Stern
Date: Fri Sep 02 2022 - 10:54:36 EST


On Fri, Sep 02, 2022 at 10:40:34AM +0200, Paul Heidekrüger wrote:
> On 31. Aug 2022, at 19:38, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
> >>> Finally, a read event X and another memory access event Y are linked by a
> >>> control dependency if Y syntactically lies within an arm of an if
> >>> statement and X affects the evaluation of the if condition via a data or
> >>> address dependency. Similarly for switch statements.
> >>
> >> What do you think?
> >
> > I like the second one. How about combining the last two sentences?
> >
> > ... via a data or address dependency (or similarly for a switch
> > statement).
>
> Yes, sounds good!
>
> > Now I suppose someone will pipe up and ask about the conditional
> > expressions in "for", "while" and "do" statements... :-)
>
> Happy to have obliged :-)
> https://lore.kernel.org/all/20F4C097-24B4-416B-95EE-AC11F5952B44@xxxxxxxxx/
>
> Do you think the text should explicitly address control dependencies in the
> context of loops as well? If yes, would it be a separate patch, or would it
> make sense to combine it with this one?

Anything else should be a separate patch.

For the time being, I'm happy not to worry about loops. In the end
we'll probably have to describe them as though they were unrolled,
with all the complications that entails.

Alan