Re: [PATCH RFC] tools/memory-model: Adjust ctrl dependency definition

[Paul Heidekrüger]

Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:

> On Fri, Jul 15, 2022 at 09:27:26AM -0400, Alan Stern wrote:
>> On Fri, Jul 15, 2022 at 02:27:28PM +0200, Paul Heidekrüger wrote:
>>> I have just been thinking about how to word this patch; am I correct in
>>> assuming that the LKMM does not deal with loop conditions? Or in other
>>> words, there is no way for a loop condition to impose a ctrl dependency on
>>> any WRITE_ONCE's in the loop body? It are only if and switch statements the
>>> LKMM is concerned with in the case of ctrl dependencies?
>> 
>> In theory, the LKMM does say that a loop condition imposes a control 
>> dependency on any memory accesses within the loop body.  However, the 
>> herd7 tool has only very limited support for looping constructs, so in 
>> practice it's not possible to create suitable litmus tests with loops.
> 
> And Alan isn't joking.  The closest simulation that I know of is to
> combine limited loop unrolling with the "filter" clause.  The point of
> the filter clause is to eliminate from consideration executions that
> need the more iterations of the loop to be unrolled.
> 
> And that means that as far as LKMM is concerned, loop-based control
> dependencies are similar to those for nested "if" statements.
> 
> 							Thanx, Paul

Makes sense, thank you both!

Paul