LISA LISA2Rtx1GM-C
(*
* Result: Maybe (compiler!)
*
* Two RCU grace periods on a single thread and one RCU read-side critical
* section, but set up to verify that the pair of RCU read-side critical
* sections provide no ordering even when interacting with an intervening
* grace period. Cycle allowed, however, tests that don't model compiler
* interactions might have a hard time seeing this, hence the "-C".
*)
{
x0 = 0;
x1 = 0;
x2 = 0;
x3 = 0;
}
P0 | P1 | P2 ;
f[rcu_read_lock] | r[once] r0 x0 | r[once] r2 x2 ;
w[once] x1 1 | f[sync] | f[mb] ;
w[once] x0 1 | r[once] r3 x3 | r[once] r1 x1 ;
f[rcu_read_unlock] | | ;
f[rcu_read_lock] | | ;
w[once] x3 1 | | ;
w[once] x2 1 | | ;
f[rcu_read_unlock] | | ;
Observed
2:r2=1; 2:r1=1; 1:r3=0; 1:r0=1;
and 2:r2=1; 2:r1=0; 1:r3=0; 1:r0=1;
and 2:r2=1; 2:r1=0; 1:r3=0; 1:r0=0;