Open andrew-appel opened 2 months ago
In vst_on_iris branch, in lib/proof/verif_locks.v, in Lemma body_release, there is a Iris Proof Mode proof that fails:
iInv i as "((% & >p & ?) & Hown)" "Hclose".
The same exact proof succeeds, with apparently the same context, in atomics/verif_lock.v (also in the vst_on_iris branch).
Can someone familiar with IPM fix this one?
There were two problems here: 1) the definition that was being destructed was declared Opaque, and 2) the Timeless instance for atomic_int_at wasn't declared as an instance. It should work now.
Opaque
Timeless
atomic_int_at
In vst_on_iris branch, in lib/proof/verif_locks.v, in Lemma body_release, there is a Iris Proof Mode proof that fails:
The same exact proof succeeds, with apparently the same context, in atomics/verif_lock.v (also in the vst_on_iris branch).
Can someone familiar with IPM fix this one?