allow hint with warning (#195)

* allow hint with warning

* format

* add a hint test

* fix fmt issue

---------

Co-authored-by: hczphn <hczphn@polyhedra.network>

Author: zhiyong1997

circuit-std-rs/tests/logup.rs

@@ -7,8 +7,14 @@ use circuit_std_rs::{
 use expander_compiler::{
     field::{BN254Fr, Goldilocks},
     frontend::*,
-    zkcuda::{context::*, kernel::*, proving_system::*, shape::Reshape},
+    zkcuda::{
+        context::*,
+        kernel::*,
+        proving_system::{expander::config::ZKCudaBN254Hyrax, *},
+        shape::Reshape,
+    },
 };
+use serdes::ExpSerde;
 
 #[test]
 fn logup_test() {
@@ -196,3 +202,31 @@ fn rangeproof_zkcuda_test_fail() {
     );
     assert!(P::verify(&verifier_setup, &computation_graph, &proof));
 }
+
+#[test]
+fn rangeproof_zkcuda_no_oversubscribe_test() {
+    let mut hint_registry = HintRegistry::<BN254Fr>::new();
+    hint_registry.register("myhint.querycounthint", query_count_hint);
+    hint_registry.register("myhint.rangeproofhint", rangeproof_hint);
+    //compile and test
+    let kernel: KernelPrimitive<BN254Config> = compile_rangeproof_test_kernel().unwrap();
+    let mut ctx: Context<BN254Config, _> = Context::new(hint_registry);
+
+    let a = BN254Fr::from((1 << 9) as u32);
+    let a = ctx.copy_to_device(&a);
+    let a = a.reshape(&[1]);
+    call_kernel!(ctx, kernel, 1, a).unwrap();
+
+    let computation_graph = ctx.compile_computation_graph().unwrap();
+    ctx.solve_witness().unwrap();
+    let (prover_setup, _) = ExpanderNoOverSubscribe::<ZKCudaBN254Hyrax>::setup(&computation_graph);
+    let proof = ExpanderNoOverSubscribe::<ZKCudaBN254Hyrax>::prove(
+        &prover_setup,
+        &computation_graph,
+        ctx.export_device_memories(),
+    );
+    let file = std::fs::File::create("proof.txt").unwrap();
+    let writer = std::io::BufWriter::new(file);
+    proof.serialize_into(writer).expect("serialize failed");
+    <ExpanderNoOverSubscribe<ZKCudaBN254Hyrax> as ProvingSystem<BN254Config>>::post_process();
+}

expander_compiler/src/zkcuda/mpi_mem_share.rs

@@ -65,18 +65,16 @@ impl<C: Config> MPISharedMemory for ComputationGraph<C> {
 
 impl<C: Config> MPISharedMemory for Kernel<C> {
     fn bytes_size(&self) -> usize {
-        assert!(
-            self.hint_solver.is_none(),
-            "Hint solver is not supported in MPISharedMemory for Kernel"
-        );
+        if self.hint_solver.is_some() {
+            eprintln!("Warning: Shared Memory will ignore the hint solver in Kernel");
+        }
         self.layered_circuit.bytes_size() + self.layered_circuit_input.bytes_size()
     }
 
     fn to_memory(&self, ptr: &mut *mut u8) {
-        assert!(
-            self.hint_solver.is_none(),
-            "Hint solver is not supported in MPISharedMemory for Kernel"
-        );
+        if self.hint_solver.is_some() {
+            eprintln!("Warning: Shared Memory will ignore the hint solver in Kernel");
+        }
         self.layered_circuit.to_memory(ptr);
         self.layered_circuit_input.to_memory(ptr);
     }