PVF: re-preparing artifact on failed runtime construction (#3187)

resolve https://github.com/paritytech/polkadot-sdk/issues/3139

- [x] use a distinguishable error for `execute_artifact`
- [x] remove artifact in case of a `RuntimeConstruction` error during
the execution
- [x] augment the `validate_candidate_with_retry` of `ValidationBackend`
with the case of retriable `RuntimeConstruction` error during the
execution
- [x] update the book
(https://paritytech.github.io/polkadot-sdk/book/node/utility/pvf-host-and-workers.html#retrying-execution-requests)
- [x] add a test
- [x] run zombienet tests

---------

Co-authored-by: s0me0ne-unkn0wn <48632512+s0me0ne-unkn0wn@users.noreply.github.com>

polkadot/node/core/pvf/common/src/error.rs

@@ -16,6 +16,7 @@
 
 use crate::prepare::{PrepareSuccess, PrepareWorkerSuccess};
 use parity_scale_codec::{Decode, Encode};
+pub use sc_executor_common::error::Error as ExecuteError;
 
 /// Result of PVF preparation from a worker, with checksum of the compiled PVF and stats of the
 /// preparation if successful.

polkadot/node/core/pvf/common/src/execute.rs

@@ -40,6 +40,9 @@ pub enum WorkerResponse {
 	},
 	/// The candidate is invalid.
 	InvalidCandidate(String),
+	/// Instantiation of the WASM module instance failed during an execution.
+	/// Possibly related to local issues or dirty node update. May be retried with re-preparation.
+	RuntimeConstruction(String),
 	/// The job timed out.
 	JobTimedOut,
 	/// The job process has died. We must kill the worker just in case.
@@ -68,6 +71,9 @@ pub enum JobResponse {
 		/// The result of parachain validation.
 		result_descriptor: ValidationResult,
 	},
+	/// A possibly transient runtime instantiation error happened during the execution; may be
+	/// retried with re-preparation
+	RuntimeConstruction(String),
 	/// The candidate is invalid.
 	InvalidCandidate(String),
 }
@@ -81,6 +87,15 @@ impl JobResponse {
 			Self::InvalidCandidate(format!("{}: {}", ctx, msg))
 		}
 	}
+
+	/// Creates a may retry response from a context `ctx` and a message `msg` (which can be empty).
+	pub fn runtime_construction(ctx: &'static str, msg: &str) -> Self {
+		if msg.is_empty() {
+			Self::RuntimeConstruction(ctx.to_string())
+		} else {
+			Self::RuntimeConstruction(format!("{}: {}", ctx, msg))
+		}
+	}
 }
 
 /// An unexpected error occurred in the execution job process. Because this comes from the job,

polkadot/node/core/pvf/common/src/executor_interface.rs

@@ -16,6 +16,7 @@
 
 //! Interface to the Substrate Executor
 
+use crate::error::ExecuteError;
 use polkadot_primitives::{
 	executor_params::{DEFAULT_LOGICAL_STACK_MAX, DEFAULT_NATIVE_STACK_MAX},
 	ExecutorParam, ExecutorParams,
@@ -109,7 +110,7 @@ pub unsafe fn execute_artifact(
 	compiled_artifact_blob: &[u8],
 	executor_params: &ExecutorParams,
 	params: &[u8],
-) -> Result<Vec<u8>, String> {
+) -> Result<Vec<u8>, ExecuteError> {
 	let mut extensions = sp_externalities::Extensions::new();
 
 	extensions.register(sp_core::traits::ReadRuntimeVersionExt::new(ReadRuntimeVersion));
@@ -123,7 +124,6 @@ pub unsafe fn execute_artifact(
 		Ok(Ok(ok)) => Ok(ok),
 		Ok(Err(err)) | Err(err) => Err(err),
 	}
-	.map_err(|err| format!("execute error: {:?}", err))
 }
 
 /// Constructs the runtime for the given PVF, given the artifact bytes.