Guilhem Bryant

There are already the `RuntimeManagerRequest` and `RuntimeManagerResponse` structures which are the messages that are (de)serialized by protobuf and used to communicate between the client and the runtime manager. Should we...

> What are the properties of the instances that are experiencing the failures? Have they been up a long time? Because originally, `nsm` was not plumbed into `/dev/random`. Is it...

Opened an issue upstream: https://github.com/aws/aws-nitro-enclaves-sdk-bootstrap/issues/15

`rust-mbedtls` doesn't get recompiled when making changes to `mbedtls_hardware_poll.c`: I suggest we add the following code after https://github.com/veracruz-project/rust-mbedtls/blob/c9df29d0016de1d67ce597d6669e1ebda12dfb3d/mbedtls-sys/build/build.rs#L108: ``` // hardware poll println!( "cargo:rerun-if-changed={}", self.mbedtls_src .join(Path::new("..").join("build").join("mbedtls_hardware_poll.c")) .display() ); ```

Passes end to end with the tiny YOLO model on Nitro :+1: It doesn't pass with the big model but that's orthogonal to this PR

On a related note, debugging messages should be more explicit when Veracruz is running in debug mode. As of now runtime manager errors are hidden behind generic `VMStatus` errors. As...

Alex and I struggled with the `VeracruzClientError::InvalidLengthError` error message returned by `read_cert()`. Could that message be improved and clearly explain why the certificate isn't valid?

Same goes for the file access error when there is no match for the program hash. `FileSystemError` should be replaced with `HashMismatch`

Relates to #155

Changing the log level of message dumps from `info` to `trace` would help optimize the CI. Low hanging fruits: * https://github.com/veracruz-project/veracruz/blob/59121f0c1956634fd2a869f6ac893c8ee4da9a52/runtime-manager/src/runtime_manager_linux.rs#L251 * https://github.com/veracruz-project/veracruz/blob/59121f0c1956634fd2a869f6ac893c8ee4da9a52/runtime-manager/src/runtime_manager_linux.rs#L378