The following issue is a follow-up ticket to track some remaining comments from the PR #695 - [it would be neat to think about changing the exit ABI so that when tracing is enabled we always bring along a trace structure in some extra registers and can coalesce the exits.](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2220497987) - [produce a separate file with call site -> string message mappings, and only record the call site number in the buffer](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2220497992) - [check tracing output for expected values](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2220497997) - [reconsider the `flush` macro](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2220498021) - [Unify the hypervisors registers with the `read_trace_req` to not have yet another type that represents registers](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2220498017) - [Check whether the TraceBuffer generic parameter could be replaced by a closure to improve perf](https://github.com/hyperlight-dev/hyperlight/pull/695#discussion_r2223562563)
The following issue is a follow-up ticket to track some remaining comments from the PR #695
flushmacroread_trace_reqto not have yet another type that represents registers