Reentrant logging, as it is implemented now, is barely useful. The only way reentracy could be helpful is iff there is an exception in the log function itself. However, this probably means that log state is corrupted anyways and log should not be trusted. Instead, there should be a trusted subset of log_nolock that would handle exceptions in log itself. For example, it can just output to e9.
Reentrant logging, as it is implemented now, is barely useful. The only way reentracy could be helpful is iff there is an exception in the log function itself. However, this probably means that log state is corrupted anyways and log should not be trusted. Instead, there should be a trusted subset of
log_nolockthat would handle exceptions in log itself. For example, it can just output to e9.