On Friday, May 10, 2024 9:25:56 PM GMT+2 Borislav Petkov wrote: > On Fri, May 10, 2024 at 09:00:34PM +0200, Fabio M. De Francesco wrote: > > I thought that ELOG and GHES should be modeled consistently. ghes_proc() > > prints to the console while ghes_do_proc() also uses ftrace. > > ghes_proc() calls ghes_do_proc(). I have no clue what you mean here. > My understanding is that ghes_proc() logs to the console and ghes_do_proc() calls the tracers. Therefore, GHES at the same time always reports the errors via two different means. Instead ELOG depends on the check on ras_userspace_consumers() to decide whether to call print_extlog_rcd() to print the logs. And if it print to the kernel logs, it jumps to "out" and skips the tracers. Why is it different with respect to how error reporting is made in GHES? I thought that ELOG should be modeled similarly to GHES and so it should print to the kernel logs always unconditionally and then it should also use ftrace (no goto "out" and skip tracers). (1) Is my understanding of logging and tracing in ELOG and GHES correct? (2) If it is, does it make sense for ELOG to print to the kernel log, unconditionally, and then call the tracers like ghes_proc() + ghes_do_proc() do? Fabio
