https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121222
--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> --- (In reply to Kees Cook from comment #6) > The primary difference is the compile-time guard for instrumentation that > depends on stack usage. What value do you normally pass for -fsanitize-coverage-stack-depth-callback-min= for the kernel? Is it something like 1024 or smaller like 16? But I don't see why you can't use the same mechanism of mcount for this? What is the benifit besides the not always calling the call back if you don't use a "larger" stack size? What is the overhead using mcount compared to this option? Is there a specifications on why the kernel needs yet another call back with actual benchmarking on why another one?
