https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121222
--- Comment #8 from Kees Cook <kees at outflux dot net> --- Linux uses 100 by default, but it is configurable. I have not measured the overhead lately, but mcount is pretty expensive, and with the 100 byte stack size guard, the performance overhead is, by definition, smaller with -fsanitize-coverage=stack-depth in callback mode.
