On Thu, Oct 08, 2020 at 12:06:47PM +0200, Johannes Schindelin wrote:
> My little script also finds this:
>
> -- snip --
> @@ -1231,8 +1231,8 @@ space := $(empty) $(empty)
>
> ifdef SANITIZE
> SANITIZERS := $(foreach flag,$(subst $(comma),$(space),$(SANITIZE)),$(flag))
> -BASIC_CFLAGS += -fsanitize=$(SANITIZE) -fno-sanitize-recover=$(SANITIZE)
> BASIC_CFLAGS += -fno-omit-frame-pointer
> +BASIC_CFLAGS += -fsanitize=$(SANITIZE) -fno-sanitize-recover=$(SANITIZE)
> ifneq ($(filter undefined,$(SANITIZERS)),)
> BASIC_CFLAGS += -DSHA1DC_FORCE_ALIGNED_ACCESS
> endif
> -- snap --
>
> I am not _so_ sure that we want to order `BASIC_CFLAGS`, but then, it does
> not hurt, does it?
I agree it would not be wrong to reorder here from the compiler's
perspective, but:
- the current ordering is not arbitrary; the intent was to show that
we are enabling -fsanitize, and then follow it up with any other
related options (first any that apply to all sanitizers, of which
there is only one, and then any sanitizer-specific ones). The patch
above splits that logic apart.
- I'd worry that there are cases in which order _does_ matter to the
compiler. I'm not sure if anything that goes in CFLAGS might
qualify, but certainly order can matter for other parts of the
command-line (e.g., static library order).
So it might be setting us up for confusion later.
-Peff
