This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] tests: move -D_FORTIFY_SOURCE to CPPFLAGS
On Mon, May 7, 2012 at 12:55 AM, Mike Frysinger <vapier@gentoo.org> wrote:
> On Sunday 06 May 2012 19:53:53 Carlos O'Donell wrote:
>> On Sun, May 6, 2012 at 6:51 PM, Mike Frysinger wrote:
>> > The -D_FORTIFY_SOURCE flags are being added to the test-specific CFLAGS
>> > values when they should be the test-specific CPPFLAGS values. ?Otherwise
>> > the values don't override the user-specified CPPFLAGS and might end up
>> > breaking the tests.
>>
>> Is this because the tests in question are being compiled with g++?
>
> no. ?the current compile order is:
> ? ? ? ?$(CC) $(CFLAGS) $(CPPFLAGS) ...
>
> and CFLAGS expands into:
> ? ? ? ?$(CFLAGS-config) ... $($(subdir)-CFLAGS) ... $(CFLAGS-$(<F)) ...
>
> and CPPFLAGS expands into:
> ? ? ? ?$(CPPFLAGS-config) ... $($(subdir)-CPPFLAGS) ... $(CPPFLAGS-$(<F)) ...
>
> so the user's settings are in CPPFLAGS-config, but the test-specific overrides
> are in CFLAGS-xxx, so the -D flags we want to use don't override the user's
> settings.
Shucks, that sucks. The intent is clear though the users flags should
be overridden by the test flags.
I'm OK with this patch provided you include a comment in
debug/Makefile stating why you can't just use CFLAGS.
Doesn't this effect *every* test that adds their own test-specific flags?
It seems like the compile order should expand into:
$(CC) $(CFLAGS) $(CPPFLAGS) $(TESTFLAGS)
I'd like to see incremental progress, but fixing the testsuite first
is the right way forward.
Please check this in on Wednesday if nobody else objects.
Cheers,
Carlos.