This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH]: testsuite fixes for loadpre5.c/loadpre8.c with -fpic/-fPIC


From: "Hans-Peter Nilsson" <hp@bitrange.com>

Looks like all those functions are pure/const and the calls likely removed (no that's *not* inlining. ;) Even more so, now that they're static. A perfect point of the caveat presented in <http://gcc.gnu.org/ml/gcc/2008-01/msg00165.html>.

(Stick an asm ("") in there or add -fno-ipa-pure-const; you
can't make a static fn weak.)

I tried both of these. Either solution you suggest makes the testcase fail "make check" under all circumstances on x86_64-unknown-linux-gnu. I.e. pic or nonpic, -m32 or -m64.


Looking at the logfile it compiles, but it fails the scan. So I left it as-is.

It's not entirely clear to me whether const-call elimination is occuring here, and whether it wasn't already occuring in the original testcase before I make the function(s) static, and if so whether that already invalidated the testcase or not.

       --Kaveh
--
Kaveh R. Ghazi


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]