[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Request for review: Enlightenment DR17 + EFL



Didier Casse wrote:

On 5/31/05, Tom 'spot' Callaway <tcallawa redhat com> wrote:


On Mon, 2005-05-30 at 13:35 +0800, Didier Casse wrote:


The solution, as discussed with Carsten, to avoid this in the future
would be to set:

CFLAGS="-O2 -fomit-frame-pointer"
export CFLAGS

in the spec file for a generic binary.


Umm, lets not be so hasty to blast away the RPM_OPT_FLAGS. If you need
-fomit-frame-pointer for a package, do this instead:

CFLAGS="$RPM_OPT_FLAGS -fomit-frame-pointer"

~spot



Hey John, For EVAS we require the full: CFLAGS="-O2 -march=pentium4 -fomit-frame-pointer -msse -mmmx -pipe"

But for x86_64 we can't use -march=pentium4, can you try the following
in the spec file and tell me if it compiles correctly:

%build
CFLAGS="$RPM_OPT_FLAGS -fomit-frame-pointer -msse -mmx -pipe"
export CFLAGS

If not, then we remove the "-msse -mmx -pipe" options. Thanks a lot.




Didier.

It was rpmbuild'ing OK without changing CFLAGS once I removed /usr/lib from PKG_CONFIG_PATH
and stopped it picking up the 32bit version of valgrind. I think the PKG_CONFIG_PATH
was just my local bug and not one that should hit Fedora x86_64 users on systems with unmodified environments.


I tried adding:

	%build
	CFLAGS="$RPM_OPT_FLAGS -fomit-frame-pointer -msse -mmmx -pipe"
	export CFLAGS

(-mmmx with 3 m's)
as you requested and it still compiles OK, but I'm not what that proves? Is there some kind of
test I should do to verify the effect of the flags?


John


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]