tinycc-devel
[Top][All Lists]
Advanced

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

[Tinycc-devel] x86 optimization flags


From: Milutin Jovanović
Subject: [Tinycc-devel] x86 optimization flags
Date: Sun, 10 Jun 2012 21:22:34 -0400


Hi all,

I am not sure if you noticed an email today about a seg fault on osx. I have traced the problem to some optimization flags. The makefile section is:

ifeq ($(ARCH),i386)
CFLAGS+=-mpreferred-stack-boundary=2
ifeq ($(GCC_MAJOR),2)
CFLAGS+=-m386 -malign-functions=0
else
CFLAGS+=-march=i386 -falign-functions=0
endif
endif

For some reason when using -mpreferred-stack-boundary=2 causes this problem.

Any way, my questions really is this, do we need this whole section at all? The stack-boundary is already not optimized on 64 bit builds. Is tcc being used in environments where this will make a difference? I think this actually slows down the code, but...

Aligning functions? Pretty much the same question. This cannot save more then a few bytes.

-march=i386, -m386? I think that -m32, which is applied later in the makefile, already does when we really need. Are we trying to produce tcc that can run on really old machines?

For now I have disabled this block for osx builds. Can we remove completely? Your thoughts?

Miki.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]