On 07/08/19 14:51, Alex Bennée wrote:
>> For i386 specifically, this allows using the host GCC
>> to compile the i386 tests. But, it should really be
>> done for all targets, unless we want to pass $(EXTRA_CFLAGS)
>> directly as part of $(CC).
> Hmm well for softmmu the tests take the decision:
>
> # For softmmu targets we include a different Makefile fragement as the
> # build options for bare programs are usually pretty different. They
> # are expected to provide their own build recipes.
>
> So we are not expecting to handle multi-classing the system compiler to
> generate different binaries with a common build string.
We almost do, since we have
case "$cpu" in
i386)
CPU_CFLAGS="-m32"
LDFLAGS="-m32 $LDFLAGS"
cross_cc_i386=$cc
cross_cc_cflags_i386="$CPU_CFLAGS"
So you can compile with "./configure --cpu=i386" and have it use "gcc
-m32" as a cross compiler on x86_64, also for TCG tests. So it seemed
half-baked to me.
> To be honest I
> wonder if we should just drop the EXTRA_CFLAGS shenanigans in favour of
> an explicit compiler per target?
I don't know. I think supporting "gcc -m32" on x86_64 is a good idea
though. Yet another possibility is to use "--cross-cc-i386='gcc -m32'"
directly, but that would be a separate patch. For these I really wanted
to have no semantic change.
Paolo
