Ed Hill wrote:
On Tue, 2005-08-23 at 14:04 -0600, Orion Poplawski wrote:
Jakub Jelinek wrote:
But in that case not just for x86_64, but for all architectures.
If the library is small and is expected to be only linked into small shared
libraries, -fpic can be used instead.
What small means is architecture dependent, on i386/x86_64 -fpic/-fPIC are
identical, on other arches it can mean e.g. 4K got entries limit or similar.
But I don't know what shared libraries it will be link to, so shouldn't
I use -fPIC to be safe? Or are there other drawbacks?
Hi Orion,
According to the GCC docs, -fpic and -fPIC are basically identical in
i386 systems since they have no inherent limits on the GOT size. There
is, however, a limit on PowerPC systems! So, in the interest of not
wasting time on these fiddly annoyances that will only potentially bite
Fedora ppc users, I'd skip -fpic and go straight to -fPIC. Please! ;-)
And I think it would be nice if adding -fPIC was strongly recommended
for all *.a libs since people almost inevitably want to use them (at
some later date) to build shared libs.
While we're on this topic, I have a related question. I've been looking
at the Debian version of a package I'm working on, and in their
buildscripts they build a static library without -fPIC and then rebuild
with -fPIC for the shared library. It seems like unnecessary extra
compilation. Is there any possible performance loss (or any other side
effect) that can come from using -fPIC?
-Quentin
