This is the mail archive of the
gcc-help@gcc.gnu.org
mailing list for the GCC project.
Re: 32 bit float emulation on 16 bit machine
Thank you.
> The linker must be complaining about something. What is it complaining
> about?
I don't know more.
> That seems odd. ?If gcc can handle an unsigned 32-bit integer comparison
> in a normal program without calling __ucmpsi2, then it should be able to
> handle it in fp-bit.c as well.
I tried to add a pattern for cbranchsi4 in the machine description
with an empty template. Then no more problem with __ucmpsi2.
The anoying thing is that now I should complete the implementation of
cbranchsi4 so that GCC generates the right code,
but what I will do is to describe what GCC automatically does in
"normal" programs.
It really seems like GCC has not the same behavior when compiling
libgcc2 and when compiling normal programs.
When I look at AVR or MSP430 backends, I can some tricks to compile
fp-bit well. These tricks do not work in my case.
Regards,
Florent.