This is the mail archive of the gcc-help@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

AW: AW: EABI


Hi Ian,

> >  1) Does is make a difference (w.r.t. ABI use?) if building a 
> > toolchain for i686 depending on whether I pass --target=i686-mytoolchain-linux-gnu or --target= i686-mytoolchain-linux-gnueabi to configure?
> > 2) Does is make a difference (w.r.t. ABI use?) if building a toolchain 
> > for powerpc depending on whether I pass --target=powerpc-mytoolchain-linux-gnu or --target= powerpc-mytoolchain-linux-gnueabi to configure?
> > 3) Again, does is make a difference (w.r.t. ABI use?) if building a 
> > toolchain for powerpc depending on whether I pass --target=powerpc-mytoolchain-linux-eabi or --target= powerpc-mytoolchain-linux-gnueabi to configure?
> 
> Why are you asking these questions?  Is this really what you want to know?

Well, I more or less inherited a poorly structured toolchain build environment for the target architectures i686, mibsel, mipseb, arm and powerpc, all for Linux, and I urgently need to do some cleanup work, as new versions for these architectures are to be supported in the future. And even when using crosstool-ng to do this task, I wanted to know which changes in the target tuple do really affect toolchain generation and which ones don't, but are to be recommended from the conventional point of view.
 
> To me these questions seem the wrong way around.  What matters is what ABIs are out there and what systems support them.  The way the toolchain handles them is secondary.

That's exactly what I was trying to find out, so sorry if I asked in a more laborious way, I just tried to make things easier :-).

> You can answer these questions yourself by looking at gcc/config.gcc.
> For example, in there I see
> 
> i[34567]86-*-linux* | i[34567]86-*-kfreebsd*-gnu | i[34567]86-*-knetbsd*-gnu | i[34567]86-*-gnu* |
> i[34567]86-*-kopensolaris*-gnu)
> 
> The i[34567]86-*-gnu* string tells me that i686-mytoolchain-linux-gnu and i686-mytoolchain-linux-gnueabi are identical.  Or, to put it another way, i686-mytoolchain-linux-gnueabi does > not exist as a separate configuration.

And that's the perfect answer I was looking for, thanks a lot :-). I was only looking at config.sub in the root directory of gcc (and binutils and glibc), so I wasn't aware that there is an additional location to handle these type of things for gcc.

Cheers,

Chris


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]