tech-toolchain archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: lib/csu and compilers other than GCC
In article <20100731224956.GA27782%britannica.bec.de@localhost>,
Joerg Sonnenberger <joerg%britannica.bec.de@localhost> wrote:
>-=-=-=-=-=-
>
>Hi all,
>this is the second iteration of moving ahead with the CSU code.
>
>- Split crt0.c into the assembler stub and the shared C module.
>- Slightly modify the ELF linker script to provide __eprol directly.
>- Use genassym to compute most of sysident code. I plan to drop the
> copy in ld.so_elf at a some point later, it is essentially the same.
> The only hard-coded constants are some strings as genassym can't
> extract strings yet.
>- crti.o and crtn.o are now assembler files as well. This replaces the
> dot_init.h file which had essentially the same content.
>- crtbegin.S and crtend.S are implemented for i386 and amd64. I'm going
> with the plain csu/i386, but I have no strong opinion on the name.
> The Makefile diff is primary intended to increase robustness.
>
>I have committed test cases for the major functions
>(/usr/tests/lib/csu).
>
>Not included is the change to GCC's and PCC's build system to skip the
>crtstuff in those places for architectures that use the new layout.
>
Can we make an arch subdirectory now that everything moved around?
christos
Home |
Main Index |
Thread Index |
Old Index