This is the mail archive of the libc-alpha@sources.redhat.com mailing list for the glibc 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]

Re: PPC startup code


Roland McGrath writes:

> This makes it sound like it's part of a formal spec, which it's not.  The
> registers at startup are part of the ELF ABI spec.  But __libc_start_main
> and its calling sequence are just part of the glibc ABI, which can and does
> change between glibc releases.  There is just the set of practical issues I
> mentioned with the part of the ABI involving crt1.o changing, which is rare
> but not intrinsically different from other glibc ABI changes that are fine.

True.  If we assume (a) we are not going to change the calling
sequence for _start, and (b) we want to use the generic
__libc_start_main, then it is apparent that we need a shim to
rearrange the arguments of _start for the generic __libc_start_main.
The question is, should that be in _start itself (and therefore linked
into every executable) or in libc.so?

I am tending to think that it should be in libc.so, mainly because
putting it in start.S will make it a little larger, and introduce some
potential problems for not very much gain.  IMHO.

Paul.


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