This is the mail archive of the
mailing list for the glibc project.
Re: enable n32 and n64, and move o32 into mips/mips32
- From: Roland McGrath <roland at redhat dot com>
- To: Alexandre Oliva <aoliva at redhat dot com>
- Cc: Daniel Jacobowitz <drow at mvista dot com>, Andreas Jaeger <aj at suse dot de>,libc-alpha at sources dot redhat dot com
- Date: Mon, 31 Mar 2003 02:00:38 -0800
- Subject: Re: enable n32 and n64, and move o32 into mips/mips32
> On Mar 29, 2003, Roland McGrath <roland at redhat dot com> wrote:
> > Nested in what? The definitions are at top level.
> Nope. dynamic-link.h is only #included inside other functions, so any
> functions defined in there are nested functions, e.g.
Ah, righteyo. I'm pretty sure I wrote that code that way so I suppose I
ought to have remembered that, but it was many years ago. The distinction
only matters in theory when the inline functions use local variables of
parent functions. But in fact these ones do. Nonetheless, in the compiler
I have (current RHL 3.2 branch) `static' here seems to work like `auto'
should, and `auto' makes the compiler crash. So go figure.
Given that the compiler is broken now and might get fixed so that what
works now will no longer work (static, which is wrong since the function
actually uses the parent function's locals), I think we need to forego the
__typeof solution and either macros for the argument type or change all the
dl-machine.h files. The latter is actually cleaner, I think. The current
type is not right for plenty of other relocs on other machines, that just
haven't been bitten so far.