This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: Problem with referenceable sections ("link sets")
On Sun, Nov 17, 2002 at 10:00:35PM -0800, Ian Lance Taylor wrote:
> Well, yeah, an AT statement that applies to the .rodata section won't
> be applied to other sections which are sorted with the .rodata
> section.
Bleh. That kinda sucks :-/ Is there an easy way to make orphans
sorted into .rodata (or any other section) honor the AT? Would that
be desirable? Seems to me that it makes sense to do so ... the behavior
can't possibly be right otherwise (at least for LOAD sections).
> For that to work I think you will need to explicitly mention the
> section in your linker script. Don't make it part of the .rodata
> section, but sort it after the .rodata section.
Well, the problem with this is that the __start_SECNAME and __stop_SECNAME
symbols are not implicitly entered if I list the "link set" sections as
input sections. "link sets" are a generic feature, and I can't possibly
list all the symbols that might end up in any given kernel (fwiw, FreeBSD
uses this feature extensively in their kernel, as well).
Could you point me at where to add the code to insert these symbols for
explicitly-listed C-referenceable sections? This is where my eyes began
to glaze over before :-)
> Or you might be able to use memory regions.
...but to use memory regions, I still need to explcitly list the
"link set" sections, right? This brings me back to the marker symbols
problem...
--
-- Jason R. Thorpe <thorpej@wasabisystems.com>