This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: make .eh_frame read-only
- From: Hans-Peter Nilsson <hp at bitrange dot com>
- To: Richard Henderson <rth at redhat dot com>
- Cc: binutils at sources dot redhat dot com
- Date: Fri, 6 Jun 2003 09:09:39 -0400 (EDT)
- Subject: Re: make .eh_frame read-only
On Thu, 5 Jun 2003, Richard Henderson wrote:
> On Thu, Jun 05, 2003 at 10:45:11PM -0400, Hans-Peter Nilsson wrote:
> > Unless "I have no ideas what I'm talking about", on some
> > architectures there'll still have to be run-time relocations for
> > DSO:s there, so forcing read-only for all would be wrong; that'd
> > force DT_TEXTREL for those targets.
>
> Nope. Elsewhere I unconditionally force pc-relative relocations:
>
> out_one (DW_EH_PE_pcrel | DW_EH_PE_sdata4);
Right; my point was that .eh_frame also contains output from
gcc, and cfi_finish is called unconditionally. But I see that
cfi_finish returns early if no CFI directives were used.
So, I guess targets that want to use the new CFI directives
should only emit pc-relative relocs in their other (i.e. GCC)
.eh_frame output.
Anyway, none of this seem to matter, since the writable bit is
or:ed with a section merge and (regarding flags setting)
.eh_frame in elf.sc is still put after data with no target
override. Never mind.
brgds, H-P