This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RFC PATCH] Don't use ELF weak for entities with vague linkagewhen COMDAT groups are available
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Julian Brown <julian at codesourcery dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>, GCJ Patches <java-patches at gcc dot gnu dot org>
- Date: Tue, 12 Apr 2005 19:27:38 -0700
- Subject: Re: [RFC PATCH] Don't use ELF weak for entities with vague linkagewhen COMDAT groups are available
- Organization: CodeSourcery, LLC
- References: <4256FCD3.5030004@codesourcery.com>
Julian Brown wrote:
Hi,
This is a patch to disable ELF weak for entities with vague linkage when
HAVE_GAS_COMDAT_GROUP is true. It was previously posted in two parts and
applied, then reverted soon after for causing breakage:
If people would try this patch and run the testsuite on their machines,
I'd be grateful. I'd appreciate it if you post information about
failures which show up for you, with version numbers of binutils bits,
so I can attempt to track down whatever broke before.
You've been met with deafening silence, which is disappointing.
As a result, I'm going to approve the patch again. After you check it
in, please watch carefully for failure reports. If you're reporting a
failure with this patch, *please* include assembler and linker versions,
so that Julian has a shot at reproducing the problem.
(I am suspicious that there are systems out there that have an assembler
with COMDAT support, but a linker without, and that this is the cause of
the problem. In that case, the fix will be to tweak the configure
script not to define HAVE_GAS_COMDAT in that case. Or, at least, to
provide a --disable-comdat option.)
Julian, if problems appear, please do not revert the patch immediately.
Instead, please immediately disable it on the affected target system
by temporarily adding the old definition of MAKE_DECL_ONE_ONLY to the
affected target system. And keep track so that we can undo this hackery
once the problem is resolved.
If you get more reports of problems on more than two systems, then
dorevert your patch -- but put the new definition of MAKE_DECL_ONE_ONLY
in an appropriate config/arm header file, so that at least we have it
there, where it originated.
I realize that this is all a bit irregular, but this is a broadly useful
patch, and it really *should* work. I can't see how to get
information about what's going on without checking the patch, given your
diligent testing, and the lack of response to your request-for-testing.
Thanks,
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304