This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Properly do the LTO bytecode version check
- From: Richard Biener <rguenther at suse dot de>
- To: gcc-patches at gcc dot gnu dot org
- Date: Tue, 4 Mar 2014 15:26:07 +0100 (CET)
- Subject: Re: [PATCH] Properly do the LTO bytecode version check
- Authentication-results: sourceware.org; auth=none
- References: <alpine dot LSU dot 2 dot 11 dot 1403041220480 dot 11121 at zhemvz dot fhfr dot qr> <alpine dot DEB dot 2 dot 10 dot 1403041509400 dot 4627 at stedding dot saclay dot inria dot fr>
On Tue, 4 Mar 2014, Marc Glisse wrote:
> On Tue, 4 Mar 2014, Richard Biener wrote:
>
> > We're doing the LTO bytecode version check only for two section
> > types at the moment - specifically _not_ for the first section
> > we read. Which causes us to crash instead of reporting a
> > version mismatch ...
>
> Not for 4.9, but when the object is fat, could we downgrade this version
> mismatch error to a warning and use the non-LTO code? (Or does that need to be
> done in ld?) The current workaround is to strip the objects involved I think.
Yes, that needs to be done in the linker plugin - it simply should
not claim the object (or issue it to a proper GCC version ...).
For this reason the check is still in the wrong place ... but how
we identify LTO objects in the linker-plugin is simply gross.
Richard.