This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: gold: Support for signalling reloc overflows?
On Tue, Jul 15, 2008 at 3:20 PM, Ian Lance Taylor <iant@google.com> wrote:
> "Will Newton" <will.newton@gmail.com> writes:
>
>> It doesn't appear to me that any of the current ports of gold check
>> for overflows in relocation processing. Is there some place I'm not
>> seeing where this is happening? It seems like it should be necessary.
>
> You are correct on both counts. gold doesn't check for relocation
> overflows, and it should.
>
> This hasn't been a priority for me because 1) relocation overflows
> don't happen in correct programs; 2) relocation overflows don't happen
They are more likely to happen on my architecture than perhaps others
- some of our instructions take 6 bit signed offsets!
> in C/C++ code on x86/x86_64. But it's definitely something that needs
> to be addressed. It's not hard.
Would it be sufficient to print an error within
Target::Relocate::relocate and return true when overflow occurs or is
there some higher level management I need to take care of?
> Ian
>