[llvm-commits] arm mc: fix relocation of an alias to an alias

[Ivan Krasin]

On Fri, Oct 7, 2011 at 2:32 PM, Nick Lewycky <nlewycky at google.com> wrote:
> On 7 October 2011 14:00, Ivan Krasin <krasin at google.com> wrote:
>>
>> Nick,
>>
>> could you please take a quick look? I see your recent change to this file.
>
> It's not clear to me that this is correct. Aliases can be weak and you
> shouldn't follow them if they are.
If I don't follow them, llc will crash.
(And the IR-level optimizers should
> probably be folding away aliases that are marked 'internal', so there's no
> sense doing it here.)
Sorry, I have not got the point. At this stage ELF file is being
emitted. There's no IR-level optimizers ahead. Could you please
explain it a little bit?

> Nick
>>
>> On Fri, Oct 7, 2011 at 9:49 AM, Jim Grosbach <grosbach at apple.com> wrote:
>> > This looks OK to me, but I'd really like if someone more familiar with
>> > the ELF bits would review the details.
>> >
>> > -Jim
>> >
>> > On Oct 5, 2011, at 5:33 PM, Ivan Krasin wrote:
>> >
>> >> Hi llvm team!
>> >>
>> >> The patch attached fixes the problem that
>> >> ARMELFObjectWriter::ExplicitRelSym didn't resolve aliases.
>> >> For some reason, the single alias works just fine, but the alias to
>> >> alias crashes llc with the following message:
>> >>
>> >> llc: /usr/local/google/home/krasin/llvm/include/llvm/MC/MCSymbol.h:111:
>> >> const llvm::MCSection &llvm::MCSymbol::getSection() const: Assertion
>> >> `isInSection() && "Invalid accessor!"' failed.
>> >>
>> >> The full background of this issue is available here:
>> >> http://code.google.com/p/nativeclient/issues/detail?id=2340
>> >>
>> >> OK to commit?
>> >>
>> >> krasin
>> >>
>> >> <arm_reloc_alias_alias.patch>_______________________________________________
>> >> llvm-commits mailing list
>> >> llvm-commits at cs.uiuc.edu
>> >> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> >
>> >
>
>