llvm dev - Jun 2014 - [LLVMdev] cross-section differences in MC generation

Attached.  You can also find it at
http://people.freebsd.org/~jhibbits/hello_llvm.s .

- Justin

On Thu, Jun 26, 2014 at 1:26 PM, Rafael Avila de Espindola
<rafael.espindola at gmail.com> wrote:
>
>
> Sent from my iPhone
>
>> On Jun 26, 2014, at 12:33, Justin Hibbits <jrh29 at
alumni.cwru.edu> wrote:
>>
>> I'm working on Position-independent code for 32-bit PowerPC, but
>> running into a problem.  At the beginning of each function, there's
a
>> pre-word that's the difference between the PICBase (.L1$pb) and the
>> GOT.  This works fine when generating assembly output, but it fails
>> when generating ELF output, with the error "Cannot represent a
>> difference across sections" (line 847,
lib/MC/ELFObjectWriter.cpp).
>>
>> Roman Divacky suggested I try something with 'let EncoderMethod =
...'
>> similar to how the TOC entries are done for ppc64.  However, I'm
not
>> sure if that would work, given that the TOC on ppc64 is handled
>> differently from the GOT on ppc32, and I'm not sure how to
represent
>> that if it could work.
>>
>> Any ideas on how a cross-section diff and relocation could be
>> represented for this?
>>
>
>
> Can you post the generated assembly that works with gas but fails with
llvm-mc?
>
>> Thanks,
>> Justin
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hello_llvm.s
Type: application/octet-stream
Size: 1618 bytes
Desc: not available
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20140626/5e90ac54/attachment.obj>

This reduces to

.section .foo
.L1:
.L2 = .L1
.section .bar
.long .L1-.L2


Which is fairly silly :-(

I think this is the some root issue as pr20119. I will add the
testcase to that PR.

Cheers,
Rafael

I think that's incorrect.  It should to:

.section .foo
.L1:
.L2 = .L1
.section .bar
.long .L3-.L2
.L3:

Because .L3 and .L2 are in different sections.

- Justin

On Thu, Jun 26, 2014 at 2:46 PM, Rafael Espíndola
<rafael.espindola at gmail.com> wrote:
> This reduces to
>
> .section .foo
> .L1:
> .L2 = .L1
> .section .bar
> .long .L1-.L2
>
>
> Which is fairly silly :-(
>
> I think this is the some root issue as pr20119. I will add the
> testcase to that PR.
>
> Cheers,
> Rafael