[PATCH] D20378: [lld] Wrong section VMA/LMA in case of using linker scripts

[Dima Stepanov via llvm-commits]

Just want to add some more clarification here. It looks like for the
current lld code it is an expected behaviour to create two sections. The
main function to create the section key is ELF/Writer.cpp:createKey():
 return SectionKey<ELFT::Is64Bits>{OutsecName, Type, Flags, Alignment};
So there are 4 arguments to generate the key. And even more, there is a
small comment about it:
  // For SHF_MERGE we create different output sections for each alignment.
  // This makes each output section simple and keeps a single level mapping
from
  // input to output.
I can't say if it is good in general, but looks like an expected behaviour.

Also two sections can be created even without a linker script. I tried to
describe this case in https://llvm.org/bugs/show_bug.cgi?id=27805. In this
case there are two .rodata sections with different alignment values. The
good thing here is that without linker script another procedure is used to
set the VMA/LMA fields (it just go through all the sections). This approach
can't be used as is in case of the linker script.


On Thu, May 19, 2016 at 9:50 AM, Rui Ueyama <ruiu at google.com> wrote:

> ruiu accepted this revision.
> ruiu added a reviewer: ruiu.
> ruiu added a comment.
> This revision is now accepted and ready to land.
>
> LGTM
>
> I'm singing off since this patch itself doesn't at least make things
> worse, and and the new test make it explicit what we are doing.
>
>
> Repository:
>   rL LLVM
>
> http://reviews.llvm.org/D20378
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160519/6b88f7a8/attachment.html>