[PATCH] Skip MachO scattered relocations in RuntimeDyld.

Jim Grosbach grosbach at apple.com
Thu Aug 8 16:30:16 PDT 2013


Hi Lang,

I agree this is fine to do, but the comment could use a bit more explanation of why it’s right to keep going rather than bail out like we do on other unhanded relocations. Adding a FIXME for any cases which actually will require applying the relocation at runtime rather than just using the pre-computed value would be good, too.

With that change, this LGTM.

-Jim

On Aug 8, 2013, at 12:36 PM, Lang Hames <lhames at gmail.com> wrote:

> Aaand I forgot to attach the patch. Here it is.
> 
> - Lang.
> 
> 
> On Thu, Aug 8, 2013 at 12:35 PM, Lang Hames <lhames at gmail.com> wrote:
> Hi All,
> 
> At the moment a scattered relocations will cause llvm-rtdyld to go off into the weeds and eventually assert while trying to look up a section that doesn't exist.
> 
> Eventually we want to handle these relocations, but in practice they often aren't required. For example, section differences only matter if symbols are moved relative to one another, which RuntimeDyld doesn't do.
> 
> This patch skips scattered relocations, which will unbreak some simple llvm-rtdyld use-cases while I work on proper support.
> 
> Cheers,
> Lang.
> 
> <skip_scattered_relocations.patch>_______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130808/248ff217/attachment.html>


More information about the llvm-commits mailing list