[llvm] r210814 - CodeGen: enable mov.w/mov.t pairs with minsize for WoA
Saleem Abdulrasool
abdulras at fb.com
Thu Jun 12 17:11:06 PDT 2014
On Jun 12, 2014, at 1:46 PM, Tim Northover <t.p.northover at gmail.com> wrote:
>> + // NOTE Windows on ARM needs to use mov.w/mov.t pairs to materialise 32-bit
>> + // immediates as it is inherently position independent, and may be out of
>> + // range otherwise.
>
> What does "inherently position independent" mean here? Because I can
> see no way in which the fragment you're testing could be loaded at an
> arbitrary address without runtime fixup.
>
> And if you are relying on runtime fixup, "inherently
> position-dependent" would probably be a better description.
Yes, these are accompanied by a IMAGE_REL_ARM_MOV32T relocation which will be adjusted if loaded an address other than the preferred base address. I suppose that its a perspective thing: Windows requires that everything can be rebased, which makes it position independent. The load itself is position dependent though. I can adjust the comment if you like.
> Cheers.
>
> Tim.
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> https://urldefense.proofpoint.com/v1/url?u=http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits&k=ZVNjlDMF0FElm4dQtryO4A%3D%3D%0A&r=CchYc4lrV44%2BZqxZADw0BQ%3D%3D%0A&m=771Q22JHVG64Ylagr7M3zPyY7E%2FqX0GwkR51T5cpnrU%3D%0A&s=3beff91608b689be043d9c0206f7744765b4f46bf03faea94c41c49e499dfc1a
--
Saleem Abdulrasool
abdulras (at) fb (dot) com
More information about the llvm-commits
mailing list