Avoid caching the relocation model on the subtarget, this is for
authorEric Christopher <echristo@gmail.com>
Fri, 18 Jul 2014 22:34:20 +0000 (22:34 +0000)
committerEric Christopher <echristo@gmail.com>
Fri, 18 Jul 2014 22:34:20 +0000 (22:34 +0000)
commitf74faf42fe964d12cf31bf25f55501b26d600f24
tree283a84aeb2ae774720820cc43df1a391d89d281a
parent396a649014460776c64d75aa6455debfc40872da
Avoid caching the relocation model on the subtarget, this is for
two reasons:

a) we're already caching the target machine which contains it,
b) which relocation model you get is dependent upon whether or
not you ask before MCCodeGenInfo is constructed on the target
machine, so avoid any latent issues there.

llvm-svn: 213420
llvm/lib/Target/Mips/MipsSubtarget.cpp
llvm/lib/Target/Mips/MipsSubtarget.h
llvm/lib/Target/Mips/MipsTargetMachine.cpp