projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 4829b1c) | patch
the result of an inline asm copy can be an arbitrary VT that the register
authorChris Lattner <sabre@nondot.org>
Thu, 12 Apr 2007 06:00:20 +0000 (06:00 +0000)
committerChris Lattner <sabre@nondot.org>
Thu, 12 Apr 2007 06:00:20 +0000 (06:00 +0000)
commit3a508c94a6c6f124e8b0549358ce76deb92ab662
tree4c15d4dd65336edd43b24a8bafc61d708cd69f8etree | snapshot
parent4829b1c6ab810ee274a104985915988b4f92aa7acommit | diff
the result of an inline asm copy can be an arbitrary VT that the register
class supports.  In the case of vectors, this means we often get the wrong
type (e.g. we get v4f32 instead of v8i16).  Make sure to convert the vector
result to the right type.  This fixes CodeGen/X86/2007-04-11-InlineAsmVectorResult.ll

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@35944 91177308-0d34-0410-b5e6-96231b3b80d8
lib/CodeGen/SelectionDAG/SelectionDAGISel.cpp diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors