diff options
| author | Reid Kleckner <rnk@google.com> | 2019-02-26 19:48:16 +0000 |
|---|---|---|
| committer | Reid Kleckner <rnk@google.com> | 2019-02-26 19:48:16 +0000 |
| commit | 8fda7e15e6249d675b6852c60ab191b3e3cda8ba (patch) | |
| tree | 75a700d50d82b40702b1b24cbb455ab250685fc2 /llvm/lib/Target/X86 | |
| parent | 6d6afb72b3a2c8b85c3c6e3acd39d96baa6e443e (diff) | |
| download | bcm5719-llvm-8fda7e15e6249d675b6852c60ab191b3e3cda8ba.tar.gz bcm5719-llvm-8fda7e15e6249d675b6852c60ab191b3e3cda8ba.zip | |
[X86] Fix bug in vectorcall calling convention
Original implementation can't correctly handle __m256 and __m512 types
passed by reference through stack. This patch fixes it.
Patch by Wei Xiao!
Differential Revision: https://reviews.llvm.org/D57643
llvm-svn: 354921
Diffstat (limited to 'llvm/lib/Target/X86')
| -rw-r--r-- | llvm/lib/Target/X86/X86CallingConv.cpp | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/llvm/lib/Target/X86/X86CallingConv.cpp b/llvm/lib/Target/X86/X86CallingConv.cpp index 9be1147df3c..aee344a2676 100644 --- a/llvm/lib/Target/X86/X86CallingConv.cpp +++ b/llvm/lib/Target/X86/X86CallingConv.cpp @@ -162,7 +162,10 @@ static bool CC_X86_64_VectorCall(unsigned &ValNo, MVT &ValVT, MVT &LocVT, // created on top of the basic 32 bytes of win64. // It can happen if the fifth or sixth argument is vector type or HVA. // At that case for each argument a shadow stack of 8 bytes is allocated. - if (Reg == X86::XMM4 || Reg == X86::XMM5) + const TargetRegisterInfo *TRI = + State.getMachineFunction().getSubtarget().getRegisterInfo(); + if (TRI->regsOverlap(Reg, X86::XMM4) || + TRI->regsOverlap(Reg, X86::XMM5)) State.AllocateStack(8, 8); if (!ArgFlags.isHva()) { |
