diff --git a/lib/CodeGen/TargetInfo.cpp b/lib/CodeGen/TargetInfo.cpp index 88f8579b9aa2ff98378402fda0fff84df390d781..33291dd5527b1a92894434b9b371d4a2675e3b58 100644 --- a/lib/CodeGen/TargetInfo.cpp +++ b/lib/CodeGen/TargetInfo.cpp @@ -1983,19 +1983,17 @@ static llvm::Value *EmitVAArgFromMemory(llvm::Value *VAListAddr, // AMD64-ABI 3.5.7p5: Step 7. Align l->overflow_arg_area upwards to a 16 // byte boundary if alignment needed by type exceeds 8 byte boundary. + // It isn't stated explicitly in the standard, but in practice we use + // alignment greater than 16 where necessary. uint64_t Align = CGF.getContext().getTypeAlign(Ty) / 8; if (Align > 8) { - // Note that we follow the ABI & gcc here, even though the type - // could in theory have an alignment greater than 16. This case - // shouldn't ever matter in practice. - - // overflow_arg_area = (overflow_arg_area + 15) & ~15; + // overflow_arg_area = (overflow_arg_area + align - 1) & -align; llvm::Value *Offset = - llvm::ConstantInt::get(CGF.Int32Ty, 15); + llvm::ConstantInt::get(CGF.Int64Ty, Align - 1); overflow_arg_area = CGF.Builder.CreateGEP(overflow_arg_area, Offset); llvm::Value *AsInt = CGF.Builder.CreatePtrToInt(overflow_arg_area, CGF.Int64Ty); - llvm::Value *Mask = llvm::ConstantInt::get(CGF.Int64Ty, ~15LL); + llvm::Value *Mask = llvm::ConstantInt::get(CGF.Int64Ty, -(uint64_t)Align); overflow_arg_area = CGF.Builder.CreateIntToPtr(CGF.Builder.CreateAnd(AsInt, Mask), overflow_arg_area->getType(), diff --git a/test/CodeGen/x86_64-arguments.c b/test/CodeGen/x86_64-arguments.c index 8571ac9655528052ff7d5912157ea6c54ca85a5a..ea7dd79c790919dd8c1520e6b7dbc0fb98ad89aa 100644 --- a/test/CodeGen/x86_64-arguments.c +++ b/test/CodeGen/x86_64-arguments.c @@ -304,3 +304,17 @@ extern void func42(SA s); void func43(SA s) { func42(s); } + +// CHECK: define i32 @f44 +// CHECK: ptrtoint +// CHECK-NEXT: and {{.*}}, -32 +// CHECK-NEXT: inttoptr +typedef int T44 __attribute((vector_size(32))); +struct s44 { T44 x; int y; }; +int f44(int i, ...) { + __builtin_va_list ap; + __builtin_va_start(ap, i); + struct s44 s = __builtin_va_arg(ap, struct s44); + __builtin_va_end(ap); + return s.y; +}