1; Linux doesn't support stack realignment for functions with allocas (PR2888). 2; Until it does, we shouldn't use movaps to access the stack. On targets with 3; sufficiently aligned stack (e.g. darwin) we should. 4; PR8969 - make 32-bit linux have a 16-byte aligned stack 5; RUN: llc < %s -mtriple=i386-pc-linux-gnu -mcpu=yonah | grep movaps | count 2 6; RUN: llc < %s -mtriple=i686-apple-darwin9 -mcpu=yonah | grep movaps | count 2 7 8 9target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" 10target triple = "i386-pc-linux-gnu" 11 12define void @foo(i32 %t) nounwind { 13 %tmp1210 = alloca i8, i32 32, align 4 14 call void @llvm.memset.p0i8.i64(i8* %tmp1210, i8 0, i64 32, i32 4, i1 false) 15 %x = alloca i8, i32 %t 16 call void @dummy(i8* %x) 17 ret void 18} 19 20declare void @dummy(i8*) 21 22declare void @llvm.memset.p0i8.i64(i8* nocapture, i8, i64, i32, i1) nounwind 23