llvm-mirror/test/CodeGen/X86/2009-06-12-x86_64-tail-call-conv-out-of-sync-bug.ll

; RUN: llc < %s -tailcallopt -march=x86-64 -mattr=+sse2 -mtriple=x86_64-apple-darwin | grep fstpt
; RUN: llc < %s -tailcallopt -march=x86-64 -mattr=+sse2 -mtriple=x86_64-apple-darwin | grep xmm

; Check that x86-64 tail calls support x86_fp80 and v2f32 types. (Tail call
; calling convention out of sync with standard c calling convention on x86_64)
; Bug 4278.

declare fastcc double @tailcallee(x86_fp80, <2 x float>) 
	
define fastcc double @tailcall() {
entry:
  %tmp = fpext float 1.000000e+00 to x86_fp80
	%tmp2 = tail call fastcc double @tailcallee( x86_fp80 %tmp,  <2 x float> <float 1.000000e+00, float 1.000000e+00>)
	ret double %tmp2
}
Eliminate more uses of llvm-as and llvm-dis. llvm-svn: 81290 2009-09-09 01:54:48 +02:00			`; RUN: llc < %s -tailcallopt -march=x86-64 -mattr=+sse2 -mtriple=x86_64-apple-darwin \| grep fstpt`
			`; RUN: llc < %s -tailcallopt -march=x86-64 -mattr=+sse2 -mtriple=x86_64-apple-darwin \| grep xmm`
Fix Bug 4278: X86-64 with -tailcallopt calling convention out of sync with regular cc. The only difference between the tail call cc and the normal cc was that one parameter register - R9 - was reserved for calling functions through a function pointer. After time the tail call cc has gotten out of sync with the regular cc. We can use R11 which is also caller saved but not used as parameter register for potential function pointers and remove the special tail call cc on x86-64. llvm-svn: 73233 2009-06-12 18:26:57 +02:00
			`; Check that x86-64 tail calls support x86_fp80 and v2f32 types. (Tail call`
			`; calling convention out of sync with standard c calling convention on x86_64)`
			`; Bug 4278.`

			`declare fastcc double @tailcallee(x86_fp80, <2 x float>)`

			`define fastcc double @tailcall() {`
			`entry:`
			`%tmp = fpext float 1.000000e+00 to x86_fp80`
			`%tmp2 = tail call fastcc double @tailcallee( x86_fp80 %tmp, <2 x float> <float 1.000000e+00, float 1.000000e+00>)`
			`ret double %tmp2`
			`}`