aboutsummaryrefslogtreecommitdiff
path: root/llvm/test/CodeGen/PowerPC
diff options
context:
space:
mode:
Diffstat (limited to 'llvm/test/CodeGen/PowerPC')
-rw-r--r--llvm/test/CodeGen/PowerPC/compare-vector-with-zero.ll (renamed from llvm/test/CodeGen/PowerPC/check-zero-vector.ll)77
-rw-r--r--llvm/test/CodeGen/PowerPC/fmf-propagation.ll90
2 files changed, 117 insertions, 50 deletions
diff --git a/llvm/test/CodeGen/PowerPC/check-zero-vector.ll b/llvm/test/CodeGen/PowerPC/compare-vector-with-zero.ll
index 0f7e0c7..1325abf 100644
--- a/llvm/test/CodeGen/PowerPC/check-zero-vector.ll
+++ b/llvm/test/CodeGen/PowerPC/compare-vector-with-zero.ll
@@ -95,3 +95,80 @@ declare i4 @llvm.ctpop.i4(i4) #1
!6 = !{!"short", !7, i64 0}
!7 = !{!"omnipotent char", !8, i64 0}
!8 = !{!"Simple C/C++ TBAA"}
+
+; Function to lockdown changes for floating point vector comparisons
+define range(i32 0, 5) i32 @cols_needed(ptr %colauths){
+; POWERPC_64LE-LABEL: cols_needed:
+; POWERPC_64LE: # %bb.0: # %entry
+; POWERPC_64LE-NEXT: lxv vs0, 0(r3)
+; POWERPC_64LE-NEXT: xxlxor vs1, vs1, vs1
+; POWERPC_64LE-NEXT: li r4, 4
+; POWERPC_64LE-NEXT: li r3, 0
+; POWERPC_64LE-NEXT: xvcmpeqsp vs0, vs0, vs1
+; POWERPC_64LE-NEXT: xxlnor v2, vs0, vs0
+; POWERPC_64LE-NEXT: vextuwrx r4, r4, v2
+; POWERPC_64LE-NEXT: vextuwrx r3, r3, v2
+; POWERPC_64LE-NEXT: rlwinm r4, r4, 1, 30, 30
+; POWERPC_64LE-NEXT: sub r3, r4, r3
+; POWERPC_64LE-NEXT: mfvsrwz r4, v2
+; POWERPC_64LE-NEXT: rlwinm r4, r4, 2, 29, 29
+; POWERPC_64LE-NEXT: or r3, r3, r4
+; POWERPC_64LE-NEXT: li r4, 12
+; POWERPC_64LE-NEXT: vextuwrx r4, r4, v2
+; POWERPC_64LE-NEXT: slwi r4, r4, 3
+; POWERPC_64LE-NEXT: or r3, r3, r4
+; POWERPC_64LE-NEXT: clrlwi r3, r3, 28
+; POWERPC_64LE-NEXT: stb r3, -1(r1)
+; POWERPC_64LE-NEXT: lbz r3, -1(r1)
+; POWERPC_64LE-NEXT: popcntd r3, r3
+; POWERPC_64LE-NEXT: blr
+;
+; POWERPC_64-LABEL: cols_needed:
+; POWERPC_64: # %bb.0: # %entry
+; POWERPC_64-NEXT: lxv vs0, 0(r3)
+; POWERPC_64-NEXT: xxlxor vs1, vs1, vs1
+; POWERPC_64-NEXT: li r4, 8
+; POWERPC_64-NEXT: xvcmpeqsp vs0, vs0, vs1
+; POWERPC_64-NEXT: xxlnor v2, vs0, vs0
+; POWERPC_64-NEXT: vextuwlx r4, r4, v2
+; POWERPC_64-NEXT: mfvsrwz r3, v2
+; POWERPC_64-NEXT: rlwinm r4, r4, 1, 30, 30
+; POWERPC_64-NEXT: rlwimi r4, r3, 2, 29, 29
+; POWERPC_64-NEXT: li r3, 0
+; POWERPC_64-NEXT: vextuwlx r3, r3, v2
+; POWERPC_64-NEXT: rlwimi r4, r3, 3, 0, 28
+; POWERPC_64-NEXT: li r3, 12
+; POWERPC_64-NEXT: vextuwlx r3, r3, v2
+; POWERPC_64-NEXT: sub r3, r4, r3
+; POWERPC_64-NEXT: clrlwi r3, r3, 28
+; POWERPC_64-NEXT: stb r3, -1(r1)
+; POWERPC_64-NEXT: lbz r3, -1(r1)
+; POWERPC_64-NEXT: popcntd r3, r3
+; POWERPC_64-NEXT: blr
+;
+; POWERPC_32-LABEL: cols_needed:
+; POWERPC_32: # %bb.0: # %entry
+; POWERPC_32-NEXT: lxv vs0, 0(r3)
+; POWERPC_32-NEXT: xxlxor vs1, vs1, vs1
+; POWERPC_32-NEXT: xvcmpeqsp vs0, vs0, vs1
+; POWERPC_32-NEXT: xxlnor vs0, vs0, vs0
+; POWERPC_32-NEXT: stxv vs0, -32(r1)
+; POWERPC_32-NEXT: lwz r3, -24(r1)
+; POWERPC_32-NEXT: lwz r4, -28(r1)
+; POWERPC_32-NEXT: rlwinm r3, r3, 1, 30, 30
+; POWERPC_32-NEXT: rlwimi r3, r4, 2, 29, 29
+; POWERPC_32-NEXT: lwz r4, -32(r1)
+; POWERPC_32-NEXT: rlwimi r3, r4, 3, 0, 28
+; POWERPC_32-NEXT: lwz r4, -20(r1)
+; POWERPC_32-NEXT: sub r3, r3, r4
+; POWERPC_32-NEXT: clrlwi r3, r3, 28
+; POWERPC_32-NEXT: popcntw r3, r3
+; POWERPC_32-NEXT: blr
+entry:
+ %0 = load <4 x float>, ptr %colauths, align 4, !tbaa !5
+ %1 = fcmp une <4 x float> %0, zeroinitializer
+ %2 = bitcast <4 x i1> %1 to i4
+ %3 = tail call range(i4 0, 5) i4 @llvm.ctpop.i4(i4 %2)
+ %4 = zext nneg i4 %3 to i32
+ ret i32 %4
+}
diff --git a/llvm/test/CodeGen/PowerPC/fmf-propagation.ll b/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
index e71f59c..cad684e 100644
--- a/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
+++ b/llvm/test/CodeGen/PowerPC/fmf-propagation.ll
@@ -325,24 +325,21 @@ define float @sqrt_afn_ieee(float %x) #0 {
;
; GLOBAL-LABEL: sqrt_afn_ieee:
; GLOBAL: # %bb.0:
-; GLOBAL-NEXT: addis 3, 2, .LCPI11_1@toc@ha
-; GLOBAL-NEXT: xsabsdp 0, 1
-; GLOBAL-NEXT: lfs 2, .LCPI11_1@toc@l(3)
-; GLOBAL-NEXT: fcmpu 0, 0, 2
-; GLOBAL-NEXT: xxlxor 0, 0, 0
-; GLOBAL-NEXT: blt 0, .LBB11_2
-; GLOBAL-NEXT: # %bb.1:
; GLOBAL-NEXT: xsrsqrtesp 0, 1
; GLOBAL-NEXT: vspltisw 2, -3
; GLOBAL-NEXT: addis 3, 2, .LCPI11_0@toc@ha
-; GLOBAL-NEXT: xvcvsxwdp 2, 34
-; GLOBAL-NEXT: xsmulsp 1, 1, 0
-; GLOBAL-NEXT: xsmaddasp 2, 1, 0
+; GLOBAL-NEXT: xvcvsxwdp 3, 34
+; GLOBAL-NEXT: xsmulsp 2, 1, 0
+; GLOBAL-NEXT: xsabsdp 1, 1
+; GLOBAL-NEXT: xsmaddasp 3, 2, 0
; GLOBAL-NEXT: lfs 0, .LCPI11_0@toc@l(3)
-; GLOBAL-NEXT: xsmulsp 0, 1, 0
-; GLOBAL-NEXT: xsmulsp 0, 0, 2
-; GLOBAL-NEXT: .LBB11_2:
-; GLOBAL-NEXT: fmr 1, 0
+; GLOBAL-NEXT: addis 3, 2, .LCPI11_1@toc@ha
+; GLOBAL-NEXT: xsmulsp 0, 2, 0
+; GLOBAL-NEXT: lfs 2, .LCPI11_1@toc@l(3)
+; GLOBAL-NEXT: xssubsp 1, 1, 2
+; GLOBAL-NEXT: xxlxor 2, 2, 2
+; GLOBAL-NEXT: xsmulsp 0, 0, 3
+; GLOBAL-NEXT: fsel 1, 1, 0, 2
; GLOBAL-NEXT: blr
%rt = call afn ninf float @llvm.sqrt.f32(float %x)
ret float %rt
@@ -393,21 +390,19 @@ define float @sqrt_afn_preserve_sign(float %x) #1 {
;
; GLOBAL-LABEL: sqrt_afn_preserve_sign:
; GLOBAL: # %bb.0:
-; GLOBAL-NEXT: xxlxor 0, 0, 0
-; GLOBAL-NEXT: fcmpu 0, 1, 0
-; GLOBAL-NEXT: beq 0, .LBB13_2
-; GLOBAL-NEXT: # %bb.1:
; GLOBAL-NEXT: xsrsqrtesp 0, 1
; GLOBAL-NEXT: vspltisw 2, -3
; GLOBAL-NEXT: addis 3, 2, .LCPI13_0@toc@ha
-; GLOBAL-NEXT: xvcvsxwdp 2, 34
-; GLOBAL-NEXT: xsmulsp 1, 1, 0
-; GLOBAL-NEXT: xsmaddasp 2, 1, 0
+; GLOBAL-NEXT: xvcvsxwdp 3, 34
+; GLOBAL-NEXT: xsmulsp 2, 1, 0
+; GLOBAL-NEXT: xsmaddasp 3, 2, 0
; GLOBAL-NEXT: lfs 0, .LCPI13_0@toc@l(3)
-; GLOBAL-NEXT: xsmulsp 0, 1, 0
-; GLOBAL-NEXT: xsmulsp 0, 0, 2
-; GLOBAL-NEXT: .LBB13_2:
-; GLOBAL-NEXT: fmr 1, 0
+; GLOBAL-NEXT: xsmulsp 0, 2, 0
+; GLOBAL-NEXT: xxlxor 2, 2, 2
+; GLOBAL-NEXT: xsmulsp 0, 0, 3
+; GLOBAL-NEXT: fsel 2, 1, 2, 0
+; GLOBAL-NEXT: xsnegdp 1, 1
+; GLOBAL-NEXT: fsel 1, 1, 2, 0
; GLOBAL-NEXT: blr
%rt = call afn ninf float @llvm.sqrt.f32(float %x)
ret float %rt
@@ -462,24 +457,21 @@ define float @sqrt_fast_ieee(float %x) #0 {
;
; GLOBAL-LABEL: sqrt_fast_ieee:
; GLOBAL: # %bb.0:
-; GLOBAL-NEXT: addis 3, 2, .LCPI15_1@toc@ha
-; GLOBAL-NEXT: xsabsdp 0, 1
-; GLOBAL-NEXT: lfs 2, .LCPI15_1@toc@l(3)
-; GLOBAL-NEXT: fcmpu 0, 0, 2
-; GLOBAL-NEXT: xxlxor 0, 0, 0
-; GLOBAL-NEXT: blt 0, .LBB15_2
-; GLOBAL-NEXT: # %bb.1:
; GLOBAL-NEXT: xsrsqrtesp 0, 1
; GLOBAL-NEXT: vspltisw 2, -3
; GLOBAL-NEXT: addis 3, 2, .LCPI15_0@toc@ha
-; GLOBAL-NEXT: xvcvsxwdp 2, 34
-; GLOBAL-NEXT: xsmulsp 1, 1, 0
-; GLOBAL-NEXT: xsmaddasp 2, 1, 0
+; GLOBAL-NEXT: xvcvsxwdp 3, 34
+; GLOBAL-NEXT: xsmulsp 2, 1, 0
+; GLOBAL-NEXT: xsabsdp 1, 1
+; GLOBAL-NEXT: xsmaddasp 3, 2, 0
; GLOBAL-NEXT: lfs 0, .LCPI15_0@toc@l(3)
-; GLOBAL-NEXT: xsmulsp 0, 1, 0
-; GLOBAL-NEXT: xsmulsp 0, 0, 2
-; GLOBAL-NEXT: .LBB15_2:
-; GLOBAL-NEXT: fmr 1, 0
+; GLOBAL-NEXT: addis 3, 2, .LCPI15_1@toc@ha
+; GLOBAL-NEXT: xsmulsp 0, 2, 0
+; GLOBAL-NEXT: lfs 2, .LCPI15_1@toc@l(3)
+; GLOBAL-NEXT: xssubsp 1, 1, 2
+; GLOBAL-NEXT: xxlxor 2, 2, 2
+; GLOBAL-NEXT: xsmulsp 0, 0, 3
+; GLOBAL-NEXT: fsel 1, 1, 0, 2
; GLOBAL-NEXT: blr
%rt = call contract reassoc afn ninf float @llvm.sqrt.f32(float %x)
ret float %rt
@@ -517,21 +509,19 @@ define float @sqrt_fast_preserve_sign(float %x) #1 {
;
; GLOBAL-LABEL: sqrt_fast_preserve_sign:
; GLOBAL: # %bb.0:
-; GLOBAL-NEXT: xxlxor 0, 0, 0
-; GLOBAL-NEXT: fcmpu 0, 1, 0
-; GLOBAL-NEXT: beq 0, .LBB16_2
-; GLOBAL-NEXT: # %bb.1:
; GLOBAL-NEXT: xsrsqrtesp 0, 1
; GLOBAL-NEXT: vspltisw 2, -3
; GLOBAL-NEXT: addis 3, 2, .LCPI16_0@toc@ha
-; GLOBAL-NEXT: xvcvsxwdp 2, 34
-; GLOBAL-NEXT: xsmulsp 1, 1, 0
-; GLOBAL-NEXT: xsmaddasp 2, 1, 0
+; GLOBAL-NEXT: xvcvsxwdp 3, 34
+; GLOBAL-NEXT: xsmulsp 2, 1, 0
+; GLOBAL-NEXT: xsmaddasp 3, 2, 0
; GLOBAL-NEXT: lfs 0, .LCPI16_0@toc@l(3)
-; GLOBAL-NEXT: xsmulsp 0, 1, 0
-; GLOBAL-NEXT: xsmulsp 0, 0, 2
-; GLOBAL-NEXT: .LBB16_2:
-; GLOBAL-NEXT: fmr 1, 0
+; GLOBAL-NEXT: xsmulsp 0, 2, 0
+; GLOBAL-NEXT: xxlxor 2, 2, 2
+; GLOBAL-NEXT: xsmulsp 0, 0, 3
+; GLOBAL-NEXT: fsel 2, 1, 2, 0
+; GLOBAL-NEXT: xsnegdp 1, 1
+; GLOBAL-NEXT: fsel 1, 1, 2, 0
; GLOBAL-NEXT: blr
%rt = call contract reassoc ninf afn float @llvm.sqrt.f32(float %x)
ret float %rt