shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity

[PowerPC] Fix issue where vsrq is given incorrect shift vector

Browse Source 
The new Power10 instruction vsrq was being given the wrong shift vector.
The original code assumed that the shift would be found in bits 121 to 127.
This is not correct. The shift is found in bits 57 to 63.
This can be fixed by swaping the first and second double words.

Reviewed By: nemanjai, #powerpc

Differential Revision: https://reviews.llvm.org/D94113
This commit is contained in:
Stefan Pintilie 2021-01-05 14:42:53 -06:00
parent 816dba48af
commit cb0c034edc
2 changed files with 35 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

41
llvm/lib/Target/PowerPC/PPCInstrPrefix.td
View File

@ -2521,18 +2521,6 @@ let Predicates = [IsISA3_1] in {
(EXTRACT_SUBREG (XVTLSBB (COPY_TO_REGCLASS $XB, VSRC)), sub_lt)>;
def : Pat<(i32 (int_ppc_vsx_xvtlsbb v16i8:$XB, 0)),
(EXTRACT_SUBREG (XVTLSBB (COPY_TO_REGCLASS $XB, VSRC)), sub_eq)>;
def : Pat<(v1i128 (shl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSLQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat<(v1i128 (PPCshl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSLQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat<(v1i128 (srl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat<(v1i128 (PPCsrl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat<(v1i128 (sra v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRAQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat<(v1i128 (PPCsra v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRAQ v1i128:$VRA, v1i128:$VRB))>;
def : Pat <(v1i128 (PPClxvrzx xoaddr:$src, 8)),
(v1i128 (COPY_TO_REGCLASS (LXVRBX xoaddr:$src), VRRC))>;
@ -2570,6 +2558,35 @@ let AddedComplexity = 400, Predicates = [IsISA3_1, IsLittleEndian] in {
(STXVRDX $src, xoaddr:$dst)>;
}
// FIXME: The swap is overkill when the shift amount is a constant.
// We should just fix the constant in the DAG.
let AddedComplexity = 400, Predicates = [IsISA3_1, HasVSX] in {
def : Pat<(v1i128 (shl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSLQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
def : Pat<(v1i128 (PPCshl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSLQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
def : Pat<(v1i128 (srl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
def : Pat<(v1i128 (PPCsrl v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
def : Pat<(v1i128 (sra v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRAQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
def : Pat<(v1i128 (PPCsra v1i128:$VRA, v1i128:$VRB)),
(v1i128 (VSRAQ v1i128:$VRA,
(XXPERMDI (COPY_TO_REGCLASS $VRB, VSRC),
(COPY_TO_REGCLASS $VRB, VSRC), 2)))>;
}
class xxevalPattern <dag pattern, bits<8> imm> :
Pat<(v4i32 pattern), (XXEVAL $vA, $vB, $vC, imm)> {}

6
llvm/test/CodeGen/PowerPC/p10-vector-shift.ll
View File

@ -13,6 +13,7 @@
define dso_local <1 x i128> @test_vec_vslq(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vslq:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vslq v2, v2, v3
; CHECK-NEXT: blr
entry:
@ -24,6 +25,7 @@ entry:
define dso_local <1 x i128> @test_vec_vsrq(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vsrq:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vsrq v2, v2, v3
; CHECK-NEXT: blr
entry:
@ -35,6 +37,7 @@ entry:
define dso_local <1 x i128> @test_vec_vsraq(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vsraq:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vsraq v2, v2, v3
; CHECK-NEXT: blr
entry:
@ -46,6 +49,7 @@ entry:
define dso_local <1 x i128> @test_vec_vslq2(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vslq2:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vslq v2, v2, v3
; CHECK-NEXT: blr
entry:
@ -56,6 +60,7 @@ entry:
define dso_local <1 x i128> @test_vec_vsrq2(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vsrq2:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vsrq v2, v2, v3
; CHECK-NEXT: blr
entry:
@ -66,6 +71,7 @@ entry:
define dso_local <1 x i128> @test_vec_vsraq2(<1 x i128> %a, <1 x i128> %b) {
; CHECK-LABEL: test_vec_vsraq2:
; CHECK: # %bb.0: # %entry
; CHECK-NEXT: xxswapd v3, v3
; CHECK-NEXT: vsraq v2, v2, v3
; CHECK-NEXT: blr
entry: