Skip to content

[LoongArch][NFC] Pre-commit for converting vector mask to vXi1 using [X]VMSKLTZ #142977

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 1 commit into from
Jun 6, 2025
Merged

[LoongArch][NFC] Pre-commit for converting vector mask to vXi1 using [X]VMSKLTZ #142977

merged 1 commit into from
Jun 6, 2025

Conversation

heiher
Copy link
Member

@heiher heiher commented Jun 5, 2025

No description provided.

@llvmbot
Copy link
Member

llvmbot commented Jun 5, 2025

@llvm/pr-subscribers-backend-loongarch

Author: hev (heiher)

Changes

Patch is 94.70 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/142977.diff

2 Files Affected:

  • (modified) llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll (+1576)
  • (modified) llvm/test/CodeGen/LoongArch/lsx/vmskcond.ll (+979)
diff --git a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
index fb3937c476b03..a02da74855795 100644
--- a/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
+++ b/llvm/test/CodeGen/LoongArch/lasx/xvmskcond.ll
@@ -16,6 +16,148 @@ entry:
   ret i32 %2
 }
 
+define i32 @xmsk_sgt_allzeros_i8(<32 x i8 > %a) {
+; CHECK-LABEL: xmsk_sgt_allzeros_i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    addi.d $sp, $sp, -64
+; CHECK-NEXT:    .cfi_def_cfa_offset 64
+; CHECK-NEXT:    st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset 1, -8
+; CHECK-NEXT:    .cfi_offset 22, -16
+; CHECK-NEXT:    addi.d $fp, $sp, 64
+; CHECK-NEXT:    .cfi_def_cfa 22, 0
+; CHECK-NEXT:    bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT:    xvrepli.b $xr1, 0
+; CHECK-NEXT:    xvslt.b $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvst $xr0, $sp, 0
+; CHECK-NEXT:    vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 8
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 9
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 10
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 11
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 12
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 13
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 14
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 16
+; CHECK-NEXT:    slli.d $a1, $a1, 15
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 17
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 16
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 18
+; CHECK-NEXT:    slli.d $a1, $a1, 17
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 19
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 18
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 20
+; CHECK-NEXT:    slli.d $a1, $a1, 19
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 21
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 20
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 22
+; CHECK-NEXT:    slli.d $a1, $a1, 21
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 23
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 22
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 24
+; CHECK-NEXT:    slli.d $a1, $a1, 23
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 25
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 24
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 26
+; CHECK-NEXT:    slli.d $a1, $a1, 25
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 27
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 26
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 28
+; CHECK-NEXT:    slli.d $a1, $a1, 27
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 29
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 28
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 30
+; CHECK-NEXT:    slli.d $a1, $a1, 29
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.b $a1, $sp, 31
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 30
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    slli.d $a1, $a1, 31
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT:    addi.d $sp, $fp, -64
+; CHECK-NEXT:    ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT:    addi.d $sp, $sp, 64
+; CHECK-NEXT:    ret
+entry:
+  %1 = icmp sgt <32 x i8> %a, splat (i8 0)
+  %2 = bitcast <32 x i1> %1 to i32
+  ret i32 %2
+}
+
 define i32 @xmsk_sgt_allones_i8(<32 x i8 > %a) {
 ; CHECK-LABEL: xmsk_sgt_allones_i8:
 ; CHECK:       # %bb.0: # %entry
@@ -100,6 +242,147 @@ entry:
   ret i4 %2
 }
 
+define i32 @xmsk_sle_allzeros_i8(<32 x i8 > %a) {
+; CHECK-LABEL: xmsk_sle_allzeros_i8:
+; CHECK:       # %bb.0: # %entry
+; CHECK-NEXT:    addi.d $sp, $sp, -64
+; CHECK-NEXT:    .cfi_def_cfa_offset 64
+; CHECK-NEXT:    st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset 1, -8
+; CHECK-NEXT:    .cfi_offset 22, -16
+; CHECK-NEXT:    addi.d $fp, $sp, 64
+; CHECK-NEXT:    .cfi_def_cfa 22, 0
+; CHECK-NEXT:    bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT:    xvslei.b $xr0, $xr0, 0
+; CHECK-NEXT:    xvst $xr0, $sp, 0
+; CHECK-NEXT:    vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 8
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 9
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 10
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 11
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 12
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 13
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 13
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 14
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 14
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 15
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 16
+; CHECK-NEXT:    slli.d $a1, $a1, 15
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 17
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 16
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 18
+; CHECK-NEXT:    slli.d $a1, $a1, 17
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 19
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 18
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 20
+; CHECK-NEXT:    slli.d $a1, $a1, 19
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 21
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 20
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 22
+; CHECK-NEXT:    slli.d $a1, $a1, 21
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 23
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 22
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 24
+; CHECK-NEXT:    slli.d $a1, $a1, 23
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 25
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 24
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 26
+; CHECK-NEXT:    slli.d $a1, $a1, 25
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 27
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 26
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 28
+; CHECK-NEXT:    slli.d $a1, $a1, 27
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.bu $a1, $sp, 29
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 28
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.bu $a2, $sp, 30
+; CHECK-NEXT:    slli.d $a1, $a1, 29
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.b $a1, $sp, 31
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 30
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    slli.d $a1, $a1, 31
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    bstrpick.d $a0, $a0, 31, 0
+; CHECK-NEXT:    addi.d $sp, $fp, -64
+; CHECK-NEXT:    ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT:    addi.d $sp, $sp, 64
+; CHECK-NEXT:    ret
+entry:
+  %1 = icmp sle <32 x i8> %a, splat (i8 0)
+  %2 = bitcast <32 x i1> %1 to i32
+  ret i32 %2
+}
+
 define i32 @xmsk_sle_allones_i8(<32 x i8 > %a) {
 ; CHECK-LABEL: xmsk_sle_allones_i8:
 ; CHECK:       # %bb.0: # %entry
@@ -169,3 +452,1296 @@ entry:
   %2 = bitcast <32 x i1> %1 to i32
   ret i32 %2
 }
+
+define i4 @xvmsk_sgt_v4i64(<4 x i64> %a, <4 x i64> %b) {
+; CHECK-LABEL: xvmsk_sgt_v4i64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    xvslt.d $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 1
+; CHECK-NEXT:    sub.d $a0, $a1, $a0
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT:    slli.d $a1, $a1, 3
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    andi $a0, $a0, 15
+; CHECK-NEXT:    addi.d $sp, $sp, 16
+; CHECK-NEXT:    ret
+  %x = icmp sgt <4 x i64> %a, %b
+  %res = bitcast <4 x i1> %x to i4
+  ret i4 %res
+}
+
+define i4 @xvmsk_ogt_v4f64(<4 x double> %a, <4 x double> %b) {
+; CHECK-LABEL: xvmsk_ogt_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    xvfcmp.clt.d $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvpickve2gr.d $a0, $xr0, 0
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 1
+; CHECK-NEXT:    sub.d $a0, $a1, $a0
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    xvpickve2gr.d $a1, $xr0, 3
+; CHECK-NEXT:    slli.d $a1, $a1, 3
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    andi $a0, $a0, 15
+; CHECK-NEXT:    addi.d $sp, $sp, 16
+; CHECK-NEXT:    ret
+  %x = fcmp ogt <4 x double> %a, %b
+  %res = bitcast <4 x i1> %x to i4
+  ret i4 %res
+}
+
+define i8 @xvmsk_sgt_v8i32(<8 x i32> %a, <8 x i32> %b) {
+; CHECK-LABEL: xvmsk_sgt_v8i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    xvslt.w $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    andi $a0, $a0, 255
+; CHECK-NEXT:    addi.d $sp, $sp, 16
+; CHECK-NEXT:    ret
+  %x = icmp sgt <8 x i32> %a, %b
+  %res = bitcast <8 x i1> %x to i8
+  ret i8 %res
+}
+
+define i8 @xvmsk_ogt_v8f32(<8 x float> %a, <8 x float> %b) {
+; CHECK-LABEL: xvmsk_ogt_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -16
+; CHECK-NEXT:    .cfi_def_cfa_offset 16
+; CHECK-NEXT:    xvfcmp.clt.s $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvpickve2gr.w $a0, $xr0, 0
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    xvpickve2gr.w $a1, $xr0, 7
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    andi $a0, $a0, 255
+; CHECK-NEXT:    addi.d $sp, $sp, 16
+; CHECK-NEXT:    ret
+  %x = fcmp ogt <8 x float> %a, %b
+  %res = bitcast <8 x i1> %x to i8
+  ret i8 %res
+}
+
+define i16 @xvmsk_sgt_v16i16(<16 x i16> %a, <16 x i16> %b) {
+; CHECK-LABEL: xvmsk_sgt_v16i16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -64
+; CHECK-NEXT:    .cfi_def_cfa_offset 64
+; CHECK-NEXT:    st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset 1, -8
+; CHECK-NEXT:    .cfi_offset 22, -16
+; CHECK-NEXT:    addi.d $fp, $sp, 64
+; CHECK-NEXT:    .cfi_def_cfa 22, 0
+; CHECK-NEXT:    bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT:    xvslt.h $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvst $xr0, $sp, 0
+; CHECK-NEXT:    vpickve2gr.h $a0, $vr0, 0
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.h $a1, $vr0, 7
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.hu $a2, $sp, 16
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.hu $a1, $sp, 18
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 8
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.hu $a2, $sp, 20
+; CHECK-NEXT:    slli.d $a1, $a1, 9
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.hu $a1, $sp, 22
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 10
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.hu $a2, $sp, 24
+; CHECK-NEXT:    slli.d $a1, $a1, 11
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.hu $a1, $sp, 26
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 12
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    ld.hu $a2, $sp, 28
+; CHECK-NEXT:    slli.d $a1, $a1, 13
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    ld.h $a1, $sp, 30
+; CHECK-NEXT:    andi $a2, $a2, 1
+; CHECK-NEXT:    slli.d $a2, $a2, 14
+; CHECK-NEXT:    or $a0, $a0, $a2
+; CHECK-NEXT:    slli.d $a1, $a1, 15
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    bstrpick.d $a0, $a0, 15, 0
+; CHECK-NEXT:    addi.d $sp, $fp, -64
+; CHECK-NEXT:    ld.d $fp, $sp, 48 # 8-byte Folded Reload
+; CHECK-NEXT:    ld.d $ra, $sp, 56 # 8-byte Folded Reload
+; CHECK-NEXT:    addi.d $sp, $sp, 64
+; CHECK-NEXT:    ret
+  %x = icmp sgt <16 x i16> %a, %b
+  %res = bitcast <16 x i1> %x to i16
+  ret i16 %res
+}
+
+define i32 @xvmsk_sgt_v32i8(<32 x i8> %a, <32 x i8> %b) {
+; CHECK-LABEL: xvmsk_sgt_v32i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    addi.d $sp, $sp, -64
+; CHECK-NEXT:    .cfi_def_cfa_offset 64
+; CHECK-NEXT:    st.d $ra, $sp, 56 # 8-byte Folded Spill
+; CHECK-NEXT:    st.d $fp, $sp, 48 # 8-byte Folded Spill
+; CHECK-NEXT:    .cfi_offset 1, -8
+; CHECK-NEXT:    .cfi_offset 22, -16
+; CHECK-NEXT:    addi.d $fp, $sp, 64
+; CHECK-NEXT:    .cfi_def_cfa 22, 0
+; CHECK-NEXT:    bstrins.d $sp, $zero, 4, 0
+; CHECK-NEXT:    xvslt.b $xr0, $xr1, $xr0
+; CHECK-NEXT:    xvst $xr0, $sp, 0
+; CHECK-NEXT:    vpickve2gr.b $a0, $vr0, 0
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 1
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    bstrins.d $a0, $a1, 63, 1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 2
+; CHECK-NEXT:    bstrins.d $a0, $a1, 2, 2
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 3
+; CHECK-NEXT:    bstrins.d $a0, $a1, 3, 3
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 4
+; CHECK-NEXT:    bstrins.d $a0, $a1, 4, 4
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 5
+; CHECK-NEXT:    bstrins.d $a0, $a1, 5, 5
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 6
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 6
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 7
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 7
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 8
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 8
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 9
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 9
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 10
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 10
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 11
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 11
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 12
+; CHECK-NEXT:    andi $a1, $a1, 1
+; CHECK-NEXT:    slli.d $a1, $a1, 12
+; CHECK-NEXT:    or $a0, $a0, $a1
+; CHECK-NEXT:    vpickve2gr.b $a1, $vr0, 1...
[truncated]

@github-actions GitHub Actions
Copy link

github-actions bot commented Jun 5, 2025 •
edited
Loading

✅ With the latest revision this PR passed the undef deprecator.

@heiher heiher requested a review from tangaac June 6, 2025 00:35
@heiher heiher merged commit 182c1c2 into llvm:main Jun 6, 2025
9 of 11 checks passed
@heiher heiher deleted the vmsk-3 branch June 6, 2025 08:26
rorth pushed a commit to rorth/llvm-project that referenced this pull request Jun 11, 2025
@heiher @rorth
[LoongArch][NFC] Pre-commit for converting vector mask to vXi1 usin…
027a0c8 
…g `[X]VMSKLTZ` (llvm#142977)
DhruvSrivastavaX pushed a commit to DhruvSrivastavaX/lldb-for-aix that referenced this pull request Jun 12, 2025
@heiher @DhruvSrivastavaX
[LoongArch][NFC] Pre-commit for converting vector mask to vXi1 usin…
23900e8 
…g `[X]VMSKLTZ` (llvm#142977)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tangaac tangaac tangaac approved these changes

Assignees
No one assigned
Labels
backend:loongarch
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@heiher @llvmbot @tangaac