[llvm] cad5328 - [X86] Add late tail duplication tests with computed gotos.
Florian Hahn via llvm-commits
llvm-commits at lists.llvm.org
Mon Jul 28 01:39:23 PDT 2025
Author: Florian Hahn
Date: 2025-07-28T09:37:16+01:00
New Revision: cad5328b009aab73e68afc7b61fe7aa0af29c594
URL: https://github.com/llvm/llvm-project/commit/cad5328b009aab73e68afc7b61fe7aa0af29c594
DIFF: https://github.com/llvm/llvm-project/commit/cad5328b009aab73e68afc7b61fe7aa0af29c594.diff
LOG: [X86] Add late tail duplication tests with computed gotos.
Add a new test for post-regalloc tail duplication with computed gotos to
complement llvm/test/CodeGen/X86/tail-dup-computed-goto.mir.
Added:
llvm/test/CodeGen/X86/late-tail-dup-computed-goto.mir
Modified:
Removed:
################################################################################
diff --git a/llvm/test/CodeGen/X86/late-tail-dup-computed-goto.mir b/llvm/test/CodeGen/X86/late-tail-dup-computed-goto.mir
new file mode 100644
index 0000000000000..e272e7ee3cb0f
--- /dev/null
+++ b/llvm/test/CodeGen/X86/late-tail-dup-computed-goto.mir
@@ -0,0 +1,128 @@
+# NOTE: Assertions have been autogenerated by utils/update_mir_test_checks.py UTC_ARGS: --version 5
+# RUN: llc -mtriple=x86_64-unknown-linux-gnu -run-pass=tailduplication -tail-dup-pred-size=1 -tail-dup-succ-size=1 %s -o - | FileCheck %s
+#
+# Check that only the computed gotos are duplicated aggressively.
+#
+--- |
+ @computed_goto.dispatch = constant [5 x ptr] [ptr null, ptr blockaddress(@computed_goto, %bb1), ptr blockaddress(@computed_goto, %bb2), ptr blockaddress(@computed_goto, %bb3), ptr blockaddress(@computed_goto, %bb4)]
+ declare i64 @f0()
+ declare i64 @f1()
+ declare i64 @f2()
+ declare i64 @f3()
+ declare i64 @f4()
+ declare i64 @f5()
+ define void @computed_goto() {
+ start:
+ ret void
+ bb1:
+ ret void
+ bb2:
+ ret void
+ bb3:
+ ret void
+ bb4:
+ ret void
+ }
+ define void @jump_table() { ret void }
+ define void @jump_table_pic() { ret void }
+...
+---
+name: computed_goto
+alignment: 1
+exposesReturnsTwice: false
+legalized: false
+regBankSelected: false
+selected: false
+failedISel: false
+tracksRegLiveness: true
+body: |
+ ; CHECK-LABEL: name: computed_goto
+ ; CHECK: bb.0:
+ ; CHECK-NEXT: successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ ; CHECK-NEXT: [[COPY:%[0-9]+]]:gr64 = COPY $rax
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY]]
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
+ ; CHECK-NEXT: JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.1.bb1 (ir-block-address-taken %ir-block.bb1):
+ ; CHECK-NEXT: successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ ; CHECK-NEXT: [[COPY3:%[0-9]+]]:gr64 = COPY $rax
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY3]]
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
+ ; CHECK-NEXT: JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.2.bb2 (ir-block-address-taken %ir-block.bb2):
+ ; CHECK-NEXT: successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ ; CHECK-NEXT: [[COPY4:%[0-9]+]]:gr64 = COPY $rax
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY4]]
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
+ ; CHECK-NEXT: JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.3.bb3 (ir-block-address-taken %ir-block.bb3):
+ ; CHECK-NEXT: successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ ; CHECK-NEXT: [[COPY5:%[0-9]+]]:gr64 = COPY $rax
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY5]]
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
+ ; CHECK-NEXT: JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: bb.4.bb4 (ir-block-address-taken %ir-block.bb4):
+ ; CHECK-NEXT: successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+ ; CHECK-NEXT: {{ $}}
+ ; CHECK-NEXT: CALL64pcrel32 target-flags(x86-plt) @f4, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ ; CHECK-NEXT: [[COPY6:%[0-9]+]]:gr64 = COPY $rax
+ ; CHECK-NEXT: [[COPY1:%[0-9]+]]:gr64_nosp = COPY [[COPY6]]
+ ; CHECK-NEXT: [[COPY2:%[0-9]+]]:gr64_nosp = COPY [[COPY1]]
+ ; CHECK-NEXT: JMP64m $noreg, 8, [[COPY2]], @computed_goto.dispatch, $noreg
+ bb.0:
+ successors: %bb.5(0x80000000)
+
+ CALL64pcrel32 target-flags(x86-plt) @f0, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ %0:gr64 = COPY $rax
+ %6:gr64_nosp = COPY %0
+ JMP_1 %bb.5
+
+ bb.1.bb1 (ir-block-address-taken %ir-block.bb1):
+ successors: %bb.5(0x80000000)
+
+ CALL64pcrel32 target-flags(x86-plt) @f1, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ %1:gr64 = COPY $rax
+ %6:gr64_nosp = COPY %1
+ JMP_1 %bb.5
+
+ bb.2.bb2 (ir-block-address-taken %ir-block.bb2):
+ successors: %bb.5(0x80000000)
+
+ CALL64pcrel32 target-flags(x86-plt) @f2, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ %2:gr64 = COPY $rax
+ %6:gr64_nosp = COPY %2
+ JMP_1 %bb.5
+
+ bb.3.bb3 (ir-block-address-taken %ir-block.bb3):
+ successors: %bb.5(0x80000000)
+
+ CALL64pcrel32 target-flags(x86-plt) @f3, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ %3:gr64 = COPY $rax
+ %6:gr64_nosp = COPY %3
+ JMP_1 %bb.5
+
+ bb.4.bb4 (ir-block-address-taken %ir-block.bb4):
+ successors: %bb.5(0x80000000)
+
+ CALL64pcrel32 target-flags(x86-plt) @f4, csr_64, implicit $rsp, implicit $ssp, implicit-def $rsp, implicit-def $ssp, implicit-def $rax
+ %4:gr64 = COPY $rax
+ %6:gr64_nosp = COPY %4
+
+ bb.5:
+ successors: %bb.1(0x20000000), %bb.2(0x20000000), %bb.3(0x20000000), %bb.4(0x20000000)
+
+ %5:gr64_nosp = COPY %6
+ JMP64m $noreg, 8, %5, @computed_goto.dispatch, $noreg
+...
More information about the llvm-commits
mailing list