Skip to content
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.

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

[RISCV] Add additional fence for amocas when required by recent ABI change #101023

Merged
merged 4 commits into from
Sep 19, 2024
Merged
Show file tree
Hide file tree
Changes from 1 commit
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 16 additions & 0 deletions llvm/lib/Target/RISCV/RISCVCodeGenPrepare.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -10,6 +10,8 @@
// It munges the code in the input function to better prepare it for
// SelectionDAG-based code generation. This works around limitations in it's
// basic-block-at-a-time approach.
// It additionally implements a fence insertion for an atomic cmpxchg in a
// case that isn't easy to do with the current AtomicExpandPass hooks API.
//
//===----------------------------------------------------------------------===//

Expand Down Expand Up @@ -59,6 +61,7 @@ class RISCVCodeGenPrepare : public FunctionPass,
bool visitAnd(BinaryOperator &BO);
bool visitIntrinsicInst(IntrinsicInst &I);
bool expandVPStrideLoad(IntrinsicInst &I);
bool visitAtomicCmpXchgInst(AtomicCmpXchgInst &I);
};

} // end anonymous namespace
Expand Down Expand Up @@ -212,6 +215,19 @@ bool RISCVCodeGenPrepare::expandVPStrideLoad(IntrinsicInst &II) {
return true;
}

// Insert a leading fence (needed for broadest atomics ABI compatibility)
// only if the Zacas extension is enabled and the AtomicCmpXchgInst has a
// SequentiallyConsistent failure ordering.
bool RISCVCodeGenPrepare::visitAtomicCmpXchgInst(AtomicCmpXchgInst &I) {
IRBuilder<> Builder(&I);
if (!ST->hasStdExtZacas() ||
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Can we ever reach here with an AtomicCmpXchgInst that won't be lowered to an AMOCAS, e.g. if it's oversized?

Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I guess that's cmpxchg_i64_seq_cst_seq_cst, and this works because AtomicExpandPass is in addIRPasses not addCodeGenPrepare so comes first and we thus only expect legal cmpxchg instructions here?

I.getFailureOrdering() != AtomicOrdering::SequentiallyConsistent)
return false;

Builder.CreateFence(AtomicOrdering::SequentiallyConsistent);
return true;
}

bool RISCVCodeGenPrepare::runOnFunction(Function &F) {
if (skipFunction(F))
return false;
Expand Down
4 changes: 4 additions & 0 deletions llvm/lib/Target/RISCV/RISCVISelLowering.h
Original file line number Diff line number Diff line change
Expand Up @@ -680,6 +680,10 @@ class RISCVTargetLowering : public TargetLowering {

bool preferZeroCompareBranch() const override { return true; }

// Note that one specific case requires fence insertion for an
// AtomicCmpXchgInst but is handled via RISCVCodeGenPrepare rather than this
// hook due to limitations in the interface here (see RISCVCodeGenPrepare
// for more information).
bool shouldInsertFencesForAtomic(const Instruction *I) const override {
return isa<LoadInst>(I) || isa<StoreInst>(I);
}
Expand Down
5 changes: 5 additions & 0 deletions llvm/test/CodeGen/RISCV/atomic-cmpxchg-branch-on-result.ll
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,7 @@ define void @cmpxchg_and_branch1(ptr %ptr, i32 signext %cmp, i32 signext %val) n
; ZACAS: # %bb.0: # %entry
; ZACAS-NEXT: .LBB0_1: # %do_cmpxchg
; ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; ZACAS-NEXT: fence rw, rw
; ZACAS-NEXT: mv a3, a1
; ZACAS-NEXT: amocas.w.aqrl a3, a2, (a0)
; ZACAS-NEXT: bne a3, a1, .LBB0_1
Expand Down Expand Up @@ -76,6 +77,7 @@ define void @cmpxchg_and_branch2(ptr %ptr, i32 signext %cmp, i32 signext %val) n
; ZACAS: # %bb.0: # %entry
; ZACAS-NEXT: .LBB1_1: # %do_cmpxchg
; ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; ZACAS-NEXT: fence rw, rw
; ZACAS-NEXT: mv a3, a1
; ZACAS-NEXT: amocas.w.aqrl a3, a2, (a0)
; ZACAS-NEXT: beq a3, a1, .LBB1_1
Expand Down Expand Up @@ -216,6 +218,7 @@ define void @cmpxchg_masked_and_branch1(ptr %ptr, i8 signext %cmp, i8 signext %v
; RV64IA-ZABHA: # %bb.0: # %entry
; RV64IA-ZABHA-NEXT: .LBB2_1: # %do_cmpxchg
; RV64IA-ZABHA-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-ZABHA-NEXT: fence rw, rw
; RV64IA-ZABHA-NEXT: mv a3, a1
; RV64IA-ZABHA-NEXT: amocas.b.aqrl a3, a2, (a0)
; RV64IA-ZABHA-NEXT: bne a3, a1, .LBB2_1
Expand Down Expand Up @@ -368,6 +371,7 @@ define void @cmpxchg_masked_and_branch2(ptr %ptr, i8 signext %cmp, i8 signext %v
; RV64IA-ZABHA: # %bb.0: # %entry
; RV64IA-ZABHA-NEXT: .LBB3_1: # %do_cmpxchg
; RV64IA-ZABHA-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-ZABHA-NEXT: fence rw, rw
; RV64IA-ZABHA-NEXT: mv a3, a1
; RV64IA-ZABHA-NEXT: amocas.b.aqrl a3, a2, (a0)
; RV64IA-ZABHA-NEXT: beq a3, a1, .LBB3_1
Expand Down Expand Up @@ -408,6 +412,7 @@ define void @cmpxchg_and_irrelevant_branch(ptr %ptr, i32 signext %cmp, i32 signe
; ZACAS: # %bb.0: # %entry
; ZACAS-NEXT: .LBB4_1: # %do_cmpxchg
; ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; ZACAS-NEXT: fence rw, rw
; ZACAS-NEXT: mv a4, a1
; ZACAS-NEXT: amocas.w.aqrl a4, a2, (a0)
; ZACAS-NEXT: beqz a3, .LBB4_1
Expand Down
14 changes: 14 additions & 0 deletions llvm/test/CodeGen/RISCV/atomic-cmpxchg.ll
Original file line number Diff line number Diff line change
Expand Up @@ -1857,6 +1857,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
;
; RV64IA-WMO-ZABHA-LABEL: cmpxchg_i8_seq_cst_seq_cst:
; RV64IA-WMO-ZABHA: # %bb.0:
; RV64IA-WMO-ZABHA-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-NEXT: amocas.b.aqrl a1, a2, (a0)
; RV64IA-WMO-ZABHA-NEXT: ret
;
Expand Down Expand Up @@ -1885,6 +1886,7 @@ define void @cmpxchg_i8_seq_cst_seq_cst(ptr %ptr, i8 %cmp, i8 %val) nounwind {
;
; RV64IA-TSO-ZABHA-LABEL: cmpxchg_i8_seq_cst_seq_cst:
; RV64IA-TSO-ZABHA: # %bb.0:
; RV64IA-TSO-ZABHA-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-NEXT: amocas.b a1, a2, (a0)
; RV64IA-TSO-ZABHA-NEXT: ret
%res = cmpxchg ptr %ptr, i8 %cmp, i8 %val seq_cst seq_cst
Expand Down Expand Up @@ -3787,6 +3789,7 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
;
; RV64IA-WMO-ZABHA-LABEL: cmpxchg_i16_seq_cst_seq_cst:
; RV64IA-WMO-ZABHA: # %bb.0:
; RV64IA-WMO-ZABHA-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-NEXT: amocas.h.aqrl a1, a2, (a0)
; RV64IA-WMO-ZABHA-NEXT: ret
;
Expand Down Expand Up @@ -3816,6 +3819,7 @@ define void @cmpxchg_i16_seq_cst_seq_cst(ptr %ptr, i16 %cmp, i16 %val) nounwind
;
; RV64IA-TSO-ZABHA-LABEL: cmpxchg_i16_seq_cst_seq_cst:
; RV64IA-TSO-ZABHA: # %bb.0:
; RV64IA-TSO-ZABHA-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-NEXT: amocas.h a1, a2, (a0)
; RV64IA-TSO-ZABHA-NEXT: ret
%res = cmpxchg ptr %ptr, i16 %cmp, i16 %val seq_cst seq_cst
Expand Down Expand Up @@ -4788,6 +4792,7 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
;
; RV32IA-WMO-ZACAS-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV32IA-WMO-ZACAS: # %bb.0:
; RV32IA-WMO-ZACAS-NEXT: fence rw, rw
; RV32IA-WMO-ZACAS-NEXT: amocas.w.aqrl a1, a2, (a0)
; RV32IA-WMO-ZACAS-NEXT: ret
;
Expand All @@ -4804,6 +4809,7 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
;
; RV32IA-TSO-ZACAS-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV32IA-TSO-ZACAS: # %bb.0:
; RV32IA-TSO-ZACAS-NEXT: fence rw, rw
; RV32IA-TSO-ZACAS-NEXT: amocas.w a1, a2, (a0)
; RV32IA-TSO-ZACAS-NEXT: ret
;
Expand Down Expand Up @@ -4834,11 +4840,13 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
;
; RV64IA-WMO-ZACAS-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV64IA-WMO-ZACAS: # %bb.0:
; RV64IA-WMO-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZACAS-NEXT: amocas.w.aqrl a1, a2, (a0)
; RV64IA-WMO-ZACAS-NEXT: ret
;
; RV64IA-WMO-ZABHA-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV64IA-WMO-ZABHA: # %bb.0:
; RV64IA-WMO-ZABHA-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-NEXT: amocas.w.aqrl a1, a2, (a0)
; RV64IA-WMO-ZABHA-NEXT: ret
;
Expand All @@ -4856,11 +4864,13 @@ define void @cmpxchg_i32_seq_cst_seq_cst(ptr %ptr, i32 %cmp, i32 %val) nounwind
;
; RV64IA-TSO-ZACAS-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV64IA-TSO-ZACAS: # %bb.0:
; RV64IA-TSO-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZACAS-NEXT: amocas.w a1, a2, (a0)
; RV64IA-TSO-ZACAS-NEXT: ret
;
; RV64IA-TSO-ZABHA-LABEL: cmpxchg_i32_seq_cst_seq_cst:
; RV64IA-TSO-ZABHA: # %bb.0:
; RV64IA-TSO-ZABHA-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-NEXT: amocas.w a1, a2, (a0)
; RV64IA-TSO-ZABHA-NEXT: ret
%res = cmpxchg ptr %ptr, i32 %cmp, i32 %val seq_cst seq_cst
Expand Down Expand Up @@ -5753,11 +5763,13 @@ define void @cmpxchg_i64_seq_cst_seq_cst(ptr %ptr, i64 %cmp, i64 %val) nounwind
;
; RV64IA-WMO-ZACAS-LABEL: cmpxchg_i64_seq_cst_seq_cst:
; RV64IA-WMO-ZACAS: # %bb.0:
; RV64IA-WMO-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZACAS-NEXT: amocas.d.aqrl a1, a2, (a0)
; RV64IA-WMO-ZACAS-NEXT: ret
;
; RV64IA-WMO-ZABHA-LABEL: cmpxchg_i64_seq_cst_seq_cst:
; RV64IA-WMO-ZABHA: # %bb.0:
; RV64IA-WMO-ZABHA-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-NEXT: amocas.d.aqrl a1, a2, (a0)
; RV64IA-WMO-ZABHA-NEXT: ret
;
Expand All @@ -5774,11 +5786,13 @@ define void @cmpxchg_i64_seq_cst_seq_cst(ptr %ptr, i64 %cmp, i64 %val) nounwind
;
; RV64IA-TSO-ZACAS-LABEL: cmpxchg_i64_seq_cst_seq_cst:
; RV64IA-TSO-ZACAS: # %bb.0:
; RV64IA-TSO-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZACAS-NEXT: amocas.d a1, a2, (a0)
; RV64IA-TSO-ZACAS-NEXT: ret
;
; RV64IA-TSO-ZABHA-LABEL: cmpxchg_i64_seq_cst_seq_cst:
; RV64IA-TSO-ZABHA: # %bb.0:
; RV64IA-TSO-ZABHA-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-NEXT: amocas.d a1, a2, (a0)
; RV64IA-TSO-ZABHA-NEXT: ret
%res = cmpxchg ptr %ptr, i64 %cmp, i64 %val seq_cst seq_cst
Expand Down
14 changes: 14 additions & 0 deletions llvm/test/CodeGen/RISCV/atomic-rmw.ll
Original file line number Diff line number Diff line change
Expand Up @@ -4437,6 +4437,7 @@ define i8 @atomicrmw_nand_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64IA-WMO-ZABHA-ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: and a3, a0, a1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: not a3, a3
; RV64IA-WMO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-ZACAS-NEXT: slli a4, a0, 56
; RV64IA-WMO-ZABHA-ZACAS-NEXT: amocas.b.aqrl a0, a3, (a2)
; RV64IA-WMO-ZABHA-ZACAS-NEXT: srai a4, a4, 56
Expand All @@ -4452,6 +4453,7 @@ define i8 @atomicrmw_nand_i8_seq_cst(ptr %a, i8 %b) nounwind {
; RV64IA-TSO-ZABHA-ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: and a3, a0, a1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: not a3, a3
; RV64IA-TSO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-ZACAS-NEXT: slli a4, a0, 56
; RV64IA-TSO-ZABHA-ZACAS-NEXT: amocas.b a0, a3, (a2)
; RV64IA-TSO-ZABHA-ZACAS-NEXT: srai a4, a4, 56
Expand Down Expand Up @@ -14410,6 +14412,7 @@ define i16 @atomicrmw_nand_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64IA-WMO-ZABHA-ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: and a3, a0, a1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: not a3, a3
; RV64IA-WMO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-ZACAS-NEXT: slli a4, a0, 48
; RV64IA-WMO-ZABHA-ZACAS-NEXT: amocas.h.aqrl a0, a3, (a2)
; RV64IA-WMO-ZABHA-ZACAS-NEXT: srai a4, a4, 48
Expand All @@ -14425,6 +14428,7 @@ define i16 @atomicrmw_nand_i16_seq_cst(ptr %a, i16 %b) nounwind {
; RV64IA-TSO-ZABHA-ZACAS-NEXT: # =>This Inner Loop Header: Depth=1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: and a3, a0, a1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: not a3, a3
; RV64IA-TSO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-ZACAS-NEXT: slli a4, a0, 48
; RV64IA-TSO-ZABHA-ZACAS-NEXT: amocas.h a0, a3, (a2)
; RV64IA-TSO-ZABHA-ZACAS-NEXT: srai a4, a4, 48
Expand Down Expand Up @@ -21637,6 +21641,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV32IA-WMO-ZACAS-NEXT: mv a3, a0
; RV32IA-WMO-ZACAS-NEXT: and a4, a0, a1
; RV32IA-WMO-ZACAS-NEXT: not a4, a4
; RV32IA-WMO-ZACAS-NEXT: fence rw, rw
; RV32IA-WMO-ZACAS-NEXT: amocas.w.aqrl a0, a4, (a2)
; RV32IA-WMO-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV32IA-WMO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -21651,6 +21656,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV32IA-TSO-ZACAS-NEXT: mv a3, a0
; RV32IA-TSO-ZACAS-NEXT: and a4, a0, a1
; RV32IA-TSO-ZACAS-NEXT: not a4, a4
; RV32IA-TSO-ZACAS-NEXT: fence rw, rw
; RV32IA-TSO-ZACAS-NEXT: amocas.w a0, a4, (a2)
; RV32IA-TSO-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV32IA-TSO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -21665,6 +21671,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64IA-WMO-ZACAS-NEXT: mv a3, a0
; RV64IA-WMO-ZACAS-NEXT: and a4, a0, a1
; RV64IA-WMO-ZACAS-NEXT: not a4, a4
; RV64IA-WMO-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZACAS-NEXT: amocas.w.aqrl a0, a4, (a2)
; RV64IA-WMO-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV64IA-WMO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -21679,6 +21686,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64IA-TSO-ZACAS-NEXT: mv a3, a0
; RV64IA-TSO-ZACAS-NEXT: and a4, a0, a1
; RV64IA-TSO-ZACAS-NEXT: not a4, a4
; RV64IA-TSO-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZACAS-NEXT: amocas.w a0, a4, (a2)
; RV64IA-TSO-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV64IA-TSO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand Down Expand Up @@ -21717,6 +21725,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64IA-WMO-ZABHA-ZACAS-NEXT: mv a3, a0
; RV64IA-WMO-ZABHA-ZACAS-NEXT: and a4, a0, a1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: not a4, a4
; RV64IA-WMO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-ZACAS-NEXT: amocas.w.aqrl a0, a4, (a2)
; RV64IA-WMO-ZABHA-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -21731,6 +21740,7 @@ define i32 @atomicrmw_nand_i32_seq_cst(ptr %a, i32 %b) nounwind {
; RV64IA-TSO-ZABHA-ZACAS-NEXT: mv a3, a0
; RV64IA-TSO-ZABHA-ZACAS-NEXT: and a4, a0, a1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: not a4, a4
; RV64IA-TSO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-ZACAS-NEXT: amocas.w a0, a4, (a2)
; RV64IA-TSO-ZABHA-ZACAS-NEXT: bne a0, a3, .LBB154_1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand Down Expand Up @@ -25546,6 +25556,7 @@ define i64 @atomicrmw_nand_i64_seq_cst(ptr %a, i64 %b) nounwind {
; RV64IA-WMO-ZACAS-NEXT: mv a3, a0
; RV64IA-WMO-ZACAS-NEXT: and a4, a0, a1
; RV64IA-WMO-ZACAS-NEXT: not a4, a4
; RV64IA-WMO-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZACAS-NEXT: amocas.d.aqrl a0, a4, (a2)
; RV64IA-WMO-ZACAS-NEXT: bne a0, a3, .LBB209_1
; RV64IA-WMO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -25560,6 +25571,7 @@ define i64 @atomicrmw_nand_i64_seq_cst(ptr %a, i64 %b) nounwind {
; RV64IA-TSO-ZACAS-NEXT: mv a3, a0
; RV64IA-TSO-ZACAS-NEXT: and a4, a0, a1
; RV64IA-TSO-ZACAS-NEXT: not a4, a4
; RV64IA-TSO-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZACAS-NEXT: amocas.d a0, a4, (a2)
; RV64IA-TSO-ZACAS-NEXT: bne a0, a3, .LBB209_1
; RV64IA-TSO-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand Down Expand Up @@ -25598,6 +25610,7 @@ define i64 @atomicrmw_nand_i64_seq_cst(ptr %a, i64 %b) nounwind {
; RV64IA-WMO-ZABHA-ZACAS-NEXT: mv a3, a0
; RV64IA-WMO-ZABHA-ZACAS-NEXT: and a4, a0, a1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: not a4, a4
; RV64IA-WMO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-WMO-ZABHA-ZACAS-NEXT: amocas.d.aqrl a0, a4, (a2)
; RV64IA-WMO-ZABHA-ZACAS-NEXT: bne a0, a3, .LBB209_1
; RV64IA-WMO-ZABHA-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand All @@ -25612,6 +25625,7 @@ define i64 @atomicrmw_nand_i64_seq_cst(ptr %a, i64 %b) nounwind {
; RV64IA-TSO-ZABHA-ZACAS-NEXT: mv a3, a0
; RV64IA-TSO-ZABHA-ZACAS-NEXT: and a4, a0, a1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: not a4, a4
; RV64IA-TSO-ZABHA-ZACAS-NEXT: fence rw, rw
; RV64IA-TSO-ZABHA-ZACAS-NEXT: amocas.d a0, a4, (a2)
; RV64IA-TSO-ZABHA-ZACAS-NEXT: bne a0, a3, .LBB209_1
; RV64IA-TSO-ZABHA-ZACAS-NEXT: # %bb.2: # %atomicrmw.end
Expand Down
2 changes: 2 additions & 0 deletions llvm/test/CodeGen/RISCV/atomic-signext.ll
Original file line number Diff line number Diff line change
Expand Up @@ -5562,6 +5562,7 @@ define signext i32 @cmpxchg_i32_monotonic_crossbb(ptr %ptr, i32 signext %cmp, i3
; RV32IA-ZACAS: # %bb.0:
; RV32IA-ZACAS-NEXT: beqz a3, .LBB64_2
; RV32IA-ZACAS-NEXT: # %bb.1: # %then
; RV32IA-ZACAS-NEXT: fence rw, rw
; RV32IA-ZACAS-NEXT: amocas.w.aqrl a1, a2, (a0)
; RV32IA-ZACAS-NEXT: mv a0, a1
; RV32IA-ZACAS-NEXT: ret
Expand Down Expand Up @@ -5612,6 +5613,7 @@ define signext i32 @cmpxchg_i32_monotonic_crossbb(ptr %ptr, i32 signext %cmp, i3
; RV64IA-ZACAS: # %bb.0:
; RV64IA-ZACAS-NEXT: beqz a3, .LBB64_2
; RV64IA-ZACAS-NEXT: # %bb.1: # %then
; RV64IA-ZACAS-NEXT: fence rw, rw
; RV64IA-ZACAS-NEXT: amocas.w.aqrl a1, a2, (a0)
; RV64IA-ZACAS-NEXT: mv a0, a1
; RV64IA-ZACAS-NEXT: ret
Expand Down
Loading