[LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions #113255

tangaac · 2024-10-22T02:53:11Z

Two options for clang: -mlam-bh & -mno-lam-bh.
Enable or disable amswap[__db].{b/h} and amadd[__db].{b/h} instructions.
The default is -mno-lam-bh.
Only works on LoongArch64.

llvmbot · 2024-10-22T02:53:44Z

@llvm/pr-subscribers-backend-loongarch
@llvm/pr-subscribers-clang-driver

@llvm/pr-subscribers-clang

Author: None (tangaac)

Changes

Two options for clang: -mlam-bh & -mno-lam-bh.
Enable or disable amswap[__db].{b/h} and amadd[__db].{b/h} instructions.
The default is -mno-lam-bh.
Only works on LoongArch64.

Patch is 76.68 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/113255.diff

14 Files Affected:

(modified) clang/include/clang/Driver/Options.td (+4)
(modified) clang/lib/Basic/Targets/LoongArch.cpp (+6-1)
(modified) clang/lib/Basic/Targets/LoongArch.h (+2)
(modified) clang/lib/Driver/ToolChains/Arch/LoongArch.cpp (+9)
(modified) clang/test/Driver/loongarch-march.c (+4-4)
(added) clang/test/Driver/loongarch-mlam-bh.c (+30)
(modified) clang/test/Preprocessor/init-loongarch.c (+17-6)
(modified) llvm/include/llvm/TargetParser/LoongArchTargetParser.def (+2-1)
(modified) llvm/include/llvm/TargetParser/LoongArchTargetParser.h (+3)
(modified) llvm/lib/Target/LoongArch/LoongArch.td (+8-1)
(modified) llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp (+6)
(modified) llvm/lib/Target/LoongArch/LoongArchInstrInfo.td (+18)
(modified) llvm/lib/TargetParser/LoongArchTargetParser.cpp (+3-1)
(added) llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-lam-bh.ll (+1674)

diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 152c43d7908ff8..3cbab84b6bf986 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -5394,6 +5394,10 @@ def mfrecipe : Flag<["-"], "mfrecipe">, Group<m_loongarch_Features_Group>,
   HelpText<"Enable frecipe.{s/d} and frsqrte.{s/d}">;
 def mno_frecipe : Flag<["-"], "mno-frecipe">, Group<m_loongarch_Features_Group>,
   HelpText<"Disable frecipe.{s/d} and frsqrte.{s/d}">;
+def mlam_bh : Flag<["-"], "mlam-bh">, Group<m_loongarch_Features_Group>,
+  HelpText<"Enable amswap_[db].{b/h} and amadd_[db].{b/h}">;
+def mno_lam_bh : Flag<["-"], "mno-lam-bh">, Group<m_loongarch_Features_Group>,
+  HelpText<"Disable amswap_[db].{b/h} and amadd_[db].{b/h}">;
 def mannotate_tablejump : Flag<["-"], "mannotate-tablejump">, Group<m_loongarch_Features_Group>,
   HelpText<"Enable annotate table jump instruction to correlate it with the jump table.">;
 def mno_annotate_tablejump : Flag<["-"], "mno-annotate-tablejump">, Group<m_loongarch_Features_Group>,
diff --git a/clang/lib/Basic/Targets/LoongArch.cpp b/clang/lib/Basic/Targets/LoongArch.cpp
index cb3fd12c48ddb6..3f07dcc3b9c374 100644
--- a/clang/lib/Basic/Targets/LoongArch.cpp
+++ b/clang/lib/Basic/Targets/LoongArch.cpp
@@ -205,7 +205,7 @@ void LoongArchTargetInfo::getTargetDefines(const LangOptions &Opts,
       // TODO: As more features of the V1.1 ISA are supported, a unified "v1.1"
       // arch feature set will be used to include all sub-features belonging to
       // the V1.1 ISA version.
-      if (HasFeatureFrecipe)
+      if (HasFeatureFrecipe || HasFeatureLAM_BH)
         Builder.defineMacro("__loongarch_arch",
                             Twine('"') + "la64v1.1" + Twine('"'));
       else
@@ -236,6 +236,9 @@ void LoongArchTargetInfo::getTargetDefines(const LangOptions &Opts,
   if (HasFeatureFrecipe)
     Builder.defineMacro("__loongarch_frecipe", Twine(1));
 
+  if (HasFeatureLAM_BH)
+    Builder.defineMacro("__loongarch_lam_bh", Twine(1));
+
   StringRef ABI = getABI();
   if (ABI == "lp64d" || ABI == "lp64f" || ABI == "lp64s")
     Builder.defineMacro("__loongarch_lp64");
@@ -312,6 +315,8 @@ bool LoongArchTargetInfo::handleTargetFeatures(
       HasUnalignedAccess = false;
     else if (Feature == "+frecipe")
       HasFeatureFrecipe = true;
+    else if (Feature == "+lam-bh")
+      HasFeatureLAM_BH = true;
   }
   return true;
 }
diff --git a/clang/lib/Basic/Targets/LoongArch.h b/clang/lib/Basic/Targets/LoongArch.h
index c668ca7eca047a..3585e9f7968b4b 100644
--- a/clang/lib/Basic/Targets/LoongArch.h
+++ b/clang/lib/Basic/Targets/LoongArch.h
@@ -30,6 +30,7 @@ class LLVM_LIBRARY_VISIBILITY LoongArchTargetInfo : public TargetInfo {
   bool HasFeatureLSX;
   bool HasFeatureLASX;
   bool HasFeatureFrecipe;
+  bool HasFeatureLAM_BH;
 
 public:
   LoongArchTargetInfo(const llvm::Triple &Triple, const TargetOptions &)
@@ -39,6 +40,7 @@ class LLVM_LIBRARY_VISIBILITY LoongArchTargetInfo : public TargetInfo {
     HasFeatureLSX = false;
     HasFeatureLASX = false;
     HasFeatureFrecipe = false;
+    HasFeatureLAM_BH = false;
     LongDoubleWidth = 128;
     LongDoubleAlign = 128;
     LongDoubleFormat = &llvm::APFloat::IEEEquad();
diff --git a/clang/lib/Driver/ToolChains/Arch/LoongArch.cpp b/clang/lib/Driver/ToolChains/Arch/LoongArch.cpp
index 355253e4b3b07c..e69a5562137ccd 100644
--- a/clang/lib/Driver/ToolChains/Arch/LoongArch.cpp
+++ b/clang/lib/Driver/ToolChains/Arch/LoongArch.cpp
@@ -260,6 +260,15 @@ void loongarch::getLoongArchTargetFeatures(const Driver &D,
     else
       Features.push_back("-frecipe");
   }
+
+  // Select lam-bh feature determined by -m[no-]lam-bh.
+  if (const Arg *A =
+          Args.getLastArg(options::OPT_mlam_bh, options::OPT_mno_lam_bh)) {
+    if (A->getOption().matches(options::OPT_mlam_bh))
+      Features.push_back("+lam-bh");
+    else
+      Features.push_back("-lam-bh");
+  }
 }
 
 std::string loongarch::postProcessTargetCPUString(const std::string &CPU,
diff --git a/clang/test/Driver/loongarch-march.c b/clang/test/Driver/loongarch-march.c
index 2d5b315d962a1e..d4cd5b07ae905f 100644
--- a/clang/test/Driver/loongarch-march.c
+++ b/clang/test/Driver/loongarch-march.c
@@ -39,21 +39,21 @@
 
 // CC1-LA64V1P1: "-target-cpu" "loongarch64"
 // CC1-LA64V1P1-NOT: "-target-feature"
-// CC1-LA64V1P1: "-target-feature" "+64bit" "-target-feature" "+d" "-target-feature" "+lsx" "-target-feature" "+ual" "-target-feature" "+frecipe"
+// CC1-LA64V1P1: "-target-feature" "+64bit" "-target-feature" "+d" "-target-feature" "+lsx" "-target-feature" "+ual" "-target-feature" "+frecipe" "-target-feature" "+lam-bh"
 // CC1-LA64V1P1-NOT: "-target-feature"
 // CC1-LA64V1P1: "-target-abi" "lp64d"
 
 // CC1-LA664: "-target-cpu" "la664"
 // CC1-LA664-NOT: "-target-feature"
-// CC1-LA664: "-target-feature" "+64bit" "-target-feature" "+f" "-target-feature" "+d" "-target-feature" "+lsx" "-target-feature" "+lasx" "-target-feature" "+ual" "-target-feature" "+frecipe"
+// CC1-LA664: "-target-feature" "+64bit" "-target-feature" "+f" "-target-feature" "+d" "-target-feature" "+lsx" "-target-feature" "+lasx" "-target-feature" "+ual" "-target-feature" "+frecipe" "-target-feature" "+lam-bh"
 // CC1-LA664-NOT: "-target-feature"
 // CC1-LA664: "-target-abi" "lp64d"
 
 // IR-LOONGARCH64: attributes #[[#]] ={{.*}}"target-cpu"="loongarch64" {{.*}}"target-features"="+64bit,+d,+f,+ual"
 // IR-LA464: attributes #[[#]] ={{.*}}"target-cpu"="la464" {{.*}}"target-features"="+64bit,+d,+f,+lasx,+lsx,+ual"
 // IR-LA64V1P0: attributes #[[#]] ={{.*}}"target-cpu"="loongarch64" {{.*}}"target-features"="+64bit,+d,+lsx,+ual"
-// IR-LA64V1P1: attributes #[[#]] ={{.*}}"target-cpu"="loongarch64" {{.*}}"target-features"="+64bit,+d,+frecipe,+lsx,+ual"
-// IR-LA664: attributes #[[#]] ={{.*}}"target-cpu"="la664" {{.*}}"target-features"="+64bit,+d,+f,+frecipe,+lasx,+lsx,+ual"
+// IR-LA64V1P1: attributes #[[#]] ={{.*}}"target-cpu"="loongarch64" {{.*}}"target-features"="+64bit,+d,+frecipe,+lam-bh,+lsx,+ual"
+// IR-LA664: attributes #[[#]] ={{.*}}"target-cpu"="la664" {{.*}}"target-features"="+64bit,+d,+f,+frecipe,+lam-bh,+lasx,+lsx,+ual"
 
 int foo(void) {
   return 3;
diff --git a/clang/test/Driver/loongarch-mlam-bh.c b/clang/test/Driver/loongarch-mlam-bh.c
new file mode 100644
index 00000000000000..6f2901e594dfcc
--- /dev/null
+++ b/clang/test/Driver/loongarch-mlam-bh.c
@@ -0,0 +1,30 @@
+/// Test -m[no]lam-bh options.
+
+// RUN: %clang --target=loongarch64 -mlam-bh -fsyntax-only %s -### 2>&1 | \
+// RUN:     FileCheck %s --check-prefix=CC1-LAM-BH
+// RUN: %clang --target=loongarch64 -mno-lam-bh -fsyntax-only %s -### 2>&1 | \
+// RUN:     FileCheck %s --check-prefix=CC1-NO-LAM-BH
+// RUN: %clang --target=loongarch64 -mno-lam-bh -mlam-bh -fsyntax-only %s -### 2>&1 | \
+// RUN:     FileCheck %s --check-prefix=CC1-LAM-BH
+// RUN: %clang --target=loongarch64  -mlam-bh -mno-lam-bh -fsyntax-only %s -### 2>&1 | \
+// RUN:     FileCheck %s --check-prefix=CC1-NO-LAM-BH
+
+// RUN: %clang --target=loongarch64 -mlam-bh -S -emit-llvm %s -o - | \
+// RUN: FileCheck %s --check-prefix=IR-LAM-BH
+// RUN: %clang --target=loongarch64 -mno-lam-bh -S -emit-llvm %s -o - | \
+// RUN: FileCheck %s --check-prefix=IR-NO-LAM-BH
+// RUN: %clang --target=loongarch64 -mno-lam-bh -mlam-bh -S -emit-llvm %s -o - | \
+// RUN: FileCheck %s --check-prefix=IR-LAM-BH
+// RUN: %clang --target=loongarch64 -mlam-bh -mno-lam-bh -S -emit-llvm %s -o - | \
+// RUN: FileCheck %s --check-prefix=IR-NO-LAM-BH
+
+
+// CC1-LAM-BH: "-target-feature" "+lam-bh"
+// CC1-NO-LAM-BH: "-target-feature" "-lam-bh"
+
+// IR-LAM-BH: attributes #[[#]] ={{.*}}"target-features"="{{(.*,)?}}+lam-bh{{(,.*)?}}"
+// IR-NO-LAM-BH: attributes #[[#]] ={{.*}}"target-features"="{{(.*,)?}}-lam-bh{{(,.*)?}}"
+
+int foo(void) {
+  return 42;
+}
\ No newline at end of file
diff --git a/clang/test/Preprocessor/init-loongarch.c b/clang/test/Preprocessor/init-loongarch.c
index 771d56ffb1c1b9..85c42bea6c7ca5 100644
--- a/clang/test/Preprocessor/init-loongarch.c
+++ b/clang/test/Preprocessor/init-loongarch.c
@@ -798,7 +798,7 @@
 // LA64-FPU0-LP64S-NOT: #define __loongarch_single_float
 // LA64-FPU0-LP64S: #define __loongarch_soft_float 1
 
-/// Check __loongarch_arch{_tune/_frecipe}.
+/// Check __loongarch_arch{_tune/_frecipe/_lam_bh}.
 
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - | \
 // RUN:   FileCheck --match-full-lines --check-prefix=ARCH-TUNE -DARCH=la64v1.0 -DTUNE=loongarch64 %s
@@ -823,26 +823,37 @@
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -Xclang -target-feature -Xclang +lsx | \
 // RUN:   FileCheck --match-full-lines --check-prefix=ARCH-TUNE -DARCH=la64v1.0 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.1 | \
-// RUN:   FileCheck --match-full-lines  --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=la64v1.1 -DTUNE=loongarch64 %s
+// RUN:   FileCheck --match-full-lines  --check-prefixes=ARCH-TUNE,FRECIPE,LAM-BH -DARCH=la64v1.1 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.1 -Xclang -target-feature -Xclang -frecipe | \
-// RUN:   FileCheck --match-full-lines --check-prefix=ARCH-TUNE -DARCH=la64v1.0 -DTUNE=loongarch64 %s
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,LAM-BH -DARCH=la64v1.1 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.1 -Xclang -target-feature -Xclang -lsx | \
-// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=loongarch64 -DTUNE=loongarch64 %s
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE,LAM-BH -DARCH=loongarch64 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -Xclang -target-feature -Xclang +frecipe | \
 // RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=loongarch64 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -Xclang -target-feature -Xclang +lsx -Xclang -target-feature -Xclang +frecipe | \
 // RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=la64v1.1 -DTUNE=loongarch64 %s
+// RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.0 -Xclang -target-feature -Xclang +lam-bh | \
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,LAM-BH -DARCH=la64v1.1 -DTUNE=loongarch64 %s
+// RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.1 -Xclang -target-feature -Xclang -lam-bh | \
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=la64v1.1 -DTUNE=loongarch64 %s
+// RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -Xclang -target-feature -Xclang +lam-bh | \
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,LAM-BH -DARCH=loongarch64 -DTUNE=loongarch64 %s
+// RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -Xclang -target-feature -Xclang +lsx -Xclang -target-feature -Xclang +lam-bh | \
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,LAM-BH -DARCH=la64v1.1 -DTUNE=loongarch64 %s
+// RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la64v1.1 -Xclang -target-feature -Xclang -frecipe -Xclang -target-feature -Xclang -lam-bh | \
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE -DARCH=la64v1.0 -DTUNE=loongarch64 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la664 | \
-// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=la664 -DTUNE=la664 %s
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE,LAM-BH -DARCH=la664 -DTUNE=la664 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -mtune=la664 | \
 // RUN:   FileCheck --match-full-lines --check-prefix=ARCH-TUNE -DARCH=la64v1.0 -DTUNE=la664 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=loongarch64 -mtune=la664 | \
 // RUN:   FileCheck --match-full-lines --check-prefix=ARCH-TUNE -DARCH=loongarch64 -DTUNE=la664 %s
 // RUN: %clang --target=loongarch64 -x c -E -dM %s -o - -march=la664 -mtune=loongarch64 | \
-// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE -DARCH=la664 -DTUNE=loongarch64 %s
+// RUN:   FileCheck --match-full-lines --check-prefixes=ARCH-TUNE,FRECIPE,LAM-BH -DARCH=la664 -DTUNE=loongarch64 %s
 
 // ARCH-TUNE: #define __loongarch_arch "[[ARCH]]"
 // FRECIPE: #define __loongarch_frecipe 1
+// LAM-BH: #define __loongarch_lam_bh 1
 // ARCH-TUNE: #define __loongarch_tune "[[TUNE]]"
 
 // RUN: %clang --target=loongarch64 -mlsx -x c -E -dM %s -o - \
diff --git a/llvm/include/llvm/TargetParser/LoongArchTargetParser.def b/llvm/include/llvm/TargetParser/LoongArchTargetParser.def
index 101a48cbd53994..6cd2018b7b59cb 100644
--- a/llvm/include/llvm/TargetParser/LoongArchTargetParser.def
+++ b/llvm/include/llvm/TargetParser/LoongArchTargetParser.def
@@ -11,6 +11,7 @@ LOONGARCH_FEATURE("+lbt", FK_LBT)
 LOONGARCH_FEATURE("+lvz", FK_LVZ)
 LOONGARCH_FEATURE("+ual", FK_UAL)
 LOONGARCH_FEATURE("+frecipe", FK_FRECIPE)
+LOONGARCH_FEATURE("+lam-bh", FK_LAM_BH)
 
 #undef LOONGARCH_FEATURE
 
@@ -20,6 +21,6 @@ LOONGARCH_FEATURE("+frecipe", FK_FRECIPE)
 
 LOONGARCH_ARCH("loongarch64", AK_LOONGARCH64, FK_64BIT | FK_FP32 | FK_FP64 | FK_UAL)
 LOONGARCH_ARCH("la464", AK_LA464, FK_64BIT | FK_FP32 | FK_FP64 | FK_LSX | FK_LASX | FK_UAL)
-LOONGARCH_ARCH("la664", AK_LA664, FK_64BIT | FK_FP32 | FK_FP64 | FK_LSX | FK_LASX | FK_UAL | FK_FRECIPE)
+LOONGARCH_ARCH("la664", AK_LA664, FK_64BIT | FK_FP32 | FK_FP64 | FK_LSX | FK_LASX | FK_UAL | FK_FRECIPE | FK_LAM_BH)
 
 #undef LOONGARCH_ARCH
diff --git a/llvm/include/llvm/TargetParser/LoongArchTargetParser.h b/llvm/include/llvm/TargetParser/LoongArchTargetParser.h
index c0bb15a5163b12..1b90e1d9ade39b 100644
--- a/llvm/include/llvm/TargetParser/LoongArchTargetParser.h
+++ b/llvm/include/llvm/TargetParser/LoongArchTargetParser.h
@@ -49,6 +49,9 @@ enum FeatureKind : uint32_t {
 
   // Floating-point approximate reciprocal instructions are available.
   FK_FRECIPE = 1 << 9,
+
+  // Atomic memory swap and add instructions for byte and half word are available.
+  FK_LAM_BH = 1 << 10,
 };
 
 struct FeatureInfo {
diff --git a/llvm/lib/Target/LoongArch/LoongArch.td b/llvm/lib/Target/LoongArch/LoongArch.td
index ddb27dc6404fa8..54ebf86666abf9 100644
--- a/llvm/lib/Target/LoongArch/LoongArch.td
+++ b/llvm/lib/Target/LoongArch/LoongArch.td
@@ -112,6 +112,12 @@ def FeatureFrecipe
                        "Support frecipe.{s/d} and frsqrte.{s/d} instructions.">;
 def HasFrecipe : Predicate<"Subtarget->hasFrecipe()">;
 
+// Atomic memory swap and add instructions for byte and half word
+def FeatureLAM_BH
+    : SubtargetFeature<"lam-bh", "HasLAM_BH", "true",
+                        "Support amswap[_db].{b/h} and amadd[_db].{b/h} instructions.">;
+def HasLAM_BH : Predicate<"Subtarget->hasLAM_BH()">;
+
 def TunePreferWInst
     : SubtargetFeature<"prefer-w-inst", "PreferWInst", "true",
                        "Prefer instructions with W suffix">;
@@ -151,7 +157,8 @@ def : ProcessorModel<"la664", NoSchedModel, [Feature64Bit,
                                              FeatureExtLASX,
                                              FeatureExtLVZ,
                                              FeatureExtLBT,
-                                             FeatureFrecipe]>;
+                                             FeatureFrecipe,
+                                             FeatureLAM_BH]>;
 
 //===----------------------------------------------------------------------===//
 // Define the LoongArch target.
diff --git a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
index 676d43ef22c47b..2d9e4e839650f0 100644
--- a/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
+++ b/llvm/lib/Target/LoongArch/LoongArchISelLowering.cpp
@@ -5735,6 +5735,12 @@ LoongArchTargetLowering::shouldExpandAtomicRMWInIR(AtomicRMWInst *AI) const {
       AI->getOperation() == AtomicRMWInst::USubSat)
     return AtomicExpansionKind::CmpXChg;
 
+  if (Subtarget.hasLAM_BH() && Subtarget.is64Bit() &&
+      (AI->getOperation() == AtomicRMWInst::Xchg ||
+       AI->getOperation() == AtomicRMWInst::Add)) {
+    return AtomicExpansionKind::None;
+  }
+
   unsigned Size = AI->getType()->getPrimitiveSizeInBits();
   if (Size == 8 || Size == 16)
     return AtomicExpansionKind::MaskedIntrinsic;
diff --git a/llvm/lib/Target/LoongArch/LoongArchInstrInfo.td b/llvm/lib/Target/LoongArch/LoongArchInstrInfo.td
index f97aace363cb7a..03d629e44f6d97 100644
--- a/llvm/lib/Target/LoongArch/LoongArchInstrInfo.td
+++ b/llvm/lib/Target/LoongArch/LoongArchInstrInfo.td
@@ -2088,8 +2088,26 @@ multiclass binary_atomic_op_wd<string inst, string op, string signed = ""> {
             (!cast<Instruction>(inst#"__DB_D"#signed) GPR:$rk, GPR:$rj)>;
 }
 
+// Atomic operation for byte and half word
+multiclass binary_atomic_op_bh<string inst, string op> {
+  def : Pat<(!cast<PatFrag>(op#"_i8_monotonic") GPR:$rj, GPR:$rk),
+            (!cast<Instruction>(inst#"_B") GPR:$rk, GPR:$rj)>;
+  def : Pat<(!cast<PatFrag>(op#"_i16_monotonic") GPR:$rj, GPR:$rk),
+            (!cast<Instruction>(inst#"_H") GPR:$rk, GPR:$rj)>;
+
+  def : Pat<(!cast<PatFrag>(op#"_i8") GPR:$rj, GPR:$rk),
+            (!cast<Instruction>(inst#"__DB_B") GPR:$rk, GPR:$rj)>;
+  def : Pat<(!cast<PatFrag>(op#"_i16") GPR:$rj, GPR:$rk),
+            (!cast<Instruction>(inst#"__DB_H") GPR:$rk, GPR:$rj)>;
+}
+
 let Predicates = [IsLA64] in {
 
+let Predicates = [ HasLAM_BH ] in {
+defm : binary_atomic_op_bh<"AMSWAP", "atomic_swap">;
+defm : binary_atomic_op_bh<"AMADD", "atomic_load_add">;
+}
+
 defm : binary_atomic_op_wd<"AMSWAP", "atomic_swap">;
 defm : binary_atomic_op_wd<"AMADD", "atomic_load_add">;
 defm : binary_atomic_op_wd<"AMAND", "atomic_load_and">;
diff --git a/llvm/lib/TargetParser/LoongArchTargetParser.cpp b/llvm/lib/TargetParser/LoongArchTargetParser.cpp
index 8e86d18de2ad9a..27e3b5683c5a6e 100644
--- a/llvm/lib/TargetParser/LoongArchTargetParser.cpp
+++ b/llvm/lib/TargetParser/LoongArchTargetParser.cpp
@@ -50,8 +50,10 @@ bool LoongArch::getArchFeatures(StringRef Arch,
     Features.push_back("+d");
     Features.push_back("+lsx");
     Features.push_back("+ual");
-    if (Arch == "la64v1.1")
+    if (Arch == "la64v1.1") {
       Features.push_back("+frecipe");
+      Features.push_back("+lam-bh");
+    }
     return true;
   }
 
diff --git a/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-lam-bh.ll b/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-lam-bh.ll
new file mode 100644
index 00000000000000..e71c87daed74ea
--- /dev/null
+++ b/llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-lam-bh.ll
@@ -0,0 +1,1674 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc --mtriple=loongarch32 --mattr=+d,-lam-bh < %s | FileCheck %s --check-prefix=LA32
+; RUN: llc --mtriple=loongarch32 --mattr=+d,+lam-bh < %s | FileCheck %s --check-prefix=LA32-LAM-BH
+; RUN: llc --mtriple=loongarch64 --mattr=+d,-lam-bh < %s | FileCheck %s --check-prefix=LA64
+; RUN: llc --mtriple=loongarch64 --mattr=+d,+lam-bh < %s | FileCheck %s --check-prefix=LA64-LAM-BH
+
+
+define i8 @atomicrmw_xchg_i8_acquire(ptr %a, i8 %b) nounwind {
+; LA32-LABEL: atomicrmw_xchg_i8_acquire:
+; LA32:       # %bb.0:
+; LA32-NEXT:    slli.w	$a2, $a0, 3
+; LA32-NEXT:    bstrins.w	$a0, $zero, 1, 0
+; LA32-NEXT:    ori	$a3, $zero, 255
+; LA32-NEXT:    sll.w	$a3, $a3, $a2
+; LA32-NEXT:    andi	$a1, $a1, 255
+; LA32-NEXT:    sll.w	$a1, $a1, $a2
+; LA32-NEXT:  .LBB0_1: # =>This Inner Loop Header: Depth=1
+; LA32-NEXT:    ll.w	$a4, $a0, 0
+; LA32-NEXT:    addi.w	$a5, $a1, 0
+; LA32-NEXT:    xor	$a5, $a4, $a5
+; LA32-NEXT:    and	$a5, $a5, $a3
+; LA32-NEXT:    xor	$a5, $a4, $a5
+; LA32-NEXT:    sc.w	$a5, $a0, 0
+; LA32-NEXT:    beqz	$a5, .LBB0_1
+; LA32-NEXT:  # %bb.2:
+; LA32-NEXT:    srl.w	$a0, $a4, $a2
+; LA32-NEXT:    ret
+;
+; LA32-LAM-BH-LABEL: atomicrmw_xchg_i8_acquire:
+; LA32-LAM-BH:       # %bb.0:
+; LA32-LAM-BH-NEXT:    slli.w	$a2, $a0, 3
+; LA32-LAM-BH-NEXT:    bstrins.w	$a0, $zero, 1, 0
+; LA32-LAM-BH-NEXT:    ori	$a3, $zero, 255
+; LA32-LAM-BH-NEXT:    sll.w	$a3, $a3, $a2
+; LA32-LAM-BH-NEXT:    andi	$a1, $a1, 255
+; LA32-LAM-BH-NEXT:    sll.w	$a1, $a1, $a2
+; LA32-LAM-BH-NEXT:  .LBB0_1: # =>Th...
[truncated]

github-actions · 2024-10-22T02:56:39Z

✅ With the latest revision this PR passed the C/C++ code formatter.

…ngArch.

clang/lib/Basic/Targets/LoongArch.cpp

llvm/lib/Target/LoongArch/LoongArchInstrInfo.td

llvm/test/CodeGen/LoongArch/ir-instruction/atomicrmw-lam-bh.ll

cjappl · 2024-10-24T16:05:27Z

Just a heads up, I think this broke documentation builds. After bisecting it points to this commit.

ninja -C build docs-clang-html
FAILED: tools/clang/docs/CMakeFiles/docs-clang-html ~/code/radsan_cjappl/llvm-project/build/tools/clang/docs/CMakeFiles/docs-clang-html
cd ~/code/radsan_cjappl/llvm-project/build/tools/clang/docs && /opt/homebrew/Cellar/cmake/3.28.3/bin/cmake -E env ~/.virtualenvs/llvm/bin/sphinx-build -b htm
l -d ~/code/radsan_cjappl/llvm-project/build/tools/clang/docs/_doctrees-clang-html -q -t builder-html -D version=20 -D release=20.0.0git -tPreRelease -W ~co
de/radsan_cjappl/llvm-project/build/tools/clang/docs ~code/radsan_cjappl/llvm-project/build/tools/clang/docs/html

Warning, treated as error:
~/code/radsan_cjappl/llvm-project/build/tools/clang/docs/ClangCommandLineReference.rst:4819:Unknown target name: "amswap". [docutils]

See an example here:
https://github.com/llvm/llvm-project/actions/runs/11502212789/job/32016658426?pr=112727

SixWeining · 2024-10-25T04:01:30Z

Just a heads up, I think this broke documentation builds. After bisecting it points to this commit.

Thanks. Fixed by #113632.

cjappl · 2024-10-25T04:29:40Z

Thanks for the quick fix, working for me after I pulled latest main.

llvmbot added clang Clang issues not falling into any other category clang:driver 'clang' and 'clang++' user-facing binaries. Not 'clang-cl' clang:frontend Language frontend issues, e.g. anything involving "Sema" backend:loongarch labels Oct 22, 2024

support amswap[__db].{b/h} and amadd[__db].{b/h} instructions for Loo…

9628b7d

…ngArch.

tangaac force-pushed the feature/lam-bh branch from 51c4019 to 9628b7d Compare October 22, 2024 02:59

SixWeining reviewed Oct 22, 2024

View reviewed changes

improve the code and update some tests

8233a04

SixWeining changed the title ~~[LoongArch] [CodeGen] Support LoongArch-specific amswap[__db].{b/h} and amadd[__db].{b/h} instructions.~~ [LoongArch] [CodeGen] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions. Oct 23, 2024

SixWeining changed the title ~~[LoongArch] [CodeGen] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions.~~ [LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions Oct 23, 2024

SixWeining approved these changes Oct 23, 2024

View reviewed changes

SixWeining merged commit 5b9c76b into llvm:main Oct 23, 2024
6 of 8 checks passed

cjappl mentioned this pull request Oct 24, 2024

[rtsan][NFC] Documentation of suppression flag #112727

Merged

frobtech mentioned this pull request Oct 25, 2024

p/libc fexcept t #113664

Closed

tangaac deleted the feature/lam-bh branch November 19, 2024 08:19

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions #113255

[LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions #113255

tangaac commented Oct 22, 2024

llvmbot commented Oct 22, 2024 •

edited

Loading

github-actions bot commented Oct 22, 2024 •

edited

Loading

cjappl commented Oct 24, 2024

SixWeining commented Oct 25, 2024

cjappl commented Oct 25, 2024

[LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions #113255

[LoongArch] Support LoongArch-specific amswap[_db].{b/h} and amadd[_db].{b/h} instructions #113255

Conversation

tangaac commented Oct 22, 2024

llvmbot commented Oct 22, 2024 • edited Loading

github-actions bot commented Oct 22, 2024 • edited Loading

cjappl commented Oct 24, 2024

SixWeining commented Oct 25, 2024

cjappl commented Oct 25, 2024

llvmbot commented Oct 22, 2024 •

edited

Loading

github-actions bot commented Oct 22, 2024 •

edited

Loading