Skip to content

[NFC][clang] Split clang/lib/CodeGen/CGBuiltin.cpp into target-specific files #132252

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 7 commits into from
Mar 21, 2025
Merged

[NFC][clang] Split clang/lib/CodeGen/CGBuiltin.cpp into target-specific files #132252

merged 7 commits into from
Mar 21, 2025

Conversation

jthackray
Copy link
Contributor

clang/lib/CodeGen/CGBuiltin.cpp is over 1MB long (>23k LoC), and can take minutes to recompile (depending on compiler and host system) when modified, and 5 seconds for clangd to update for every edit. Splitting this file was discussed in this thread:

https://discourse.llvm.org/t/splitting-clang-s-cgbuiltin-cpp-over-23k-lines-long-takes-1min-to-compile/

and the idea has received a number of +1 votes, hence this change.

@llvmbot llvmbot added clang Clang issues not falling into any other category backend:AMDGPU backend:RISC-V clang:codegen IR generation bugs: mangling, exceptions, etc. labels Mar 20, 2025
@llvmbot
Copy link
Member

llvmbot commented Mar 20, 2025 •
edited
Loading

@llvm/pr-subscribers-backend-amdgpu

@llvm/pr-subscribers-backend-risc-v

Author: Jonathan Thackray (jthackray)

Changes

clang/lib/CodeGen/CGBuiltin.cpp is over 1MB long (>23k LoC), and can take minutes to recompile (depending on compiler and host system) when modified, and 5 seconds for clangd to update for every edit. Splitting this file was discussed in this thread:

https://discourse.llvm.org/t/splitting-clang-s-cgbuiltin-cpp-over-23k-lines-long-takes-1min-to-compile/

and the idea has received a number of +1 votes, hence this change.

Patch is 1.50 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/132252.diff

13 Files Affected:

  • (added) clang/lib/CodeGen/BuiltinTargets/AArch64.cpp (+8038)
  • (added) clang/lib/CodeGen/BuiltinTargets/AMDGPU.cpp (+1884)
  • (added) clang/lib/CodeGen/BuiltinTargets/Hexagon.cpp (+267)
  • (added) clang/lib/CodeGen/BuiltinTargets/NVPTX.cpp (+1168)
  • (added) clang/lib/CodeGen/BuiltinTargets/PPC.cpp (+1363)
  • (added) clang/lib/CodeGen/BuiltinTargets/RISCV.cpp (+425)
  • (added) clang/lib/CodeGen/BuiltinTargets/SPIR.cpp (+67)
  • (added) clang/lib/CodeGen/BuiltinTargets/SystemZ.cpp (+442)
  • (added) clang/lib/CodeGen/BuiltinTargets/WebAssembly.cpp (+625)
  • (added) clang/lib/CodeGen/BuiltinTargets/X86.cpp (+3399)
  • (modified) clang/lib/CodeGen/CGBuiltin.cpp (+154-17657)
  • (added) clang/lib/CodeGen/CGBuiltin.h (+102)
  • (modified) clang/lib/CodeGen/CMakeLists.txt (+10)
diff --git a/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp b/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp
new file mode 100644
index 0000000000000..a1fe34f8a562a
--- /dev/null
+++ b/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp
@@ -0,0 +1,8038 @@
+//===------ AArch64.cpp - Emit LLVM Code for builtins ---------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This contains code to emit Builtin calls as LLVM code.
+//
+//===----------------------------------------------------------------------===//
+
+#include "ABIInfo.h"
+#include "CGBuiltin.h"
+#include "TargetInfo.h"
+#include "clang/Basic/TargetBuiltins.h"
+#include "llvm/IR/InlineAsm.h"
+#include "llvm/IR/Intrinsics.h"
+#include "llvm/IR/IntrinsicsAArch64.h"
+#include "llvm/IR/IntrinsicsARM.h"
+#include "llvm/IR/IntrinsicsBPF.h"
+#include "llvm/TargetParser/AArch64TargetParser.h"
+
+using namespace clang;
+using namespace CodeGen;
+using namespace llvm;
+
+static std::optional<CodeGenFunction::MSVCIntrin>
+translateAarch64ToMsvcIntrin(unsigned BuiltinID) {
+  using MSVCIntrin = CodeGenFunction::MSVCIntrin;
+  switch (BuiltinID) {
+  default:
+    return std::nullopt;
+  case clang::AArch64::BI_BitScanForward:
+  case clang::AArch64::BI_BitScanForward64:
+    return MSVCIntrin::_BitScanForward;
+  case clang::AArch64::BI_BitScanReverse:
+  case clang::AArch64::BI_BitScanReverse64:
+    return MSVCIntrin::_BitScanReverse;
+  case clang::AArch64::BI_InterlockedAnd64:
+    return MSVCIntrin::_InterlockedAnd;
+  case clang::AArch64::BI_InterlockedExchange64:
+    return MSVCIntrin::_InterlockedExchange;
+  case clang::AArch64::BI_InterlockedExchangeAdd64:
+    return MSVCIntrin::_InterlockedExchangeAdd;
+  case clang::AArch64::BI_InterlockedExchangeSub64:
+    return MSVCIntrin::_InterlockedExchangeSub;
+  case clang::AArch64::BI_InterlockedOr64:
+    return MSVCIntrin::_InterlockedOr;
+  case clang::AArch64::BI_InterlockedXor64:
+    return MSVCIntrin::_InterlockedXor;
+  case clang::AArch64::BI_InterlockedDecrement64:
+    return MSVCIntrin::_InterlockedDecrement;
+  case clang::AArch64::BI_InterlockedIncrement64:
+    return MSVCIntrin::_InterlockedIncrement;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_acq:
+    return MSVCIntrin::_InterlockedExchangeAdd_acq;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_rel:
+    return MSVCIntrin::_InterlockedExchangeAdd_rel;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_nf:
+    return MSVCIntrin::_InterlockedExchangeAdd_nf;
+  case clang::AArch64::BI_InterlockedExchange8_acq:
+  case clang::AArch64::BI_InterlockedExchange16_acq:
+  case clang::AArch64::BI_InterlockedExchange_acq:
+  case clang::AArch64::BI_InterlockedExchange64_acq:
+  case clang::AArch64::BI_InterlockedExchangePointer_acq:
+    return MSVCIntrin::_InterlockedExchange_acq;
+  case clang::AArch64::BI_InterlockedExchange8_rel:
+  case clang::AArch64::BI_InterlockedExchange16_rel:
+  case clang::AArch64::BI_InterlockedExchange_rel:
+  case clang::AArch64::BI_InterlockedExchange64_rel:
+  case clang::AArch64::BI_InterlockedExchangePointer_rel:
+    return MSVCIntrin::_InterlockedExchange_rel;
+  case clang::AArch64::BI_InterlockedExchange8_nf:
+  case clang::AArch64::BI_InterlockedExchange16_nf:
+  case clang::AArch64::BI_InterlockedExchange_nf:
+  case clang::AArch64::BI_InterlockedExchange64_nf:
+  case clang::AArch64::BI_InterlockedExchangePointer_nf:
+    return MSVCIntrin::_InterlockedExchange_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange8_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange16_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange64_acq:
+  case clang::AArch64::BI_InterlockedCompareExchangePointer_acq:
+    return MSVCIntrin::_InterlockedCompareExchange_acq;
+  case clang::AArch64::BI_InterlockedCompareExchange8_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange16_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange64_rel:
+  case clang::AArch64::BI_InterlockedCompareExchangePointer_rel:
+    return MSVCIntrin::_InterlockedCompareExchange_rel;
+  case clang::AArch64::BI_InterlockedCompareExchange8_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange16_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange64_nf:
+    return MSVCIntrin::_InterlockedCompareExchange_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange128:
+    return MSVCIntrin::_InterlockedCompareExchange128;
+  case clang::AArch64::BI_InterlockedCompareExchange128_acq:
+    return MSVCIntrin::_InterlockedCompareExchange128_acq;
+  case clang::AArch64::BI_InterlockedCompareExchange128_nf:
+    return MSVCIntrin::_InterlockedCompareExchange128_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange128_rel:
+    return MSVCIntrin::_InterlockedCompareExchange128_rel;
+  case clang::AArch64::BI_InterlockedOr8_acq:
+  case clang::AArch64::BI_InterlockedOr16_acq:
+  case clang::AArch64::BI_InterlockedOr_acq:
+  case clang::AArch64::BI_InterlockedOr64_acq:
+    return MSVCIntrin::_InterlockedOr_acq;
+  case clang::AArch64::BI_InterlockedOr8_rel:
+  case clang::AArch64::BI_InterlockedOr16_rel:
+  case clang::AArch64::BI_InterlockedOr_rel:
+  case clang::AArch64::BI_InterlockedOr64_rel:
+    return MSVCIntrin::_InterlockedOr_rel;
+  case clang::AArch64::BI_InterlockedOr8_nf:
+  case clang::AArch64::BI_InterlockedOr16_nf:
+  case clang::AArch64::BI_InterlockedOr_nf:
+  case clang::AArch64::BI_InterlockedOr64_nf:
+    return MSVCIntrin::_InterlockedOr_nf;
+  case clang::AArch64::BI_InterlockedXor8_acq:
+  case clang::AArch64::BI_InterlockedXor16_acq:
+  case clang::AArch64::BI_InterlockedXor_acq:
+  case clang::AArch64::BI_InterlockedXor64_acq:
+    return MSVCIntrin::_InterlockedXor_acq;
+  case clang::AArch64::BI_InterlockedXor8_rel:
+  case clang::AArch64::BI_InterlockedXor16_rel:
+  case clang::AArch64::BI_InterlockedXor_rel:
+  case clang::AArch64::BI_InterlockedXor64_rel:
+    return MSVCIntrin::_InterlockedXor_rel;
+  case clang::AArch64::BI_InterlockedXor8_nf:
+  case clang::AArch64::BI_InterlockedXor16_nf:
+  case clang::AArch64::BI_InterlockedXor_nf:
+  case clang::AArch64::BI_InterlockedXor64_nf:
+    return MSVCIntrin::_InterlockedXor_nf;
+  case clang::AArch64::BI_InterlockedAnd8_acq:
+  case clang::AArch64::BI_InterlockedAnd16_acq:
+  case clang::AArch64::BI_InterlockedAnd_acq:
+  case clang::AArch64::BI_InterlockedAnd64_acq:
+    return MSVCIntrin::_InterlockedAnd_acq;
+  case clang::AArch64::BI_InterlockedAnd8_rel:
+  case clang::AArch64::BI_InterlockedAnd16_rel:
+  case clang::AArch64::BI_InterlockedAnd_rel:
+  case clang::AArch64::BI_InterlockedAnd64_rel:
+    return MSVCIntrin::_InterlockedAnd_rel;
+  case clang::AArch64::BI_InterlockedAnd8_nf:
+  case clang::AArch64::BI_InterlockedAnd16_nf:
+  case clang::AArch64::BI_InterlockedAnd_nf:
+  case clang::AArch64::BI_InterlockedAnd64_nf:
+    return MSVCIntrin::_InterlockedAnd_nf;
+  case clang::AArch64::BI_InterlockedIncrement16_acq:
+  case clang::AArch64::BI_InterlockedIncrement_acq:
+  case clang::AArch64::BI_InterlockedIncrement64_acq:
+    return MSVCIntrin::_InterlockedIncrement_acq;
+  case clang::AArch64::BI_InterlockedIncrement16_rel:
+  case clang::AArch64::BI_InterlockedIncrement_rel:
+  case clang::AArch64::BI_InterlockedIncrement64_rel:
+    return MSVCIntrin::_InterlockedIncrement_rel;
+  case clang::AArch64::BI_InterlockedIncrement16_nf:
+  case clang::AArch64::BI_InterlockedIncrement_nf:
+  case clang::AArch64::BI_InterlockedIncrement64_nf:
+    return MSVCIntrin::_InterlockedIncrement_nf;
+  case clang::AArch64::BI_InterlockedDecrement16_acq:
+  case clang::AArch64::BI_InterlockedDecrement_acq:
+  case clang::AArch64::BI_InterlockedDecrement64_acq:
+    return MSVCIntrin::_InterlockedDecrement_acq;
+  case clang::AArch64::BI_InterlockedDecrement16_rel:
+  case clang::AArch64::BI_InterlockedDecrement_rel:
+  case clang::AArch64::BI_InterlockedDecrement64_rel:
+    return MSVCIntrin::_InterlockedDecrement_rel;
+  case clang::AArch64::BI_InterlockedDecrement16_nf:
+  case clang::AArch64::BI_InterlockedDecrement_nf:
+  case clang::AArch64::BI_InterlockedDecrement64_nf:
+    return MSVCIntrin::_InterlockedDecrement_nf;
+  }
+  llvm_unreachable("must return from switch");
+}
+
+static std::optional<CodeGenFunction::MSVCIntrin>
+translateArmToMsvcIntrin(unsigned BuiltinID) {
+  using MSVCIntrin = CodeGenFunction::MSVCIntrin;
+  switch (BuiltinID) {
+  default:
+    return std::nullopt;
+  case clang::ARM::BI_BitScanForward:
+  case clang::ARM::BI_BitScanForward64:
+    return MSVCIntrin::_BitScanForward;
+  case clang::ARM::BI_BitScanReverse:
+  case clang::ARM::BI_BitScanReverse64:
+    return MSVCIntrin::_BitScanReverse;
+  case clang::ARM::BI_InterlockedAnd64:
+    return MSVCIntrin::_InterlockedAnd;
+  case clang::ARM::BI_InterlockedExchange64:
+    return MSVCIntrin::_InterlockedExchange;
+  case clang::ARM::BI_InterlockedExchangeAdd64:
+    return MSVCIntrin::_InterlockedExchangeAdd;
+  case clang::ARM::BI_InterlockedExchangeSub64:
+    return MSVCIntrin::_InterlockedExchangeSub;
+  case clang::ARM::BI_InterlockedOr64:
+    return MSVCIntrin::_InterlockedOr;
+  case clang::ARM::BI_InterlockedXor64:
+    return MSVCIntrin::_InterlockedXor;
+  case clang::ARM::BI_InterlockedDecrement64:
+    return MSVCIntrin::_InterlockedDecrement;
+  case clang::ARM::BI_InterlockedIncrement64:
+    return MSVCIntrin::_InterlockedIncrement;
+  case clang::ARM::BI_InterlockedExchangeAdd8_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd16_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd64_acq:
+    return MSVCIntrin::_InterlockedExchangeAdd_acq;
+  case clang::ARM::BI_InterlockedExchangeAdd8_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd16_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd64_rel:
+    return MSVCIntrin::_InterlockedExchangeAdd_rel;
+  case clang::ARM::BI_InterlockedExchangeAdd8_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd16_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd64_nf:
+    return MSVCIntrin::_InterlockedExchangeAdd_nf;
+  case clang::ARM::BI_InterlockedExchange8_acq:
+  case clang::ARM::BI_InterlockedExchange16_acq:
+  case clang::ARM::BI_InterlockedExchange_acq:
+  case clang::ARM::BI_InterlockedExchange64_acq:
+  case clang::ARM::BI_InterlockedExchangePointer_acq:
+    return MSVCIntrin::_InterlockedExchange_acq;
+  case clang::ARM::BI_InterlockedExchange8_rel:
+  case clang::ARM::BI_InterlockedExchange16_rel:
+  case clang::ARM::BI_InterlockedExchange_rel:
+  case clang::ARM::BI_InterlockedExchange64_rel:
+  case clang::ARM::BI_InterlockedExchangePointer_rel:
+    return MSVCIntrin::_InterlockedExchange_rel;
+  case clang::ARM::BI_InterlockedExchange8_nf:
+  case clang::ARM::BI_InterlockedExchange16_nf:
+  case clang::ARM::BI_InterlockedExchange_nf:
+  case clang::ARM::BI_InterlockedExchange64_nf:
+  case clang::ARM::BI_InterlockedExchangePointer_nf:
+    return MSVCIntrin::_InterlockedExchange_nf;
+  case clang::ARM::BI_InterlockedCompareExchange8_acq:
+  case clang::ARM::BI_InterlockedCompareExchange16_acq:
+  case clang::ARM::BI_InterlockedCompareExchange_acq:
+  case clang::ARM::BI_InterlockedCompareExchange64_acq:
+  case clang::ARM::BI_InterlockedCompareExchangePointer_acq:
+    return MSVCIntrin::_InterlockedCompareExchange_acq;
+  case clang::ARM::BI_InterlockedCompareExchange8_rel:
+  case clang::ARM::BI_InterlockedCompareExchange16_rel:
+  case clang::ARM::BI_InterlockedCompareExchange_rel:
+  case clang::ARM::BI_InterlockedCompareExchange64_rel:
+  case clang::ARM::BI_InterlockedCompareExchangePointer_rel:
+    return MSVCIntrin::_InterlockedCompareExchange_rel;
+  case clang::ARM::BI_InterlockedCompareExchange8_nf:
+  case clang::ARM::BI_InterlockedCompareExchange16_nf:
+  case clang::ARM::BI_InterlockedCompareExchange_nf:
+  case clang::ARM::BI_InterlockedCompareExchange64_nf:
+    return MSVCIntrin::_InterlockedCompareExchange_nf;
+  case clang::ARM::BI_InterlockedOr8_acq:
+  case clang::ARM::BI_InterlockedOr16_acq:
+  case clang::ARM::BI_InterlockedOr_acq:
+  case clang::ARM::BI_InterlockedOr64_acq:
+    return MSVCIntrin::_InterlockedOr_acq;
+  case clang::ARM::BI_InterlockedOr8_rel:
+  case clang::ARM::BI_InterlockedOr16_rel:
+  case clang::ARM::BI_InterlockedOr_rel:
+  case clang::ARM::BI_InterlockedOr64_rel:
+    return MSVCIntrin::_InterlockedOr_rel;
+  case clang::ARM::BI_InterlockedOr8_nf:
+  case clang::ARM::BI_InterlockedOr16_nf:
+  case clang::ARM::BI_InterlockedOr_nf:
+  case clang::ARM::BI_InterlockedOr64_nf:
+    return MSVCIntrin::_InterlockedOr_nf;
+  case clang::ARM::BI_InterlockedXor8_acq:
+  case clang::ARM::BI_InterlockedXor16_acq:
+  case clang::ARM::BI_InterlockedXor_acq:
+  case clang::ARM::BI_InterlockedXor64_acq:
+    return MSVCIntrin::_InterlockedXor_acq;
+  case clang::ARM::BI_InterlockedXor8_rel:
+  case clang::ARM::BI_InterlockedXor16_rel:
+  case clang::ARM::BI_InterlockedXor_rel:
+  case clang::ARM::BI_InterlockedXor64_rel:
+    return MSVCIntrin::_InterlockedXor_rel;
+  case clang::ARM::BI_InterlockedXor8_nf:
+  case clang::ARM::BI_InterlockedXor16_nf:
+  case clang::ARM::BI_InterlockedXor_nf:
+  case clang::ARM::BI_InterlockedXor64_nf:
+    return MSVCIntrin::_InterlockedXor_nf;
+  case clang::ARM::BI_InterlockedAnd8_acq:
+  case clang::ARM::BI_InterlockedAnd16_acq:
+  case clang::ARM::BI_InterlockedAnd_acq:
+  case clang::ARM::BI_InterlockedAnd64_acq:
+    return MSVCIntrin::_InterlockedAnd_acq;
+  case clang::ARM::BI_InterlockedAnd8_rel:
+  case clang::ARM::BI_InterlockedAnd16_rel:
+  case clang::ARM::BI_InterlockedAnd_rel:
+  case clang::ARM::BI_InterlockedAnd64_rel:
+    return MSVCIntrin::_InterlockedAnd_rel;
+  case clang::ARM::BI_InterlockedAnd8_nf:
+  case clang::ARM::BI_InterlockedAnd16_nf:
+  case clang::ARM::BI_InterlockedAnd_nf:
+  case clang::ARM::BI_InterlockedAnd64_nf:
+    return MSVCIntrin::_InterlockedAnd_nf;
+  case clang::ARM::BI_InterlockedIncrement16_acq:
+  case clang::ARM::BI_InterlockedIncrement_acq:
+  case clang::ARM::BI_InterlockedIncrement64_acq:
+    return MSVCIntrin::_InterlockedIncrement_acq;
+  case clang::ARM::BI_InterlockedIncrement16_rel:
+  case clang::ARM::BI_InterlockedIncrement_rel:
+  case clang::ARM::BI_InterlockedIncrement64_rel:
+    return MSVCIntrin::_InterlockedIncrement_rel;
+  case clang::ARM::BI_InterlockedIncrement16_nf:
+  case clang::ARM::BI_InterlockedIncrement_nf:
+  case clang::ARM::BI_InterlockedIncrement64_nf:
+    return MSVCIntrin::_InterlockedIncrement_nf;
+  case clang::ARM::BI_InterlockedDecrement16_acq:
+  case clang::ARM::BI_InterlockedDecrement_acq:
+  case clang::ARM::BI_InterlockedDecrement64_acq:
+    return MSVCIntrin::_InterlockedDecrement_acq;
+  case clang::ARM::BI_InterlockedDecrement16_rel:
+  case clang::ARM::BI_InterlockedDecrement_rel:
+  case clang::ARM::BI_InterlockedDecrement64_rel:
+    return MSVCIntrin::_InterlockedDecrement_rel;
+  case clang::ARM::BI_InterlockedDecrement16_nf:
+  case clang::ARM::BI_InterlockedDecrement_nf:
+  case clang::ARM::BI_InterlockedDecrement64_nf:
+    return MSVCIntrin::_InterlockedDecrement_nf;
+  }
+  llvm_unreachable("must return from switch");
+}
+
+// Emit an intrinsic where all operands are of the same type as the result.
+// Depending on mode, this may be a constrained floating-point intrinsic.
+static Value *emitCallMaybeConstrainedFPBuiltin(CodeGenFunction &CGF,
+                                                unsigned IntrinsicID,
+                                                unsigned ConstrainedIntrinsicID,
+                                                llvm::Type *Ty,
+                                                ArrayRef<Value *> Args) {
+  Function *F;
+  if (CGF.Builder.getIsFPConstrained())
+    F = CGF.CGM.getIntrinsic(ConstrainedIntrinsicID, Ty);
+  else
+    F = CGF.CGM.getIntrinsic(IntrinsicID, Ty);
+
+  if (CGF.Builder.getIsFPConstrained())
+    return CGF.Builder.CreateConstrainedFPCall(F, Args);
+  else
+    return CGF.Builder.CreateCall(F, Args);
+}
+
+static llvm::FixedVectorType *GetNeonType(CodeGenFunction *CGF,
+                                          NeonTypeFlags TypeFlags,
+                                          bool HasLegalHalfType = true,
+                                          bool V1Ty = false,
+                                          bool AllowBFloatArgsAndRet = true) {
+  int IsQuad = TypeFlags.isQuad();
+  switch (TypeFlags.getEltType()) {
+  case NeonTypeFlags::Int8:
+  case NeonTypeFlags::Poly8:
+  case NeonTypeFlags::MFloat8:
+    return llvm::FixedVectorType::get(CGF->Int8Ty, V1Ty ? 1 : (8 << IsQuad));
+  case NeonTypeFlags::Int16:
+  case NeonTypeFlags::Poly16:
+    return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::BFloat16:
+    if (AllowBFloatArgsAndRet)
+      return llvm::FixedVectorType::get(CGF->BFloatTy, V1Ty ? 1 : (4 << IsQuad));
+    else
+      return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::Float16:
+    if (HasLegalHalfType)
+      return llvm::FixedVectorType::get(CGF->HalfTy, V1Ty ? 1 : (4 << IsQuad));
+    else
+      return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::Int32:
+    return llvm::FixedVectorType::get(CGF->Int32Ty, V1Ty ? 1 : (2 << IsQuad));
+  case NeonTypeFlags::Int64:
+  case NeonTypeFlags::Poly64:
+    return llvm::FixedVectorType::get(CGF->Int64Ty, V1Ty ? 1 : (1 << IsQuad));
+  case NeonTypeFlags::Poly128:
+    // FIXME: i128 and f128 doesn't get fully support in Clang and llvm.
+    // There is a lot of i128 and f128 API missing.
+    // so we use v16i8 to represent poly128 and get pattern matched.
+    return llvm::FixedVectorType::get(CGF->Int8Ty, 16);
+  case NeonTypeFlags::Float32:
+    return llvm::FixedVectorType::get(CGF->FloatTy, V1Ty ? 1 : (2 << IsQuad));
+  case NeonTypeFlags::Float64:
+    return llvm::FixedVectorType::get(CGF->DoubleTy, V1Ty ? 1 : (1 << IsQuad));
+  }
+  llvm_unreachable("Unknown vector element type!");
+}
+
+static llvm::VectorType *GetFloatNeonType(CodeGenFunction *CGF,
+                                          NeonTypeFlags IntTypeFlags) {
+  int IsQuad = IntTypeFlags.isQuad();
+  switch (IntTypeFlags.getEltType()) {
+  case NeonTypeFlags::Int16:
+    return llvm::FixedVectorType::get(CGF->HalfTy, (4 << IsQuad));
+  case NeonTypeFlags::Int32:
+    return llvm::FixedVectorType::get(CGF->FloatTy, (2 << IsQuad));
+  case NeonTypeFlags::Int64:
+    return llvm::FixedVectorType::get(CGF->DoubleTy, (1 << IsQuad));
+  default:
+    llvm_unreachable("Type can't be converted to floating-point!");
+  }
+}
+
+Value *CodeGenFunction::EmitNeonSplat(Value *V, Constant *C,
+                                      const ElementCount &Count) {
+  Value *SV = llvm::ConstantVector::getSplat(Count, C);
+  return Builder.CreateShuffleVector(V, V, SV, "lane");
+}
+
+Value *CodeGenFunction::EmitNeonSplat(Value *V, Constant *C) {
+  ElementCount EC = cast<llvm::VectorType>(V->getType())->getElementCount();
+  return EmitNeonSplat(V, C, EC);
+}
+
+Value *CodeGenFunction::EmitNeonCall(Function *F, SmallVectorImpl<Value*> &Ops,
+                                     const char *name,
+                                     unsigned shift, bool rightshift) {
+  unsigned j = 0;
+  for (Function::const_arg_iterator ai = F->arg_begin(), ae = F->arg_end();
+       ai != ae; ++ai, ++j) {
+    if (F->isConstrainedFPIntr...
[truncated]

@llvmbot
Copy link
Member

llvmbot commented Mar 20, 2025

@llvm/pr-subscribers-clang

Author: Jonathan Thackray (jthackray)

Changes

clang/lib/CodeGen/CGBuiltin.cpp is over 1MB long (>23k LoC), and can take minutes to recompile (depending on compiler and host system) when modified, and 5 seconds for clangd to update for every edit. Splitting this file was discussed in this thread:

https://discourse.llvm.org/t/splitting-clang-s-cgbuiltin-cpp-over-23k-lines-long-takes-1min-to-compile/

and the idea has received a number of +1 votes, hence this change.

Patch is 1.50 MiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/132252.diff

13 Files Affected:

  • (added) clang/lib/CodeGen/BuiltinTargets/AArch64.cpp (+8038)
  • (added) clang/lib/CodeGen/BuiltinTargets/AMDGPU.cpp (+1884)
  • (added) clang/lib/CodeGen/BuiltinTargets/Hexagon.cpp (+267)
  • (added) clang/lib/CodeGen/BuiltinTargets/NVPTX.cpp (+1168)
  • (added) clang/lib/CodeGen/BuiltinTargets/PPC.cpp (+1363)
  • (added) clang/lib/CodeGen/BuiltinTargets/RISCV.cpp (+425)
  • (added) clang/lib/CodeGen/BuiltinTargets/SPIR.cpp (+67)
  • (added) clang/lib/CodeGen/BuiltinTargets/SystemZ.cpp (+442)
  • (added) clang/lib/CodeGen/BuiltinTargets/WebAssembly.cpp (+625)
  • (added) clang/lib/CodeGen/BuiltinTargets/X86.cpp (+3399)
  • (modified) clang/lib/CodeGen/CGBuiltin.cpp (+154-17657)
  • (added) clang/lib/CodeGen/CGBuiltin.h (+102)
  • (modified) clang/lib/CodeGen/CMakeLists.txt (+10)
diff --git a/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp b/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp
new file mode 100644
index 0000000000000..a1fe34f8a562a
--- /dev/null
+++ b/clang/lib/CodeGen/BuiltinTargets/AArch64.cpp
@@ -0,0 +1,8038 @@
+//===------ AArch64.cpp - Emit LLVM Code for builtins ---------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This contains code to emit Builtin calls as LLVM code.
+//
+//===----------------------------------------------------------------------===//
+
+#include "ABIInfo.h"
+#include "CGBuiltin.h"
+#include "TargetInfo.h"
+#include "clang/Basic/TargetBuiltins.h"
+#include "llvm/IR/InlineAsm.h"
+#include "llvm/IR/Intrinsics.h"
+#include "llvm/IR/IntrinsicsAArch64.h"
+#include "llvm/IR/IntrinsicsARM.h"
+#include "llvm/IR/IntrinsicsBPF.h"
+#include "llvm/TargetParser/AArch64TargetParser.h"
+
+using namespace clang;
+using namespace CodeGen;
+using namespace llvm;
+
+static std::optional<CodeGenFunction::MSVCIntrin>
+translateAarch64ToMsvcIntrin(unsigned BuiltinID) {
+  using MSVCIntrin = CodeGenFunction::MSVCIntrin;
+  switch (BuiltinID) {
+  default:
+    return std::nullopt;
+  case clang::AArch64::BI_BitScanForward:
+  case clang::AArch64::BI_BitScanForward64:
+    return MSVCIntrin::_BitScanForward;
+  case clang::AArch64::BI_BitScanReverse:
+  case clang::AArch64::BI_BitScanReverse64:
+    return MSVCIntrin::_BitScanReverse;
+  case clang::AArch64::BI_InterlockedAnd64:
+    return MSVCIntrin::_InterlockedAnd;
+  case clang::AArch64::BI_InterlockedExchange64:
+    return MSVCIntrin::_InterlockedExchange;
+  case clang::AArch64::BI_InterlockedExchangeAdd64:
+    return MSVCIntrin::_InterlockedExchangeAdd;
+  case clang::AArch64::BI_InterlockedExchangeSub64:
+    return MSVCIntrin::_InterlockedExchangeSub;
+  case clang::AArch64::BI_InterlockedOr64:
+    return MSVCIntrin::_InterlockedOr;
+  case clang::AArch64::BI_InterlockedXor64:
+    return MSVCIntrin::_InterlockedXor;
+  case clang::AArch64::BI_InterlockedDecrement64:
+    return MSVCIntrin::_InterlockedDecrement;
+  case clang::AArch64::BI_InterlockedIncrement64:
+    return MSVCIntrin::_InterlockedIncrement;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd_acq:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_acq:
+    return MSVCIntrin::_InterlockedExchangeAdd_acq;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd_rel:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_rel:
+    return MSVCIntrin::_InterlockedExchangeAdd_rel;
+  case clang::AArch64::BI_InterlockedExchangeAdd8_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd16_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd_nf:
+  case clang::AArch64::BI_InterlockedExchangeAdd64_nf:
+    return MSVCIntrin::_InterlockedExchangeAdd_nf;
+  case clang::AArch64::BI_InterlockedExchange8_acq:
+  case clang::AArch64::BI_InterlockedExchange16_acq:
+  case clang::AArch64::BI_InterlockedExchange_acq:
+  case clang::AArch64::BI_InterlockedExchange64_acq:
+  case clang::AArch64::BI_InterlockedExchangePointer_acq:
+    return MSVCIntrin::_InterlockedExchange_acq;
+  case clang::AArch64::BI_InterlockedExchange8_rel:
+  case clang::AArch64::BI_InterlockedExchange16_rel:
+  case clang::AArch64::BI_InterlockedExchange_rel:
+  case clang::AArch64::BI_InterlockedExchange64_rel:
+  case clang::AArch64::BI_InterlockedExchangePointer_rel:
+    return MSVCIntrin::_InterlockedExchange_rel;
+  case clang::AArch64::BI_InterlockedExchange8_nf:
+  case clang::AArch64::BI_InterlockedExchange16_nf:
+  case clang::AArch64::BI_InterlockedExchange_nf:
+  case clang::AArch64::BI_InterlockedExchange64_nf:
+  case clang::AArch64::BI_InterlockedExchangePointer_nf:
+    return MSVCIntrin::_InterlockedExchange_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange8_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange16_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange_acq:
+  case clang::AArch64::BI_InterlockedCompareExchange64_acq:
+  case clang::AArch64::BI_InterlockedCompareExchangePointer_acq:
+    return MSVCIntrin::_InterlockedCompareExchange_acq;
+  case clang::AArch64::BI_InterlockedCompareExchange8_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange16_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange_rel:
+  case clang::AArch64::BI_InterlockedCompareExchange64_rel:
+  case clang::AArch64::BI_InterlockedCompareExchangePointer_rel:
+    return MSVCIntrin::_InterlockedCompareExchange_rel;
+  case clang::AArch64::BI_InterlockedCompareExchange8_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange16_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange_nf:
+  case clang::AArch64::BI_InterlockedCompareExchange64_nf:
+    return MSVCIntrin::_InterlockedCompareExchange_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange128:
+    return MSVCIntrin::_InterlockedCompareExchange128;
+  case clang::AArch64::BI_InterlockedCompareExchange128_acq:
+    return MSVCIntrin::_InterlockedCompareExchange128_acq;
+  case clang::AArch64::BI_InterlockedCompareExchange128_nf:
+    return MSVCIntrin::_InterlockedCompareExchange128_nf;
+  case clang::AArch64::BI_InterlockedCompareExchange128_rel:
+    return MSVCIntrin::_InterlockedCompareExchange128_rel;
+  case clang::AArch64::BI_InterlockedOr8_acq:
+  case clang::AArch64::BI_InterlockedOr16_acq:
+  case clang::AArch64::BI_InterlockedOr_acq:
+  case clang::AArch64::BI_InterlockedOr64_acq:
+    return MSVCIntrin::_InterlockedOr_acq;
+  case clang::AArch64::BI_InterlockedOr8_rel:
+  case clang::AArch64::BI_InterlockedOr16_rel:
+  case clang::AArch64::BI_InterlockedOr_rel:
+  case clang::AArch64::BI_InterlockedOr64_rel:
+    return MSVCIntrin::_InterlockedOr_rel;
+  case clang::AArch64::BI_InterlockedOr8_nf:
+  case clang::AArch64::BI_InterlockedOr16_nf:
+  case clang::AArch64::BI_InterlockedOr_nf:
+  case clang::AArch64::BI_InterlockedOr64_nf:
+    return MSVCIntrin::_InterlockedOr_nf;
+  case clang::AArch64::BI_InterlockedXor8_acq:
+  case clang::AArch64::BI_InterlockedXor16_acq:
+  case clang::AArch64::BI_InterlockedXor_acq:
+  case clang::AArch64::BI_InterlockedXor64_acq:
+    return MSVCIntrin::_InterlockedXor_acq;
+  case clang::AArch64::BI_InterlockedXor8_rel:
+  case clang::AArch64::BI_InterlockedXor16_rel:
+  case clang::AArch64::BI_InterlockedXor_rel:
+  case clang::AArch64::BI_InterlockedXor64_rel:
+    return MSVCIntrin::_InterlockedXor_rel;
+  case clang::AArch64::BI_InterlockedXor8_nf:
+  case clang::AArch64::BI_InterlockedXor16_nf:
+  case clang::AArch64::BI_InterlockedXor_nf:
+  case clang::AArch64::BI_InterlockedXor64_nf:
+    return MSVCIntrin::_InterlockedXor_nf;
+  case clang::AArch64::BI_InterlockedAnd8_acq:
+  case clang::AArch64::BI_InterlockedAnd16_acq:
+  case clang::AArch64::BI_InterlockedAnd_acq:
+  case clang::AArch64::BI_InterlockedAnd64_acq:
+    return MSVCIntrin::_InterlockedAnd_acq;
+  case clang::AArch64::BI_InterlockedAnd8_rel:
+  case clang::AArch64::BI_InterlockedAnd16_rel:
+  case clang::AArch64::BI_InterlockedAnd_rel:
+  case clang::AArch64::BI_InterlockedAnd64_rel:
+    return MSVCIntrin::_InterlockedAnd_rel;
+  case clang::AArch64::BI_InterlockedAnd8_nf:
+  case clang::AArch64::BI_InterlockedAnd16_nf:
+  case clang::AArch64::BI_InterlockedAnd_nf:
+  case clang::AArch64::BI_InterlockedAnd64_nf:
+    return MSVCIntrin::_InterlockedAnd_nf;
+  case clang::AArch64::BI_InterlockedIncrement16_acq:
+  case clang::AArch64::BI_InterlockedIncrement_acq:
+  case clang::AArch64::BI_InterlockedIncrement64_acq:
+    return MSVCIntrin::_InterlockedIncrement_acq;
+  case clang::AArch64::BI_InterlockedIncrement16_rel:
+  case clang::AArch64::BI_InterlockedIncrement_rel:
+  case clang::AArch64::BI_InterlockedIncrement64_rel:
+    return MSVCIntrin::_InterlockedIncrement_rel;
+  case clang::AArch64::BI_InterlockedIncrement16_nf:
+  case clang::AArch64::BI_InterlockedIncrement_nf:
+  case clang::AArch64::BI_InterlockedIncrement64_nf:
+    return MSVCIntrin::_InterlockedIncrement_nf;
+  case clang::AArch64::BI_InterlockedDecrement16_acq:
+  case clang::AArch64::BI_InterlockedDecrement_acq:
+  case clang::AArch64::BI_InterlockedDecrement64_acq:
+    return MSVCIntrin::_InterlockedDecrement_acq;
+  case clang::AArch64::BI_InterlockedDecrement16_rel:
+  case clang::AArch64::BI_InterlockedDecrement_rel:
+  case clang::AArch64::BI_InterlockedDecrement64_rel:
+    return MSVCIntrin::_InterlockedDecrement_rel;
+  case clang::AArch64::BI_InterlockedDecrement16_nf:
+  case clang::AArch64::BI_InterlockedDecrement_nf:
+  case clang::AArch64::BI_InterlockedDecrement64_nf:
+    return MSVCIntrin::_InterlockedDecrement_nf;
+  }
+  llvm_unreachable("must return from switch");
+}
+
+static std::optional<CodeGenFunction::MSVCIntrin>
+translateArmToMsvcIntrin(unsigned BuiltinID) {
+  using MSVCIntrin = CodeGenFunction::MSVCIntrin;
+  switch (BuiltinID) {
+  default:
+    return std::nullopt;
+  case clang::ARM::BI_BitScanForward:
+  case clang::ARM::BI_BitScanForward64:
+    return MSVCIntrin::_BitScanForward;
+  case clang::ARM::BI_BitScanReverse:
+  case clang::ARM::BI_BitScanReverse64:
+    return MSVCIntrin::_BitScanReverse;
+  case clang::ARM::BI_InterlockedAnd64:
+    return MSVCIntrin::_InterlockedAnd;
+  case clang::ARM::BI_InterlockedExchange64:
+    return MSVCIntrin::_InterlockedExchange;
+  case clang::ARM::BI_InterlockedExchangeAdd64:
+    return MSVCIntrin::_InterlockedExchangeAdd;
+  case clang::ARM::BI_InterlockedExchangeSub64:
+    return MSVCIntrin::_InterlockedExchangeSub;
+  case clang::ARM::BI_InterlockedOr64:
+    return MSVCIntrin::_InterlockedOr;
+  case clang::ARM::BI_InterlockedXor64:
+    return MSVCIntrin::_InterlockedXor;
+  case clang::ARM::BI_InterlockedDecrement64:
+    return MSVCIntrin::_InterlockedDecrement;
+  case clang::ARM::BI_InterlockedIncrement64:
+    return MSVCIntrin::_InterlockedIncrement;
+  case clang::ARM::BI_InterlockedExchangeAdd8_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd16_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd_acq:
+  case clang::ARM::BI_InterlockedExchangeAdd64_acq:
+    return MSVCIntrin::_InterlockedExchangeAdd_acq;
+  case clang::ARM::BI_InterlockedExchangeAdd8_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd16_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd_rel:
+  case clang::ARM::BI_InterlockedExchangeAdd64_rel:
+    return MSVCIntrin::_InterlockedExchangeAdd_rel;
+  case clang::ARM::BI_InterlockedExchangeAdd8_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd16_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd_nf:
+  case clang::ARM::BI_InterlockedExchangeAdd64_nf:
+    return MSVCIntrin::_InterlockedExchangeAdd_nf;
+  case clang::ARM::BI_InterlockedExchange8_acq:
+  case clang::ARM::BI_InterlockedExchange16_acq:
+  case clang::ARM::BI_InterlockedExchange_acq:
+  case clang::ARM::BI_InterlockedExchange64_acq:
+  case clang::ARM::BI_InterlockedExchangePointer_acq:
+    return MSVCIntrin::_InterlockedExchange_acq;
+  case clang::ARM::BI_InterlockedExchange8_rel:
+  case clang::ARM::BI_InterlockedExchange16_rel:
+  case clang::ARM::BI_InterlockedExchange_rel:
+  case clang::ARM::BI_InterlockedExchange64_rel:
+  case clang::ARM::BI_InterlockedExchangePointer_rel:
+    return MSVCIntrin::_InterlockedExchange_rel;
+  case clang::ARM::BI_InterlockedExchange8_nf:
+  case clang::ARM::BI_InterlockedExchange16_nf:
+  case clang::ARM::BI_InterlockedExchange_nf:
+  case clang::ARM::BI_InterlockedExchange64_nf:
+  case clang::ARM::BI_InterlockedExchangePointer_nf:
+    return MSVCIntrin::_InterlockedExchange_nf;
+  case clang::ARM::BI_InterlockedCompareExchange8_acq:
+  case clang::ARM::BI_InterlockedCompareExchange16_acq:
+  case clang::ARM::BI_InterlockedCompareExchange_acq:
+  case clang::ARM::BI_InterlockedCompareExchange64_acq:
+  case clang::ARM::BI_InterlockedCompareExchangePointer_acq:
+    return MSVCIntrin::_InterlockedCompareExchange_acq;
+  case clang::ARM::BI_InterlockedCompareExchange8_rel:
+  case clang::ARM::BI_InterlockedCompareExchange16_rel:
+  case clang::ARM::BI_InterlockedCompareExchange_rel:
+  case clang::ARM::BI_InterlockedCompareExchange64_rel:
+  case clang::ARM::BI_InterlockedCompareExchangePointer_rel:
+    return MSVCIntrin::_InterlockedCompareExchange_rel;
+  case clang::ARM::BI_InterlockedCompareExchange8_nf:
+  case clang::ARM::BI_InterlockedCompareExchange16_nf:
+  case clang::ARM::BI_InterlockedCompareExchange_nf:
+  case clang::ARM::BI_InterlockedCompareExchange64_nf:
+    return MSVCIntrin::_InterlockedCompareExchange_nf;
+  case clang::ARM::BI_InterlockedOr8_acq:
+  case clang::ARM::BI_InterlockedOr16_acq:
+  case clang::ARM::BI_InterlockedOr_acq:
+  case clang::ARM::BI_InterlockedOr64_acq:
+    return MSVCIntrin::_InterlockedOr_acq;
+  case clang::ARM::BI_InterlockedOr8_rel:
+  case clang::ARM::BI_InterlockedOr16_rel:
+  case clang::ARM::BI_InterlockedOr_rel:
+  case clang::ARM::BI_InterlockedOr64_rel:
+    return MSVCIntrin::_InterlockedOr_rel;
+  case clang::ARM::BI_InterlockedOr8_nf:
+  case clang::ARM::BI_InterlockedOr16_nf:
+  case clang::ARM::BI_InterlockedOr_nf:
+  case clang::ARM::BI_InterlockedOr64_nf:
+    return MSVCIntrin::_InterlockedOr_nf;
+  case clang::ARM::BI_InterlockedXor8_acq:
+  case clang::ARM::BI_InterlockedXor16_acq:
+  case clang::ARM::BI_InterlockedXor_acq:
+  case clang::ARM::BI_InterlockedXor64_acq:
+    return MSVCIntrin::_InterlockedXor_acq;
+  case clang::ARM::BI_InterlockedXor8_rel:
+  case clang::ARM::BI_InterlockedXor16_rel:
+  case clang::ARM::BI_InterlockedXor_rel:
+  case clang::ARM::BI_InterlockedXor64_rel:
+    return MSVCIntrin::_InterlockedXor_rel;
+  case clang::ARM::BI_InterlockedXor8_nf:
+  case clang::ARM::BI_InterlockedXor16_nf:
+  case clang::ARM::BI_InterlockedXor_nf:
+  case clang::ARM::BI_InterlockedXor64_nf:
+    return MSVCIntrin::_InterlockedXor_nf;
+  case clang::ARM::BI_InterlockedAnd8_acq:
+  case clang::ARM::BI_InterlockedAnd16_acq:
+  case clang::ARM::BI_InterlockedAnd_acq:
+  case clang::ARM::BI_InterlockedAnd64_acq:
+    return MSVCIntrin::_InterlockedAnd_acq;
+  case clang::ARM::BI_InterlockedAnd8_rel:
+  case clang::ARM::BI_InterlockedAnd16_rel:
+  case clang::ARM::BI_InterlockedAnd_rel:
+  case clang::ARM::BI_InterlockedAnd64_rel:
+    return MSVCIntrin::_InterlockedAnd_rel;
+  case clang::ARM::BI_InterlockedAnd8_nf:
+  case clang::ARM::BI_InterlockedAnd16_nf:
+  case clang::ARM::BI_InterlockedAnd_nf:
+  case clang::ARM::BI_InterlockedAnd64_nf:
+    return MSVCIntrin::_InterlockedAnd_nf;
+  case clang::ARM::BI_InterlockedIncrement16_acq:
+  case clang::ARM::BI_InterlockedIncrement_acq:
+  case clang::ARM::BI_InterlockedIncrement64_acq:
+    return MSVCIntrin::_InterlockedIncrement_acq;
+  case clang::ARM::BI_InterlockedIncrement16_rel:
+  case clang::ARM::BI_InterlockedIncrement_rel:
+  case clang::ARM::BI_InterlockedIncrement64_rel:
+    return MSVCIntrin::_InterlockedIncrement_rel;
+  case clang::ARM::BI_InterlockedIncrement16_nf:
+  case clang::ARM::BI_InterlockedIncrement_nf:
+  case clang::ARM::BI_InterlockedIncrement64_nf:
+    return MSVCIntrin::_InterlockedIncrement_nf;
+  case clang::ARM::BI_InterlockedDecrement16_acq:
+  case clang::ARM::BI_InterlockedDecrement_acq:
+  case clang::ARM::BI_InterlockedDecrement64_acq:
+    return MSVCIntrin::_InterlockedDecrement_acq;
+  case clang::ARM::BI_InterlockedDecrement16_rel:
+  case clang::ARM::BI_InterlockedDecrement_rel:
+  case clang::ARM::BI_InterlockedDecrement64_rel:
+    return MSVCIntrin::_InterlockedDecrement_rel;
+  case clang::ARM::BI_InterlockedDecrement16_nf:
+  case clang::ARM::BI_InterlockedDecrement_nf:
+  case clang::ARM::BI_InterlockedDecrement64_nf:
+    return MSVCIntrin::_InterlockedDecrement_nf;
+  }
+  llvm_unreachable("must return from switch");
+}
+
+// Emit an intrinsic where all operands are of the same type as the result.
+// Depending on mode, this may be a constrained floating-point intrinsic.
+static Value *emitCallMaybeConstrainedFPBuiltin(CodeGenFunction &CGF,
+                                                unsigned IntrinsicID,
+                                                unsigned ConstrainedIntrinsicID,
+                                                llvm::Type *Ty,
+                                                ArrayRef<Value *> Args) {
+  Function *F;
+  if (CGF.Builder.getIsFPConstrained())
+    F = CGF.CGM.getIntrinsic(ConstrainedIntrinsicID, Ty);
+  else
+    F = CGF.CGM.getIntrinsic(IntrinsicID, Ty);
+
+  if (CGF.Builder.getIsFPConstrained())
+    return CGF.Builder.CreateConstrainedFPCall(F, Args);
+  else
+    return CGF.Builder.CreateCall(F, Args);
+}
+
+static llvm::FixedVectorType *GetNeonType(CodeGenFunction *CGF,
+                                          NeonTypeFlags TypeFlags,
+                                          bool HasLegalHalfType = true,
+                                          bool V1Ty = false,
+                                          bool AllowBFloatArgsAndRet = true) {
+  int IsQuad = TypeFlags.isQuad();
+  switch (TypeFlags.getEltType()) {
+  case NeonTypeFlags::Int8:
+  case NeonTypeFlags::Poly8:
+  case NeonTypeFlags::MFloat8:
+    return llvm::FixedVectorType::get(CGF->Int8Ty, V1Ty ? 1 : (8 << IsQuad));
+  case NeonTypeFlags::Int16:
+  case NeonTypeFlags::Poly16:
+    return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::BFloat16:
+    if (AllowBFloatArgsAndRet)
+      return llvm::FixedVectorType::get(CGF->BFloatTy, V1Ty ? 1 : (4 << IsQuad));
+    else
+      return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::Float16:
+    if (HasLegalHalfType)
+      return llvm::FixedVectorType::get(CGF->HalfTy, V1Ty ? 1 : (4 << IsQuad));
+    else
+      return llvm::FixedVectorType::get(CGF->Int16Ty, V1Ty ? 1 : (4 << IsQuad));
+  case NeonTypeFlags::Int32:
+    return llvm::FixedVectorType::get(CGF->Int32Ty, V1Ty ? 1 : (2 << IsQuad));
+  case NeonTypeFlags::Int64:
+  case NeonTypeFlags::Poly64:
+    return llvm::FixedVectorType::get(CGF->Int64Ty, V1Ty ? 1 : (1 << IsQuad));
+  case NeonTypeFlags::Poly128:
+    // FIXME: i128 and f128 doesn't get fully support in Clang and llvm.
+    // There is a lot of i128 and f128 API missing.
+    // so we use v16i8 to represent poly128 and get pattern matched.
+    return llvm::FixedVectorType::get(CGF->Int8Ty, 16);
+  case NeonTypeFlags::Float32:
+    return llvm::FixedVectorType::get(CGF->FloatTy, V1Ty ? 1 : (2 << IsQuad));
+  case NeonTypeFlags::Float64:
+    return llvm::FixedVectorType::get(CGF->DoubleTy, V1Ty ? 1 : (1 << IsQuad));
+  }
+  llvm_unreachable("Unknown vector element type!");
+}
+
+static llvm::VectorType *GetFloatNeonType(CodeGenFunction *CGF,
+                                          NeonTypeFlags IntTypeFlags) {
+  int IsQuad = IntTypeFlags.isQuad();
+  switch (IntTypeFlags.getEltType()) {
+  case NeonTypeFlags::Int16:
+    return llvm::FixedVectorType::get(CGF->HalfTy, (4 << IsQuad));
+  case NeonTypeFlags::Int32:
+    return llvm::FixedVectorType::get(CGF->FloatTy, (2 << IsQuad));
+  case NeonTypeFlags::Int64:
+    return llvm::FixedVectorType::get(CGF->DoubleTy, (1 << IsQuad));
+  default:
+    llvm_unreachable("Type can't be converted to floating-point!");
+  }
+}
+
+Value *CodeGenFunction::EmitNeonSplat(Value *V, Constant *C,
+                                      const ElementCount &Count) {
+  Value *SV = llvm::ConstantVector::getSplat(Count, C);
+  return Builder.CreateShuffleVector(V, V, SV, "lane");
+}
+
+Value *CodeGenFunction::EmitNeonSplat(Value *V, Constant *C) {
+  ElementCount EC = cast<llvm::VectorType>(V->getType())->getElementCount();
+  return EmitNeonSplat(V, C, EC);
+}
+
+Value *CodeGenFunction::EmitNeonCall(Function *F, SmallVectorImpl<Value*> &Ops,
+                                     const char *name,
+                                     unsigned shift, bool rightshift) {
+  unsigned j = 0;
+  for (Function::const_arg_iterator ai = F->arg_begin(), ae = F->arg_end();
+       ai != ae; ++ai, ++j) {
+    if (F->isConstrainedFPIntr...
[truncated]

@jthackray jthackray force-pushed the refactor-cgbuiltin branch from 7b7ad22 to c389db7 Compare March 20, 2025 16:38
@github-actions GitHub Actions

This comment was marked as off-topic.

Sign in to view
shiltian
shiltian reviewed Mar 20, 2025
View reviewed changes
Copy link
Contributor

@shiltian shiltian left a comment

Choose a reason for hiding this comment

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

I'm super happy to see this change. The AMDGPU part looks good to me! Thanks!

@github-actions GitHub Actions

This comment was marked as outdated.

Sign in to view
@jthackray jthackray force-pushed the refactor-cgbuiltin branch from c389db7 to eeb7668 Compare March 20, 2025 16:57
Sirraide
Sirraide reviewed Mar 20, 2025
View reviewed changes
Copy link
Member

@Sirraide Sirraide left a comment

Choose a reason for hiding this comment

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

I think splitting this up by target is a good idea.

jhuber6
jhuber6 approved these changes Mar 20, 2025
View reviewed changes
Copy link
Contributor

@jhuber6 jhuber6 left a comment

Choose a reason for hiding this comment

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

LG, thanks

MacDue
MacDue reviewed Mar 20, 2025
View reviewed changes
@arichardson
Copy link
Member

This makes a lot of sense to me, just wondering why you chose clang/lib/CodeGen/BuiltinTargets/ rather than clang/lib/CodeGen/TargetBuiltins/? The former shounds like it contains the list of all builtin targets, rather than the codegen for target-specific builtins.

@MacDue
Copy link
Member

MacDue commented Mar 20, 2025

This makes a lot of sense to me, just wondering why you chose clang/lib/CodeGen/BuiltinTargets/ rather than clang/lib/CodeGen/TargetBuiltins/? The former shounds like it contains the list of all builtin targets, rather than the codegen for target-specific builtins.

+1 to this

@jhuber6
Copy link
Contributor

jhuber6 commented Mar 20, 2025

Seems something with AArch64 has made the CI unhappy as well.

@efriedma-quic efriedma-quic mentioned this pull request Mar 20, 2025
Merged
@jthackray
Copy link
Contributor Author

jthackray commented Mar 20, 2025 •
edited
Loading

Seems something with AArch64 has made the CI unhappy as well.

@jhuber6 Thanks. It appears clang-format breaks some of the fragile #defines around NEONMAP0 and similar. If I revert the clang-format, it passes.

@jthackray jthackray force-pushed the refactor-cgbuiltin branch from 9f16212 to 1c207e8 Compare March 20, 2025 20:33
@jthackray
Copy link
Contributor Author

jthackray commented Mar 20, 2025 •
edited
Loading

This makes a lot of sense to me, just wondering why you chose clang/lib/CodeGen/BuiltinTargets/ rather than clang/lib/CodeGen/TargetBuiltins/? The former shounds like it contains the list of all builtin targets, rather than the codegen for target-specific builtins.

@arichardson @MacDue Thanks. Now updated with your suggestion.

@jthackray jthackray mentioned this pull request Mar 22, 2025
Merged
jthackray added a commit to jthackray/llvm-project that referenced this pull request Mar 22, 2025
@jthackray
[NFC][clang] Remove superfluous header files after refactor in llvm#1…
a55f760 
…32252
@jthackray
Copy link
Contributor Author

I've opened #132495 to remove quite a lot of superfluous #includes, which might improve compile time.

@jurahul
Copy link
Contributor

jurahul commented Mar 24, 2025

I see that some of the HLSL builtin code (like EmitHLSLBuiltinExpr) ended up in AMDGPU.cpp. Does that seem misplaced?

@jthackray
Copy link
Contributor Author

@jurahul

I see that some of the HLSL builtin code (like EmitHLSLBuiltinExpr) ended up in AMDGPU.cpp. Does that seem misplaced?

@jurahul Yes, apologies. Will write another PR to move it.

jthackray added a commit that referenced this pull request Mar 26, 2025
@jthackray
[NFC][clang] Remove superfluous header files after refactor in #132252 (
a1a74c9 
#132495)

Remove superfluous header files after refactor in #132252
@farzonl
Copy link
Member

farzonl commented Mar 27, 2025 •
edited
Loading

In the future It would be good to get feedback from the HLSL team before moving EmitHLSLBuiltinExpr to clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp. HLSL is a language and does not fall into the TargetBuiltin categorization. It supports all gpus.

@farzonl
Copy link
Member

farzonl commented Mar 27, 2025

Would it be ok if we prefixed all of these files in TargetBuiltins with CG? #133199

Building using AppleClang seems to be upset that a cpp file with this name already exists. I'm assuming its talking about all the files in clang/lib/Basic/Targets/. For example there is now a clang/lib/Basic/Targets/AMDGPU.cpp and clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp.

@jhuber6
Copy link
Contributor

jhuber6 commented Mar 27, 2025

Would it be ok if we prefixed all of these files in TargetBuiltins with CG? #133199

Building using AppleClang seems to be upset that a cpp file with this name already exists. I'm assuming its talking about all the files in clang/lib/Basic/Targets/. For example there is now a clang/lib/Basic/Targets/AMDGPU.cpp and clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp.

We currently have four, so I'm unsure why one more makes a difference. I guess because they go in the same library component or something?

clang/lib/Basic/Targets/AMDGPU.cpp
clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp
clang/lib/CodeGen/Targets/AMDGPU.cpp
clang/lib/Driver/ToolChains/AMDGPU.cpp

@farzonl farzonl mentioned this pull request Mar 27, 2025
Merged
farzonl added a commit that referenced this pull request Mar 27, 2025
@farzonl
[NFC][HLSL] Move emitter out of AMDGPU.cpp (#133251)
59d0607 
- Move all HLSL code out of AMDGPU.cpp to CGHLSLBuiltins.cpp
- Fixes accidental reorganization of HLSL code into AMDGPU caused by
(#132252,
7f920e2e5f70b)
llvm-sync bot pushed a commit to arm/arm-toolchain that referenced this pull request Mar 27, 2025
@farzonl @github-actions
Automerge: [NFC][HLSL] Move emitter out of AMDGPU.cpp (#133251)
4775336 
- Move all HLSL code out of AMDGPU.cpp to CGHLSLBuiltins.cpp
- Fixes accidental reorganization of HLSL code into AMDGPU caused by
(llvm/llvm-project#132252,
llvm/llvm-project@7f920e2e5f70b)
@farzonl farzonl mentioned this pull request Mar 30, 2025
Open
farzonl added a commit to farzonl/llvm-project that referenced this pull request Mar 30, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
1bb39c5 
PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.
farzonl added a commit to farzonl/llvm-project that referenced this pull request Mar 30, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
5d69e50 
PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.

fixes llvm#133199
farzonl added a commit to farzonl/llvm-project that referenced this pull request Mar 30, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
b768d3f 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
@farzonl farzonl mentioned this pull request Mar 30, 2025
Merged
farzonl added a commit that referenced this pull request Mar 31, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings (#133619)
cc2b432 
fixes #133199
 
PR #132252 Created a second file that shared `<TargetName>.cpp` in
`clang/lib/CodeGen/CMakeLists.txt`

For example There were two `AMDGPU.cpp`'s one in `TargetBuiltins` and
the other in `Targets`. Even though these were in different directories
`libtool` warns that it might not distinguish them because they share
the same base name.

There are two potential fixes. The easy fix is to rename one of them and
keep one cmake file. That solution though doesn't future proof this
problem in the event of a third `<TargetName>.cpp` and it seems teams
want to just use the target name

#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries. It might of been
possible to build an OBJECT, but I only saw examples of this in
compiler-rt and test directories so assumed there was a reason it wasn't
used.
llvm-sync bot pushed a commit to arm/arm-toolchain that referenced this pull request Mar 31, 2025
@farzonl @github-actions
Automerge: [Clang][Cmake] fix libtool duplicate member name warnings …
e6ff049 
…(#133619)

fixes #133199

PR #132252 Created a second file that shared `<TargetName>.cpp` in
`clang/lib/CodeGen/CMakeLists.txt`

For example There were two `AMDGPU.cpp`'s one in `TargetBuiltins` and
the other in `Targets`. Even though these were in different directories
`libtool` warns that it might not distinguish them because they share
the same base name.

There are two potential fixes. The easy fix is to rename one of them and
keep one cmake file. That solution though doesn't future proof this
problem in the event of a third `<TargetName>.cpp` and it seems teams
want to just use the target name

llvm/llvm-project#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries. It might of been
possible to build an OBJECT, but I only saw examples of this in
compiler-rt and test directories so assumed there was a reason it wasn't
used.
farzonl added a commit to farzonl/llvm-project that referenced this pull request Apr 1, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
a3c7c5b 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
@farzonl farzonl mentioned this pull request Apr 1, 2025
Merged
@nikic
Copy link
Contributor

nikic commented Apr 2, 2025

I looked into why the tiny files are slow to compile, and it seems to mostly come down to CodeGenFunction.h being expensive (including it adds ~3s to the build time). I submitted #134100 to improve this a bit.

farzonl added a commit to farzonl/llvm-project that referenced this pull request Apr 2, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
ac51727 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
farzonl added a commit to farzonl/llvm-project that referenced this pull request Apr 3, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
cd2f665 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
farzonl added a commit to farzonl/llvm-project that referenced this pull request Apr 3, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
beba62a 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
farzonl added a commit to farzonl/llvm-project that referenced this pull request Apr 4, 2025
@farzonl
[Clang][Cmake] fix libtool duplicate member name warnings
d4bfa07 
fixes llvm#133199

PR llvm#132252 Created a second file that shared <TargetName>.cpp in `clang/lib/CodeGen/CMakeLists.txt`

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the same base name.

There are two fixes. The easy fix is to rename one of them and keep one
cmake file. That solution though doesn't future proof this problem in
the event of a third <TargetName>.cpp and it seems teams want to
just use the target name
llvm#132252 (comment).

The alternative fix is to seperate the cmake files into their own sub
directories. I chose to create static libraries.  It might of been
possible to build an OBJECT, but I only saw examples of $<TARGET_OBJECTS:>
in compiler-rt and test directories so assumed there was a reason it wasn't used.
farzonl added a commit that referenced this pull request Apr 7, 2025
@farzonl
Reland [Clang][Cmake] fix libtool duplicate member name warnings (#13…
0d71d9a 
…3850)

fixes #133199

As of the third commit the fix to the linker missing references in
`Targets/DirectX.cpp` found in
#133776 was fixed by moving
`HLSLBufferLayoutBuilder.cpp` to `clang/lib/CodeGen/Targets/`.

It fixes the circular reference issue found in
#133619 for all
`-DBUILD_SHARED_LIBS=ON` builds by removing `target_link_libraries` from
the sub directory cmake files.

testing for amdgpu offload was done via
`cmake -B ../llvm_amdgpu -S llvm -GNinja -C
offload/cmake/caches/Offload.cmake -DCMAKE_BUILD_TYPE=Release`

PR #132252 Created a second
file that shared <TargetName>.cpp in clang/lib/CodeGen/CMakeLists.txt

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the
same base name.

There are two potential fixes. The easy fix is to rename one of them and
keep one cmake file. That solution though doesn't future proof this
problem in the event of a third <TargetName>.cpp and it seems teams want
to just use the target name

#132252 (comment).

The alternative fix that this PR went with is to seperate the cmake
files into their own sub directories as static libs.
llvm-sync bot pushed a commit to arm/arm-toolchain that referenced this pull request Apr 7, 2025
@farzonl @github-actions
Automerge: Reland [Clang][Cmake] fix libtool duplicate member name wa…
5fa816d 
…rnings (#133850)

fixes llvm/llvm-project#133199

As of the third commit the fix to the linker missing references in
`Targets/DirectX.cpp` found in
llvm/llvm-project#133776 was fixed by moving
`HLSLBufferLayoutBuilder.cpp` to `clang/lib/CodeGen/Targets/`.

It fixes the circular reference issue found in
llvm/llvm-project#133619 for all
`-DBUILD_SHARED_LIBS=ON` builds by removing `target_link_libraries` from
the sub directory cmake files.

testing for amdgpu offload was done via
`cmake -B ../llvm_amdgpu -S llvm -GNinja -C
offload/cmake/caches/Offload.cmake -DCMAKE_BUILD_TYPE=Release`

PR llvm/llvm-project#132252 Created a second
file that shared <TargetName>.cpp in clang/lib/CodeGen/CMakeLists.txt

For example There were two AMDGPU.cpp's one in TargetBuiltins and the
other in Targets. Even though these were in different directories
libtool warns that it might not distinguish them because they share the
same base name.

There are two potential fixes. The easy fix is to rename one of them and
keep one cmake file. That solution though doesn't future proof this
problem in the event of a third <TargetName>.cpp and it seems teams want
to just use the target name

llvm/llvm-project#132252 (comment).

The alternative fix that this PR went with is to seperate the cmake
files into their own sub directories as static libs.
@jthackray jthackray deleted the refactor-cgbuiltin branch April 8, 2025 10:55
@topperc topperc mentioned this pull request Apr 11, 2025
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@shiltian shiltian shiltian left review comments

@MacDue MacDue MacDue left review comments

@davemgreen davemgreen davemgreen left review comments

@Sirraide Sirraide Sirraide left review comments

@jhuber6 jhuber6 jhuber6 approved these changes

@s-barannikov s-barannikov s-barannikov approved these changes

@AaronBallman AaronBallman Awaiting requested review from AaronBallman

@wangpc-pp wangpc-pp Awaiting requested review from wangpc-pp

@CarolineConcatto CarolineConcatto Awaiting requested review from CarolineConcatto

@tmatheson-arm tmatheson-arm Awaiting requested review from tmatheson-arm

@rgwott rgwott Awaiting requested review from rgwott

@mstorsjo mstorsjo Awaiting requested review from mstorsjo

@erichkeane erichkeane Awaiting requested review from erichkeane

@jurahul jurahul Awaiting requested review from jurahul

Assignees
No one assigned
Labels
backend:AMDGPU backend:RISC-V clang:codegen IR generation bugs: mangling, exceptions, etc. clang Clang issues not falling into any other category
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.