[COFF, Mach-O] Include -mllvm options in thinlto cache key

[thakis]

Summary

Like D134013, but for COFF and Mach-O.

Also expand the ELF test a bit. I at first didn't realize that getValue() for
-mllvm -foo=bar would return -foo=bar instead of just bar, and so
I wrote the test to check if we indeed get this wrong. We don't, but
having the test for it seems nice, so I'm including it.

[thakis]

fix COFF test

[thakis]

clang-format

[thakis]

actually clang-format

[MaskRay]

Looks great!

lld/COFF/Config.h

 //===- Config.h -------------------------------------------------*- C++ -*-===//
 //
 // 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
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef LLD_COFF_CONFIG_H
 #define LLD_COFF_CONFIG_H
 
 #include "llvm/ADT/MapVector.h"
 #include "llvm/ADT/SetVector.h"
+#include "llvm/ADT/SmallVector.h"
 #include "llvm/ADT/StringMap.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/Object/COFF.h"
 #include "llvm/Support/CachePruning.h"
 #include "llvm/Support/VirtualFileSystem.h"
 #include <cstdint>
 #include <map>
 #include <set>
struct Configuration {
   bool debugGHashes = false;
   bool debugSymtab = false;
   bool driver = false;
   bool driverUponly = false;
   bool driverWdm = false;
   bool showTiming = false;
   bool showSummary = false;
   unsigned debugTypes = static_cast<unsigned>(DebugType::None);
+  llvm::SmallVector<llvm::StringRef, 0> mllvmOpts;
   std::vector<std::string> natvisFiles;
   llvm::StringMap<std::string> namedStreams;
   llvm::SmallString<128> pdbAltPath;
   int pdbPageSize = 4096;
   llvm::SmallString<128> pdbPath;
   llvm::SmallString<128> pdbSourcePath;
   std::vector<llvm::StringRef> argv;
 

lld/COFF/Driver.cpp

void LinkerDriver::linkerMain(ArrayRef<const char *> argsArr) {
 
   // Parse command line options.
   ArgParser parser;
   opt::InputArgList args = parser.parse(argsArr);
 
   // Parse and evaluate -mllvm options.
   std::vector<const char *> v;
   v.push_back("lld-link (LLVM option parsing)");
-  for (auto *arg : args.filtered(OPT_mllvm))
+  for (const auto *arg : args.filtered(OPT_mllvm)) {
     v.push_back(arg->getValue());
+    config->mllvmOpts.emplace_back(arg->getValue());
+  }
   cl::ResetAllOptionOccurrences();
   cl::ParseCommandLineOptions(v.size(), v.data());
 
   // Handle /errorlimit early, because error() depends on it.
   if (auto *arg = args.getLastArg(OPT_errorlimit)) {
     int n = 20;
     StringRef s = arg->getValue();
     if (s.getAsInteger(10, n))

lld/COFF/LTO.cpp

return lto::getThinLTOOutputFile(
       std::string(path), std::string(config->thinLTOPrefixReplace.first),
       std::string(config->thinLTOPrefixReplace.second));
 }
 
 static lto::Config createConfig() {
   lto::Config c;
   c.Options = initTargetOptionsFromCodeGenFlags();
   c.Options.EmitAddrsig = true;
+  for (StringRef C : config->mllvmOpts)
+    c.MllvmArgs.emplace_back(C.str());
 
   // Always emit a section per function/datum with LTO. LLVM LTO should get most
   // of the benefit of linker GC, but there are still opportunities for ICF.
   c.Options.FunctionSections = true;
   c.Options.DataSections = true;
 
   // Use static reloc model on 32-bit x86 because it usually results in more
   // compact code, and because there are also known code generation bugs when

lld/ELF/Driver.cpp

static void readConfigs(opt::InputArgList &args) {
   for (opt::Arg *arg : args.filtered(OPT_plugin_opt_eq))
     if (!StringRef(arg->getValue()).endswith("lto-wrapper"))
       error(arg->getSpelling() + ": unknown plugin option '" + arg->getValue() +
             "'");
 
   config->passPlugins = args::getStrings(args, OPT_load_pass_plugins);
 
   // Parse -mllvm options.
-  for (auto *arg : args.filtered(OPT_mllvm)) {
+  for (const auto *arg : args.filtered(OPT_mllvm)) {
     parseClangOption(arg->getValue(), arg->getSpelling());
     config->mllvmOpts.emplace_back(arg->getValue());
   }
 
   // --threads= takes a positive integer and provides the default value for
   // --thinlto-jobs=.
   if (auto *arg = args.getLastArg(OPT_threads)) {
     StringRef v(arg->getValue());

lld/MachO/Config.h

struct Configuration {
   SymbolPatterns exportedSymbols;
   SymbolPatterns unexportedSymbols;
   SymbolPatterns whyLive;
 
   std::vector<std::pair<llvm::StringRef, llvm::StringRef>> aliasedSymbols;
 
   SymtabPresence localSymbolsPresence = SymtabPresence::All;
   SymbolPatterns localSymbolPatterns;
+  llvm::SmallVector<llvm::StringRef, 0> mllvmOpts;
 
   bool zeroModTime = true;
 
   llvm::StringRef osoPrefix;
 
   std::vector<llvm::StringRef> dyldEnvs;
 
   llvm::MachO::Architecture arch() const { return platformInfo.target.Arch; }

lld/MachO/Driver.cpp

if (config->timeTraceEnabled)
 
     cl::ResetAllOptionOccurrences();
 
     // Parse LTO options.
     if (const Arg *arg = args.getLastArg(OPT_mcpu))
       parseClangOption(saver().save("-mcpu=" + StringRef(arg->getValue())),
                        arg->getSpelling());
 
-    for (const Arg *arg : args.filtered(OPT_mllvm))
+    for (const Arg *arg : args.filtered(OPT_mllvm)) {
       parseClangOption(arg->getValue(), arg->getSpelling());
+      config->mllvmOpts.emplace_back(arg->getValue());
+    }
 
     createSyntheticSections();
     createSyntheticSymbols();
     addSynthenticMethnames();
 
     createAliases();
     // If we are in "explicit exports" mode, hide everything that isn't
     // explicitly exported. Do this before running LTO so that LTO can better

lld/MachO/LTO.cpp

 using namespace llvm;
 using namespace llvm::MachO;
 using namespace llvm::sys;
 
 static lto::Config createConfig() {
   lto::Config c;
   c.Options = initTargetOptionsFromCodeGenFlags();
   c.Options.EmitAddrsig = config->icfLevel == ICFLevel::safe;
+  for (StringRef C : config->mllvmOpts)
+    c.MllvmArgs.emplace_back(C.str());
   c.CodeModel = getCodeModelFromCMModel();
   c.CPU = getCPUStr();
   c.MAttrs = getMAttrs();
   c.DiagHandler = diagnosticHandler;
   c.PreCodeGenPassesHook = [](legacy::PassManager &pm) {
     pm.add(createObjCARCContractPass());
   };
   c.TimeTraceEnabled = config->timeTraceEnabled;

lld/test/COFF/lto-cache.ll

 ; REQUIRES: x86
 ; NetBSD: noatime mounts currently inhibit 'touch' from updating atime
 ; UNSUPPORTED: system-netbsd
 
 ; RUN: opt -module-hash -module-summary %s -o %t.o
 ; RUN: opt -module-hash -module-summary %p/Inputs/lto-cache.ll -o %t2.o
 
 ; RUN: rm -Rf %t.cache && mkdir %t.cache
 ; Create two files that would be removed by cache pruning due to age.
 ; We should only remove files matching "llvmcache-*" or "Thin-*".
 ; RUN: touch -t 197001011200 %t.cache/llvmcache-foo %t.cache/Thin-123.tmp.o %t.cache/foo
 ; RUN: lld-link /lldltocache:%t.cache /lldltocachepolicy:prune_after=1h /out:%t3 /entry:main %t2.o %t.o
 
 ; Two cached objects, plus a timestamp file and "foo", minus the file we removed.
 ; RUN: ls %t.cache | count 4
 
+;; Check that mllvm options participate in the cache key
+; RUN: rm -rf %t.cache && mkdir %t.cache
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o
+; RUN: ls %t.cache | count 3
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 5
+
+;; Adding another option resuls in 2 more cache entries
+; RUN: rm -rf %t.cache && mkdir %t.cache
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o
+; RUN: ls %t.cache | count 3
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 5
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default /mllvm:-max-devirt-iterations=1
+; RUN: ls %t.cache | count 7
+
+;; Changing order may matter - e.g. if overriding /mllvm:options - so we get 2 more entries
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-max-devirt-iterations=1 /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 9
+
+;; Going back to a pre-cached order doesn't create more entries.
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default /mllvm:-max-devirt-iterations=1
+; RUN: ls %t.cache | count 9
+
+;; Different flag values matter
+; RUN: rm -rf %t.cache && mkdir %t.cache
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default /mllvm:-max-devirt-iterations=2
+; RUN: ls %t.cache | count 3
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default /mllvm:-max-devirt-iterations=1
+; RUN: ls %t.cache | count 5
+
+;; Same flag value passed to different flags matters, and switching the order
+;; of the two flags matters.
+; RUN: rm -rf %t.cache && mkdir %t.cache
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 3
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-emit-dwarf-unwind=default
+; RUN: ls %t.cache | count 5
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 5
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-enable-ml-inliner=default /mllvm:-emit-dwarf-unwind=default
+; RUN: ls %t.cache | count 7
+; RUN: lld-link /lldltocache:%t.cache /entry:main /out:%t3 %t2.o %t.o /mllvm:-emit-dwarf-unwind=default /mllvm:-enable-ml-inliner=default
+; RUN: ls %t.cache | count 9
+
 target datalayout = "e-m:w-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-pc-windows-msvc"
 
 define void @globalfunc() #0 {
 entry:
   ret void
 }

lld/test/ELF/lto/cache.ll

 
 ;; Different flag values matter
 ; RUN: rm -rf %t.cache && mkdir %t.cache
 ; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=2
 ; RUN: ls %t.cache | count 3
 ; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=1
 ; RUN: ls %t.cache | count 5
 
+;; Same flag value passed to different flags matters, and switching the order
+;; of the two flags matters.
+; RUN: rm -rf %t.cache && mkdir %t.cache
+; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t.cache | count 3
+; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -emit-dwarf-unwind=default
+; RUN: ls %t.cache | count 5
+; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t.cache | count 5
+; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -enable-ml-inliner=default -mllvm -emit-dwarf-unwind=default
+; RUN: ls %t.cache | count 7
+; RUN: ld.lld --thinlto-cache-dir=%t.cache -o %t3 %t2.o %t.o -mllvm -emit-dwarf-unwind=default -mllvm -enable-ml-inliner=default
+; RUN: ls %t.cache | count 9
+
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
 define void @globalfunc() #0 {
 entry:
   ret void
 }

lld/test/MachO/lto-cache.ll

 ;; Check that  `-max_relative_cache_size_lto` is a legal argument.
 ; RUN: %lld -cache_path_lto %t/cache -max_relative_cache_size_lto 10 \
 ; RUN:   -o %t/test %t/foo.o %t/bar.o
 
 ; CHECK-NOT: llvmcache-old
 ; CHECK: llvmcache-newer
 ; CHECK-NOT: llvmcache-old
 
+;; Check that mllvm options participate in the cache key
+; RUN: rm -rf %t/cache && mkdir %t/cache
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o
+; RUN: ls %t/cache | count 3
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 5
+
+;; Adding another option resuls in 2 more cache entries
+; RUN: rm -rf %t/cache && mkdir %t/cache
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o
+; RUN: ls %t/cache | count 3
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 5
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=1
+; RUN: ls %t/cache | count 7
+
+;; Changing order may matter - e.g. if overriding -mllvm options - so we get 2 more entries
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -max-devirt-iterations=1 -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 9
+
+;; Going back to a pre-cached order doesn't create more entries.
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=1
+; RUN: ls %t/cache | count 9
+
+;; Different flag values matter
+; RUN: rm -rf %t/cache && mkdir %t/cache
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=2
+; RUN: ls %t/cache | count 3
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default -mllvm -max-devirt-iterations=1
+; RUN: ls %t/cache | count 5
+
+;; Same flag value passed to different flags matters, and switching the order
+;; of the two flags matters.
+; RUN: rm -rf %t/cache && mkdir %t/cache
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 3
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -emit-dwarf-unwind=default
+; RUN: ls %t/cache | count 5
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 5
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -enable-ml-inliner=default -mllvm -emit-dwarf-unwind=default
+; RUN: ls %t/cache | count 7
+; RUN: %lld -cache_path_lto %t/cache -o %t/test %t/foo.o %t/bar.o -mllvm -emit-dwarf-unwind=default -mllvm -enable-ml-inliner=default
+; RUN: ls %t/cache | count 9
+
 ;--- foo.ll
 
 target triple = "x86_64-apple-macosx10.15.0"
 target datalayout = "e-m:o-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 
 define void @globalfunc() #0 {
 entry:
   ret void