[lld-macho] Use fixed chunk size for UUID

[MaskRay]

Summary

Chunk size decided by the thread count makes the UUID less deterministic
(e.g. across machines with different core counts.)
Follow ELF and just use a fixed chunksize.

Fixes: https://github.com/llvm/llvm-project/issues/63961

[MaskRay]

Didn't see D155733. abandoning

[MaskRay]

update description

[aaron.ballman]

FWIW, I think this may have broken the lld bot for Windows: https://lab.llvm.org/buildbot/#/builders/123/builds/20027

[MaskRay]

In D155761#4519510, @aaron.ballman wrote:

FWIW, I think this may have broken the lld bot for Windows: https://lab.llvm.org/buildbot/#/builders/123/builds/20027

https://lab.llvm.org/buildbot/#/builders/123/builds/20027 is about lld/test/ELF/export-dynamic-symbol.s, which is last modified in 2021 by me. The ld.lld (lld/ELF) test is unrelated to lld-macho (lld/MachO).

Unfortunately I cannot spot anything with the information. Another Windows bot (that I check more frequently) http://45.33.8.238/ is happy with the test.
Dumping all output from llvm-objdump may help identify the root cause.

$ "c:\b\slave\clang-x64-windows-msvc\build\stage1\bin\filecheck.exe" "--check-prefix=PLT1" "C:\b\slave\clang-x64-windows-msvc\llvm-project\lld\test\ELF\export-dynamic-symbol.s"
# command stderr:
C:\b\slave\clang-x64-windows-msvc\llvm-project\lld\test\ELF\export-dynamic-symbol.s:64:9: error: PLT1: expected string not found in input
# PLT1: <foo@plt>

[aaron.ballman]

In D155761#4519565, @MaskRay wrote:

In D155761#4519510, @aaron.ballman wrote:

FWIW, I think this may have broken the lld bot for Windows: https://lab.llvm.org/buildbot/#/builders/123/builds/20027

https://lab.llvm.org/buildbot/#/builders/123/builds/20027 is about lld/test/ELF/export-dynamic-symbol.s, which is last modified in 2021 by me. The ld.lld (lld/ELF) test is unrelated to lld-macho (lld/MachO).

Oh! Interesting coincidence then. There were a few changes on the blamelist for https://lab.llvm.org/buildbot/#/builders/123/builds/20027 and this was the only one for lld that changed something other than tests. I don't build lld myself, I just noticed the bot went down for the better part of a day and nobody was addressing it.

Unfortunately I cannot spot anything with the information. Another Windows bot (that I check more frequently) http://45.33.8.238/ is happy with the test.
Dumping all output from llvm-objdump may help identify the root cause.

$ "c:\b\slave\clang-x64-windows-msvc\build\stage1\bin\filecheck.exe" "--check-prefix=PLT1" "C:\b\slave\clang-x64-windows-msvc\llvm-project\lld\test\ELF\export-dynamic-symbol.s"
# command stderr:
C:\b\slave\clang-x64-windows-msvc\llvm-project\lld\test\ELF\export-dynamic-symbol.s:64:9: error: PLT1: expected string not found in input
# PLT1: <foo@plt>

Hmm yeah, the arm64 bots are happy with the changes as well.

lld/MachO/Writer.cpp

 
 // In order to utilize multiple cores, we first split the buffer into chunks,
 // compute a hash for each chunk, and then compute a hash value of the hash
 // values.
 void Writer::writeUuid() {
   TimeTraceScope timeScope("Computing UUID");
 
   ArrayRef<uint8_t> data{buffer->getBufferStart(), buffer->getBufferEnd()};
-  unsigned chunkCount = parallel::strategy.compute_thread_count() * 10;
-  // Round-up integer division
-  size_t chunkSize = (data.size() + chunkCount - 1) / chunkCount;
-  std::vector<ArrayRef<uint8_t>> chunks = split(data, chunkSize);
+  std::vector<ArrayRef<uint8_t>> chunks = split(data, 1024 * 1024);
   // Leave one slot for filename
   std::vector<uint64_t> hashes(chunks.size() + 1);
   SmallVector<std::shared_future<void>> threadFutures;
   threadFutures.reserve(chunks.size());
   for (size_t i = 0; i < chunks.size(); ++i)
     threadFutures.emplace_back(threadPool.async(
         [&](size_t j) { hashes[j] = xxh3_64bits(chunks[j]); }, i));
   for (std::shared_future<void> &future : threadFutures)