[ADT] Always use 32-bit size type for SmallVector with 16-bit elements #95536

jayfoad · 2024-06-14T13:03:53Z

SmallVector has a special case to allow vector of char to exceed 4 GB in
size on 64-bit hosts. Apply this special case only for 8-bit element
types, instead of all element types < 32 bits.

This makes SmallVector<MCPhysReg> more compact because MCPhysReg is
uint16_t.

SmallVector has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes SmallVector<MCPhysReg> more compact because MCPhysReg is uint16_t.

llvmbot · 2024-06-14T13:04:27Z

@llvm/pr-subscribers-llvm-support

@llvm/pr-subscribers-llvm-adt

Author: Jay Foad (jayfoad)

Changes

SmallVector has a special case to allow vector of char to exceed 4 GB in
size on 64-bit hosts. Apply this special case only for 8-bit element
types, instead of all element types < 32 bits.

This makes SmallVector<MCPhysReg> more compact because MCPhysReg is
uint16_t.

Full diff: https://github.com/llvm/llvm-project/pull/95536.diff

1 Files Affected:

(modified) llvm/include/llvm/ADT/SmallVector.h (+1-2)

diff --git a/llvm/include/llvm/ADT/SmallVector.h b/llvm/include/llvm/ADT/SmallVector.h
index 09676d792dfeb..d7e889cc7b1e4 100644
--- a/llvm/include/llvm/ADT/SmallVector.h
+++ b/llvm/include/llvm/ADT/SmallVector.h
@@ -116,8 +116,7 @@ template <class Size_T> class SmallVectorBase {
 
 template <class T>
 using SmallVectorSizeType =
-    std::conditional_t<sizeof(T) < 4 && sizeof(void *) >= 8, uint64_t,
-                       uint32_t>;
+    std::conditional_t<sizeof(T) == 1, uintptr_t, uint32_t>;
 
 /// Figure out the offset of the first element.
 template <class T, typename = void> struct SmallVectorAlignmentAndSize {

llvm/include/llvm/ADT/SmallVector.h

Co-authored-by: Nikita Popov <[email protected]>

nikic

LGTM. To the best of my knowledge, the original motivation for this was specifically about the char/byte case, to support large binary blobs. If this is needed for other types, then we should have a mechanism to control it.

kuhar

Please add some tests for common types like char, int16_t, etc. (e.g., using std::is_same_v<T1, T2>)

jayfoad · 2024-06-15T08:16:16Z

Please add some tests for common types like char, int16_t, etc. (e.g., using std::is_same_v<T1, T2>)

There were already some static_asserts for this kind of thing. I added one for int16_t.

llvm-ci · 2024-06-26T21:00:32Z

LLVM Buildbot has detected a new failure on builder llvm-clang-x86_64-darwin running on doug-worker-3 while building llvm at step 5 "build-unified-tree".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/23/builds/398

Here is the relevant piece of the build log for the reference:

Step 5 (build-unified-tree) failure: build (failure)
...
54.104 [5597/12/186] Building CXX object lib/TableGen/CMakeFiles/LLVMTableGen.dir/Record.cpp.o
54.498 [5596/12/187] Building CXX object utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/IntrinsicEmitter.cpp.o
55.096 [5595/12/188] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/Option.cpp.o
55.120 [5594/12/189] Building CXX object lib/Bitstream/Reader/CMakeFiles/LLVMBitstreamReader.dir/BitstreamReader.cpp.o
55.152 [5593/12/190] Linking CXX static library lib/libLLVMBitstreamReader.a
55.538 [5592/12/191] Building CXX object utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/DirectiveEmitter.cpp.o
55.809 [5591/12/192] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/ArgList.cpp.o
55.886 [5590/12/193] Building CXX object lib/TableGen/CMakeFiles/LLVMTableGen.dir/TGParser.cpp.o
55.924 [5589/12/194] Linking CXX static library lib/libLLVMTableGen.a
56.087 [5588/12/195] Linking CXX executable bin/llvm-min-tblgen
FAILED: bin/llvm-min-tblgen 
: && /usr/bin/clang++ -fPIC -fvisibility-inlines-hidden -Werror=date-time -Werror=unguarded-availability-new -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -pedantic -Wno-long-long -Wc++98-compat-extra-semi -Wimplicit-fallthrough -Wcovered-switch-default -Wno-noexcept-type -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wsuggest-override -Wstring-conversion -Wmisleading-indentation -Wctad-maybe-unsupported -fdiagnostics-color -O3 -DNDEBUG -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX14.4.sdk -Wl,-search_paths_first -Wl,-headerpad_max_install_names -Wl,-dead_strip -Wl,-no_warn_duplicate_libraries utils/TableGen/Basic/CMakeFiles/obj.LLVMTableGenBasic.dir/CodeGenIntrinsics.cpp.o utils/TableGen/Basic/CMakeFiles/obj.LLVMTableGenBasic.dir/SDNodeProperties.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/TableGen.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/ARMTargetDefEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/Attributes.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/DirectiveEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/IntrinsicEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/RISCVTargetDefEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/VTEmitter.cpp.o -o bin/llvm-min-tblgen  -Wl,-rpath,@loader_path/../lib  lib/libLLVMSupport.a  lib/libLLVMTableGen.a  lib/libLLVMSupport.a  -lm  /Library/Developer/CommandLineTools/SDKs/MacOSX14.4.sdk/usr/lib/libz.tbd  /usr/local/lib/libzstd.dylib  lib/libLLVMDemangle.a && :
Undefined symbols for architecture x86_64:
  "llvm::SmallVectorBase<unsigned long>::grow_pod(void*, unsigned long, unsigned long)", referenced from:
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      ...
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
57.136 [5588/11/196] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFError.cpp.o
57.372 [5588/10/197] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFCommon.cpp.o
57.847 [5588/9/198] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CodeViewError.cpp.o
57.896 [5588/8/199] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MappedBlockStream.cpp.o
58.109 [5588/7/200] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/AppendingTypeTableBuilder.cpp.o
58.381 [5588/6/201] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/OptTable.cpp.o
58.553 [5588/5/202] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFBuilder.cpp.o
58.627 [5588/4/203] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CodeViewRecordIO.cpp.o
59.182 [5588/3/204] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/ContinuationRecordBuilder.cpp.o
59.308 [5588/2/205] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CVSymbolVisitor.cpp.o

dyung · 2024-06-26T21:09:49Z

LLVM Buildbot has detected a new failure on builder llvm-clang-x86_64-darwin running on doug-worker-3 while building llvm at step 5 "build-unified-tree".

Full details are available at: https://lab.llvm.org/buildbot/#/builders/23/builds/398

Here is the relevant piece of the build log for the reference:

Step 5 (build-unified-tree) failure: build (failure)
...
54.104 [5597/12/186] Building CXX object lib/TableGen/CMakeFiles/LLVMTableGen.dir/Record.cpp.o
54.498 [5596/12/187] Building CXX object utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/IntrinsicEmitter.cpp.o
55.096 [5595/12/188] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/Option.cpp.o
55.120 [5594/12/189] Building CXX object lib/Bitstream/Reader/CMakeFiles/LLVMBitstreamReader.dir/BitstreamReader.cpp.o
55.152 [5593/12/190] Linking CXX static library lib/libLLVMBitstreamReader.a
55.538 [5592/12/191] Building CXX object utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/DirectiveEmitter.cpp.o
55.809 [5591/12/192] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/ArgList.cpp.o
55.886 [5590/12/193] Building CXX object lib/TableGen/CMakeFiles/LLVMTableGen.dir/TGParser.cpp.o
55.924 [5589/12/194] Linking CXX static library lib/libLLVMTableGen.a
56.087 [5588/12/195] Linking CXX executable bin/llvm-min-tblgen
FAILED: bin/llvm-min-tblgen 
: && /usr/bin/clang++ -fPIC -fvisibility-inlines-hidden -Werror=date-time -Werror=unguarded-availability-new -Wall -Wextra -Wno-unused-parameter -Wwrite-strings -Wcast-qual -Wmissing-field-initializers -pedantic -Wno-long-long -Wc++98-compat-extra-semi -Wimplicit-fallthrough -Wcovered-switch-default -Wno-noexcept-type -Wnon-virtual-dtor -Wdelete-non-virtual-dtor -Wsuggest-override -Wstring-conversion -Wmisleading-indentation -Wctad-maybe-unsupported -fdiagnostics-color -O3 -DNDEBUG -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX14.4.sdk -Wl,-search_paths_first -Wl,-headerpad_max_install_names -Wl,-dead_strip -Wl,-no_warn_duplicate_libraries utils/TableGen/Basic/CMakeFiles/obj.LLVMTableGenBasic.dir/CodeGenIntrinsics.cpp.o utils/TableGen/Basic/CMakeFiles/obj.LLVMTableGenBasic.dir/SDNodeProperties.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/TableGen.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/ARMTargetDefEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/Attributes.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/DirectiveEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/IntrinsicEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/RISCVTargetDefEmitter.cpp.o utils/TableGen/CMakeFiles/llvm-min-tblgen.dir/VTEmitter.cpp.o -o bin/llvm-min-tblgen  -Wl,-rpath,@loader_path/../lib  lib/libLLVMSupport.a  lib/libLLVMTableGen.a  lib/libLLVMSupport.a  -lm  /Library/Developer/CommandLineTools/SDKs/MacOSX14.4.sdk/usr/lib/libz.tbd  /usr/local/lib/libzstd.dylib  lib/libLLVMDemangle.a && :
Undefined symbols for architecture x86_64:
  "llvm::SmallVectorBase<unsigned long>::grow_pod(void*, unsigned long, unsigned long)", referenced from:
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      llvm::detail::IEEEFloat::toString(llvm::SmallVectorImpl<char>&, unsigned int, unsigned int, bool) const in libLLVMSupport.a[13](APFloat.cpp.o)
      ...
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
57.136 [5588/11/196] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFError.cpp.o
57.372 [5588/10/197] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFCommon.cpp.o
57.847 [5588/9/198] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CodeViewError.cpp.o
57.896 [5588/8/199] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MappedBlockStream.cpp.o
58.109 [5588/7/200] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/AppendingTypeTableBuilder.cpp.o
58.381 [5588/6/201] Building CXX object lib/Option/CMakeFiles/LLVMOption.dir/OptTable.cpp.o
58.553 [5588/5/202] Building CXX object lib/DebugInfo/MSF/CMakeFiles/LLVMDebugInfoMSF.dir/MSFBuilder.cpp.o
58.627 [5588/4/203] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CodeViewRecordIO.cpp.o
59.182 [5588/3/204] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/ContinuationRecordBuilder.cpp.o
59.308 [5588/2/205] Building CXX object lib/DebugInfo/CodeView/CMakeFiles/LLVMDebugInfoCodeView.dir/CVSymbolVisitor.cpp.o

This failure also happened on my other mac buildbot https://lab.llvm.org/buildbot/#/builders/190/builds/750

It didn't get automatically flagged because the host machine ran out of disk space which I am fixing up now.

BertalanD · 2024-06-26T21:28:20Z

In SmallVectorSizeType's definition, size_t or uint32_t is used:

llvm-project/llvm/include/llvm/ADT/SmallVector.h

Lines 117 to 119 in 2582d11

    
           template <class T> 
        
           using SmallVectorSizeType = 
        
               std::conditional_t<sizeof(T) == 1, size_t, uint32_t>;

But the grow_pod function is instantiated for uint64_t or uint32_t:

llvm-project/llvm/lib/Support/SmallVector.cpp

Lines 163 to 170 in 2582d11

    
           template class llvm::SmallVectorBase<uint32_t>; 
        
           // Disable the uint64_t instantiation for 32-bit builds. 
        
           // Both uint32_t and uint64_t instantiations are needed for 64-bit builds. 
        
           // This instantiation will never be used in 32-bit builds, and will cause 
        
           // warnings when sizeof(Size_T) > sizeof(size_t). 
        
           #if SIZE_MAX > UINT32_MAX 
        
           template class llvm::SmallVectorBase<uint64_t>;

On the arm64 Apple ABI, size_t == unsigned long and uint64_t == unsigned long long.

chelcassanova · 2024-06-26T21:45:03Z

This is also breaking the GreenDragon macOS buildbots: https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console

kuhar · 2024-06-26T21:46:50Z

Revert?

chelcassanova · 2024-06-26T21:56:31Z

Sure, just to unblock the bots while this gets looked into.

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

llvm#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (llvm#96826) Reverts llvm#95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (#96826) Reverts #95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

llvm#95536) `SmallVector` has a special case to allow vector of char to exceed 4 GB in size on 64-bit hosts. Apply this special case only for 8-bit element types, instead of all element types < 32 bits. This makes `SmallVector<MCPhysReg>` more compact because `MCPhysReg` is `uint16_t`. --------- Co-authored-by: Nikita Popov <[email protected]>

… elements" (llvm#96826) Reverts llvm#95536, this is breaking macOS GreenDragon buildbots on arm64 and x86_64 https://green.lab.llvm.org/job/llvm.org/view/LLDB/job/as-lldb-cmake/6522/console, also breaks the Darwin LLVM buildbots: https://lab.llvm.org/buildbot/#/builders/23/builds/398

llvmbot added the llvm:adt label Jun 14, 2024

jayfoad mentioned this pull request Jun 14, 2024

[RFC][MC] Cache MCRegAliasIterator #93510

Merged

jayfoad requested review from nikic and Pierre-vh June 14, 2024 13:04

nikic reviewed Jun 14, 2024

View reviewed changes

llvm/include/llvm/ADT/SmallVector.h Outdated Show resolved Hide resolved

Update llvm/include/llvm/ADT/SmallVector.h

d025afa

Co-authored-by: Nikita Popov <[email protected]>

nikic approved these changes Jun 14, 2024

View reviewed changes

kuhar reviewed Jun 14, 2024

View reviewed changes

Add static_assert

04d78ce

llvmbot added the llvm:support label Jun 15, 2024

Merge remote-tracking branch 'origin/main' into smallvector-16-bit

79ce4d8

kuhar approved these changes Jun 26, 2024

View reviewed changes

jayfoad merged commit 2582d11 into llvm:main Jun 26, 2024
5 of 6 checks passed

jayfoad deleted the smallvector-16-bit branch June 26, 2024 20:48

chelcassanova mentioned this pull request Jun 26, 2024

Revert "[ADT] Always use 32-bit size type for SmallVector with 16-bit elements" #96826

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[ADT] Always use 32-bit size type for SmallVector with 16-bit elements #95536

[ADT] Always use 32-bit size type for SmallVector with 16-bit elements #95536

jayfoad commented Jun 14, 2024 •

edited

Loading

llvmbot commented Jun 14, 2024 •

edited

Loading

nikic left a comment

kuhar left a comment

jayfoad commented Jun 15, 2024

llvm-ci commented Jun 26, 2024

dyung commented Jun 26, 2024

BertalanD commented Jun 26, 2024

chelcassanova commented Jun 26, 2024

kuhar commented Jun 26, 2024

chelcassanova commented Jun 26, 2024

[ADT] Always use 32-bit size type for SmallVector with 16-bit elements #95536

[ADT] Always use 32-bit size type for SmallVector with 16-bit elements #95536

Conversation

jayfoad commented Jun 14, 2024 • edited Loading

llvmbot commented Jun 14, 2024 • edited Loading

nikic left a comment

Choose a reason for hiding this comment

kuhar left a comment

Choose a reason for hiding this comment

jayfoad commented Jun 15, 2024

llvm-ci commented Jun 26, 2024

dyung commented Jun 26, 2024

BertalanD commented Jun 26, 2024

chelcassanova commented Jun 26, 2024

kuhar commented Jun 26, 2024

chelcassanova commented Jun 26, 2024

jayfoad commented Jun 14, 2024 •

edited

Loading

llvmbot commented Jun 14, 2024 •

edited

Loading