[lldb][Language] Sync LanguageType enumeration with DWARF codes

[Michael137]

Synchronized with the values in

llvm-project/llvm/include/llvm/BinaryFormat/Dwarf.def

Lines 1012 to 1021 in c2765b7

	HANDLE_DW_LANG(0x0034, GLSL, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0035, GLSL_ES, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0036, HLSL, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0037, OpenCL_CPP, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0038, CPP_for_OpenCL, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0039, SYCL, 0, 0, DWARF)
	HANDLE_DW_LANG(0x003d, Metal, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0040, Ruby, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0041, Move, 0, 0, DWARF)
	HANDLE_DW_LANG(0x0042, Hylo, 0, 0, DWARF)

Added somewhere between DWARFv5 and DWARFv6

[llvmbot]

@llvm/pr-subscribers-lldb

Author: Michael Buch (Michael137)

Changes

---

Full diff: https://github.com/llvm/llvm-project/pull/161810.diff

2 Files Affected:

• (modified) lldb/include/lldb/lldb-enumerations.h (+10)
• (modified) lldb/source/Target/Language.cpp (+10)

diff --git a/lldb/include/lldb/lldb-enumerations.h b/lldb/include/lldb/lldb-enumerations.h
index fec9fdef44df9..fe6b66fd66957 100644
--- a/lldb/include/lldb/lldb-enumerations.h
+++ b/lldb/include/lldb/lldb-enumerations.h
@@ -522,6 +522,16 @@ enum LanguageType {
   eLanguageTypeAssembly = 0x0031,
   eLanguageTypeC_sharp = 0x0032,
   eLanguageTypeMojo = 0x0033,
+  eLanguageTypeGLSL = 0x0034,
+  eLanguageTypeGLSL_ES = 0x0035,
+  eLanguageTypeHLSL = 0x0036,
+  eLanguageTypeOpenCL_CPP = 0x0037,
+  eLanguageTypeCppForOpenCL = 0x0038,
+  eLanguageTypeSycl = 0x0039,
+  eLanguageTypeMetal = 0x003d,
+  eLanguageTypeRuby = 0x0040,
+  eLanguageTypeMove = 0x0041,
+  eLanguageTypeHylo = 0x0042,
   eLanguageTypeLastStandardLanguage = eLanguageTypeMojo,
 
   // Vendor Extensions
diff --git a/lldb/source/Target/Language.cpp b/lldb/source/Target/Language.cpp
index 484d9badde397..e291b42d7264d 100644
--- a/lldb/source/Target/Language.cpp
+++ b/lldb/source/Target/Language.cpp
@@ -244,6 +244,16 @@ struct language_name_pair language_names[] = {
     {"assembly", eLanguageTypeAssembly},
     {"c-sharp", eLanguageTypeC_sharp},
     {"mojo", eLanguageTypeMojo},
+    {"GLSL", eLanguageTypeGLSL},
+    {"GLSL_ES", eLanguageTypeGLSL_ES},
+    {"HLSL", eLanguageTypeHLSL},
+    {"OpenCL_CPP", eLanguageTypeOpenCL_CPP},
+    {"CPP_for_OpenCL", eLanguageTypeCppForOpenCL},
+    {"SYCL", eLanguageTypeSycl},
+    {"Metal", eLanguageTypeMetal},
+    {"Ruby", eLanguageTypeRuby},
+    {"Move", eLanguageTypeMove},
+    {"Hylo", eLanguageTypeHylo},
     // Vendor Extensions
     {"assembler", eLanguageTypeMipsAssembler},
     // Now synonyms, in arbitrary order

[DavidSpickett]

DWARF codes from what exactly, DWARFv5, things already allocated to be in DWARFv6?

[Michael137]

DWARF codes from what exactly, DWARFv5, things already allocated to be in DWARFv6?

With the values in Dwarf.def
Most likely DWARFv6 values

I'll put it in the PR description

[DavidSpickett]

LGTM

[DavidSpickett]

CI failing with another thing that depends on the size of this enum.

[Michael137]

CI failing with another thing that depends on the size of this enum.

Ah looks like we've hit this limit here:

/// A 64-bit SmallBitVector is only small up to 64-7 bits, and the
/// setBitsInMask interface wants to write full bytes.
static const size_t g_num_small_bitvector_bits = 64 - 8;
static_assert(eNumLanguageTypes < g_num_small_bitvector_bits,
              "Languages bit vector is no longer small on 64 bit systems");
LanguageSet::LanguageSet() : bitvector(eNumLanguageTypes, false) {}

Not sure we should really care about this though? Seems like the LanguageSet is just less efficient now. But this would have inevitably happened right?

[DavidSpickett]

If there's a stack based type to hand use it, but otherwise yeah we were always gonna hit this limit eventually.

Also double check that what the comment says is still true. Maybe bitvector got better in the meantime?

[Michael137]

Talked to @adrian-prantl and we might want to just move away from the old-style unversioned language codes in favour of the new DWARFv6 language codes. In which case maybe we can do this differently. Closing for now