[Driver] Default riscv*- triples to -fdebug-default-version=4

[MaskRay]

Summary

This adds a RISC-V special case to ToolChain::GetDefaultDwarfVersion,
affecting Linux/Haiku/RISCVToolChain.

DWARF v5 .debug_loclists/.debug_rnglists's
DW_LLE_offset_pair/DW_RLE_offset_pair entry kinds utilitize .uleb128 A-B
directives where A and B reference local labels in code sections.
When A and B are separated by a RISC-V linker-relaxable instruction,
A-B is incorrectly folded without a relocation, causing incorrect debug
information.

void ext(void);
int foo(int x) {ext(); return 0;}
// DW_AT_location [DW_FORM_loclistx] of a DW_TAG_formal_parameter references a DW_LLE_offset_pair that can be incorrect after linker relaxation.

int ext(void);
void foo() { {
  int ret = ext();
  if (__builtin_expect(ret, 0))
    ext();
} }
// DW_AT_ranges [DW_FORM_rnglistx] of a DW_TAG_lexical_block references a DW_RLE_offset_pair that can be incorrect after linker relaxation.

D157657 will implement R_RISCV_SET_ULEB128/R_RISCV_SUB_ULEB128
relocations, fixing the issue, but the relocation is only supported by
bleeding-edge binutils 2.41 and not by lld/ELF yet.

The goal is to make the emitted DWARF correct after linking.
Many users don't care about the default DWARF version, but a linker
error will be unacceptable. Let's just downgrade the default DWARF
version, before binutils>=2.41 is more widely available.

An alternative compatibility option is to add a toggle to DwarfDebug.cpp,
but that doesn't seem like a good idea.

This patch is planned for the main and release/17.x branches.

[MaskRay]

add a comment

[jrtc27]

RISCVToolchain also needs an override

[jrtc27]

Also Haiku I guess, as another OS that has a somewhat working RISC-V port but falls back on ToolChain's default implementation. So maybe this just belongs in the default implementation, given all the ones that need modifying are the ones that end up there.

[MaskRay]

Thanks for mentioning RISCVToolChain/Haiku.

Update ToolChain::

[kito-cheng]

LGTM, I am OK with that, actually our downstream toolchain has downgrade the dwarf version to 4 too, because binutils/gdb not well support before...

[asb]

LGTM, this seems like a good workaround.

clang/include/clang/Driver/ToolChain.h

public:
   /// compile unit information.
   virtual bool UseDwarfDebugFlags() const { return false; }
 
   /// Add an additional -fdebug-prefix-map entry.
   virtual std::string GetGlobalDebugPathRemapping() const { return {}; }
 
   // Return the DWARF version to emit, in the absence of arguments
   // to the contrary.
-  virtual unsigned GetDefaultDwarfVersion() const { return 5; }
+  virtual unsigned GetDefaultDwarfVersion() const;
 
   // Some toolchains may have different restrictions on the DWARF version and
   // may need to adjust it. E.g. NVPTX may need to enforce DWARF2 even when host
   // compilation uses DWARF5.
   virtual unsigned getMaxDwarfVersion() const { return UINT_MAX; }
 
   // True if the driver should assume "-fstandalone-debug"
   // in the absence of an option specifying otherwise,

clang/lib/Driver/ToolChain.cpp

std::string ToolChain::getInputFilename(const InputInfo &Input) const {
   return Input.getFilename();
 }
 
 ToolChain::UnwindTableLevel
 ToolChain::getDefaultUnwindTableLevel(const ArgList &Args) const {
   return UnwindTableLevel::None;
 }
 
+unsigned ToolChain::GetDefaultDwarfVersion() const {
+  // TODO: Remove the RISC-V special case when R_RISCV_SET_ULEB128 linker
+  // support becomes more widely available.
+  return getTriple().isRISCV() ? 4 : 5;
+}
+
 Tool *ToolChain::getClang() const {
   if (!Clang)
     Clang.reset(new tools::Clang(*this, useIntegratedBackend()));
   return Clang.get();
 }
 
 Tool *ToolChain::getFlang() const {
   if (!Flang)

clang/test/Driver/clang-g-opts.c

 
 // CHECK-WITHOUT-G-NOT: -debug-info-kind
 // CHECK-WITH-G: "-debug-info-kind=constructor"
 // CHECK-WITH-G: "-dwarf-version=5"
 // CHECK-WITH-G-DWARF2: "-dwarf-version=2"
 
 // CHECK-WITH-G-STANDALONE: "-debug-info-kind=standalone"
 // CHECK-WITH-G-STANDALONE: "-dwarf-version=2"
+
+/// TODO: Special case before R_RISCV_SET_ULEB128 linker support becomes more widely available.
+// RUN: %clang -### -S %s -g --target=riscv64-linux-gnu 2>&1 | FileCheck --check-prefix=VERSION4 %s
+// RUN: %clang -### -S %s -g --target=riscv64-unknown-elf 2>&1 | FileCheck --check-prefix=VERSION4 %s
+// VERSION4: "-dwarf-version=4"