[RISCV] Allow parsing of bare symbols with offsets

[lewis-revill]

Summary

This patch allows symbols followed by an expression for an offset to be parsed as bare symbols.

[lewis-revill]

Should this only support constant offsets to a symbol? It seems sensible to only support offsets to the symbol, IE add/sub rather than more complex expressions, but the RHS could be any expression at the moment.

[asb]

Is this being used in the wild and does it work sensibly with binutils? I'm seeing that binutils seems to be associating the addend with both the emitted R_RISCV_PCREL_HI20 relocation and R_RISCV_RELAX relocation. Maybe this is just a binutils bug and the addend is ignored by the linker but this does seem surprising...

$ cat t.s 
lla a5, a_symbol + (0xFF << 3)
$ ./riscv32-unknown-elf-as t.s
$ ./riscv32-unknown-elf-readelf -r a.out 

Relocation section '.rela.text' at offset 0xec contains 4 entries:
 Offset     Info    Type            Sym.Value  Sym. Name + Addend
00000000  00000617 R_RISCV_PCREL_HI2 00000000   a_symbol + 7f8
00000000  00000033 R_RISCV_RELAX                7f8
00000004  00000418 R_RISCV_PCREL_LO1 00000000   .L0  + 0
00000004  00000033 R_RISCV_RELAX                0

[jrtc27]

In D57332#1549661, @asb wrote:

Is this being used in the wild and does it work sensibly with binutils? I'm seeing that binutils seems to be associating the addend with both the emitted R_RISCV_PCREL_HI20 relocation and R_RISCV_RELAX relocation. Maybe this is just a binutils bug and the addend is ignored by the linker but this does seem surprising...

$ cat t.s 
lla a5, a_symbol + (0xFF << 3)
$ ./riscv32-unknown-elf-as t.s
$ ./riscv32-unknown-elf-readelf -r a.out 

Relocation section '.rela.text' at offset 0xec contains 4 entries:
 Offset     Info    Type            Sym.Value  Sym. Name + Addend
00000000  00000617 R_RISCV_PCREL_HI2 00000000   a_symbol + 7f8
00000000  00000033 R_RISCV_RELAX                7f8
00000004  00000418 R_RISCV_PCREL_LO1 00000000   .L0  + 0
00000004  00000033 R_RISCV_RELAX                0

Yes, FreeRTOS at least uses it, and we've had to manually patch the source to refer to just the bare symbol and follow it with an addi in order to support LLVM. The RELAX addend is probably a bug but I don't think it actually ever gets read.

[asb]

This patch seems good, but needs rebasing. Thanks!

[lewis-revill]

Rebased

[lenary]

@asb are we happy for this to land? You were happy with it without explicitly approving it.

[asb]

Adding explicit LGTM in the Phabrictor UI. Thanks again Lewis!

[lenary]

@hans please can you backport rL369097 to the 9.0 branch?

[hans]

In D57332#1632995, @lenary wrote:

@hans please can you backport rL369097 to the 9.0 branch?

Merged in r369338.

lib/Target/RISCV/AsmParser/RISCVAsmParser.cpp

if (getLexer().getKind() != AsmToken::Identifier)
     return MatchOperand_NoMatch;
 
   StringRef Identifier;
   if (getParser().parseIdentifier(Identifier))
     return MatchOperand_ParseFail;
 
   MCSymbol *Sym = getContext().getOrCreateSymbol(Identifier);
   Res = MCSymbolRefExpr::create(Sym, MCSymbolRefExpr::VK_None, getContext());
+
+  MCBinaryExpr::Opcode Opcode;
+  switch (getLexer().getKind()) {
+  default:
+    Operands.push_back(RISCVOperand::createImm(Res, S, E, isRV64()));
+    return MatchOperand_Success;
+  case AsmToken::Plus:
+    Opcode = MCBinaryExpr::Add;
+    break;
+  case AsmToken::Minus:
+    Opcode = MCBinaryExpr::Sub;
+    break;
+  }
+
+  const MCExpr *Expr;
+  if (getParser().parseExpression(Expr))
+    return MatchOperand_ParseFail;
+  Res = MCBinaryExpr::create(Opcode, Res, Expr, getContext());
   Operands.push_back(RISCVOperand::createImm(Res, S, E, isRV64()));
   return MatchOperand_Success;
 }
 
 OperandMatchResultTy RISCVAsmParser::parseJALOffset(OperandVector &Operands) {
   // Parsing jal operands is fiddly due to the `jal foo` and `jal ra, foo`
   // both being acceptable forms. When parsing `jal ra, foo` this function
   // will be called for the `ra` register operand in an attempt to match the

test/MC/RISCV/rvi-pseudos.s

 # CHECK: auipc a3, %pcrel_hi(ra)
 # CHECK: addi  a3, a3, %pcrel_lo(.Lpcrel_hi3)
 lla a3, ra
 
 # CHECK: .Lpcrel_hi4:
 # CHECK: auipc a4, %pcrel_hi(f1)
 # CHECK: addi  a4, a4, %pcrel_lo(.Lpcrel_hi4)
 lla a4, f1
+
+# CHECK: .Lpcrel_hi5:
+# CHECK: auipc a5, %pcrel_hi(a_symbol+2040)
+# CHECK: addi  a5, a5, %pcrel_lo(.Lpcrel_hi5)
+lla a5, a_symbol + (0xFF << 3)