[lld][LoongArch] Relax TLS LE/GD/LD (#123600)

In local-exec form, the code sequence is converted as follows:
```
From:
  lu12i.w $rd, %le_hi20_r(sym)
    R_LARCH_TLS_LE_HI20_R, R_LARCH_RELAX
  add.w/d $rd, $rd, $tp, %le_add_r(sym)
    R_LARCH_TLS_LE_ADD_R, R_LARCH_RELAX
  addi/ld/st.w/d $rd, $rd, %le_lo12_r(sym)
    R_LARCH_TLS_LE_LO12_R, R_LARCH_RELAX
To:
  addi/ld/st.w/d $rd, $tp, %le_lo12_r(sym)
    R_LARCH_TLS_LE_LO12_R
``` 

In global-dynamic or local-dynamic, the code sequence is converted as
follows:
```
From:
  pcalau12i     $a0, %ld_pc_hi20(sym)  | %gd_pc_hi20(sym)
    R_LARCH_TLS_GD_PC_HI20 | R_LARCH_TLS_LD_PC_HI20, R_LARCH_RELAX
  addi.w/d $a0, $a0, %got_pc_lo12(sym) | %got_pc_lo12(sym)
    R_LARCH_GOT_PC_LO12, R_LARCH_RELAX
To:
  pcaddi        $a0, %got_pc_lo12(sym) | %got_pc_lo12(sym)
    R_LARCH_TLS_GD_PCREL20_S2 | R_LARCH_TLS_LD_PCREL20_S2
``` 
Note: For initial-exec form, since it involves the conversion from IE to
LE, we will implement it in a future patch.

Author: ylzsx

lld/ELF/Arch/LoongArch.cpp

@@ -154,6 +154,10 @@ static uint32_t setJ20(uint32_t insn, uint32_t imm) {
   return (insn & 0xfe00001f) | (extractBits(imm, 19, 0) << 5);
 }
 
+static uint32_t setJ5(uint32_t insn, uint32_t imm) {
+  return (insn & 0xfffffc1f) | (extractBits(imm, 4, 0) << 5);
+}
+
 static uint32_t setK12(uint32_t insn, uint32_t imm) {
   return (insn & 0xffc003ff) | (extractBits(imm, 11, 0) << 10);
 }
@@ -761,10 +765,10 @@ static bool isPairRelaxable(ArrayRef<Relocation> relocs, size_t i) {
 
 // Relax code sequence.
 // From:
-//   pcalau12i $a0, %pc_hi20(sym)
-//   addi.w/d $a0, $a0, %pc_lo12(sym)
+//   pcalau12i     $a0, %pc_hi20(sym) | %ld_pc_hi20(sym)  | %gd_pc_hi20(sym)
+//   addi.w/d $a0, $a0, %pc_lo12(sym) | %got_pc_lo12(sym) | %got_pc_lo12(sym)
 // To:
-//   pcaddi $a0, %pc_lo12(sym)
+//   pcaddi $a0, %pc_lo12(sym) | %got_pc_lo12(sym) | %got_pc_lo12(sym)
 //
 // From:
 //   pcalau12i $a0, %got_pc_hi20(sym_got)
@@ -778,6 +782,10 @@ static void relaxPCHi20Lo12(Ctx &ctx, const InputSection &sec, size_t i,
   if (!((rHi20.type == R_LARCH_PCALA_HI20 &&
          rLo12.type == R_LARCH_PCALA_LO12) ||
         (rHi20.type == R_LARCH_GOT_PC_HI20 &&
+         rLo12.type == R_LARCH_GOT_PC_LO12) ||
+        (rHi20.type == R_LARCH_TLS_GD_PC_HI20 &&
+         rLo12.type == R_LARCH_GOT_PC_LO12) ||
+        (rHi20.type == R_LARCH_TLS_LD_PC_HI20 &&
          rLo12.type == R_LARCH_GOT_PC_LO12)))
     return;
 
@@ -798,6 +806,8 @@ static void relaxPCHi20Lo12(Ctx &ctx, const InputSection &sec, size_t i,
   else if (rHi20.expr == RE_LOONGARCH_PAGE_PC ||
            rHi20.expr == RE_LOONGARCH_GOT_PAGE_PC)
     dest = rHi20.sym->getVA(ctx);
+  else if (rHi20.expr == RE_LOONGARCH_TLSGD_PAGE_PC)
+    dest = ctx.in.got->getGlobalDynAddr(*rHi20.sym);
   else {
     Err(ctx) << getErrorLoc(ctx, (const uint8_t *)loc) << "unknown expr ("
              << rHi20.expr << ") against symbol " << rHi20.sym
@@ -827,7 +837,12 @@ static void relaxPCHi20Lo12(Ctx &ctx, const InputSection &sec, size_t i,
     return;
 
   sec.relaxAux->relocTypes[i] = R_LARCH_RELAX;
-  sec.relaxAux->relocTypes[i + 2] = R_LARCH_PCREL20_S2;
+  if (rHi20.type == R_LARCH_TLS_GD_PC_HI20)
+    sec.relaxAux->relocTypes[i + 2] = R_LARCH_TLS_GD_PCREL20_S2;
+  else if (rHi20.type == R_LARCH_TLS_LD_PC_HI20)
+    sec.relaxAux->relocTypes[i + 2] = R_LARCH_TLS_LD_PCREL20_S2;
+  else
+    sec.relaxAux->relocTypes[i + 2] = R_LARCH_PCREL20_S2;
   sec.relaxAux->writes.push_back(insn(PCADDI, getD5(nextInsn), 0, 0));
   remove = 4;
 }
@@ -863,6 +878,33 @@ static void relaxCall36(Ctx &ctx, const InputSection &sec, size_t i,
   }
 }
 
+// Relax code sequence.
+// From:
+//   lu12i.w $rd, %le_hi20_r(sym)
+//   add.w/d $rd, $rd, $tp, %le_add_r(sym)
+//   addi/ld/st.w/d $rd, $rd, %le_lo12_r(sym)
+// To:
+//   addi/ld/st.w/d $rd, $tp, %le_lo12_r(sym)
+static void relaxTlsLe(Ctx &ctx, const InputSection &sec, size_t i,
+                       uint64_t loc, Relocation &r, uint32_t &remove) {
+  uint64_t val = r.sym->getVA(ctx, r.addend);
+  // Check if the val exceeds the range of addi/ld/st.
+  if (!isInt<12>(val))
+    return;
+  uint32_t currInsn = read32le(sec.content().data() + r.offset);
+  switch (r.type) {
+  case R_LARCH_TLS_LE_HI20_R:
+  case R_LARCH_TLS_LE_ADD_R:
+    sec.relaxAux->relocTypes[i] = R_LARCH_RELAX;
+    remove = 4;
+    break;
+  case R_LARCH_TLS_LE_LO12_R:
+    sec.relaxAux->writes.push_back(setJ5(currInsn, R_TP));
+    sec.relaxAux->relocTypes[i] = R_LARCH_TLS_LE_LO12_R;
+    break;
+  }
+}
+
 static bool relax(Ctx &ctx, InputSection &sec) {
   const uint64_t secAddr = sec.getVA();
   const MutableArrayRef<Relocation> relocs = sec.relocs();
@@ -903,6 +945,8 @@ static bool relax(Ctx &ctx, InputSection &sec) {
     }
     case R_LARCH_PCALA_HI20:
     case R_LARCH_GOT_PC_HI20:
+    case R_LARCH_TLS_GD_PC_HI20:
+    case R_LARCH_TLS_LD_PC_HI20:
       // The overflow check for i+2 will be carried out in isPairRelaxable.
       if (isPairRelaxable(relocs, i))
         relaxPCHi20Lo12(ctx, sec, i, loc, r, relocs[i + 2], remove);
@@ -911,6 +955,12 @@ static bool relax(Ctx &ctx, InputSection &sec) {
       if (relaxable(relocs, i))
         relaxCall36(ctx, sec, i, loc, r, remove);
       break;
+    case R_LARCH_TLS_LE_HI20_R:
+    case R_LARCH_TLS_LE_ADD_R:
+    case R_LARCH_TLS_LE_LO12_R:
+      if (relaxable(relocs, i))
+        relaxTlsLe(ctx, sec, i, loc, r, remove);
+      break;
     }
 
     // For all anchors whose offsets are <= r.offset, they are preceded by
@@ -1015,8 +1065,21 @@ void LoongArch::finalizeRelax(int passes) const {
             r.expr = r.sym->hasFlag(NEEDS_PLT) ? R_PLT_PC : R_PC;
             break;
           case R_LARCH_B26:
+          case R_LARCH_TLS_LE_LO12_R:
+            skip = 4;
+            write32le(p, aux.writes[writesIdx++]);
+            break;
+          case R_LARCH_TLS_GD_PCREL20_S2:
+            // Note: R_LARCH_TLS_LD_PCREL20_S2 must also use R_TLSGD_PC instead
+            // of R_TLSLD_PC due to historical reasons. In fact, right now TLSLD
+            // behaves exactly like TLSGD on LoongArch.
+            //
+            // This reason has also been mentioned in mold commit:
+            // https://github.com/rui314/mold/commit/5dfa1cf07c03bd57cb3d493b652ef22441bcd71c
+          case R_LARCH_TLS_LD_PCREL20_S2:
             skip = 4;
             write32le(p, aux.writes[writesIdx++]);
+            r.expr = R_TLSGD_PC;
             break;
           default:
             llvm_unreachable("unsupported type");

lld/test/ELF/loongarch-relax-emit-relocs.s

@@ -27,19 +27,45 @@
 # RELAX-NEXT:     R_LARCH_RELAX *ABS*
 # RELAX-NEXT:     R_LARCH_PCREL20_S2 _start
 # RELAX-NEXT:     R_LARCH_RELAX *ABS*
-# RELAX32-NEXT:  nop
-# RELAX32-NEXT:    R_LARCH_ALIGN *ABS*+0xc
-# RELAX32-NEXT:  nop
-# RELAX32-NEXT:  ret
 
 # RELAX64-NEXT:  bl  -8
 # RELAX64-NEXT:    R_LARCH_B26 _start
 # RELAX64-NEXT:    R_LARCH_RELAX *ABS*
 # RELAX64-NEXT:  b   -12
 # RELAX64-NEXT:    R_LARCH_B26 _start
 # RELAX64-NEXT:    R_LARCH_RELAX *ABS*
-# RELAX64-NEXT:  ret
+
+# RELAX-NEXT:   lu12i.w   $a0, 0
+# RELAX-NEXT:     R_LARCH_TLS_LE_HI20 a
+# RELAX-NEXT:   ori       $a0, $a0, 0
+# RELAX-NEXT:     R_LARCH_TLS_LE_LO12 a
+# RELAX-NEXT:   pcaddi    $a0, [[#]]
+# RELAX-NEXT:     R_LARCH_RELAX a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:     R_LARCH_TLS_GD_PCREL20_S2 a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:   pcaddi    $a0, [[#]]
+# RELAX-NEXT:     R_LARCH_RELAX a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:     R_LARCH_TLS_LD_PCREL20_S2 a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:   addi.{{[dw]}} $a0, $tp, 0
+# RELAX-NEXT:     R_LARCH_RELAX a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:     R_LARCH_RELAX a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+# RELAX-NEXT:     R_LARCH_TLS_LE_LO12_R a
+# RELAX-NEXT:     R_LARCH_RELAX *ABS*
+
+# RELAX32-NEXT:  nop
+# RELAX32-NEXT:    R_LARCH_ALIGN *ABS*+0xc
+# RELAX32-NEXT:  ret
+
+# RELAX64-NEXT:  nop
 # RELAX64-NEXT:    R_LARCH_ALIGN *ABS*+0xc
+# RELAX64-NEXT:  nop
+# RELAX64-NEXT:  nop
+# RELAX64-NEXT:  ret
 
 # NORELAX:      <_start>:
 # NORELAX-NEXT:   pcalau12i $a0, 0
@@ -62,8 +88,36 @@
 # NORELAX-NEXT:     R_LARCH_CALL36 _start
 # NORELAX-NEXT:     R_LARCH_RELAX *ABS*
 # NORELAX-NEXT:   jirl $zero, $a0, -24
-# NORELAX-NEXT:   ret
+# NORELAX-NEXT:   lu12i.w   $a0, 0
+# NORELAX-NEXT:     R_LARCH_TLS_LE_HI20 a
+# NORELAX-NEXT:   ori       $a0, $a0, 0
+# NORELAX-NEXT:     R_LARCH_TLS_LE_LO12 a
+# NORELAX-NEXT:   pcalau12i $a0, 16
+# NORELAX-NEXT:     R_LARCH_TLS_GD_PC_HI20 a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   addi.d    $a0, $a0, 8
+# NORELAX-NEXT:     R_LARCH_GOT_PC_LO12 a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   pcalau12i $a0, 16
+# NORELAX-NEXT:     R_LARCH_TLS_LD_PC_HI20 a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   addi.d    $a0, $a0, 8
+# NORELAX-NEXT:     R_LARCH_GOT_PC_LO12 a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   lu12i.w   $a0, 0
+# NORELAX-NEXT:     R_LARCH_TLS_LE_HI20_R a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   add.d     $a0, $a0, $tp
+# NORELAX-NEXT:     R_LARCH_TLS_LE_ADD_R a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   addi.d    $a0, $a0, 0
+# NORELAX-NEXT:     R_LARCH_TLS_LE_LO12_R a
+# NORELAX-NEXT:     R_LARCH_RELAX *ABS*
+# NORELAX-NEXT:   nop
 # NORELAX-NEXT:     R_LARCH_ALIGN *ABS*+0xc
+# NORELAX-NEXT:   nop
+# NORELAX-NEXT:   nop
+# NORELAX-NEXT:   ret
 
 # CHECKR:      <_start>:
 # CHECKR-NEXT:   pcalau12i $a0, 0
@@ -86,12 +140,53 @@
 # CHECKR-NEXT:     R_LARCH_CALL36 _start
 # CHECKR-NEXT:     R_LARCH_RELAX *ABS*
 # CHECKR-NEXT:   jr     $a0
+# CHECKR-NEXT:   lu12i.w   $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_LE_HI20 a
+# CHECKR-NEXT:   ori       $a0, $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_LE_LO12 a
+# CHECKR-NEXT:   pcalau12i $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_GD_PC_HI20 a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   addi.d    $a0, $a0, 0
+# CHECKR-NEXT:     R_LARCH_GOT_PC_LO12 a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   pcalau12i $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_LD_PC_HI20 a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   addi.d    $a0, $a0, 0
+# CHECKR-NEXT:     R_LARCH_GOT_PC_LO12 a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   lu12i.w   $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_LE_HI20_R a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   add.d     $a0, $a0, $tp
+# CHECKR-NEXT:     R_LARCH_TLS_LE_ADD_R a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
+# CHECKR-NEXT:   addi.d    $a0, $a0, 0
+# CHECKR-NEXT:     R_LARCH_TLS_LE_LO12_R a
+# CHECKR-NEXT:     R_LARCH_RELAX *ABS*
 # CHECKR-NEXT:   nop
 # CHECKR-NEXT:     R_LARCH_ALIGN *ABS*+0xc
 # CHECKR-NEXT:   nop
 # CHECKR-NEXT:   nop
 # CHECKR-NEXT:   ret
 
+.macro add dst, src1, src2, src3
+.ifdef ELF64
+  add.d \dst, \src1, \src2, \src3
+.else
+  add.w \dst, \src1, \src2, \src3
+.endif
+.endm
+
+.macro addi dst, src1, src2
+.ifdef ELF64
+  addi.d \dst, \src1, \src2
+.else
+  addi.w \dst, \src1, \src2
+.endif
+.endm
+
 .global _start
 _start:
   la.pcrel $a0, _start
@@ -101,5 +196,19 @@ _start:
   call36 _start
   tail36 $a0, _start
 .endif
+
+  la.tls.le $a0, a  # without R_LARCH_RELAX reloaction
+  la.tls.gd $a0, a
+  la.tls.ld $a0, a
+
+  lu12i.w $a0, %le_hi20_r(a)
+  add $a0, $a0, $tp, %le_add_r(a)
+  addi $a0, $a0, %le_lo12_r(a)
+
   .p2align 4
   ret
+
+.section .tbss,"awT",@nobits
+.globl a
+a:
+.zero 4