[ELF][AArch64] Implement the AArch64 TLSLD_LDST_LO12 family of relocs

[peter.smith]

Summary

Implement the following relocations for AArch64:
R_AARCH64_TLSLE_LDST8_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST16_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST32_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST64_TPREL_LO12_NC
R_AARCH64_TLSLE_LDST128_TPREL_LO12_NC

These are specified in ELF for the 64-bit Arm Architecture (http://infocenter.arm.com/help/topic/com.arm.doc.ihi0056b/IHI0056B_aaelf64.pdf)

These relocation directives were first used in an optimisation in https://reviews.llvm.org/D44355 . This had to be reverted as no linker (gold, ld.bfd or lld) implemented these relocations. Both gold and ld.bfd have since implemented support (https://sourceware.org/bugzilla/show_bug.cgi?id=22969 and https://sourceware.org/bugzilla/show_bug.cgi?id=22970 ) so it would be good to have LLD support as well.

Fixes pr36727

[peter.smith]

Updated diff to remove relocations from usesOnlyLowPageBits(). While this is true for these relocations the usesOnlyLowPageBits() function is never called for local exec TLS so by the same logic as applied to D46249 (don't add dead code) I've removed them from the patch.

[ruiu]

LGTM

ELF/Arch/AArch64.cpp

case R_AARCH64_TLSDESC_ADR_PAGE21:
     return R_TLSDESC_PAGE;
   case R_AARCH64_TLSDESC_LD64_LO12:
   case R_AARCH64_TLSDESC_ADD_LO12:
     return R_TLSDESC;
   case R_AARCH64_TLSDESC_CALL:
     return R_TLSDESC_CALL;
   case R_AARCH64_TLSLE_ADD_TPREL_HI12:
   case R_AARCH64_TLSLE_ADD_TPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST8_TPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST16_TPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST32_TPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST64_TPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST128_TPREL_LO12_NC:
     return R_TLS;
   case R_AARCH64_CALL26:
   case R_AARCH64_CONDBR19:
   case R_AARCH64_JUMP26:
   case R_AARCH64_TSTBR14:
     return R_PLT_PC;
   case R_AARCH64_PREL16:
   case R_AARCH64_PREL32:
case R_AARCH64_CALL26:
     break;
   case R_AARCH64_CONDBR19:
   case R_AARCH64_LD_PREL_LO19:
     checkAlignment(Loc, Val, 4, Type);
     checkInt(Loc, Val, 21, Type);
     or32le(Loc, (Val & 0x1FFFFC) << 3);
     break;
   case R_AARCH64_LDST8_ABS_LO12_NC:
+  case R_AARCH64_TLSLE_LDST8_TPREL_LO12_NC:
     or32AArch64Imm(Loc, getBits(Val, 0, 11));
     break;
   case R_AARCH64_LDST16_ABS_LO12_NC:
+  case R_AARCH64_TLSLE_LDST16_TPREL_LO12_NC:
     checkAlignment(Loc, Val, 2, Type);
     or32AArch64Imm(Loc, getBits(Val, 1, 11));
     break;
   case R_AARCH64_LDST32_ABS_LO12_NC:
+  case R_AARCH64_TLSLE_LDST32_TPREL_LO12_NC:
     checkAlignment(Loc, Val, 4, Type);
     or32AArch64Imm(Loc, getBits(Val, 2, 11));
     break;
   case R_AARCH64_LDST64_ABS_LO12_NC:
   case R_AARCH64_LD64_GOT_LO12_NC:
   case R_AARCH64_TLSIE_LD64_GOTTPREL_LO12_NC:
+  case R_AARCH64_TLSLE_LDST64_TPREL_LO12_NC:
   case R_AARCH64_TLSDESC_LD64_LO12:
     checkAlignment(Loc, Val, 8, Type);
     or32AArch64Imm(Loc, getBits(Val, 3, 11));
     break;
   case R_AARCH64_LDST128_ABS_LO12_NC:
+  case R_AARCH64_TLSLE_LDST128_TPREL_LO12_NC:
     checkAlignment(Loc, Val, 16, Type);
     or32AArch64Imm(Loc, getBits(Val, 4, 11));
     break;
   case R_AARCH64_MOVW_UABS_G0_NC:
     or32le(Loc, (Val & 0xFFFF) << 5);
     break;
   case R_AARCH64_MOVW_UABS_G1_NC:
     or32le(Loc, (Val & 0xFFFF0000) >> 11);

test/ELF/aarch64-tlsld-ldst.s

+// RUN: llvm-mc -triple=aarch64-linux-gnu -filetype=obj %s -o %t.o
+// RUN: ld.lld %t.o -o %t
+// RUN: llvm-objdump -d %t | FileCheck %s
+// RUN: llvm-readelf --symbols %t | FileCheck -check-prefix CHECK-SYMS %s
+        .text
+        .globl _start
+        .type _start, %function
+_start:  mrs x8, TPIDR_EL0
+
+        add x8, x8, :tprel_hi12:var0
+        ldr q20, [x8, :tprel_lo12_nc:var0]
+
+        add x8, x8, :tprel_hi12:var1
+        ldr x0, [x8, :tprel_lo12_nc:var1]
+
+        add x8, x8, :tprel_hi12:var2
+        ldr w0, [x8, :tprel_lo12_nc:var2]
+
+        add x8, x8, :tprel_hi12:var3
+        ldrh w0, [x8, :tprel_lo12_nc:var3]
+
+        add x8, x8, :tprel_hi12:var4
+        ldrb w0, [x8, :tprel_lo12_nc:var4]
+
+// CHECK: _start:
+// CHECK-NEXT:    20000:       48 d0 3b d5     mrs     x8, TPIDR_EL0
+// 0x0 + c10 = 0xc10       = tcb (16-bytes) + var0
+// CHECK-NEXT:    20004:       08 01 40 91     add     x8, x8, #0, lsl #12
+// CHECK-NEXT:    20008:       14 05 c3 3d     ldr     q20, [x8, #3088]
+// 0x1000 + 0x820 = 0x1820 = tcb + var1
+// CHECK-NEXT:    2000c:       08 05 40 91     add     x8, x8, #1, lsl #12
+// CHECK-NEXT:    20010:       00 11 44 f9     ldr     x0, [x8, #2080]
+// 0x2000 + 0x428 = 0x2428 = tcb + var2
+// CHECK-NEXT:    20014:       08 09 40 91     add     x8, x8, #2, lsl #12
+// CHECK-NEXT:    20018:       00 29 44 b9     ldr     w0, [x8, #1064]
+// 0x3000 + 0x2c  = 0x302c = tcb + var3
+// CHECK-NEXT:    2001c:       08 0d 40 91     add     x8, x8, #3, lsl #12
+// CHECK-NEXT:    20020:       00 59 40 79     ldrh    w0, [x8, #44]
+// 0x3000 + 0xc2e = 0x32ce = tcb + var4
+// CHECK-NEXT:    20024:       08 0d 40 91     add     x8, x8, #3, lsl #12
+// CHECK-NEXT:    20028:       00 b9 70 39     ldrb    w0, [x8, #3118]
+
+// CHECK-SYMS:      0000000000000c00     0 TLS     GLOBAL DEFAULT    2 var0
+// CHECK-SYMS-NEXT: 0000000000001810     4 TLS     GLOBAL DEFAULT    2 var1
+// CHECK-SYMS-NEXT: 0000000000002418     2 TLS     GLOBAL DEFAULT    2 var2
+// CHECK-SYMS-NEXT: 000000000000301c     1 TLS     GLOBAL DEFAULT    2 var3
+// CHECK-SYMS-NEXT: 0000000000003c1e     0 TLS     GLOBAL DEFAULT    2 var4
+
+        .globl var0
+        .globl var1
+        .globl var2
+        .globl var3
+        .globl var4
+        .type var0,@object
+        .type var1,@object
+        .type var2,@object
+        .type var3,@object
+
+.section .tbss,"awT",@nobits
+        .balign 16
+        .space 1024 * 3
+var0:
+        .quad 0
+        .quad 0
+        .size var1, 16
+        .space 1024 * 3
+var1:
+        .quad 0
+        .size var1, 8
+        .space 1024 * 3
+var2:
+        .word 0
+        .size var1, 4
+
+        .space 1024 * 3
+var3:
+        .hword 0
+        .size var2, 2
+        .space 1024 * 3
+var4:
+        .byte 0
+        .size var3, 1
+        .space 1024 * 3