[PPC64] Add TLS global dynamic to local exec relaxation

[syzaara]

Summary

This patch adds the target call back relaxTlsGdToLe to support TLS relaxation from global dynamic to local exec model.

[sfertile]

lld/ELF/Arch/PPC64.cpp

171 ↗

(On Diff #150951)

minor nit: Could use a ternary here instead of an if.

lld/ELF/Relocations.cpp

936 ↗

(On Diff #150951)

This comment should be updated to reflect that we are only skipping the non-tls hints since the tls hints need the following processing.

lld/ELF/Relocations.h

47 ↗

(On Diff #150951)

I think we are better off to call it R_TLSGD_HINT and add it with the other General dynamic relocation below.

[ruiu]

lld/ELF/Arch/PPC64.cpp

162 ↗

(On Diff #151177)

I found this comment a bit confusing. Does this mean the first instruction is one of

addis r3, r2, a@got@tlsgd@ha [R_PPC64_GOT_TLSGD16_HA]

addi r3, r3, a@got@tlsgd@l [R_PPC64_GOT_TLSGD16_LO]

bl __tls_get_addr(a@tlsgd) [R_PPC64_TLSGD]

followed by a nop? Or, do we expect the following four instructions?

addis r3, r2, a@got@tlsgd@ha [R_PPC64_GOT_TLSGD16_HA]
addi r3, r3, a@got@tlsgd@l [R_PPC64_GOT_TLSGD16_LO]
bl __tls_get_addr(a@tlsgd) [R_PPC64_TLSGD]
nop

[ruiu]

LGTM

lld/ELF/Arch/PPC64.cpp

187–188 ↗

(On Diff #152941)

I believe clang-format would vertically align these comments.

lld/trunk/ELF/Arch/PPC64.cpp

void writePlt(uint8_t *Buf, uint64_t GotPltEntryAddr, uint64_t PltEntryAddr,
                 int32_t Index, unsigned RelOff) const override;
   void relocateOne(uint8_t *Loc, RelType Type, uint64_t Val) const override;
   void writeGotHeader(uint8_t *Buf) const override;
   bool needsThunk(RelExpr Expr, RelType Type, const InputFile *File,
                   uint64_t BranchAddr, const Symbol &S) const override;
   RelExpr adjustRelaxExpr(RelType Type, const uint8_t *Data,
                           RelExpr Expr) const override;
   void relaxTlsGdToIe(uint8_t *Loc, RelType Type, uint64_t Val) const override;
+  void relaxTlsGdToLe(uint8_t *Loc, RelType Type, uint64_t Val) const override;
 };
 } // namespace
 
 // Relocation masks following the #lo(value), #hi(value), #ha(value),
 // #higher(value), #highera(value), #highest(value), and #highesta(value)
 // macros defined in section 4.5.1. Relocation Types of the PPC-elf64abi
 // document.
 static uint16_t lo(uint64_t V) { return V; }
uint32_t PPC64::calcEFlags() const {
   if (NonZeroFlag == 1) {
     error("PPC64 V1 ABI not supported");
     return 0;
   }
 
   return 2;
 }
 
+void PPC64::relaxTlsGdToLe(uint8_t *Loc, RelType Type, uint64_t Val) const {
+  // Reference: 3.7.4.2 of the 64-bit ELF V2 abi supplement.
+  // The general dynamic code sequence for a global `x` will look like:
+  // Instruction                    Relocation                Symbol
+  // addis r3, r2, x@got@tlsgd@ha   R_PPC64_GOT_TLSGD16_HA      x
+  // addi  r3, r3, x@got@tlsgd@l    R_PPC64_GOT_TLSGD16_LO      x
+  // bl __tls_get_addr(x@tlsgd)     R_PPC64_TLSGD               x
+  //                                R_PPC64_REL24               __tls_get_addr
+  // nop                            None                       None
+
+  // Relaxing to local exec entails converting:
+  // addis r3, r2, x@got@tlsgd@ha    into      nop
+  // addi  r3, r3, x@got@tlsgd@l     into      addis r3, r13, x@tprel@ha
+  // bl __tls_get_addr(x@tlsgd)      into      nop
+  // nop                             into      addi r3, r3, x@tprel@l
+
+  uint32_t EndianOffset = Config->EKind == ELF64BEKind ? 2U : 0U;
+
+  switch (Type) {
+  case R_PPC64_GOT_TLSGD16_HA:
+    write32(Loc - EndianOffset, 0x60000000); // nop
+    break;
+  case R_PPC64_GOT_TLSGD16_LO:
+    write32(Loc - EndianOffset, 0x3c6d0000); // addis r3, r13
+    relocateOne(Loc, R_PPC64_TPREL16_HA, Val);
+    break;
+  case R_PPC64_TLSGD:
+    write32(Loc, 0x60000000);     // nop
+    write32(Loc + 4, 0x38630000); // addi r3, r3
+    relocateOne(Loc + 4 + EndianOffset, R_PPC64_TPREL16_LO, Val);
+    break;
+  default:
+    llvm_unreachable("unsupported relocation for TLS GD to LE relaxation");
+  }
+}
+
 RelExpr PPC64::getRelExpr(RelType Type, const Symbol &S,
                           const uint8_t *Loc) const {
   switch (Type) {
   case R_PPC64_TOC16:
   case R_PPC64_TOC16_DS:
   case R_PPC64_TOC16_HA:
   case R_PPC64_TOC16_HI:
   case R_PPC64_TOC16_LO:

lld/trunk/test/ELF/ppc64-tls-gd-le.s

+// REQUIRES: ppc
+
+// RUN: llvm-mc -filetype=obj -triple=powerpc64le-unknown-linux %s -o %t.o
+// RUN: llvm-readelf -relocations --wide %t.o | FileCheck --check-prefix=InputRelocs %s
+// RUN: ld.lld  %t.o -o %t
+// RUN: llvm-objdump -D %t | FileCheck --check-prefix=Dis %s
+// RUN: llvm-readelf -relocations --wide %t | FileCheck --check-prefix=OutputRelocs %s
+
+// RUN: llvm-mc -filetype=obj -triple=powerpc64-unknown-linux %s -o %t.o
+// RUN: llvm-readelf -relocations --wide %t.o | FileCheck --check-prefix=InputRelocs %s
+// RUN: ld.lld  %t.o -o %t
+// RUN: llvm-objdump -D %t | FileCheck --check-prefix=Dis %s
+// RUN: llvm-readelf -relocations --wide %t | FileCheck --check-prefix=OutputRelocs %s
+
+  .text
+  .abiversion 2
+  .globl  _start                    # -- Begin function _start
+  .p2align  4
+  .type _start,@function
+_start:                                   # @_start
+.Lfunc_begin0:
+.Lfunc_gep0:
+  addis 2, 12, .TOC.-.Lfunc_gep0@ha
+  addi 2, 2, .TOC.-.Lfunc_gep0@l
+.Lfunc_lep0:
+  .localentry _start, .Lfunc_lep0-.Lfunc_gep0
+# %bb.0:                                # %entry
+  mflr 0
+  std 31, -8(1)
+  std 0, 16(1)
+  stdu 1, -64(1)
+  mr 31, 1
+  std 30, 48(31)                  # 8-byte Folded Spill
+  li 3, 0
+  stw 3, 44(31)
+  addis 3, 2, a@got@tlsgd@ha
+  addi 3, 3, a@got@tlsgd@l
+  bl __tls_get_addr(a@tlsgd)
+  nop
+  lwz 30, 0(3)
+  extsw 3, 30
+  ld 30, 48(31)                   # 8-byte Folded Reload
+  addi 1, 1, 64
+  ld 0, 16(1)
+  ld 31, -8(1)
+  mtlr 0
+  blr
+  .long 0
+  .quad 0
+.Lfunc_end0:
+  .size _start, .Lfunc_end0-.Lfunc_begin0
+
+.globl __tls_get_addr
+.type __tls_get_addr,@function
+__tls_get_addr:
+
+                                        # -- End function
+  .type a,@object               # @a
+  .section  .tdata,"awT",@progbits
+  .globl  a
+  .p2align  2
+a:
+  .long 55                      # 0x37
+  .size a, 4
+
+// Verify that the input has general-dynamic tls relocation types
+// InputRelocs:  Relocation section '.rela.text'
+// InputRelocs: R_PPC64_GOT_TLSGD16_HA  {{0+}}  a + 0
+// InputRelocs: R_PPC64_GOT_TLSGD16_LO  {{0+}}  a + 0
+// InputRelocs: R_PPC64_TLSGD           {{0+}}  a + 0
+
+// Verify that the general-dynamic sequence is  relaxed to local exec.
+// #ha(a@tprel) --> (0 - 0x7000 + 0x8000) >> 16 = 0
+// #lo(a@tprel)) --> (0 - 0x7000) &  0xFFFF =  -0x7000 = -28672
+// Dis: _start:
+// Dis: nop
+// Dis: addis 3, 13, 0
+// Dis: nop
+// Dis: addi 3, 3, -28672
+
+// Verify that no general-dynamic relocations exist for the dynamic linker.
+// OutputRelocs-NOT: R_PPC64_DTPMOD64
+// OutputRelocs-NOT: R_PPC64_DTPREL64