[BOLT] Support computed goto and allow map addrs inside functions

bolt/lib/Rewrite/RewriteInstance.cpp

@@ -2439,6 +2439,13 @@ void RewriteInstance::readDynamicRelocations(const SectionRef &Section,
     if (Symbol)
       SymbolIndex[Symbol] = getRelocationSymbol(InputFile, Rel);
 
+    const uint64_t SymAddress = SymbolAddress + Addend;
+    BinaryFunction *Func = BC->getBinaryFunctionContainingAddress(SymAddress);
+    if (Func && !Func->isInConstantIsland(SymAddress)) {
+      if (const uint64_t SymOffset = SymAddress - Func->getAddress())
+        Func->addEntryPointAtOffset(SymOffset);
+    }
+
     BC->addDynamicRelocation(Rel.getOffset(), Symbol, RType, Addend);
   }
 }
@@ -5599,7 +5606,7 @@ uint64_t RewriteInstance::getNewFunctionOrDataAddress(uint64_t OldAddress) {
         for (const BinaryBasicBlock &BB : *BF)
           if (BB.isEntryPoint() &&
               (BF->getAddress() + BB.getOffset()) == OldAddress)
-            return BF->getOutputAddress() + BB.getOffset();
+            return BB.getOutputStartAddress();
       }
       BC->errs() << "BOLT-ERROR: unable to get new address corresponding to "
                     "input address 0x"

bolt/test/AArch64/computed-goto.s

@@ -0,0 +1,67 @@
+// This test checks that BOLT creates entry points for addresses
+// referenced by dynamic relocations.
+// The test also checks that BOLT can map addresses inside functions.
+
+// Checks for error and entry points.
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown %s -o %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -Wl,-q
+# RUN: llvm-bolt %t.exe -o %t.bolt 2>&1 | FileCheck %s
+# RUN: llvm-bolt %t.exe -o %t.bolt --print-cfg | FileCheck --check-prefix=CHECK-ENTRYS %s
+
+// Checks for dynamic relocations.
+# RUN: llvm-readelf -dr %t.bolt > %t.out.txt
+# RUN: llvm-objdump -j .rela.dyn -d %t.bolt >> %t.out.txt
+# RUN: FileCheck --check-prefix=CHECK-RELOCS %s --input-file=%t.out.txt
+
+// Before bolt could handle mapping addresses within moved functions, it
+// would bail out with an error of the form:
+// BOLT-ERROR: unable to get new address corresponding to input address 0x10390 in function main. Consider adding this function to --skip-funcs=...
+// These addresses arise if computed GOTO is in use.
+// Check that bolt does not emit any error.
+# CHECK-NOT: BOLT-ERROR
+
+// Check that there are dynamic relocations.
+# CHECK-RELOCS:     Dynamic section at offset {{.*}} contains {{.*}} entries:
+# CHECK-RELOCS:     Relocation section '.rela.dyn' at offset {{.*}} contains {{.*}} entries
+
+// Check that dynamic relocations were updated
+# CHECK-RELOCS: [[#%x,OFF:]] [[#%x,INFO_DYN:]] R_AARCH64_RELATIVE [[#%x,ADDR:]]
+# CHECK-RELOCS-NEXT: [[#OFF + 8]] {{0*}}[[#INFO_DYN]] R_AARCH64_RELATIVE [[#ADDR + 8]]
+# CHECK-RELOCS: [[#ADDR]] <unknown>
+# CHECK-RELOCS: [[#ADDR + 8]] <unknown>
+
+// Check that BOLT registers extra entry points for dynamic relocations.
+# CHECK-ENTRYS: Binary Function "main" after building cfg {
+# CHECK-ENTRYS:  IsMultiEntry: 1
+# CHECK-ENTRYS: .Ltmp0 {{.*}}
+# CHECK-ENTRYS-NEXT: Secondary Entry Point: {{.*}}
+# CHECK-ENTRYS: .Ltmp1 {{.*}}
+# CHECK-ENTRYS-NEXT: Secondary Entry Point: {{.*}}
+
+.globl  main
+.p2align        2
+.type   main,@function
+main:
+.cfi_startproc
+        adrp    x8, .L__const.main.ptrs+8
+        add     x8, x8, :lo12:.L__const.main.ptrs+8
+        ldr     x9, [x8], #8
+        br      x9
+
+.Label0: // Block address taken
+        ldr     x9, [x8], #8
+        br      x9
+
+.Label1: // Block address taken
+        mov     w0, #42
+        ret
+
+.Lfunc_end0:
+.size   main, .Lfunc_end0-main
+.cfi_endproc
+        .type   .L__const.main.ptrs,@object
+        .section        .data.rel.ro,"aw",@progbits
+        .p2align        3, 0x0
+.L__const.main.ptrs:
+        .xword  .Label0
+        .xword  .Label1

bolt/test/X86/Inputs/indirect_goto.c → bolt/test/Inputs/indirect_goto.c

@@ -1,6 +1,6 @@
 int main(int argc, char *argv[]) {
-  static const void *T1[] = { &&L1, &&L2 };
-  static const void *T2[] = { &&L2, &&L3 };
+  static const void *T1[] = {&&L1, &&L2};
+  static const void *T2[] = {&&L2, &&L3};
 
   const void **T = (argc > 1) ? T1 : T2;
 

bolt/test/X86/indirect-goto.test

@@ -1,5 +1,5 @@
 ## Check llvm-bolt processes binaries compiled from sources that use indirect goto.
-RUN: %clang %cflags -no-pie %S/Inputs/indirect_goto.c -Wl,-q -o %t
+RUN: %clang %cflags -no-pie %S/../Inputs/indirect_goto.c -Wl,-q -o %t
 RUN: llvm-bolt %t -o %t.null --relocs=1 --print-cfg --print-only=main \
 RUN:   --strict \
 RUN:   2>&1 | FileCheck %s

bolt/test/dynamic-relocs.test

@@ -0,0 +1,13 @@
+// This test checks that BOLT creates entry points for addresses
+// referenced by dynamic relocations.
+
+RUN: %clang %cflags %S/Inputs/indirect_goto.c -o %t.exe -Wl,-q
+RUN: llvm-bolt %t.exe -o %t.bolt --print-cfg | FileCheck %s
+
+// Check that bolt does not emit any error.
+CHECK-NOT: BOLT-ERROR
+
+// Check that BOLT registers extra entry points for dynamic relocations.
+CHECK: Binary Function "main" after building cfg {
+CHECK: IsMultiEntry: 1
+CHECK: Secondary Entry Points : {{.*}}