[BOLT] Fix long jump negative offset issue. (#67132)

In instruction encoding, the relative offset address of the PC is
signed, that is, the number of positive offset bits and the number of
negative offset bits is asymmetric. Therefore, the maximum and minimum
values are used to replace Mask to determine the boundary.

Co-authored-by: qijitao <[email protected]>

Author: qijitao

bolt/lib/Passes/LongJmp.cpp

@@ -138,18 +138,21 @@ BinaryBasicBlock *LongJmpPass::lookupStubFromGroup(
       Cand = LeftCand;
   }
   int BitsAvail = BC.MIB->getPCRelEncodingSize(Inst) - 1;
-  uint64_t Mask = ~((1ULL << BitsAvail) - 1);
+  assert(BitsAvail < 63 && "PCRelEncodingSize is too large to use int64_t to"
+                           "check for out-of-bounds.");
+  int64_t MaxVal = (1ULL << BitsAvail) - 1;
+  int64_t MinVal = -(1ULL << BitsAvail);
   uint64_t PCRelTgtAddress = Cand->first;
-  PCRelTgtAddress = DotAddress > PCRelTgtAddress ? DotAddress - PCRelTgtAddress
-                                                 : PCRelTgtAddress - DotAddress;
+  int64_t PCOffset = (int64_t)(PCRelTgtAddress - DotAddress);
+
   LLVM_DEBUG({
     if (Candidates.size() > 1)
       dbgs() << "Considering stub group with " << Candidates.size()
              << " candidates. DotAddress is " << Twine::utohexstr(DotAddress)
              << ", chosen candidate address is "
              << Twine::utohexstr(Cand->first) << "\n";
   });
-  return PCRelTgtAddress & Mask ? nullptr : Cand->second;
+  return (PCOffset < MinVal || PCOffset > MaxVal) ? nullptr : Cand->second;
 }
 
 BinaryBasicBlock *
@@ -512,13 +515,15 @@ bool LongJmpPass::needsStub(const BinaryBasicBlock &BB, const MCInst &Inst,
   }
 
   int BitsAvail = BC.MIB->getPCRelEncodingSize(Inst) - 1;
-  uint64_t Mask = ~((1ULL << BitsAvail) - 1);
+  assert(BitsAvail < 63 && "PCRelEncodingSize is too large to use int64_t to"
+                           "check for out-of-bounds.");
+  int64_t MaxVal = (1ULL << BitsAvail) - 1;
+  int64_t MinVal = -(1ULL << BitsAvail);
 
   uint64_t PCRelTgtAddress = getSymbolAddress(BC, TgtSym, TgtBB);
-  PCRelTgtAddress = DotAddress > PCRelTgtAddress ? DotAddress - PCRelTgtAddress
-                                                 : PCRelTgtAddress - DotAddress;
+  int64_t PCOffset = (int64_t)(PCRelTgtAddress - DotAddress);
 
-  return PCRelTgtAddress & Mask;
+  return PCOffset < MinVal || PCOffset > MaxVal;
 }
 
 bool LongJmpPass::relax(BinaryFunction &Func) {

bolt/test/AArch64/Inputs/long-jmp-offset-boundary.ld

@@ -0,0 +1,11 @@
+SECTIONS {
+    . = 0;
+    . = ALIGN(0x400000);
+    .text : {
+        *(foo_section)
+        . += 0x7BFFFFC;
+        *(main_section)
+        ASSERT(foo == 0x400000, "Error: foo address is not 0x400000.");
+        ASSERT(_start == 0x8000000, "Error: _start address is not 0x8000000.");
+    }
+}

bolt/test/AArch64/long-jmp-offset-boundary.s

@@ -0,0 +1,31 @@
+# This test checks long call negative offset boundary(0x8000000) for aarch64.
+
+# REQUIRES: system-linux
+
+# RUN: llvm-mc -filetype=obj -triple aarch64-unknown-unknown \
+# RUN:   %s -o %t.o
+# RUN: %clang %cflags %t.o -o %t.exe -nostartfiles -fuse-ld=lld -Wl,-q \
+# RUN:   -Wl,--script=%p/Inputs/long-jmp-offset-boundary.ld
+# RUN: llvm-bolt %t.exe -o %t.bolt.exe -skip-funcs="foo.*"
+# RUN: llvm-objdump -d -j .text --print-imm-hex %t.bolt.exe | FileCheck %s
+
+# The default alignment of the new program header table and the new text is
+# HugePageSize(2MB).
+# CHECK: [[#%x,ADDR:]]: [[#]]     	bl
+# CHECK-SAME: 	0x[[#ADDR-0x8000000]] <foo>
+
+	.text
+	.section	foo_section,"ax",@progbits
+	.globl	foo
+	.type	foo,@function
+foo:
+	ret
+	.size	foo, .-foo
+
+	.section	main_section,"ax",@progbits
+	.globl	_start
+	.type	_start,@function
+_start:
+	bl	foo
+	ret
+	.size	_start, .-_start