[llvm-objdump] Support nonzero section addresses in addSymbolizer

[Pierre-vh]

Summary

The previous calculations seem to have assumed that the section address would be zero.
This is true for relocatable object files, but certainly not for linked files like shared libraries.

Fixed the calculations to make them identical to the "real" getInstruction call below & added a regression test.

[scott.linder]

LGTM, thank you! I wasn't aware of the change which caused the regression in https://reviews.llvm.org/D130357

I left a comment to confirm there isn't something more subtle happening, but regardless the code is clearly incorrect.

[simon_tatham]

No, I think this was exactly the mistake on my part that it looks like. Good catch, and thanks for fixing it!

llvm/test/tools/llvm-objdump/AMDGPU/elf-disassemble-symbolize-operands.yaml

-# RUN: yaml2obj %s -o %t
-# RUN: llvm-objdump %t -d --symbolize-operands --no-show-raw-insn --no-leading-addr | \
-# RUN:   FileCheck %s
+# RUN: rm -rf %t
+# RUN: %split-file %s %t
+
+# RUN: yaml2obj %t/zero-secaddr.yml -o %t/zero-secaddr
+# RUN: llvm-objdump %t/zero-secaddr -d --symbolize-operands\
+# RUN:  --no-show-raw-insn --no-leading-addr | FileCheck %s
+
+# RUN: yaml2obj %t/nonzero-secaddr.yml -o %t/nonzero-secaddr
+# RUN: llvm-objdump %t/nonzero-secaddr -d --symbolize-operands\
+# RUN:  --no-show-raw-insn --no-leading-addr | FileCheck %s
 
 ## Expect to find the branch labels.
 # CHECK: <break_cond_is_arg>:
 # CHECK:     s_branch L1
 # CHECK: <L0>:
 # CHECK:     s_cbranch_execz L2
 # CHECK: <L1>:
 # CHECK:     s_branch L0

 #   then:                                             ; preds = %bb
 #     call void @llvm.amdgcn.raw.buffer.store.f32(float undef, <4 x i32> undef, i32 0, i32 undef, i32 0)
 #     br label %endif
 #   endif:                                             ; preds = %bb28, %bb
 #     br i1 %breakcond, label %loop, label %loopexit
 #   loopexit:
 #     ret void
 #   }
-#   
+#
 #   declare void @llvm.amdgcn.raw.buffer.store.f32(float, <4 x i32>, i32, i32, i32 immarg) #0
-#   
+#
 #   attributes #0 = { nounwind writeonly }
 #
 # I compiled it to a relocatable ELF:
 #
 #   llc -march=amdgcn -mcpu=gfx1030 llvm/a.ll -filetype=obj -o a.elf
 #
 # then converted it to YAML:
 #
 #   obj2yaml a.elf
 #
 # then manually removed the BB0_1 etc local symbols.
+#
+# Note that there are two copies of the file:
+#   - One as a relocatable object file (zero section addresses)
+#   - One as a shared object file (non-zero section addresses)
 
+#--- zero-secaddr.yml
 --- !ELF
 FileHeader:
   Class:           ELFCLASS64
   Data:            ELFDATA2LSB
   Type:            ET_REL
   Machine:         EM_AMDGPU
   Flags:           [ EF_AMDGPU_MACH_AMDGCN_GFX1030 ]
 Sections:
Sections:
       - Name:            .symtab
 Symbols:
   - Name:            break_cond_is_arg
     Type:            STT_FUNC
     Section:         .text
     Binding:         STB_GLOBAL
     Size:            0x5C
 ...
+
+#--- nonzero-secaddr.yml
+--- !ELF
+FileHeader:
+  Class:           ELFCLASS64
+  Data:            ELFDATA2LSB
+  Type:            ET_DYN
+  Machine:         EM_AMDGPU
+  Flags:           [ EF_AMDGPU_MACH_AMDGCN_GFX1030 ]
+Sections:
+  - Name:            .text
+    Type:            SHT_PROGBITS
+    Flags:           [ SHF_ALLOC, SHF_EXECINSTR ]
+    Address:         0x1000
+    AddressAlign:    0x4
+    Content:         00008CBF0000FDBB81020236810385BE800384BE8102847D6AC10689040082BF7E077E88058105817E047E8A080088BF0500887D7E060787070404886A3C87BEF7FF88BF000070E000000104F4FF82BF7E047E880000FDBB1E2080BE
+  - Name:            .AMDGPU.config
+    Type:            SHT_PROGBITS
+    Address:         0x2000
+    AddressAlign:    0x1
+    Content:         48B80000000000004CB800000000000060B800000000000004000000000000000800000000000000
+  - Name:            .note.GNU-stack
+    Type:            SHT_PROGBITS
+    Address:         0x3000
+    AddressAlign:    0x1
+  - Name:            .note
+    Type:            SHT_NOTE
+    Address:         0x4000
+    AddressAlign:    0x4
+    Notes:
+      - Name:            AMD
+        Desc:            616D6467636E2D756E6B6E6F776E2D6C696E75782D676E752D67667831303330
+        Type:            NT_FREEBSD_PROCSTAT_GROUPS
+  - Type:            SectionHeaderTable
+    Sections:
+      - Name:            .strtab
+      - Name:            .shstrtab
+      - Name:            .text
+      - Name:            .AMDGPU.config
+      - Name:            .note.GNU-stack
+      - Name:            .note
+      - Name:            .symtab
+Symbols:
+  - Name:            break_cond_is_arg
+    Type:            STT_FUNC
+    Section:         .text
+    Binding:         STB_GLOBAL
+    Size:            0x5C
+    Value:           0x1000
+...

llvm/tools/llvm-objdump/llvm-objdump.cpp

if (!SymbolizeOperands)
     return;
 
   // Synthesize labels referenced by branch instructions by
   // disassembling, discarding the output, and collecting the referenced
   // addresses from the symbolizer.
   for (size_t Index = 0; Index != Bytes.size();) {
     MCInst Inst;
     uint64_t Size;
-    ArrayRef<uint8_t> ThisBytes = Bytes.slice(Index - SectionAddr);
-    DisAsm->getInstruction(Inst, Size, ThisBytes, Index, nulls());
+    ArrayRef<uint8_t> ThisBytes = Bytes.slice(Index);
+    const uint64_t ThisAddr = SectionAddr + Index;
+    DisAsm->getInstruction(Inst, Size, ThisBytes, ThisAddr, nulls());
     if (Size == 0)
       Size = std::min<uint64_t>(ThisBytes.size(),
                                 DisAsm->suggestBytesToSkip(ThisBytes, Index));
     Index += Size;
   }
   ArrayRef<uint64_t> LabelAddrsRef = SymbolizerPtr->getReferencedAddresses();
   // Copy and sort to remove duplicates.
   std::vector<uint64_t> LabelAddrs;