[llvm-readelf]Print filename for multiple inputs and fix formatting regression

[jhenderson]

Summary

This patch addresses two closely related bugs: https://bugs.llvm.org/show_bug.cgi?id=42930 and https://bugs.llvm.org/show_bug.cgi?id=42931. GNU readelf prints the file name for every input unless there is only one input and that input is not an archive. This patch adds the printing for multiple inputs. A previous change did it for archives, but introduced a regression with GNU compatibility for single-output formatting, resulting in a spurious initial blank line. This is fixed in this patch too.

[MaskRay]

test/tools/llvm-readobj/file-name.test

28

To test the first line is blank: # NAME1: {{^$}}

This is what readelf does (since its initial import in 1999):

/* Process the file.  */
if (show_name)
  printf (_("\nFile: %s\n"), filedata->file_name);

I think the blank line is pretty arbitrary. It is there just because it was easy to implement...

[grimar]

LGTM.

[jhenderson]

test/tools/llvm-readobj/file-name.test

28

I agree that the line is arbitrary. Unfortunately, I've noticed a local parsing script that relies on it in some cases, so I'd be surprised if there aren't others out there doing the same thing.

To test the first line is blank: # NAME1: {{^$}}

I knew that somebody would comment on this :)

I was just typing out this response, when I realised that we don't need the added complexity:

All your suggested pattern does is test that there is an empty line somewhere in the input (and that later checks follow it). It doesn't guarantee it's the first line. Just as we want one empty line at the start, we also don't want 2+

I realised however that FileCheck ensures that the first check matches the first line, and then the CHECK-NEXT matches the next, so it can't be an arbitrary location. I'll update the test before committing.

test/tools/llvm-readobj/elf-dynamic-malformed.test

 # WARN:      ]
 # WARN-EMPTY:
 ## A warning is printed at the place where a normal dynamic table should be. 
 # WARN-NEXT: warning: invalid section size (4) or entity size (16)
 # WARN-NEXT: ProgramHeaders [
 
 # WARN-GNU-NOT:  warning
 # WARN-GNU:      warning: invalid section size (4) or entity size (16)
-# WARN-GNU-EMPTY:
 # WARN-GNU-NEXT: ELF Header:
 # WARN-GNU:      Symbol table '.symtab' contains 1 entries:
 # WARN-GNU:        0:
 # WARN-GNU-EMPTY:
 ## A warning is printed at the place where a normal dynamic table should be. 
 # WARN-GNU:      warning: invalid section size (4) or entity size (16)
 # WARN-GNU-EMPTY:
 

test/tools/llvm-readobj/file-name.test

+## This test shows that the name of the file is printed under the right
+## circumstances and with the correct formatting for object inputs.
+
+# RUN: yaml2obj %s -o %t1
+# RUN: cp %t1 %t2
+
+## Show that the file name is not printed for a single input for GNU output.
+# RUN: llvm-readelf --file-headers %t1 \
+# RUN:   | FileCheck %s --implicit-check-not=File: --check-prefix=GNU-SINGLE
+
+## Show that the very first line of the output is the one with "ELF Header" on.
+# GNU-SINGLE:      {{^}}
+# GNU-SINGLE-SAME: ELF Header:
+
+## Show that the file names are printed for all inputs for GNU output.
+# RUN: llvm-readelf --file-headers %t1 %t2 \
+# RUN:   | FileCheck %s --check-prefixes=NAME1,GNU,NAME2 -DFILE1=%t1 -DFILE2=%t2
+
+## Show that the file name is printed for a single input for LLVM output.
+# RUN: llvm-readobj --file-headers %t1 | FileCheck %s --check-prefix=NAME1 -DFILE1=%t1
+
+## Show that the file name is printed with correct spacing for multiple inputs
+## with LLVM output.
+# RUN: llvm-readobj --file-headers %t1 %t2 \
+# RUN:   | FileCheck %s --check-prefixes=NAME1,LLVM,NAME2 -DFILE1=%t1 -DFILE2=%t2
+
+## The very first line should be blank. CHECK-EMPTY isn't allowed on the first line.
+# NAME1:      {{^}}
+# NAME1-NOT:  {{.}}
+# NAME1-SAME: {{$}}
+# NAME1-NEXT: {{^}}File: [[FILE1]]{{$}}
+
+# GNU:          Section header string table index:
+# LLVM:         StringTableSectionIndex:
+# LLVM-NEXT:  }
+# NAME2-EMPTY:
+# NAME2-NEXT: {{^}}File: [[FILE2]]{{$}}
+
+--- !ELF
+FileHeader:
+  Class:    ELFCLASS64
+  Data:     ELFDATA2LSB
+  Type:     ET_REL
+  Machine:  EM_X86_64

test/tools/llvm-readobj/stack-sizes.test

- Name:    .stack_sizes
     Link:    .text
     Content: "000000000000000008"
 
 ## Check that we handle multiple object files, separately and when they
 ## are in an archive. This also checks whether we have blank lines between the
 ## tables.
 
 # RUN: llvm-ar rc %t1.a %t01 %t02
-# RUN: llvm-readelf --stack-sizes %t01 %t02 | FileCheck %s --check-prefix=MULTIPLE
+# RUN: llvm-readelf --stack-sizes %t01 %t02 \
+# RUN:   | FileCheck %s --check-prefixes=MULTIPLE,OBJECT -DFILE1=%t01 -DFILE2=%t02
 # RUN: llvm-readelf --stack-sizes %t1.a \
-# RUN:   | FileCheck %s --check-prefix=MULTIPLE --check-prefix=ARCHIVE --strict-whitespace\
+# RUN:   | FileCheck %s --check-prefixes=MULTIPLE,ARCHIVE --strict-whitespace\
 # RUN:   --match-full-lines -DFILE=%t1.a
 
+#        OBJECT:File: [[FILE1]]
 #       ARCHIVE:File: [[FILE]]({{.*01}})
 #      MULTIPLE:Stack Sizes:
 # MULTIPLE-NEXT:         Size     Function
 # MULTIPLE-NEXT:           16     referenced_by_symbol_foo
 # MULTIPLE-NEXT:           32     referenced_via_section_bar
 # MULTIPLE-NEXT:            8     separate_text_section_baz
 # MULTIPLE-EMPTY:
+#        OBJECT:File: [[FILE2]]
 #       ARCHIVE:File: [[FILE]]({{.*02}})
 # MULTIPLE-EMPTY:
 # MULTIPLE-NEXT:Stack Sizes:
 # MULTIPLE-NEXT:         Size     Function
 # MULTIPLE-NEXT:           16     foo
 # MULTIPLE-NEXT:           32     bar
 
 ## Check that we do not consider symbols that are not function symbols, even though

tools/llvm-readobj/llvm-readobj.cpp

static void dumpObject(const ObjectFile *Obj, ScopedPrinter &Writer,
   std::string FileStr =
           A ? Twine(A->getFileName() + "(" + Obj->getFileName() + ")").str()
             : Obj->getFileName().str();
 
   std::unique_ptr<ObjDumper> Dumper;
   if (std::error_code EC = createDumper(Obj, Writer, Dumper))
     reportError(FileStr, EC);
 
+  if (opts::Output == opts::LLVM || opts::InputFilenames.size() > 1 || A) {
     Writer.startLine() << "\n";
-  if (opts::Output == opts::LLVM) {
     Writer.printString("File", FileStr);
+  }
+  if (opts::Output == opts::LLVM) {
     Writer.printString("Format", Obj->getFileFormatName());
     Writer.printString("Arch", Triple::getArchTypeName(
                                    (llvm::Triple::ArchType)Obj->getArch()));
     Writer.printString("AddressSize",
                        formatv("{0}bit", 8 * Obj->getBytesInAddress()));
     Dumper->printLoadName();
-  } else if (opts::Output == opts::GNU && A) {
-    Writer.printString("File", FileStr);
   }
 
   if (opts::FileHeaders)
     Dumper->printFileHeaders();
   if (opts::SectionHeaders)
     Dumper->printSectionHeaders();
   if (opts::Relocations)
     Dumper->printRelocations();