[RFA,1/2] Use a distinguishing name for minidebug objfile

Message ID 20170413041504.14435-2-tom@tromey.com
State New, archived
Headers

Commit Message

Tom Tromey April 13, 2017, 4:15 a.m. UTC
  One part of PR cli/19951 is that the mini debug info objfile reuses the
name of the main objfile from which it comes.  This can be seen because
gdb claims to be reading symbols from the same file two times, like:

Reading symbols from /bin/gdb...Reading symbols from /bin/gdb...(no debugging symbols found)...done.

I think this would be less confusing if the minidebug objfile were given
a different name.  That is what this patch implements.  It also arranges
for the minidebug objfile to be marked OBJF_NOT_FILENAME.

gdb/ChangeLog
2017-04-12  Tom Tromey  <tom@tromey.com>

	PR cli/19551:
	* symfile-add-flags.h (enum symfile_add_flags)
	<SYMFILE_NOT_FILENAME>: New constant.
	* symfile.c (read_symbols): Use SYMFILE_NOT_FILENAME.  Get
	objfile name from BFD.
	(symbol_file_add_with_addrs): Check SYMFILE_NOT_FILENAME.
	* minidebug.c (find_separate_debug_file_in_section): Put
	".gnu_debugdata" into BFD's file name.
---
 gdb/ChangeLog           | 11 +++++++++++
 gdb/minidebug.c         |  5 ++++-
 gdb/symfile-add-flags.h |  7 +++++--
 gdb/symfile.c           |  4 +++-
 4 files changed, 23 insertions(+), 4 deletions(-)
  

Comments

Pedro Alves April 18, 2017, 5:42 p.m. UTC | #1
On 04/13/2017 05:15 AM, Tom Tromey wrote:
> One part of PR cli/19951 is that the mini debug info objfile reuses the
> name of the main objfile from which it comes.  This can be seen because
> gdb claims to be reading symbols from the same file two times, like:
> 
> Reading symbols from /bin/gdb...Reading symbols from /bin/gdb...(no debugging symbols found)...done.
> 
> I think this would be less confusing if the minidebug objfile were given
> a different name.  That is what this patch implements.  It also arranges
> for the minidebug objfile to be marked OBJF_NOT_FILENAME.

Could you update the commit log to also show how things look
after the patch?  I could guess it after staring at the patch for
a bit, but it'd be nice to be explicit.

Otherwise LGTM.

Thanks,
Pedro Alves
  
Tom Tromey April 19, 2017, 3:11 a.m. UTC | #2
>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:

Pedro> Could you update the commit log to also show how things look
Pedro> after the patch?  I could guess it after staring at the patch for
Pedro> a bit, but it'd be nice to be explicit.

Thanks for suggesting this, as it pointed out that during one of the
rebases I'd apparently regressed the patch, and it no longer printed the
desired thing.

I'll see if I can write a test case.

Tom
  

Patch

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 315932e..c56de68 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,14 @@ 
+2017-04-12  Tom Tromey  <tom@tromey.com>
+
+	PR cli/19551:
+	* symfile-add-flags.h (enum symfile_add_flags)
+	<SYMFILE_NOT_FILENAME>: New constant.
+	* symfile.c (read_symbols): Use SYMFILE_NOT_FILENAME.  Get
+	objfile name from BFD.
+	(symbol_file_add_with_addrs): Check SYMFILE_NOT_FILENAME.
+	* minidebug.c (find_separate_debug_file_in_section): Put
+	".gnu_debugdata" into BFD's file name.
+
 2017-04-13  Pedro Alves  <palves@redhat.com>
 
 	* breakpoint.h (struct breakpoint): Reindent.
diff --git a/gdb/minidebug.c b/gdb/minidebug.c
index d4687dd..c2ada46 100644
--- a/gdb/minidebug.c
+++ b/gdb/minidebug.c
@@ -270,7 +270,10 @@  find_separate_debug_file_in_section (struct objfile *objfile)
     return NULL;
 
 #ifdef HAVE_LIBLZMA
-  abfd = gdb_bfd_openr_iovec (objfile_name (objfile), gnutarget, lzma_open,
+  std::string filename = (std::string (".gnu_debugdata for ")
+			  + objfile_name (objfile));
+
+  abfd = gdb_bfd_openr_iovec (filename.c_str (), gnutarget, lzma_open,
 			      section, lzma_pread, lzma_close, lzma_stat);
   if (abfd == NULL)
     return NULL;
diff --git a/gdb/symfile-add-flags.h b/gdb/symfile-add-flags.h
index 4d5f58a..c2c6fd2 100644
--- a/gdb/symfile-add-flags.h
+++ b/gdb/symfile-add-flags.h
@@ -40,8 +40,11 @@  enum symfile_add_flag
 
     /* Do not immediately read symbols for this file.  By default,
        symbols are read when the objfile is created.  */
-    SYMFILE_NO_READ = 1 << 4
-  };
+    SYMFILE_NO_READ = 1 << 4,
+
+    /* The new objfile should be marked OBJF_NOT_FILENAME.  */
+    SYMFILE_NOT_FILENAME = 1 << 5,
+ };
 
 DEF_ENUM_FLAGS_TYPE (enum symfile_add_flag, symfile_add_flags);
 
diff --git a/gdb/symfile.c b/gdb/symfile.c
index 846aabe..ff1e726 100644
--- a/gdb/symfile.c
+++ b/gdb/symfile.c
@@ -876,7 +876,7 @@  read_symbols (struct objfile *objfile, symfile_add_flags add_flags)
 	     section.  Therefore use also non-canonical name form for the same
 	     file containing the section.  */
 	  symbol_file_add_separate (abfd.get (), objfile->original_name,
-				    add_flags, objfile);
+				    add_flags | SYMFILE_NOT_FILENAME, objfile);
 	}
     }
   if ((add_flags & SYMFILE_NO_READ) == 0)
@@ -1143,6 +1143,8 @@  symbol_file_add_with_addrs (bfd *abfd, const char *name,
       flags |= OBJF_READNOW;
       add_flags &= ~SYMFILE_NO_READ;
     }
+  if ((add_flags & SYMFILE_NOT_FILENAME) != 0)
+    flags |= OBJF_NOT_FILENAME;
 
   /* Give user a chance to burp if we'd be
      interactively wiping out any existing symbols.  */