[rfa/dwarf] Handle debug_line extended ops correctly

[rfa/dwarf] Handle debug_line extended ops correctly

[Daniel Jacobowitz]

This has been hanging around in my tree for a while.  The HP-UX ia64
compiler generates things like this in .debug_line:
  Extended opcode 32: UNKNOWN: length 51
  Extended opcode 32: UNKNOWN: length 146

146 is different as a byte and as a uleb128, so we get confused.  Of course,
we're going to bail out shortly anyway because we don't accept unknown
extended ops (probably for the best); but we might as well parse the length
right.

OK?

-- 
Daniel Jacobowitz

2004-06-16  Daniel Jacobowitz  <dan@debian.org>

	PR gdb/1658
	* dwarf2read.c (dwarf_decode_lines): Read the length of the extended
	operation as a uleb128.  Found by Michael Coulter.

Index: gdb/dwarf2read.c
===================================================================
RCS file: /cvs/src/src/gdb/dwarf2read.c,v
retrieving revision 1.153
diff -u -p -r1.153 dwarf2read.c
--- gdb/dwarf2read.c	15 Jun 2004 01:04:19 -0000	1.153
+++ gdb/dwarf2read.c	16 Jun 2004 15:38:49 -0000
@@ -5986,7 +5986,8 @@ dwarf_decode_lines (struct line_header *
 	  else switch (op_code)
 	    {
 	    case DW_LNS_extended_op:
-	      line_ptr += 1;	/* ignore length */
+	      read_unsigned_leb128 (abfd, line_ptr, &bytes_read);
+	      line_ptr += bytes_read;
 	      extended_op = read_1_byte (abfd, line_ptr);
 	      line_ptr += 1;
 	      switch (extended_op)

Re: [rfa/dwarf] Handle debug_line extended ops correctly

[Jim Blandy]

Daniel Jacobowitz <drow@false.org> writes:

> This has been hanging around in my tree for a while.  The HP-UX ia64
> compiler generates things like this in .debug_line:
>   Extended opcode 32: UNKNOWN: length 51
>   Extended opcode 32: UNKNOWN: length 146
> 
> 146 is different as a byte and as a uleb128, so we get confused.  Of course,
> we're going to bail out shortly anyway because we don't accept unknown
> extended ops (probably for the best); but we might as well parse the length
> right.
> 
> OK?

Okay.

Re: [rfa/dwarf] Handle debug_line extended ops correctly

[Daniel Jacobowitz]

On Wed, Jun 16, 2004 at 12:07:30PM -0500, Jim Blandy wrote:
> Daniel Jacobowitz <drow@false.org> writes:
> 
> > This has been hanging around in my tree for a while.  The HP-UX ia64
> > compiler generates things like this in .debug_line:
> >   Extended opcode 32: UNKNOWN: length 51
> >   Extended opcode 32: UNKNOWN: length 146
> > 
> > 146 is different as a byte and as a uleb128, so we get confused.  Of course,
> > we're going to bail out shortly anyway because we don't accept unknown
> > extended ops (probably for the best); but we might as well parse the length
> > right.
> > 
> > OK?
> 
> Okay.

Committed, thanks.

-- 
Daniel Jacobowitz