Hi David,
> The bug in readelf (I think) is at the end of display_debug_frames() in
> dwarf.c (binutils directory).
>
> The frame_display_row call is a bit odd if all that exist are nops and
> results in the rax output
I am currently testing out the attached patch. I am not sure however if
it is doing the right thing by just silently ignoring DW_CFA_NOPs. Maybe
it should issue an interpretation along the lines of:
LOC CFA ra
0000000000020480 nop -
What do you think ?
Cheers
Nick
diff --git a/binutils/dwarf.c b/binutils/dwarf.c
index 770fa6b..282e069 100644
--- a/binutils/dwarf.c
+++ b/binutils/dwarf.c
@@ -5780,6 +5780,7 @@ display_debug_frames (struct dwarf_section *section,
unsigned int encoded_ptr_size = saved_eh_addr_size;
unsigned int offset_size;
unsigned int initial_length_size;
+ bfd_boolean all_nops;
saved_start = start;
@@ -6213,6 +6214,8 @@ display_debug_frames (struct dwarf_section *section,
start = tmp;
}
+ all_nops = TRUE;
+
/* Now we know what registers are used, make a second pass over
the chunk, this time actually printing out the info. */
@@ -6231,6 +6234,10 @@ display_debug_frames (struct dwarf_section *section,
if (op & 0xc0)
op &= 0xc0;
+ /* Make a note if something other than DW_CFA_nop happens. */
+ if (op != DW_CFA_nop)
+ all_nops = FALSE;
+
/* Warning: if you add any more cases to this switch, be
sure to add them to the corresponding switch above. */
switch (op)
@@ -6661,7 +6668,8 @@ display_debug_frames (struct dwarf_section *section,
}
}
- if (do_debug_frames_interp)
+ /* Interpret the CFA - as long as it is not completely full of NOPs. */
+ if (do_debug_frames_interp && ! all_nops)
frame_display_row (fc, &need_col_headers, &max_regs);
start = block_end;
_______________________________________________
Dwarf-Discuss mailing list
Dwarf-Discuss@lists.dwarfstd.org
http://lists.dwarfstd.org/listinfo.cgi/dwarf-discuss-dwarfstd.org
