Hello,

Testing reveals that the test failed on some platforms:

        (gdb) b world
        Breakpoint 1 at 0x4005c4

Upon investigating it, it turns out that we were accessing the value
of the attribute using:

    high = DW_ADDR (attr_high);

This is not correct, since DW_ADDR assumes the attribute's form
is an address, whereas the test was using a data form, which means
we should have been using DW_UNSND. On some platforms, DW_ADDR
happens to work, but that's only pure luck related to the two
components being stored within the same union. But on ppc-linux,
the attribute value was stored in the DW_UNSND component, and
reading it back using DW_ADDR returned zero/garbage.

In addition to revealing a bug in my patch, this revealed a slightly
more general issue, which is the assumption that address attributes
are always encoded using address forms. So, I amended my patch to
read addresses in a way that takes the attribute's form into account.
Apart from increasing the number of edits, all mostly mechanical,
I find that it simplified the handling for DW_AT_high_pc quite
a bit, thanks to the introduction of a function getting the attribute
value.

Changes compared to the previous version:

  1. The introduction of attr_value_as_address, to be used
     in place of DW_ADDR when dealing with address attributes.
     I left a few uses of this macro in the situations where
     we actually know that the form is an address form.

  2. Instead of duplicating everywhere the conditions for
     non-address forms in the handling of DW_AT_high_pc
     attributes, I used the attr_form_is_constant function
     instead.  It's not stricly the same, but I think it is
     closer to the DWARF reference.

  3. Some very minor modifications of the assembly files in our
     testcase to make it compilable on a larger number of targets:
     Removing some of the extra parameters for the .section
     pseudo-op, as well as using .2byte instead of .value.

  4. In our testcase, I also adjusted the DW_AT_low_pc attributes
     to be encoded using DW_FORM_data4, to be consistent with
     the DW_AT_high_pc attributes, as well as to test reading
     of that attribute in constant form.

gdb/ChangeLog:

        * dwarf2read.c (attr_value_as_address): New function.
        (dwarf2_find_base_address, read_call_site_scope): Use
        attr_value_as_address in place of DW_ADDR.
        (dwarf2_get_pc_bounds): Use attr_value_as_address to get
        the low and high addresses.  Slight rework of the handling
        of the high pc being a constant form, and limit it to
        DWARF verson 4 or higher.
        (dwarf2_record_block_ranges): Likewise.
        (read_partial_die): Likewise.
        (new_symbol_full): Use attr_value_as_address in place of DW_ADDR.

gdb/testsuite/ChangeLog:

        * gdb.dwarf2/dw2-abs-hi-pc-hello-dbg.S: New file.
        * gdb.dwarf2/dw2-abs-hi-pc-hello.c: New file.
        * gdb.dwarf2/dw2-abs-hi-pc-world-dbg.S: New file.
        * gdb.dwarf2/dw2-abs-hi-pc-world.c: New file.
        * gdb.dwarf2/dw2-abs-hi-pc.c: New file.
        * gdb.dwarf2/dw2-abs-hi-pc.exp: New file.

Tested on x86_64-linux. The testcase was also testing on ppc-linux,
where it was failing prior to the second iteration of this patch.

OK to commit?

Thanks,
-- 
Joel