[PATCH 4/9] Remove addr_size field from struct piece_closure

[Simon Marchi]

On 2017-04-18 13:24, Andreas Arnez wrote:
>> Since you are planning on making a v2, I think it's better if you 
>> check it
>> in right now to get it out of the way (it could be obvious anyway).
> 
> I would, but it depends on the patch before -- "[PATCH 3/9] PR
> gdb/21226: Take DWARF stack value pieces from LSB end".
> 
> (Maybe it wasn't clear, but the current code for handling DWARF stack
> values is slightly more broken than just not taking them from the
> correct end.  It also uses the relic "addr_size" instead of the actual
> DWARF stack value size to determine whether the piece is contained in
> its underlying object.  And if it's *partially* contained, like taking 
> a
> 1MB piece from a 64-bit stack value, the current code accesses invalid
> memory anyway.)

Ah yes indeed, my bad.