This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Proposal for merging scalar-storage-order branch into mainline
- From: Jakub Jelinek <jakub at redhat dot com>
- To: Eric Botcazou <ebotcazou at adacore dot com>
- Cc: Mark Wielaard <mjw at redhat dot com>, gcc at gcc dot gnu dot org
- Date: Tue, 9 Jun 2015 12:33:05 +0200
- Subject: Re: Proposal for merging scalar-storage-order branch into mainline
- Authentication-results: sourceware.org; auth=none
- References: <2354857 dot 0uXrE6NL1R at polaris> <1433771820 dot 4538 dot 379 dot camel at bordewijk dot wildebeest dot org> <1815334 dot 33LqHF3RXV at polaris>
- Reply-to: Jakub Jelinek <jakub at redhat dot com>
On Tue, Jun 09, 2015 at 12:17:49PM +0200, Eric Botcazou wrote:
> > How is this represented in DWARF?
>
> This is not represented on the branch, because this cannot be done in pure
> DWARF. DW_AT_endianity only applies to base types or stand-alone objects and
> we would need it for DW_TAG_member (and even DW_TAG_array_type in Ada). But
> this could easily be implemented once the representation is agreed on.
DW_AT_endianity attribute is listed in DWARF4 for:
DW_TAG_base_type
DW_TAG_constant
DW_TAG_formal_parameter
DW_TAG_variable
Not really sure how to interpret it e.g. on DW_TAG_variable (or
formal_parameter) if it has DW_TAG_reference_type type - does it talk
about the endianity of what it refers to, or the reference itself, both?
Anyway, the DWARF standard doesn't forbid using it on other kinds of DIEs
and I think emitting it on DW_TAG_member would be natural.
Not sure why you would want it on DW_TAG_array_type, the endianity for
arrays should be specified on the element type, shouldn't it?
Or is array indexing endian dependent?
Jakub