This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [9/9] Support __float128 format on Intel and Power
- From: Joseph Myers <joseph at codesourcery dot com>
- To: Ulrich Weigand <uweigand at de dot ibm dot com>
- Cc: <gdb-patches at sourceware dot org>
- Date: Thu, 25 Aug 2016 23:27:45 +0000
- Subject: Re: [9/9] Support __float128 format on Intel and Power
- Authentication-results: sourceware.org; auth=none
- References: <20160825150829.9A434C9FB@oc7340732750.ibm.com>
On Thu, 25 Aug 2016, Ulrich Weigand wrote:
> Since at this point we do not yet have any indication in the debug info to
> distinguish different floating-point formats of the same length, we simply
> use the type name as hint. Types named "__float128" get the IEEE format.
Note that now (GCC 7) __float128 is a built-in typedef (i.e. a name
registered with the register_builtin_type langhook) for _Float128, and
_Float128 is the primary name used when the type is created. If I build a
program with GCC 7 that uses __float128, the debug info I get references
_Float128 (and not __float128 as far as I can see).
Obviously debugging should work whether someone uses _Float128 or
__float128 (and in C++, right now they can only use __float128, as the
keyword _Float128 is C-specific). Then on Power, _Float64x is also
binary128 when available (but is a distinct type at the C level from
_Float128), but on x86 it has the Intel extended format (but is a distinct
type from long double).
--
Joseph S. Myers
joseph@codesourcery.com