This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
[patch 0/4] decimal floating point support
- From: Thiago Jung Bauermann <bauerman at br dot ibm dot com>
- To: gdb-patches at sourceware dot org
- Date: Thu, 20 Sep 2007 18:54:10 -0300
- Subject: [patch 0/4] decimal floating point support
:ADDPATCH language-feature:
Hi folks,
This is a continuation of the Decimal Floating Point support saga. :-)
It works with binaries generated by GCC 4.2.0 (which has software DFP
support) and also binaries using hardware DFP as found in POWER6
processors.
I refreshed for today's CVS HEAD the DFP patches I submitted previously
here:
http://sourceware.org/ml/gdb-patches/2007-03/msg00247.html
Please see the link above for more detailed discussion and references.
The main change from the previous patch series is in the dfp.c file,
which interfaces with libdecnumber. The reason is that libdecnumber in
gcc's svn trunk was updated to version 3.53, bringing some changes with
it.
Some .h files changed location, but more importantly the library now
stores decimal floats in host byte order. In previous versions,
libdecnumber always stored them in big endian, regardless of the host
byte order. The function match_endianness in dfp.c takes care of fixing
byte order before/after calling libdecnumber functions (decimal floats
are stored in target byte order in GDB data structures).
I also tidied up some minor things like fixing one or two ChangeLog
entries and updating copyright year in testcase files. When refreshing,
there were two rejects in gdbtypes because of the Great Builtin-Types
Reorganization that took place, but they were trivial to fix.
These patches have been tested on ppc32-linux, ppc64-linux and with
remote debugging using i386 host and ppc32 target (to test endianness
handling). I also checked that there are no regressions both in ppc32
and ppc64.
Is this ok for CVS HEAD?
--
[]'s
Thiago Jung Bauermann
Software Engineer
IBM Linux Technology Center