| Message ID | 20180110215357.GA2785@ibm-tiger.the-meissners.org |
|---|---|
| State | New |
| Headers | show |
| Series | Add PowerPC configuration option --with-long-double-format={ibm,ieee} | expand |
On Wed, 10 Jan 2018, Michael Meissner wrote: > This patch is next in my series of patches to enable us to configure the long > double type on PowerPC systems. This patch is only about the configuration > option. A future patch will contain the multilib support. In general we expect configure options to be documented in install.texi. Is this one being deliberately omitted because it is only actually usable for toolchain development at present, with various ways in which the support for IEEE long double is not yet completely functional and library support is missing - with documentation intended to be added later once fully functional?
Hi Mike, On Wed, Jan 10, 2018 at 04:53:57PM -0500, Michael Meissner wrote: > This patch is next in my series of patches to enable us to configure the long > double type on PowerPC systems. This patch is only about the configuration > option. A future patch will contain the multilib support. > > This patch needs the previous patch I submitted today (January 10th, 2018) > applied before it will work. > https://gcc.gnu.org/ml/gcc-patches/2018-01/msg00806.html > > I built a bootstrap configuratrion on a little endian power8 system using the > --with-long-double-format-ibm option and it passed with no regressions. I also > built a non-bootstrap version using --with-long-double-format-ieee and I > verified that it defaulted to IEEE 128-bit for long double. Can I check this > patch into the trunk once the previous patch has been checked in? > > 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com> > > * configure.ac (--with-long-double-format): Add support for the > configuration option to change the default long double format on > PowerPC systems. > * config.gcc (powerpc*-linux*-*): Likewise. > * configure: Regenerate. > * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long > double is IEEE, define __KC__ and __KF__ to allow floatn.h to be > used without modification. > +# Check if TFmode long double target should use the IBM extended double or IEEE > +# 128-bit floating point formats if long doubles are 128-bits long. The long > +# double type can only be switched on powerpc64 bit Linux systems where VSX is > +# supported. Other PowerPC systems do not build the IEEE 128-bit emulator in > +# liggcc. Typo ("libgcc"). > +AC_ARG_WITH([long-double-format], > + [AS_HELP_STRING([--with-long-double-format={ieee,ibm}] > + [Specify whether PowerPC long double uses IEEE or IBM format])],[ > +case x"$target:$with_long_double_format" in (This does not need the "x" thing because the string can never be empty: there always is the ":", for starters). > + xpowerpc64le-*-linux*:ieee | xpowerpc64le-*-linux*:ibm) > + : > + ;; > + xpowerpc64-*-linux*:ieee | xpowerpc64-*-linux*:*:ibm) I think you have an extra ":*" here. > + # IEEE 128-bit emulationr is only built on 64-bit VSX Linux systems Typo ("emulation"). The configure script looks more complicated than it has to (you can first check for valid ieee|ibm and then do the rest, for example). Okay for trunk with the typos fixed. Thanks, Segher
On Wed, Jan 10, 2018 at 11:48:30PM +0000, Joseph Myers wrote: > On Wed, 10 Jan 2018, Michael Meissner wrote: > > > This patch is next in my series of patches to enable us to configure the long > > double type on PowerPC systems. This patch is only about the configuration > > option. A future patch will contain the multilib support. > > In general we expect configure options to be documented in install.texi. > Is this one being deliberately omitted because it is only actually usable > for toolchain development at present, with various ways in which the > support for IEEE long double is not yet completely functional and library > support is missing - with documentation intended to be added later once > fully functional? Yes, I'll try to come up with documentation for it next week if I don't get to it today (I'll be out the next 4 days).
On Wed, Jan 10, 2018 at 11:48:30PM +0000, Joseph Myers wrote: > On Wed, 10 Jan 2018, Michael Meissner wrote: > > > This patch is next in my series of patches to enable us to configure the long > > double type on PowerPC systems. This patch is only about the configuration > > option. A future patch will contain the multilib support. > > In general we expect configure options to be documented in install.texi. > Is this one being deliberately omitted because it is only actually usable > for toolchain development at present, with various ways in which the > support for IEEE long double is not yet completely functional and library > support is missing - with documentation intended to be added later once > fully functional? Here is the first pass at documentation. Segher is this ok? 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com> * doc/install.texi (--with-long-double-format=): Document the switch, and note it isn't for general use at the moment.
On Thu, Jan 11, 2018 at 05:35:59PM -0500, Michael Meissner wrote: > On Wed, Jan 10, 2018 at 11:48:30PM +0000, Joseph Myers wrote: > > On Wed, 10 Jan 2018, Michael Meissner wrote: > > > > > This patch is next in my series of patches to enable us to configure the long > > > double type on PowerPC systems. This patch is only about the configuration > > > option. A future patch will contain the multilib support. > > > > In general we expect configure options to be documented in install.texi. > > Is this one being deliberately omitted because it is only actually usable > > for toolchain development at present, with various ways in which the > > support for IEEE long double is not yet completely functional and library > > support is missing - with documentation intended to be added later once > > fully functional? > > Here is the first pass at documentation. Segher is this ok? Sure, thanks! One tiny thing: > +Specify what format 128-bit @code{long double} should use on PowerPC > +VSX systems, If you use @code{--with-long-double-format=ibm}, +VSX systems. If you use @code{--with-long-double-format=ibm}, ? Segher
Index: gcc/configure.ac =================================================================== --- gcc/configure.ac (revision 256356) +++ gcc/configure.ac (working copy) @@ -5885,6 +5885,43 @@ if test x$gcc_cv_target_ldbl128 = xyes; [Define if TFmode long double should be the default]) fi +# Check if TFmode long double target should use the IBM extended double or IEEE +# 128-bit floating point formats if long doubles are 128-bits long. The long +# double type can only be switched on powerpc64 bit Linux systems where VSX is +# supported. Other PowerPC systems do not build the IEEE 128-bit emulator in +# liggcc. +AC_ARG_WITH([long-double-format], + [AS_HELP_STRING([--with-long-double-format={ieee,ibm}] + [Specify whether PowerPC long double uses IEEE or IBM format])],[ +case x"$target:$with_long_double_format" in + xpowerpc64le-*-linux*:ieee | xpowerpc64le-*-linux*:ibm) + : + ;; + xpowerpc64-*-linux*:ieee | xpowerpc64-*-linux*:*:ibm) + # IEEE 128-bit emulationr is only built on 64-bit VSX Linux systems + case "$with_cpu" in + power7 | power8 | power9 | power1*) + : + ;; + *) + AC_MSG_ERROR([Configuration option --with-long-double-format is only \ +supported if the default cpu is power7 or newer]) + with_long_double_format="" + ;; + esac + ;; + xpowerpc64*-*-linux*:*) + AC_MSG_ERROR([--with-long-double-format argument should be ibm or ieee]) + with_long_double_format="" + ;; + *) + AC_MSG_ERROR([Configure option --with-long-double-format is only supported \ +on 64-bit PowerPC VSX Linux systems]) + with_long_double_format="" + ;; +esac], + []) + # Check if the target LIBC supports exporting the AT_PLATFORM and AT_HWCAP # values in the TCB. Currently, only GLIBC 2.23 and later support this. gcc_cv_libc_provides_hwcap_in_tcb=no Index: gcc/configure =================================================================== --- gcc/configure (revision 256356) +++ gcc/configure (working copy) @@ -945,6 +945,7 @@ enable_linker_build_id enable_libssp enable_default_ssp with_long_double_128 +with_long_double_format with_gc with_system_zlib enable_maintainer_mode @@ -1738,6 +1739,9 @@ Optional Packages: --with-glibc-version=M.N assume GCC used with glibc version M.N or later --with-long-double-128 use 128-bit long double by default + --with-long-double-format={ieee,ibm} + Specify whether PowerPC long double uses IEEE or IBM format + --with-gc={page,zone} this option is not supported anymore. It used to choose the garbage collection mechanism to use with the compiler @@ -18442,7 +18446,7 @@ else lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext <<_LT_EOF -#line 18445 "configure" +#line 18449 "configure" #include "confdefs.h" #if HAVE_DLFCN_H @@ -18548,7 +18552,7 @@ else lt_dlunknown=0; lt_dlno_uscore=1; lt_dlneed_uscore=2 lt_status=$lt_dlunknown cat > conftest.$ac_ext <<_LT_EOF -#line 18551 "configure" +#line 18555 "configure" #include "confdefs.h" #if HAVE_DLFCN_H @@ -29185,6 +29189,45 @@ $as_echo "#define TARGET_DEFAULT_LONG_DO fi +# Check if TFmode long double target should use the IBM extended double or IEEE +# 128-bit floating point formats if long doubles are 128-bits long. The long +# double type can only be switched on powerpc64 bit Linux systems where VSX is +# supported. Other PowerPC systems do not build the IEEE 128-bit emulator in +# liggcc. + +# Check whether --with-long-double-format was given. +if test "${with_long_double_format+set}" = set; then : + withval=$with_long_double_format; +case x"$target:$with_long_double_format" in + xpowerpc64le-*-linux*:ieee | xpowerpc64le-*-linux*:ibm) + : + ;; + xpowerpc64-*-linux*:ieee | xpowerpc64-*-linux*:*:ibm) + # IEEE 128-bit emulationr is only built on 64-bit VSX Linux systems + case "$with_cpu" in + power7 | power8 | power9 | power1*) + : + ;; + *) + as_fn_error "Configuration option --with-long-double-format is only \ +supported if the default cpu is power7 or newer" "$LINENO" 5 + with_long_double_format="" + ;; + esac + ;; + xpowerpc64*-*-linux*:*) + as_fn_error "--with-long-double-format argument should be ibm or ieee" "$LINENO" 5 + with_long_double_format="" + ;; + *) + as_fn_error "Configure option --with-long-double-format is only supported \ +on 64-bit PowerPC VSX Linux systems" "$LINENO" 5 + with_long_double_format="" + ;; +esac +fi + + # Check if the target LIBC supports exporting the AT_PLATFORM and AT_HWCAP # values in the TCB. Currently, only GLIBC 2.23 and later support this. gcc_cv_libc_provides_hwcap_in_tcb=no Index: gcc/config.gcc =================================================================== --- gcc/config.gcc (revision 256356) +++ gcc/config.gcc (working copy) @@ -4400,6 +4400,14 @@ case "${target}" in exit 1 fi fi + + # Set up the default long double format if the user changed it. + if test x$with_long_double_format = xieee; then + tm_defines="${tm_defines} TARGET_IEEEQUAD_DEFAULT=1" + + elif test x$with_long_double_format = xibm; then + tm_defines="${tm_defines} TARGET_IEEEQUAD_DEFAULT=0" + fi ;; s390*-*-*) Index: gcc/config/rs6000/rs6000-c.c =================================================================== --- gcc/config/rs6000/rs6000-c.c (revision 256356) +++ gcc/config/rs6000/rs6000-c.c (working copy) @@ -708,7 +708,18 @@ rs6000_cpu_cpp_builtins (cpp_reader *pfi builtin_define ("__LONGDOUBLE128"); if (TARGET_IEEEQUAD) - builtin_define ("__LONG_DOUBLE_IEEE128__"); + { + /* Older versions of GLIBC used __attribute__((__KC__)) to create the + IEEE 128-bit floating point complex type for C++ (which does not + support _Float128 _Complex). If the default for long double is + IEEE 128-bit mode, the library would need to use + __attribute__((__TC__)) instead. Defining __KF__ and __KC__ + is a stop-gap to build with the older libraries, until we + get an updated library. */ + builtin_define ("__LONG_DOUBLE_IEEE128__"); + builtin_define ("__KF__=__TF__"); + builtin_define ("__KC__=__TC__"); + } else builtin_define ("__LONG_DOUBLE_IBM128__"); }