[Bug libfortran/104233] New: [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[Bug libfortran/104233] New: [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

            Bug ID: 104233
           Summary: [12 regression] ieee/issignaling_fallback.h broke
                    Solaris/SPARC bootstrap
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libfortran
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ro at gcc dot gnu.org
                CC: fxcoudert at gcc dot gnu.org
  Target Milestone: ---
            Target: sparc-sun-solaris2.11

The introduction of ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap:

In file included from
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:32:
/vol/gcc/src/hg/master/local/libgfortran/ieee/issignaling_fallback.h:204:3:
error: unknown type name ‘__float128’
  204 |   __float128 value;
      |   ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/issignaling_fallback.h:218:20:
error: unknown type name ‘__float128’; did you mean ‘_Float128’?
  218 | __issignalingf128 (__float128 x)
      |                    ^~~~~~~~~~
      |                    _Float128
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c: In function
‘ieee_class_helper_4’:
/vol/gcc/src/hg/master/local/libgfortran/ieee/issignaling_fallback.h:240:13:
error: unknown type name ‘__float128’
  240 |             __float128: __issignalingf128, \
      |             ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:97:11: note: in
expansion of macro ‘issignaling’
   97 |       if (issignaling (*value)) \
      |           ^~~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:106:1: note: in
expansion of macro ‘CLASSMACRO’
  106 | CLASSMACRO(4)
      | ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c: In function
‘ieee_class_helper_8’:
/vol/gcc/src/hg/master/local/libgfortran/ieee/issignaling_fallback.h:240:13:
error: unknown type name ‘__float128’
  240 |             __float128: __issignalingf128, \
      |             ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:97:11: note: in
expansion of macro ‘issignaling’
   97 |       if (issignaling (*value)) \
      |           ^~~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:107:1: note: in
expansion of macro ‘CLASSMACRO’
  107 | CLASSMACRO(8)
      | ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c: In function
‘ieee_class_helper_16’:
/vol/gcc/src/hg/master/local/libgfortran/ieee/issignaling_fallback.h:240:13:
error: unknown type name ‘__float128’
  240 |             __float128: __issignalingf128, \
      |             ^~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:97:11: note: in
expansion of macro ‘issignaling’
   97 |       if (issignaling (*value)) \
      |           ^~~~~~~~~~~
/vol/gcc/src/hg/master/local/libgfortran/ieee/ieee_helper.c:114:1: note: in
expansion of macro ‘CLASSMACRO’
  114 | CLASSMACRO(16)
      | ^~~~~~~~~~

While __FLT128_IS_IEC_60559__ is defined as 2, __float128 isn't supported on
SPARC.

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

Rainer Orth <ro at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.0

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #1 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
If __FLT128_IS_IEC_60559__ is defined, does it mean that the associated type is
accessible as _Float128?

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #2 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Created attachment 52289
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52289&action=edit
Tentative patch

Does the attached patch fix the issue?

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jakub at gcc dot gnu.org

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Comment on attachment 52289
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52289
Tentative patch

I'm not sure this will work properly, I think in C _Float128 is a distinct
type, so in _Generic it will probably not match __float128.
I think libgfortran uses __float128, not _Float128.
Shouldn't the __float128 stuff be guarded with
defined(GFC_REAL_16_IS_FLOAT128) || defined(HAVE_GFC_REAL_17)
?

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #4 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
> Shouldn't the __float128 stuff be guarded with
> defined(GFC_REAL_16_IS_FLOAT128) || defined(HAVE_GFC_REAL_17)

We could. My goal was for that header not to depend on libgfortran macros
(meaning it's generic), but GFC_REAL_16_IS_FLOAT128 would work in practice.

Re. HAVE_GFC_REAL_17 it is currently not supported in the IEEE files, so I'll
avoid that for now.



Does that restore Solaris/SPARC bootstrap?

diff --git a/libgfortran/ieee/issignaling_fallback.h
b/libgfortran/ieee/issignaling_fallback.h
index 440a3f6c9a5..98b0b2819ac 100644
--- a/libgfortran/ieee/issignaling_fallback.h
+++ b/libgfortran/ieee/issignaling_fallback.h
@@ -195,9 +195,9 @@ __issignalingl (long double x)
 #endif


-#if __FLT128_IS_IEC_60559__
+#if defined(GFC_REAL_16_IS_FLOAT128)

-/* We have a _Float128 type.  */
+/* We have a __float128 type.  */

 typedef union
 {
@@ -234,7 +234,7 @@ __issignalingf128 (__float128 x)

 /* Define the type-generic macro based on the functions above.  */

-#if __FLT128_IS_IEC_60559__
+#if defined(GFC_REAL_16_IS_FLOAT128)
 # define issignaling(X) \
   _Generic ((X), \
            __float128: __issignalingf128, \

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think testing __FLT128_IS_IEC_60559__ next to it is fine too, if __float128
and _Float128 exists, they are on all supported targets the same IEEE quad
types, just _Generic doesn't treat them as the same type.

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

Eric Botcazou <ebotcazou at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
                 CC|                            |ebotcazou at gcc dot gnu.org
   Last reconfirmed|                            |2022-01-26
             Status|UNCONFIRMED                 |NEW

--- Comment #6 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
The canonical 128-bit floating-point type is "long double" on SPARC/Solaris.

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #7 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
Pushed the fix as logical after checking on aarch64-apple-darwin.
https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=8769f32b645eb08b9f07ec3171cc21c3420f8616

Please let me know if this fixes SPARC/Solaris bootstrap.

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

--- Comment #8 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #7 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
> Pushed the fix as logical after checking on aarch64-apple-darwin.
> https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=8769f32b645eb08b9f07ec3171cc21c3420f8616
>
> Please let me know if this fixes SPARC/Solaris bootstrap.

It did indeed.  Thanks.

[Bug libfortran/104233] [12 regression] ieee/issignaling_fallback.h broke Solaris/SPARC bootstrap

[fxcoudert at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104233

Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |FIXED

--- Comment #9 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> ---
> It did indeed.  Thanks.

Great! Thanks for letting me know, and apologies for breaking the build in the
first place.