[Bug target/104208] New: -mlong-double-64 should override a previous -mabi=ibmlongdouble

[Bug target/104208] New: -mlong-double-64 should override a previous -mabi=ibmlongdouble

[fw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

            Bug ID: 104208
           Summary: -mlong-double-64 should override a previous
                    -mabi=ibmlongdouble
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: fw at gcc dot gnu.org
  Target Milestone: ---
             Build: powerpc64le-*-linux-gnu

We have trouble building glibc 2.35 with GCC 12 because -mlong-double-64 and
-mabi=ibmlongdouble conflict:

ppc64le: gcc 12 vs -mabi=ibmlongdouble
https://sourceware.org/pipermail/libc-alpha/2022-January/135647.html

glibc needs to build itself with -mabi=ibmlongdouble because that's how the
source tree defaults are laid out, and given our build system, it's very
difficult to remove that for individual files. So I'd appreciate a GCC-side fix
for this.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[fw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

Florian Weimer <fw at gcc dot gnu.org> changed:

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

--- Comment #1 from Florian Weimer <fw at gcc dot gnu.org> ---
Setting P1 based on off-bug discussions.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #2 from Segher Boessenkool <segher at gcc dot gnu.org> ---
If you want -mlong-double-64 to override -mabi={ibm,ieee}longdouble, you need
make sure that the last of those options on the command line wins.  And what
should -mlong-double-128 do in that scheme?

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

Peter Bergner <bergner at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |bergner at gcc dot gnu.org
   Last reconfirmed|                            |2022-03-01

--- Comment #3 from Peter Bergner <bergner at gcc dot gnu.org> ---
So the change that is hitting us here was an (overeager?) attempt by me in
PR87496 to catch conflicting options:

+  else if (global_options_set.x_rs6000_ieeequad)
+    {
+      if (global_options.x_rs6000_ieeequad)
+       error ("%qs requires %qs", "-mabi=ieeelongdouble",
"-mlong-double-128");
+      else
+       error ("%qs requires %qs", "-mabi=ibmlongdouble", "-mlong-double-128");
+    }

In talking with the clang developers about this issue, they ignore the
-mabi={ibm,ieee}longdouble option in the presence of -mlong-double-64 like we
used to do, so I think the most convenient solution for everyone is to just
revert the hunk above from my previous patch, which does fix the issue seen in
glibc.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
That is certainly one option.
Another option would be to move this diagnostics to rs6000_handle_option
and diagnose if -mabi=ieeelongdouble or -mabi=ibmlongdouble is seen while
-mlong-double-64 is in effect and not when -mlong-double-128 is in effect.
So, in a compiler that defaults to -mlong-double-64,
-mabi=ieeelongdouble
or
-mlong-double-128 -mlong-double-64 -mabi=ibmlongdouble
would error, but
-mlong-double-128 -mabi=ieeelongdouble
would not.
And in a compiler that defaults to -mlong-double128,
-mabi=ieeelongdouble
or
-mabi=ibmlongdouble
would be ok, but
-mlong-double-64 -mabi=ibmlongdouble
would not.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #5 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #4)
> And in a compiler that defaults to -mlong-double128,
> -mabi=ieeelongdouble
> or
> -mabi=ibmlongdouble
> would be ok, but
> -mlong-double-64 -mabi=ibmlongdouble
> would not.

But isn't this last case exactly the scenario that Florian mentions they have
in the glibc build issue?  Meaning they have a gcc that defaults to
-mlong-double-128 and they're building individual files using -mlong-double-64
-mabi=ibmlongdouble which you state should not be ok, but as it stands now,
it's already not ok and that is what is causing the build to error out.  Or am
I misunderstanding you?

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think the glibc problem is in a compiler defaulting to -mabi=ieeelongdouble
they want to use -mabi=ibmlongdouble -mlong-double-64 (because
-mabi=ibmlongdouble is added into CFLAGS everywhere early, and -mlong-double-64
is needed for a few TUs).
That wouldn't be diagnosed in the #c4 case, -mabi=ibmlongdouble is used while
-mlong-double-128 is in effect, so valid case.
-mlong-double-64 -mabi=ibmlongdouble
would be diagnosed though (-mabi=*double while -mlong-double-64 in effect).

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #7 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #6)
> I think the glibc problem is in a compiler defaulting to -mabi=ieeelongdouble
> they want to use -mabi=ibmlongdouble -mlong-double-64 (because
> -mabi=ibmlongdouble is added into CFLAGS everywhere early, and
> -mlong-double-64
> is needed for a few TUs).
> That wouldn't be diagnosed in the #c4 case, -mabi=ibmlongdouble is used while
> -mlong-double-128 is in effect, so valid case.
> -mlong-double-64 -mabi=ibmlongdouble
> would be diagnosed though (-mabi=*double while -mlong-double-64 in effect).

If that's the case, then I'll spend some time trying your suggestion.  Florian,
can you confirm that -mlong-double-64 comes after the -mabi=ibmlongdouble
option in the problematical glibc build?

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[fw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #8 from Florian Weimer <fw at gcc dot gnu.org> ---
(In reply to Peter Bergner from comment #7)
> Florian, can you confirm that -mlong-double-64 comes after the
> -mabi=ibmlongdouble option in the problematical glibc build?

The mailing list post referenced in comment 0 quotes the full command line.
Stripped to the vaguely ABI-ish flags, it's 

-std=gnu11 -fgnu89-inline -O2 -g -m64 -mabi=ibmlongdouble -frounding-math
-fmath-errno  -mno-gnu-attribute -mlong-double-64

So yes, it comes afterwards for this particular command line (and also in
general I think, but we'll know for sure only once we can get past this
particular error).

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #9 from Peter Bergner <bergner at gcc dot gnu.org> ---
Created attachment 52559
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52559&action=edit
Proposed patch

Attaching a patch using Jakub's suggestion.  I'm having problems posting it to
the mailing list at the moment, so I'm putting it here for people to look at,
comment on and try if they like.  I'll submit the patch once I can send email
again.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #10 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Peter Bergner <bergner@gcc.gnu.org>:

https://gcc.gnu.org/g:cb16bc3b5f34733ef9bbf8d2e3acacdecb099a62

commit r12-7485-gcb16bc3b5f34733ef9bbf8d2e3acacdecb099a62
Author: Peter Bergner <bergner@linux.ibm.com>
Date:   Fri Mar 4 09:03:44 2022 -0600

    rs6000: Allow -mlong-double-64 after -mabi={ibm,ieee}longdouble [PR104208,
PR87496]

    The glibc build is showing a build error due to extra "error" checking from
my
    PR87496 fix.  That checking was overeager, disallowing setting the long
double
    size to 64-bits if the 128-bit long double ABI had already been specified.
    Now we only emit an error if we specify a 128-bit long double ABI if our
    long double size is not 128 bits.  This also fixes an erroneous error when
    -mabi=ieeelongdouble is used and ISA 2.06 is not enabled, but the long
double
    size has been changed to 64 bits.

    2022-03-04  Peter Bergner  <bergner@linux.ibm.com>

    gcc/
            PR target/87496
            PR target/104208
            * config/rs6000/rs6000.cc (rs6000_option_override_internal): Make
the
            ISA 2.06 requirement for -mabi=ieeelongdouble conditional on
            -mlong-double-128.
            Move the -mabi=ieeelongdouble and -mabi=ibmlongdouble error
checking
            from here...
            * common/config/rs6000/rs6000-common.cc (rs6000_handle_option):
            ... to here.

    gcc/testsuite/
            PR target/87496
            PR target/104208
            * gcc.target/powerpc/pr104208-1.c: New test.
            * gcc.target/powerpc/pr104208-2.c: Likewise.
            * gcc.target/powerpc/pr87496-2.c: Swap long double options to
trigger
            the expected error.
            * gcc.target/powerpc/pr87496-3.c: Likewise.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #11 from Peter Bergner <bergner at gcc dot gnu.org> ---
I tested GCC 11 and it fails the same way, so we'll need to backport this fix
there.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #12 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to Peter Bergner from comment #11)
> I tested GCC 11 and it fails the same way, so we'll need to backport this
> fix there.

Ditto for GCC 10.

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[fw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #13 from Florian Weimer <fw at gcc dot gnu.org> ---
Thanks, I can confirm that we can build the glibc test suite once more in
Fedora rawhide.

(Fedora only needs the GCC 12 fix, it's our first GCC version with the float128
default).

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #14 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Peter Bergner
<bergner@gcc.gnu.org>:

https://gcc.gnu.org/g:51a34fdff2af99e2bb096436d6872c8ea434d177

commit r11-9668-g51a34fdff2af99e2bb096436d6872c8ea434d177
Author: Peter Bergner <bergner@linux.ibm.com>
Date:   Fri Mar 18 14:08:44 2022 -0500

    rs6000: Allow -mlong-double-64 after -mabi={ibm,ieee}longdouble [PR104208,
PR87496]

    The glibc build is showing a build error due to extra "error" checking from
my
    PR87496 fix.  That checking was overeager, disallowing setting the long
double
    size to 64-bits if the 128-bit long double ABI had already been specified.
    Now we only emit an error if we specify a 128-bit long double ABI if our
    long double size is not 128 bits.  This also fixes an erroneous error when
    -mabi=ieeelongdouble is used and ISA 2.06 is not enabled, but the long
double
    size has been changed to 64 bits.

    2022-03-04  Peter Bergner  <bergner@linux.ibm.com>

    gcc/
            PR target/87496
            PR target/104208
            * config/rs6000/rs6000.c (rs6000_option_override_internal): Make
the
            ISA 2.06 requirement for -mabi=ieeelongdouble conditional on
            -mlong-double-128.
            Move the -mabi=ieeelongdouble and -mabi=ibmlongdouble error
checking
            from here...
            * common/config/rs6000/rs6000-common.c (rs6000_handle_option):
            ... to here.

    gcc/testsuite/
            PR target/87496
            PR target/104208
            * gcc.target/powerpc/pr104208-1.c: New test.
            * gcc.target/powerpc/pr104208-2.c: Likewise.
            * gcc.target/powerpc/pr87496-2.c: Swap long double options to
trigger
            the expected error.
            * gcc.target/powerpc/pr87496-3.c: Likewise.

    (cherry picked from commit cb16bc3b5f34733ef9bbf8d2e3acacdecb099a62)

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

--- Comment #15 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Peter Bergner
<bergner@gcc.gnu.org>:

https://gcc.gnu.org/g:063ef668c716ae0bf9c0f79ae62243e89075b475

commit r10-10504-g063ef668c716ae0bf9c0f79ae62243e89075b475
Author: Peter Bergner <bergner@linux.ibm.com>
Date:   Fri Mar 18 14:09:26 2022 -0500

    rs6000: Allow -mlong-double-64 after -mabi={ibm,ieee}longdouble [PR104208,
PR87496]

    The glibc build is showing a build error due to extra "error" checking from
my
    PR87496 fix.  That checking was overeager, disallowing setting the long
double
    size to 64-bits if the 128-bit long double ABI had already been specified.
    Now we only emit an error if we specify a 128-bit long double ABI if our
    long double size is not 128 bits.  This also fixes an erroneous error when
    -mabi=ieeelongdouble is used and ISA 2.06 is not enabled, but the long
double
    size has been changed to 64 bits.

    2022-03-04  Peter Bergner  <bergner@linux.ibm.com>

    gcc/
            PR target/87496
            PR target/104208
            * config/rs6000/rs6000.c (rs6000_option_override_internal): Make
the
            ISA 2.06 requirement for -mabi=ieeelongdouble conditional on
            -mlong-double-128.
            Move the -mabi=ieeelongdouble and -mabi=ibmlongdouble error
checking
            from here...
            * common/config/rs6000/rs6000-common.c (rs6000_handle_option):
            ... to here.

    gcc/testsuite/
            PR target/87496
            PR target/104208
            * gcc.target/powerpc/pr104208-1.c: New test.
            * gcc.target/powerpc/pr104208-2.c: Likewise.
            * gcc.target/powerpc/pr87496-2.c: Swap long double options to
trigger
            the expected error.
            * gcc.target/powerpc/pr87496-3.c: Likewise.

    (cherry picked from commit cb16bc3b5f34733ef9bbf8d2e3acacdecb099a62)

[Bug target/104208] -mlong-double-64 should override a previous -mabi=ibmlongdouble

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104208

Peter Bergner <bergner at gcc dot gnu.org> changed:

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

--- Comment #16 from Peter Bergner <bergner at gcc dot gnu.org> ---
The fix was backported to GCC11 and GCC10.