[Bug target/104829] New: Pure 32-bit PowerPC build broken

[Bug target/104829] New: Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

            Bug ID: 104829
           Summary: Pure 32-bit PowerPC build broken
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: segher at gcc dot gnu.org
  Target Milestone: ---

In <https://gcc.gnu.org/pipermail/gcc-patches/2022-March/591322.html> Joseph
reports that I have broken the default build for powerpc-linux.  Whoops.

[Bug target/104829] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Segher Boessenkool <segher at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Target|                            |powerpc-linux
           Assignee|unassigned at gcc dot gnu.org      |segher at gcc dot gnu.org
           Priority|P3                          |P1

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|Pure 32-bit PowerPC build   |[12 Regression] Pure 32-bit
                   |broken                      |PowerPC build broken
   Target Milestone|---                         |12.0

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-03-07
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |ASSIGNED

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

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

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

--- Comment #2 from Peter Bergner <bergner at gcc dot gnu.org> ---
I did a 32-bit build using top of tree sources for gcc and binutils and cannot
recreate the reported error.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #3 from Segher Boessenkool <segher at gcc dot gnu.org> ---
I cannot reproduce it either.  The machine instruction that gives the error is
lfiwzx, a power7 insn; GCC will not generate this instruction unless you are
compiling for power7 or later.  That is not the default of course.  Building
with --with-cpu=power7 does not reproduce it either.

Help?

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[sebastian.huber@embedded-brains.de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Sebastian Huber <sebastian.huber@embedded-brains.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |sebastian.huber@embedded-br
                   |                            |ains.de

--- Comment #4 from Sebastian Huber <sebastian.huber@embedded-brains.de> ---
Maybe you can reproduce it with Binutils 2.38. There was a recent change in
Binutils which could prevent the error:

commit cebc89b9328eab994f6b0314c263f94e7949a553
Author: Alan Modra <amodra@gmail.com>
Date:   Mon Feb 21 10:58:57 2022 +1030

    binutils 2.38 vs. ppc32 linux kernel

    Commit b25f942e18d6 made .machine more strict.  Weaken it again.

            * config/tc-ppc.c (ppc_machine): Treat an early .machine specially,
            keeping sticky options to work around gcc bugs.

diff --git a/gas/config/tc-ppc.c b/gas/config/tc-ppc.c
index 054f9c72161..89bc7d3f9b9 100644
--- a/gas/config/tc-ppc.c
+++ b/gas/config/tc-ppc.c
@@ -5965,7 +5965,30 @@ ppc_machine (int ignore ATTRIBUTE_UNUSED)
             options do not count as a new machine, instead they add
             to currently selected opcodes.  */
          ppc_cpu_t machine_sticky = 0;
-         new_cpu = ppc_parse_cpu (ppc_cpu, &machine_sticky, cpu_string);
+         /* Unfortunately, some versions of gcc emit a .machine
+            directive very near the start of the compiler's assembly
+            output file.  This is bad because it overrides user -Wa
+            cpu selection.  Worse, there are versions of gcc that
+            emit the *wrong* cpu, not even respecting the -mcpu given
+            to gcc.  See gcc pr101393.  And to compound the problem,
+            as of 20220222 gcc doesn't pass the correct cpu option to
+            gas on the command line.  See gcc pr59828.  Hack around
+            this by keeping sticky options for an early .machine.  */
+         asection *sec;
+         for (sec = stdoutput->sections; sec != NULL; sec = sec->next)
+           {
+             segment_info_type *info = seg_info (sec);
+             /* Are the frags for this section perturbed from their
+                initial state?  Even .align will count here.  */
+             if (info != NULL
+                 && (info->frchainP->frch_root != info->frchainP->frch_last
+                     || info->frchainP->frch_root->fr_type != rs_fill
+                     || info->frchainP->frch_root->fr_fix != 0))
+               break;
+           }
+         new_cpu = ppc_parse_cpu (ppc_cpu,
+                                  sec == NULL ? &sticky : &machine_sticky,
+                                  cpu_string);
          if (new_cpu != 0)
            ppc_cpu = new_cpu;
          else

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #5 from Segher Boessenkool <segher at gcc dot gnu.org> ---
No difference with binutils-2.38 .

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #6 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
The compilation command is:

/scratch/jmyers/glibc/many12/build/compilers/powerpc-linux-gnu/gcc/./gcc/xgcc 
-B/scratch/jmyers/glibc/many12/build/compilers/powerpc-linux-gnu/gcc/./gcc/ 
-B/scratch/jmyers/glibc/many12/install/compilers/powerpc-linux-gnu/powerpc-glibc-linux-gnu/bin/ 
-B/scratch/jmyers/glibc/many12/install/compilers/powerpc-linux-gnu/powerpc-glibc-linux-gnu/lib/ 
-isystem 
/scratch/jmyers/glibc/many12/install/compilers/powerpc-linux-gnu/powerpc-glibc-linux-gnu/include 
-isystem 
/scratch/jmyers/glibc/many12/install/compilers/powerpc-linux-gnu/powerpc-glibc-linux-gnu/sys-include 
-g -O2 -O2  -g -O2 -DIN_GCC  -DCROSS_DIRECTORY_STRUCTURE  -W -Wall 
-Wno-narrowing -Wwrite-strings -Wcast-qual -Wstrict-prototypes 
-Wmissing-prototypes -Wold-style-definition  -isystem ./include   -fPIC 
-mlong-double-128 -mno-minimal-toc -g -DIN_LIBGCC2 -fbuilding-libgcc 
-fno-stack-protector   -fPIC -mlong-double-128 -mno-minimal-toc -I. -I. 
-I../.././gcc -I/scratch/jmyers/glibc/many12/src/gcc/libgcc 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/. 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/../gcc 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/../include 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/../libdecnumber/dpd 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/../libdecnumber 
-DHAVE_CC_TLS  -Wno-type-limits -mvsx -mfloat128 -mno-float128-hardware 
-mno-gnu-attribute -I/scratch/jmyers/glibc/many12/src/gcc/libgcc/soft-fp 
-I/scratch/jmyers/glibc/many12/src/gcc/libgcc/config/rs6000  
-mno-gnu-attribute -Wno-psabi -mabi=ieeelongdouble  -o _kf_to_sd.o -MT 
_kf_to_sd.o -MD -MP -MF _kf_to_sd.dep  -c 
/scratch/jmyers/glibc/many12/src/gcc/libgcc/config/rs6000/_kf_to_sd.c 
-fvisibility=hidden -DHIDE_EXPORTS

The generated .s file has ".machine ppc".  Maybe there is some 
inconsistency arising from the use of -mvsx -mfloat128 for this 32-bit 
configuration with no --with option to change the default CPU?  Should 
these KFmode functions really be being built for 32-bit libgcc at all?

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #7 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to joseph@codesourcery.com from comment #6)
> The compilation command is:

Can you attach the _kf_to_sd.i preprocessed file from your command so I can
more easily try your exact compile options without having all of your -B and -I
options pointing at your glibc build?

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[jsm28 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #8 from Joseph S. Myers <jsm28 at gcc dot gnu.org> ---
Created attachment 52597
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52597&action=edit
preprocessed source

Preprocessed source attached.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[jsm28 at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #9 from Joseph S. Myers <jsm28 at gcc dot gnu.org> ---
Created attachment 52598
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52598&action=edit
generated assembly

Assembly output attached.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #10 from Segher Boessenkool <segher at gcc dot gnu.org> ---
(In reply to joseph@codesourcery.com from comment #6)
> The generated .s file has ".machine ppc".  Maybe there is some 
> inconsistency arising from the use of -mvsx -mfloat128 for this 32-bit 
> configuration with no --with option to change the default CPU?  Should 
> these KFmode functions really be being built for 32-bit libgcc at all?

Using -mvsx with a CPU that does not support VSX insns is problematic.

All this code should use -mcpu=power7.

But, I'll make a patch for rs6000_machine_from_flags to make this work
the way it did before, I now know what the problem is.  Thanks!

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #11 from Segher Boessenkool <segher at gcc dot gnu.org> ---
Created attachment 52599
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52599&action=edit
proposed patch

This patch should restore the previous behaviour.  Joseph, can you test it
please?

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #12 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
I still get the same error (and the same ".machine ppc") with that patch 
applied.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #13 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to joseph@codesourcery.com from comment #12)
> I still get the same error (and the same ".machine ppc") with that patch 
> applied.

Confirmed, the patch doesn't fix the issue for me either:

[bergner@makalu-lp1 PR104829]$ cat bug.i 
_Decimal32
foo (_Decimal32 *f)
{
  return *f;
}
[bergner@makalu-lp1 PR104829]$
/home/bergner/gcc/build/gcc-fsf-mainline-32bit/gcc/xgcc
-B/home/bergner/gcc/build/gcc-fsf-mainline-32bit/gcc -O2 -mcpu=powerpc -mvsx -c
bug.i 
/tmp/cc1CLPZR.s: Assembler messages:
/tmp/cc1CLPZR.s:10: Error: unrecognized opcode: `lfiwzx'

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Segher Boessenkool <segher at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
  Attachment #52599|0                           |1
        is obsolete|                            |

--- Comment #14 from Segher Boessenkool <segher at gcc dot gnu.org> ---
Created attachment 52605
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=52605&action=edit
proposed patch v2

I added this some time ago, but it disappeared again?

This restores more of the previous not-so-good behaviour.  "Old bugs are better
than new".

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #15 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
I confirm that the second patch does fix the problem I see.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #16 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Segher Boessenkool <segher@gcc.gnu.org>:

https://gcc.gnu.org/g:80fcc4b6afee72443bef551064826b3b4b6785e6

commit r12-7628-g80fcc4b6afee72443bef551064826b3b4b6785e6
Author: Segher Boessenkool <segher@kernel.crashing.org>
Date:   Fri Mar 11 21:15:18 2022 +0000

    rs6000: Do not use rs6000_cpu for .machine ppc and ppc64 (PR104829)

    Fixes: 77eccbf39ed5

    rs6000.h has
      #define PROCESSOR_POWERPC   PROCESSOR_PPC604
      #define PROCESSOR_POWERPC64 PROCESSOR_RS64A
    which means that if you use things like  -mcpu=powerpc -mvsx  it will no
    longer work after my latest .machine patch.  This causes GCC build errors
    in some cases, not a good idea (even if the errors are actually
    pre-existing: using -mvsx with a machine that does not have VSX cannot
    work properly).

    2022-03-11  Segher Boessenkool  <segher@kernel.crashing.org>

            PR target/104829
            * config/rs6000/rs6000.cc (rs6000_machine_from_flags): Don't output
            "ppc" and "ppc64" based on rs6000_cpu.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[segher at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

Segher Boessenkool <segher at gcc dot gnu.org> changed:

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

--- Comment #17 from Segher Boessenkool <segher at gcc dot gnu.org> ---
Fixed.

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #18 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Sebastian Huber
<sh@gcc.gnu.org>:

https://gcc.gnu.org/g:58465d673bb21fc234f642166ec0cf9e621787c3

commit r10-10859-g58465d673bb21fc234f642166ec0cf9e621787c3
Author: Segher Boessenkool <segher@kernel.crashing.org>
Date:   Fri Mar 11 21:15:18 2022 +0000

    rs6000: Do not use rs6000_cpu for .machine ppc and ppc64 (PR104829)

    Fixes: 77eccbf39ed5

    rs6000.h has
      #define PROCESSOR_POWERPC   PROCESSOR_PPC604
      #define PROCESSOR_POWERPC64 PROCESSOR_RS64A
    which means that if you use things like  -mcpu=powerpc -mvsx  it will no
    longer work after my latest .machine patch.  This causes GCC build errors
    in some cases, not a good idea (even if the errors are actually
    pre-existing: using -mvsx with a machine that does not have VSX cannot
    work properly).

    2022-03-11  Segher Boessenkool  <segher@kernel.crashing.org>

            PR target/104829
            * config/rs6000/rs6000.c (rs6000_machine_from_flags): Don't output
            "ppc" and "ppc64" based on rs6000_cpu.

    (cherry picked from commit 80fcc4b6afee72443bef551064826b3b4b6785e6)

[Bug target/104829] [12 Regression] Pure 32-bit PowerPC build broken

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=104829

--- Comment #19 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Sebastian Huber
<sh@gcc.gnu.org>:

https://gcc.gnu.org/g:d87e0e297b1cba73a0c055d2a3e9267d288f435a

commit r11-10084-gd87e0e297b1cba73a0c055d2a3e9267d288f435a
Author: Segher Boessenkool <segher@kernel.crashing.org>
Date:   Fri Mar 11 21:15:18 2022 +0000

    rs6000: Do not use rs6000_cpu for .machine ppc and ppc64 (PR104829)

    Fixes: 77eccbf39ed5

    rs6000.h has
      #define PROCESSOR_POWERPC   PROCESSOR_PPC604
      #define PROCESSOR_POWERPC64 PROCESSOR_RS64A
    which means that if you use things like  -mcpu=powerpc -mvsx  it will no
    longer work after my latest .machine patch.  This causes GCC build errors
    in some cases, not a good idea (even if the errors are actually
    pre-existing: using -mvsx with a machine that does not have VSX cannot
    work properly).

    2022-03-11  Segher Boessenkool  <segher@kernel.crashing.org>

            PR target/104829
            * config/rs6000/rs6000.c (rs6000_machine_from_flags): Don't output
            "ppc" and "ppc64" based on rs6000_cpu.

    (cherry picked from commit 80fcc4b6afee72443bef551064826b3b4b6785e6)