[Bug bootstrap/102831] New: [12 regression] Spurious -Wparentheses etc. warnings

[Bug bootstrap/102831] New: [12 regression] Spurious -Wparentheses etc. warnings

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

            Bug ID: 102831
           Summary: [12 regression] Spurious -Wparentheses etc. warnings
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: bootstrap
          Assignee: unassigned at gcc dot gnu.org
          Reporter: ro at gcc dot gnu.org
  Target Milestone: ---
            Target: *-*-solaris2.11

Recently, Solaris bootstrap has repeatedly been broken in very confusing ways:

* 

/vol/gcc/src/hg/master/local/gcc/lra-constraints.c: In function 'bool
curr_insn_transform(bool)':
/vol/gcc/src/hg/master/local/gcc/lra-constraints.c:4367:21: error: suggest
parentheses around assignment used as truth value [-Werror=parentheses]
 4367 |                  (c = *constraint) && c != ',' && c != '#';
      |                  ~~~^~~~~~~~~~~~~~

  Solaris 11.4/SPARC, once 20211006, again 20211018

* 

/vol/gcc/src/hg/master/local/libgomp/oacc-init.c:876:1: error: optimization
attribute on ‘acc_on_device’ follows definition but the attribute doesn’t match
[-Werror=attributes]
  876 | {
      | ^
/vol/gcc/src/hg/master/local/libgomp/oacc-init.c:875:1: note: previous
definition of ‘acc_on_device’ with type ‘int(acc_device_t)’
  875 | acc_on_device (acc_device_t dev)
      | ^~~~~~~~~~~~~
/vol/gcc/src/hg/master/local/libgomp/oacc-init.c:954:1: internal compiler
error: in dwarf2out_register_main_translation_unit, at dwarf2out.c:28329
  954 | }
      | ^
0x109240f dwarf2out_register_main_translation_unit
        /vol/gcc/src/hg/master/local/gcc/dwarf2out.c:28328
0xc184b3 pop_scope()
        /vol/gcc/src/hg/master/local/gcc/c/c-decl.c:1217
0xc19963 pop_file_scope()
        /vol/gcc/src/hg/master/local/gcc/c/c-decl.c:1465
0xdf4be7 c_common_parse_file()
        /vol/gcc/src/hg/master/local/gcc/c-family/c-opts.c:1238

  Solaris 11.4/SPARC, 32 and 64-bit, 20211003

* 

In file included from /vol/gcc/src/hg/master/local/gcc/sparseset.c:23:
/vol/gcc/src/hg/master/local/gcc/sparseset.c: In function 'void
sparseset_ior(sparseset, sparseset, sparseset)':
/vol/gcc/src/hg/master/local/gcc/sparseset.h:215:20: error: suggest parentheses
around assignment used as truth value [-Werror=parentheses]
  215 |        && (((ITER) = sparseset_iter_elm (SPARSESET)) || 1);            
\
      |            ~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/vol/gcc/src/hg/master/local/gcc/sparseset.c:214:7: note: in expansion of macro
'EXECUTE_IF_SET_IN_SPARSESET'
  214 |       EXECUTE_IF_SET_IN_SPARSESET (b, e)
      |       ^~~~~~~~~~~~~~~~~~~~~~~~~~~

  Solaris 11.3, first SPARC, 20210924, then x86, 20211001, 20211015, probably
  more

  reghunts on sparc and x86 converged, but the results made no sense:

  sparc as of 20210924 identified

commit 4e7a92c0ff3871d955ca8fb133f869b216d7224d
Author: Richard Earnshaw <rearnsha@arm.com>
Date:   Tue Sep 21 11:14:46 2021 +0100

    arm: pass architecture extensions to assembler if supported

  x86 as of 20211001 identified

commit a11052d98db2f2a61841f0c5ee84de4ca1b3e296
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Tue Sep 28 12:35:29 2021 +0100

    libstdc++: Improve std::forward static assert message

This seems like some sort of memory corruption to me, but I'm a bit at a loss
on how to proceed.

When continuing the builds with adding -Wno-error=parentheses to the affected
files, the test results showed nothing unexpected.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[ro at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

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

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

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
     Ever confirmed|0                           |1
                 CC|                            |ebotcazou at gcc dot gnu.org,
                   |                            |msebor at gcc dot gnu.org
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2021-11-08

--- Comment #1 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
Yes, I have it this morning on the SPARC:

In file included from /homes/botcazou/gcc-head/src/gcc/sparseset.c:23:
/homes/botcazou/gcc-head/src/gcc/sparseset.c: In function 'void
sparseset_ior(sparseset, sparseset, sparseset)':
/homes/botcazou/gcc-head/src/gcc/sparseset.h:215:20: error: suggest parentheses
around assignment used as truth value [-Werror=parentheses]
  215 |        && (((ITER) = sparseset_iter_elm (SPARSESET)) || 1);            
\
      |            ~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/homes/botcazou/gcc-head/src/gcc/sparseset.c:214:7: note: in expansion of macro
'EXECUTE_IF_SET_IN_SPARSESET'
  214 |       EXECUTE_IF_SET_IN_SPARSESET (b, e)
      |       ^~~~~~~~~~~~~~~~~~~~~~~~~~~
cc1plus: all warnings being treated as errors
gmake[3]: *** [sparseset.o] Error 1
gmake[3]: *** Waiting for unfinished jobs....

See PR bootstrap/101574 for a similar report.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[tschwinge at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

Thomas Schwinge <tschwinge at gcc dot gnu.org> changed:

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

--- Comment #2 from Thomas Schwinge <tschwinge at gcc dot gnu.org> ---
(In reply to Rainer Orth from comment #0)
> Recently, Solaris bootstrap has repeatedly been broken in very confusing
> ways: [...]

(In reply to Eric Botcazou from comment #1)
> See PR bootstrap/101574 for a similar report.

Indeed.  But, hmm, that's unfortunate, as my commit
r12-3498-g6c79057fae6bbb36c4a4fd61c5b7107a16b71b17 "Don't maintain a warning
spec for 'UNKNOWN_LOCATION'/'BUILTINS_LOCATION' [PR101574]" has been meant to
generally resolve this issue (per my analysis/understanding) -- well, maybe it
did, and there really is another issue here?

Are you guys able to reliably reproduce the problem?  Asking because for me, it
was very flaky: some (seemingly random) change elsewhere in the compiler, and
PR101574 often didn't reproduce anymore (but came back later).  (But really
have not anymore seen it since the aforementioned commit.)

PR101204 and PR101292 may or may not be relevant, too.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #3 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #2 from Thomas Schwinge <tschwinge at gcc dot gnu.org> ---
> Are you guys able to reliably reproduce the problem?  Asking because for me, it
> was very flaky: some (seemingly random) change elsewhere in the compiler, and
> PR101574 often didn't reproduce anymore (but came back later).  (But really
> have not anymore seen it since the aforementioned commit.)

In my case the errors were anything but reliable: they started to show on
Solaris 11.3 builds only, sometimes sparc, sometimes x86, at others
(from the identical source tree) Solaris 11.4/SPARC only.

In the end (not having made progress to figure out what's wrong), I
ended  up with a local patch to disable -Wparentheses for
lra-constraints.c and sparseset.c as those failures didn't have any
other effect on the build or testsuite results, but just had become a
nuisance.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #4 from Martin Sebor <msebor at gcc dot gnu.org> ---
I have been testing the following changes to deal with other location and
warning related problems.  They might be worth giving a try to see if they help
with this issue as well. 

diff --git a/gcc/diagnostic-spec.c b/gcc/diagnostic-spec.c
index 85ffb725c02..09fd42bcf53 100644
--- a/gcc/diagnostic-spec.c
+++ b/gcc/diagnostic-spec.c
@@ -186,11 +186,8 @@ copy_warning (location_t to, location_t from)
     /* We cannot set no-warning dispositions for 'to', so we have no chance
but
        lose those potentially set for 'from'.  */
     ;
-  else
-    {
-      if (from_spec)
-       nowarn_map->put (to, *from_spec);
-      else
-       nowarn_map->remove (to);
-    }
+  else if (from_spec)
+    nowarn_map->put (to, *from_spec);
+  /* ...else keep the entry since it may be referenced by multiple trees
+     or GIMPLE statements at this location.  */
 }
diff --git a/gcc/warning-control.cc b/gcc/warning-control.cc
index 36a47ab6bae..e490bf28b24 100644
--- a/gcc/warning-control.cc
+++ b/gcc/warning-control.cc
@@ -198,21 +198,13 @@ void copy_warning (ToType to, FromType from)
     /* We cannot set no-warning dispositions for 'to', so we have no chance
but
        lose those potentially set for 'from'.  */
     ;
-  else
+  else if (from_spec)
     {
-      if (from_spec)
-       {
-         /* If there's an entry in the map the no-warning bit must be set.  */
-         gcc_assert (supp);
-
-         gcc_checking_assert (nowarn_map);
-         nowarn_map->put (to_loc, *from_spec);
-       }
-      else
-       {
-         if (nowarn_map)
-           nowarn_map->remove (to_loc);
-       }
+      /* If there's an entry in the map the no-warning bit must be set.  */
+      gcc_assert (supp);
+
+      gcc_checking_assert (nowarn_map);
+      nowarn_map->put (to_loc, *from_spec);
     }

   /* The no-warning bit might be set even if the map has not been consulted,
or

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bergner at gcc dot gnu.org
             Target|*-*-solaris2.11             |*-*-solaris2.11
                   |                            |powerpc*-linux

--- Comment #5 from Peter Bergner <bergner at gcc dot gnu.org> ---
I see the same error in Comment #3 on my powerpc64le-linux build, which breaks
my bootstrap too.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #6 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to Peter Bergner from comment #5)
> I see the same error in Comment #3 on my powerpc64le-linux build, which
> breaks my bootstrap too.

Hmmm, if I manually execute the failing compile command, it ICEs, but if I add
-save-temps to try and get a preprocessed source file, it compiles with no
errors.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[msebor at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #7 from Martin Sebor <msebor at gcc dot gnu.org> ---
(In reply to Peter Bergner from comment #6)

If it's reliably reproducible (ideally with a cross), can you attach a
translation unit and the options to use to reproduce it with?

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #8 from Peter Bergner <bergner at gcc dot gnu.org> ---
(In reply to Martin Sebor from comment #7)
> (In reply to Peter Bergner from comment #6)
> 
> If it's reliably reproducible (ideally with a cross), can you attach a
> translation unit and the options to use to reproduce it with?

It's somewhat repeatable. :-(  If I cd into $BUILD_DIR/gcc and execute the
command from my build log that ICEs, that will 100% always ICE.  However, if I
add -save-temps it will not ICE and adding -E and compiling the resulting .ii
file will not ICE either.  In addition, if I cd to my home dir and execute the
failing command line, with only changing the -I. options to -I/$BUILD_DIR/gcc,
then it also does not ICE.  It seems only executing the command in the correct
build directory seems to work.  I have been able to recreate the issue on the
gcc compile farm machine gcc135.  You can recreate the ICE with the following
command.  I believe you should be able to access that directory.

[bergner@gcc135 gcc]$ pwd
/home/bergner/gcc/build/gcc-fsf-mainline-pr103127-base-regtest-2/gcc

[bergner@gcc135 gcc]$
/home/bergner/gcc/build/gcc-fsf-mainline-pr103127-base-regtest-2/./prev-gcc/xg++
-B/home/bergner/gcc/build/gcc-fsf-mainline-pr103127-base-regtest-2/./prev-gcc/
-nostdinc++
-I/home/bergner/gcc/build/gcc-fsf-mainline-pr103127-base-regtest-2/prev-powerpc64le-linux/libstdc++-v3/include/powerpc64le-linux

-I/home/bergner/gcc/build/gcc-fsf-mainline-pr103127-base-regtest-2/prev-powerpc64le-linux/libstdc++-v3/include
 -I/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/libstdc++-v3/libsupc++
-fno-PIE -S -g -O2 -fno-checking -gtoggle -DIN_GCC     -fno-exceptions
-fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings
-Wcast-qual -Wno-error=format-diag -Wmissing-format-attribute
-Woverloaded-virtual -pedantic -Wno-long-long -Wno-variadic-macros
-Wno-overlength-strings -Werror -fno-common  -DHAVE_CONFIG_H -I./
-I/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/.
-I/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/../include
-I/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/../libcpp/include 
/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/lra-lives.c

In file included from
/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/lra-lives.c:43:
/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/lra-lives.c: In function
‘bool process_bb_lives(basic_block, int&, bool)’:
/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/sparseset.h:215:20: error:
suggest parentheses around assignment used as truth value [-Werror=parentheses]
  215 |        && (((ITER) = sparseset_iter_elm (SPARSESET)) || 1);            
\
      |            ~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/bergner/gcc/gcc-fsf-mainline-pr103127-base/gcc/lra-lives.c:1033:7: note:
in expansion of macro ‘EXECUTE_IF_SET_IN_SPARSESET’
 1033 |       EXECUTE_IF_SET_IN_SPARSESET (unused_set, j)
      |       ^~~~~~~~~~~~~~~~~~~~~~~~~~~

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[tschwinge at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

Thomas Schwinge <tschwinge at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |WAITING
                 CC|                            |rguenth at gcc dot gnu.org

--- Comment #9 from Thomas Schwinge <tschwinge at gcc dot gnu.org> ---
Please check if this happens to be resolved with Richard's recent PR101292
commit r12-6640-g1374d4b963a6ac2e0ec1645c09e5162e68b009d6 "middle-end/101292 -
invalid memory access with warning control".

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

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

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

--- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
Assuming fixed.

[Bug bootstrap/102831] [12 regression] Spurious -Wparentheses etc. warnings

[ro at CeBiTec dot Uni-Bielefeld.DE]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102831

--- Comment #11 from ro at CeBiTec dot Uni-Bielefeld.DE <ro at CeBiTec dot Uni-Bielefeld.DE> ---
> --- Comment #10 from Richard Biener <rguenth at gcc dot gnu.org> ---
> Assuming fixed.

I can't tell for certain because I had a local patch in my tree
selectively disabling the warnings to avoid random bootstrap failures.
I've now removed it to see if I still hit this.