[Bug target/103696] New: Lambda functions are not inlined under certain optimization pragmas

[Bug target/103696] New: Lambda functions are not inlined under certain optimization pragmas

[imachug at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

            Bug ID: 103696
           Summary: Lambda functions are not inlined under certain
                    optimization pragmas
           Product: gcc
           Version: 11.2.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: imachug at gmail dot com
  Target Milestone: ---

This seems like a very weird bug to me and I'm not even sure how to label it,
so please fix the component if needed.

Testcase (C++):


#pragma GCC optimize("finite-math-only")
#pragma GCC target("sse3")

void fn() {
}

int global_var;

int solve() {
    auto nested = []() {
        return global_var;
    };
    return nested();
}


When compiling this code via `g++ test.cpp -c -O2 -std=c++17`, I get the
following assembly:


$ objdump -d test.o
...
0000000000000000 <_ZZ5solvevENKUlvE_clEv.constprop.0>:
   0:   8b 05 00 00 00 00       mov    0x0(%rip),%eax        # 6
<_ZZ5solvevENKUlvE_clEv.constprop.0+0x6>
   6:   c3                      retq   
   7:   66 0f 1f 84 00 00 00    nopw   0x0(%rax,%rax,1)
   e:   00 00 
...
0000000000000020 <_Z5solvev>:
  20:   f3 0f 1e fa             endbr64 
  24:   e8 d7 ff ff ff          callq  0 <_ZZ5solvevENKUlvE_clEv.constprop.0>
  29:   c3                      retq   


As you can see, the nested() lambda call was not inlined into solve().

However, if I do any of the following, the lambda is inlined as expected:

- Remove `fn` definition
- Move `fn` definition under `solve`
- Replace reading `global_var` with a constant
- Make `nested` a global function
- Remove either of the two pragmas (or both)
- Add -ffinite-math-only or -msse3 or both to the compilation line (regardless
of whether the pragmas are still there)

I have absolutely no idea why a floating point optimization affects inlining or
how a pragma is different from a compilation line option wrt. this bug.

[Bug target/103696] Lambda functions are not inlined under certain optimization pragmas

[imachug at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

--- Comment #1 from Ivan Machugovskiy <imachug at gmail dot com> ---
Obligatory info dump. I managed to reproduce this on G++ 9.3.0 and G++ 10.3.0
locally, and on G++ trunk on Godbolt (see https://godbolt.org/z/Y5Kr3KfjW).
This is probably a longstanding bug.


$ g++ -v
Using built-in specs.
COLLECT_GCC=g++
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/9/lto-wrapper
OFFLOAD_TARGET_NAMES=nvptx-none:hsa
OFFLOAD_TARGET_DEFAULT=1
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu
9.3.0-17ubuntu1~20.04' --with-bugurl=file:///usr/share/doc/gcc-9/README.Bugs
--enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++,gm2 --prefix=/usr
--with-gcc-major-version-only --program-suffix=-9
--program-prefix=x86_64-linux-gnu- --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--libdir=/usr/lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug
--enable-libstdcxx-time=yes --with-default-libstdcxx-abi=new
--enable-gnu-unique-object --disable-vtable-verify --enable-plugin
--enable-default-pie --with-system-zlib --with-target-system-zlib=auto
--enable-objc-gc=auto --enable-multiarch --disable-werror --with-arch-32=i686
--with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib
--with-tune=generic
--enable-offload-targets=nvptx-none=/build/gcc-9-HskZEa/gcc-9-9.3.0/debian/tmp-nvptx/usr,hsa
--without-cuda-driver --enable-checking=release --build=x86_64-linux-gnu
--host=x86_64-linux-gnu --target=x86_64-linux-gnu
Thread model: posix
gcc version 9.3.0 (Ubuntu 9.3.0-17ubuntu1~20.04)

$ g++-10 -v
Using built-in specs.
COLLECT_GCC=g++-10
COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/10/lto-wrapper
OFFLOAD_TARGET_NAMES=nvptx-none:amdgcn-amdhsa:hsa
OFFLOAD_TARGET_DEFAULT=1
Target: x86_64-linux-gnu
Configured with: ../src/configure -v --with-pkgversion='Ubuntu
10.3.0-1ubuntu1~20.04' --with-bugurl=file:///usr/share/doc/gcc-10/README.Bugs
--enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++,m2 --prefix=/usr
--with-gcc-major-version-only --program-suffix=-10
--program-prefix=x86_64-linux-gnu- --enable-shared --enable-linker-build-id
--libexecdir=/usr/lib --without-included-gettext --enable-threads=posix
--libdir=/usr/lib --enable-nls --enable-bootstrap --enable-clocale=gnu
--enable-libstdcxx-debug --enable-libstdcxx-time=yes
--with-default-libstdcxx-abi=new --enable-gnu-unique-object
--disable-vtable-verify --enable-plugin --enable-default-pie --with-system-zlib
--enable-libphobos-checking=release --with-target-system-zlib=auto
--enable-objc-gc=auto --enable-multiarch --disable-werror --with-arch-32=i686
--with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib
--with-tune=generic
--enable-offload-targets=nvptx-none=/build/gcc-10-S4I5Pr/gcc-10-10.3.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-10-S4I5Pr/gcc-10-10.3.0/debian/tmp-gcn/usr,hsa
--without-cuda-driver --enable-checking=release --build=x86_64-linux-gnu
--host=x86_64-linux-gnu --target=x86_64-linux-gnu
--with-build-config=bootstrap-lto-lean --enable-link-mutex
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 10.3.0 (Ubuntu 10.3.0-1ubuntu1~20.04)

[Bug c++/103696] pragma optimization is not applying to Lambdas

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2021-12-14
            Summary|Lambda functions are not    |pragma optimization is not
                   |inlined under certain       |applying to Lambdas
                   |optimization pragmas        |
           Keywords|                            |wrong-code
             Blocks|                            |54367
          Component|target                      |c++
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
If I manually apply the attributes rather than use the pragma, I get the
correct results.

that is the following works:
int global_var;

[[gnu::optimize("finite-math-only"),gnu::target("sse3")]]
int solve() {
    auto nested = 
     []
     [[gnu::optimize("finite-math-only"),gnu::target("sse")]]
     () {
        return global_var;
    };
    return nested();
}


Referenced Bugs:

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=54367
[Bug 54367] [meta-bug] lambda expressions

[Bug c++/103696] pragma optimization is not applying to Lambdas

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |marxin at gcc dot gnu.org
             Status|NEW                         |ASSIGNED
                 CC|                            |marxin at gcc dot gnu.org

--- Comment #3 from Martin Liška <marxin at gcc dot gnu.org> ---
Confirmed, debugging that right now.

[Bug c++/103696] pragma optimization is not applying to Lambdas

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

Martin Liška <marxin at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch

--- Comment #4 from Martin Liška <marxin at gcc dot gnu.org> ---
All right, I've got a patch candidate:
https://gcc.gnu.org/pipermail/gcc-patches/2021-December/586999.html

[Bug c++/103696] pragma optimization is not applying to Lambdas

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Martin Liska <marxin@gcc.gnu.org>:

https://gcc.gnu.org/g:5b8f5a50a93c5fd44fa1ecb6f431fee1df267571

commit r12-6018-g5b8f5a50a93c5fd44fa1ecb6f431fee1df267571
Author: Martin Liska <mliska@suse.cz>
Date:   Wed Dec 15 17:27:56 2021 +0100

    pragma: respect pragma in lambda functions

    In g:01ad8c54fdca we started supporting target pragma changes
    that are primarily caused by optimization option. The same can happen
    in the opposite way and we need to check for changes both
    in optimization_current_node and target_option_current_node.

            PR c++/103696

    gcc/ChangeLog:

            * attribs.c (decl_attributes): Check if
            target_option_current_node is changed.

    gcc/testsuite/ChangeLog:

            * g++.target/i386/pr103696.C: New test.

[Bug c++/103696] pragma optimization is not applying to Lambdas

[marxin at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

Martin Liška <marxin at gcc dot gnu.org> changed:

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

--- Comment #6 from Martin Liška <marxin at gcc dot gnu.org> ---
Fixed on master, not planning to backport as it depends on quite some
option-related patches that were done in GCC 12 timeframe.

[Bug c++/103696] pragma optimization is not applying to Lambdas

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103696

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

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