[Bug c++/107466] New: [12 Regression] invalid -Wnarrowing error

[Bug c++/107466] New: [12 Regression] invalid -Wnarrowing error

[littlefox+gcc-bugzilla@lf-net.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

            Bug ID: 107466
           Summary: [12 Regression] invalid -Wnarrowing error
           Product: gcc
           Version: 12.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: littlefox+gcc-bugzilla@lf-net.org
  Target Milestone: ---

Created attachment 53797
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=53797&action=edit
Code triggering the error

Hi,

the attached code (and Godbolt[1]) fails in gcc 12+ with an -Wnarrowing error
in a stdlibc++ template instantiated with an uint16_t and apparently trying to
work with a 2^31:

/opt/compiler-explorer/gcc-12.2.0/include/c++/12.2.0/bits/random.tcc:545:9:
error: narrowing conversion of '2147483563' from 'unsigned int' to 'short
unsigned int' [-Wnarrowing]
  545 |         __lcg(__value == 0u ? default_seed : __value);

The code works fine in 11.3 and clang (14) and since it also works in clang
using libstdc++ 12.2, this error is probably in gcc itself and not in
libstdc++.

Tested this on my machine (debian testing), where it broke after a some system
updates where installed - including a gcc update. Then tested this in Godbolt
to check compiler versions and other configurations and asked in
libera.chat/#c++ for others to verify if my code makes sense - which it seems
to do.

Best, Mara

[1] Godbolt link: https://godbolt.org/z/sj18Mv5j9

[Bug c++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Shorter testcase (still includes):
#include <random>

void f() {
    std::subtract_with_carry_engine<unsigned short, 12, 5, 12>();
}

[Bug c++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #2 from Jonathan Wakely <redi at gcc dot gnu.org> ---
I don't think this is a compiler bug. Converting that value to uint16_t is
narrowing.

The problem is that the standard requires subtract_with_carry_engine<T> to use
linear_congruential_engine<T, 40014u, 0u, 2147483563u> where each of those
values is converted to T.

This appears to mean subtract_with_carry_engine cannot be used with uint16_t.

[Bug c++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[littlefox+gcc-bugzilla@lf-net.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #3 from Mara Sophie Grosch <littlefox+gcc-bugzilla@lf-net.org> ---
Standard explicitly allows using unsigned short though:
https://en.cppreference.com/w/cpp/numeric/random/subtract_with_carry_engine

[Bug libstdc++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|                            |2022-11-04
             Status|UNCONFIRMED                 |NEW
          Component|c++                         |libstdc++
     Ever confirmed|0                           |1

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> ---
That's not the standard. That text comes from a general requirement in
[rand.req.gen] that applies to all engines and adaptors. The more specific
requirements for subtract_with_carry_engine say it uses a specialization of
linear_congruential_engine that is ill-formed. More specific requirements
overrule more general ones.

I've created a new LWG issue so this will get fixed in the standard. The best
suggestion so far is that it should use
linear_congruential_engine<uint_least32_t,...> instead of using its own
result_type.

Reassigning to component:libstdc++.

[Bug libstdc++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |12.3

[Bug libstdc++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #5 from Jonathan Wakely <redi at gcc dot gnu.org> ---
https://cplusplus.github.io/LWG/issue3809

[Bug libstdc++/107466] [12/13 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #6 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jonathan Wakely <redi@gcc.gnu.org>:

https://gcc.gnu.org/g:a64775a0edd46980036b757041f0c065ed9f8d22

commit r13-4364-ga64775a0edd46980036b757041f0c065ed9f8d22
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Mon Nov 28 09:44:52 2022 +0000

    libstdc++: Make 16-bit std::subtract_with_carry_engine work [PR107466]

    This implements the proposed resolution of LWG 3809, so that
    std::subtract_with_carry_engine can be used with a 16-bit result_type.
    Currently this produces a narrowing error when instantiating the
    std::linear_congruential_engine to create the initial state. It also
    truncates the default_seed constant when passing it as a result_type
    argument.

    Change the type of the constant to uint_least32_t and pass 0u when the
    default_seed should be used.

    libstdc++-v3/ChangeLog:

            PR libstdc++/107466
            * include/bits/random.h (subtract_with_carry_engine): Use 32-bit
            type for default seed. Use 0u as default argument for
            subtract_with_carry_engine(result_type) constructor and
            seed(result_type) member function.
            * include/bits/random.tcc (subtract_with_carry_engine): Use
            32-bit type for default seed and engine used for initial state.
            *
testsuite/26_numerics/random/subtract_with_carry_engine/cons/lwg3809.cc:
            New test.

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

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

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

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #7 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Jonathan Wakely
<redi@gcc.gnu.org>:

https://gcc.gnu.org/g:d8def0b55572fc5bc02a0fee6a98787cd50ee885

commit r12-9485-gd8def0b55572fc5bc02a0fee6a98787cd50ee885
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Mon Nov 28 09:44:52 2022 +0000

    libstdc++: Make 16-bit std::subtract_with_carry_engine work [PR107466]

    This implements the proposed resolution of LWG 3809, so that
    std::subtract_with_carry_engine can be used with a 16-bit result_type.
    Currently this produces a narrowing error when instantiating the
    std::linear_congruential_engine to create the initial state. It also
    truncates the default_seed constant when passing it as a result_type
    argument.

    Change the type of the constant to uint_least32_t and pass 0u when the
    default_seed should be used.

    libstdc++-v3/ChangeLog:

            PR libstdc++/107466
            * include/bits/random.h (subtract_with_carry_engine): Use 32-bit
            type for default seed. Use 0u as default argument for
            subtract_with_carry_engine(result_type) constructor and
            seed(result_type) member function.
            * include/bits/random.tcc (subtract_with_carry_engine): Use
            32-bit type for default seed and engine used for initial state.
            *
testsuite/26_numerics/random/subtract_with_carry_engine/cons/lwg3809.cc:
            New test.

    (cherry picked from commit a64775a0edd46980036b757041f0c065ed9f8d22)

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

Jonathan Wakely <redi at gcc dot gnu.org> changed:

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

--- Comment #8 from Jonathan Wakely <redi at gcc dot gnu.org> ---
Fixed for 12.3

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

Jonathan Wakely <redi at gcc dot gnu.org> changed:

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

--- Comment #9 from Jonathan Wakely <redi at gcc dot gnu.org> ---
I'm reopening this because the fix changes the behaviour of some code:

https://cplusplus.github.io/LWG/issue4014

As noted there, std::ranlux48_base(UINT_MAX+1LL)() changed behaviour.

Originally that yielded 22575453646312 but after the commits above it yields
23223501020940

This is what I proposed for 4014 and restores the original behaviour:

--- a/libstdc++-v3/include/bits/random.tcc
+++ b/libstdc++-v3/include/bits/random.tcc
@@ -541,8 +541,16 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSION
     subtract_with_carry_engine<_UIntType, __w, __s, __r>::
     seed(result_type __value)
     {
+      // _GLIBCXX_RESOLVE_LIB_DEFECTS
+      // 3809. Is std::subtract_with_carry_engine<uint16_t> supposed to work?
+      // 4014. LWG 3809 changes behavior of some existing code
+      if (__value == 0u)
+       __value = default_seed;
+      else
+       __value %= 2147483563u;
+
       std::linear_congruential_engine<uint_least32_t, 40014u, 0u, 2147483563u>
-       __lcg(__value == 0u ? default_seed : __value);
+       __lcg(__value);

       const size_t __n = (__w + 31) / 32;

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #10 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jonathan Wakely <redi@gcc.gnu.org>:

https://gcc.gnu.org/g:c224dec0e7c88e7a95633023018cdcb6ee87c65f

commit r14-7216-gc224dec0e7c88e7a95633023018cdcb6ee87c65f
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Thu Jan 11 15:09:12 2024 +0000

    libstdc++: Fix non-portable results from 64-bit
std::subtract_with_carry_engine [PR107466]

    I implemented the resolution of LWG 3809 in r13-4364-ga64775a0edd469 but
    it was recently noted in the MSVC STL github repo that the change causes
    possible truncation for 64-bit seeds. Whether the truncation occurs (and
    to what value) depends on the width of uint_least32_t which is not
    portable, so the output of the PRNG for 64-bit seed values is no longer
    the same as in C++20, and no longer portable across platforms.

    That new issue was filed as LWG 4014. I proposed a new change which
    reduces the seed by the LCG's modulus before the conversion to
    uint_least32_t. This ensures that 64-bit seed values are consistently
    reduced by the modulus before any truncation. This removes the
    platform-dependent behaviour and restores the old behaviour for
    std::subtract_with_carry_engine specializations using a 64-bit result
    type (such as std::ranlux48_base).

    libstdc++-v3/ChangeLog:

            PR libstdc++/107466
            * include/bits/random.tcc (subtract_with_carry_engine::seed):
            Implement proposed resolution of LWG 4014.
            * testsuite/26_numerics/random/pr60037-neg.cc: Adjust dg-error
            line number.
            *
testsuite/26_numerics/random/subtract_with_carry_engine/cons/lwg3809.cc:
            Check for expected result of 64-bit engine with seed that
            doesn't fit in 32-bits.

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #11 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Jonathan Wakely
<redi@gcc.gnu.org>:

https://gcc.gnu.org/g:3bdd80d56aa07d5975f551e0026f3cf9411124bf

commit r13-8306-g3bdd80d56aa07d5975f551e0026f3cf9411124bf
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Thu Jan 11 15:09:12 2024 +0000

    libstdc++: Fix non-portable results from 64-bit
std::subtract_with_carry_engine [PR107466]

    I implemented the resolution of LWG 3809 in r13-4364-ga64775a0edd469 but
    it was recently noted in the MSVC STL github repo that the change causes
    possible truncation for 64-bit seeds. Whether the truncation occurs (and
    to what value) depends on the width of uint_least32_t which is not
    portable, so the output of the PRNG for 64-bit seed values is no longer
    the same as in C++20, and no longer portable across platforms.

    That new issue was filed as LWG 4014. I proposed a new change which
    reduces the seed by the LCG's modulus before the conversion to
    uint_least32_t. This ensures that 64-bit seed values are consistently
    reduced by the modulus before any truncation. This removes the
    platform-dependent behaviour and restores the old behaviour for
    std::subtract_with_carry_engine specializations using a 64-bit result
    type (such as std::ranlux48_base).

    libstdc++-v3/ChangeLog:

            PR libstdc++/107466
            * include/bits/random.tcc (subtract_with_carry_engine::seed):
            Implement proposed resolution of LWG 4014.
            * testsuite/26_numerics/random/pr60037-neg.cc: Adjust dg-error
            line number.
            *
testsuite/26_numerics/random/subtract_with_carry_engine/cons/lwg3809.cc:
            Check for expected result of 64-bit engine with seed that
            doesn't fit in 32-bits.

    (cherry picked from commit c224dec0e7c88e7a95633023018cdcb6ee87c65f)

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

--- Comment #12 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Jonathan Wakely
<redi@gcc.gnu.org>:

https://gcc.gnu.org/g:f2af87b9705d5a7e37b65bf342146ff25f025e49

commit r12-10143-gf2af87b9705d5a7e37b65bf342146ff25f025e49
Author: Jonathan Wakely <jwakely@redhat.com>
Date:   Thu Jan 11 15:09:12 2024 +0000

    libstdc++: Fix non-portable results from 64-bit
std::subtract_with_carry_engine [PR107466]

    I implemented the resolution of LWG 3809 in r13-4364-ga64775a0edd469 but
    it was recently noted in the MSVC STL github repo that the change causes
    possible truncation for 64-bit seeds. Whether the truncation occurs (and
    to what value) depends on the width of uint_least32_t which is not
    portable, so the output of the PRNG for 64-bit seed values is no longer
    the same as in C++20, and no longer portable across platforms.

    That new issue was filed as LWG 4014. I proposed a new change which
    reduces the seed by the LCG's modulus before the conversion to
    uint_least32_t. This ensures that 64-bit seed values are consistently
    reduced by the modulus before any truncation. This removes the
    platform-dependent behaviour and restores the old behaviour for
    std::subtract_with_carry_engine specializations using a 64-bit result
    type (such as std::ranlux48_base).

    libstdc++-v3/ChangeLog:

            PR libstdc++/107466
            * include/bits/random.tcc (subtract_with_carry_engine::seed):
            Implement proposed resolution of LWG 4014.
            * testsuite/26_numerics/random/pr60037-neg.cc: Adjust dg-error
            line number.
            *
testsuite/26_numerics/random/subtract_with_carry_engine/cons/lwg3809.cc:
            Check for expected result of 64-bit engine with seed that
            doesn't fit in 32-bits.

    (cherry picked from commit c224dec0e7c88e7a95633023018cdcb6ee87c65f)

[Bug libstdc++/107466] [12 Regression] invalid -Wnarrowing error with std::subtract_with_carry_engine

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107466

Jonathan Wakely <redi at gcc dot gnu.org> changed:

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

--- Comment #13 from Jonathan Wakely <redi at gcc dot gnu.org> ---
And fixed for real now, for 12.4, 13.3 and trunk.