[Bug c++/109750] New: Incorrect code generated for shifted bit field

[Bug c++/109750] New: Incorrect code generated for shifted bit field

[jim.gray at atos dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

            Bug ID: 109750
           Summary: Incorrect code generated for shifted bit field
           Product: gcc
           Version: 9.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jim.gray at atos dot net
  Target Milestone: ---

Created attachment 55008
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=55008&action=edit
Tar file with source, assembly listing, compile script, executable

Attached is a simple program that shows the problem encountered.

Bad output when compiling with GCC compiler on Ubuntu
~/gcc-bug$ ./gccbug
  Source = 000000020000
  Target = 000000000200
  Target = 1777777777760000000200

Good output when compiling with MS C++ compiler on Windows:
>gccbug
  Source = 000000020000
  Target = 000000000200
  Target = 020000000200

[Bug c++/109750] Incorrect code generated for shifted bit field

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

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

((UI64)source.sH.lower)

The question is source.sH.lower promoted to int rather than the underlying
type. Note clang has the same behavior as GCC here even. So I am thinking this
is a bug in MSVC rather than GCC ... (or some implemented defined behavior and
both are correct).

[Bug c++/109750] Incorrect code generated for shifted bit field

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |INVALID
             Status|UNCONFIRMED                 |RESOLVED

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
MSVC bug:

In [conv.prom] p5 the standard says:

"A prvalue for an integral bit-field (11.4.9) can be converted to a prvalue of
type int if int can represent all the values of the bit-field;"

[Bug c++/109750] Incorrect code generated for shifted bit field

[jim.gray at atos dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

--- Comment #3 from jim.gray at atos dot net ---
The "correct" answer is also provided by the Intel C++ compiler,
so apparently GCC is the odd one out.

[Bug c++/109750] Incorrect code generated for shifted bit field

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to jim.gray from comment #3)
> The "correct" answer is also provided by the Intel C++ compiler,
> so apparently GCC is the odd one out.

I think MSVC is the odd one out really. see below.
You can also do the testing yourself at https://godbolt.org/ to see that.

Huh? Maybe ICC on windows gives a different answer than ICC on Linux.
Which might make sense if ICC on Windows is trying to provide bug for bug
compatibility with MSVC.

ICC gives:

icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and
will be removed from product release in the second half of 2023. The Intel(R)
oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward.
Please transition to use this compiler. Use '-diag-disable=10441' to disable
this message.
ASM generation compiler returned: 0
icc: remark #10441: The Intel(R) C++ Compiler Classic (ICC) is deprecated and
will be removed from product release in the second half of 2023. The Intel(R)
oneAPI DPC++/C++ Compiler (ICX) is the recommended compiler moving forward.
Please transition to use this compiler. Use '-diag-disable=10441' to disable
this message.
Execution build compiler returned: 0
Program returned: 0
Example showing GCC bug:
  Source = 000000020000
  Target = 000000000200
  Target = 1777777777760000000200

clang gives:
<source>:103:14: warning: unused parameter 'argc' [-Wunused-parameter]
int main(int argc, char *argv[]) {
             ^
<source>:103:26: warning: unused parameter 'argv' [-Wunused-parameter]
int main(int argc, char *argv[]) {
                         ^
2 warnings generated.
ASM generation compiler returned: 0
<source>:103:14: warning: unused parameter 'argc' [-Wunused-parameter]
int main(int argc, char *argv[]) {
             ^
<source>:103:26: warning: unused parameter 'argv' [-Wunused-parameter]
int main(int argc, char *argv[]) {
                         ^
2 warnings generated.
Execution build compiler returned: 0
Program returned: 0
Example showing GCC bug:
  Source = 000000020000
  Target = 000000000200
  Target = 1777777777760000000200

[Bug c++/109750] Incorrect code generated for shifted bit field

[jim.gray at atos dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

--- Comment #5 from jim.gray at atos dot net ---
I don't know if it makes any difference, but I ran with an older version of the
Intel C++ compiler on Linux, with the results below:

Good output when compiling with the Intel C++ compiler 
on CentOS Linux release 7.7.1908 

[builder@7f936516e827 ~]$ icc --version
icc (ICC) 17.0.2 20170213
Copyright (C) 1985-2017 Intel Corporation.  All rights reserved.

[builder@af80ebc5f176 gccbug]$ icc gccbug.cpp
[builder@af80ebc5f176 gccbug]$ 

[builder@af80ebc5f176 gccbug]$ ls -l
total 28
-rwxrwxr-x 1 builder builder 22592 May  5 18:02 a.out
-rw-rw-r-- 1 builder builder  3553 May  5 18:02 gccbug.cpp
[builder@af80ebc5f176 gccbug]$ 

[builder@af80ebc5f176 gccbug]$ ./a.out
Example showing GCC bug:
  Source = 000000020000
  Target = 000000000200
  Target = 020000000200
Example showing workaround for bug:
  Source = 000000020000
  Target = 000000000200
  Temp   = 000000020000
  Temp   = 020000000000
  Target = 020000000200
[builder@af80ebc5f176 gccbug]$

[Bug c++/109750] Incorrect code generated for shifted bit field

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
icc version 2021.1.2 (gcc version 10.1.0 compatibility)
Has the same behavior as GCC.
While:
icc version 19.0.0.117 (gcc version 8.2.0 compatibility)
had the MSVC behavior

clang has the same behavior as GCC since clang 3.0.0.

[Bug c++/109750] Incorrect code generated for shifted bit field

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=109750

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
This is also definitely a bug in MSVC and should reported to them.
ICC fixed their bug already even.