[Bug c++/88086] gcc only allows valid expressions as unknown C++ attribute argument clause

2020-03-06 Thread gbuella at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88086

Gábor Buella  changed:

   What|Removed |Added

 CC||gbuella at gmail dot com

--- Comment #1 from Gábor Buella  ---
This did come up for me as well, when trying to compile:

enum [[clang::enum_extensibility(closed)]] dummy { a, b };


$ g++-8 dummy.cc
y.cc:2:34: error: ‘closed’ was not declared in this scope
 enum [[clang::enum_extensibility(closed)]] dummy { a, b };
  ^~
y.cc:2:34: note: suggested alternative: ‘class’
 enum [[clang::enum_extensibility(closed)]] dummy { a, b };
  ^~
  class
y.cc:2:44: warning: ‘clang::enum_extensibility’ scoped attribute directive
ignored [-Wattributes]
 enum [[clang::enum_extensibility(closed)]] dummy { a, b };
^


It is rather annoying, the attribute is ignored, but there is a compile error.

[Bug c++/93674] New: GCC eliminates conditions it should not, when strict-enums is on

2020-02-11 Thread gbuella at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93674

Bug ID: 93674
   Summary: GCC eliminates conditions it should not, when
strict-enums is on
   Product: gcc
   Version: 8.3.0
Status: UNCONFIRMED
  Severity: normal
  Priority: P3
 Component: c++
  Assignee: unassigned at gcc dot gnu.org
  Reporter: gbuella at gmail dot com
  Target Milestone: ---

$ cat code.cc

enum some_enum { x = 1000 };

void sink(some_enum);

void func()
{
for (int i = 0; i < 3; ++i) {
for (int j = 3; j >= 0; --j) {
sink((some_enum)(i + j));
}
}
}

$ g++-8 -Wall -Wextra -pedantic -std=c++14 code.cc   -fstrict-enums  -S -o
/dev/stdout -fno-dwarf2-cfi-asm -O3 | head -20
.file   "code.cc"
.text
.p2align 4,,15
.globl  _Z4funcv
.type   _Z4funcv, @function
_Z4funcv:
.LFB0:
pushq   %rbx
.LCFI0:
movl$3, %ebx
.p2align 4,,10
.p2align 3
.L2:
movl%ebx, %edi
subl$1, %ebx
call_Z4sink9some_enum@PLT
jmp .L2
.LFE0:
.size   _Z4funcv, .-_Z4funcv
.section.eh_frame,"a",@progbits

$ g++-8 --version
g++-8 (Ubuntu 8.3.0-6ubuntu1~18.04.1) 8.3.0
Copyright (C) 2018 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.


Notice the infinite loop in generated assembly (unconditional jmp to .L2).
This doesn't happen without "-fstruct-enums", doesn't happen without -O2 or
-O3.
I'v seen this bug with GCC 8, GCC 9 . 
GCC-7 seems to be OK.

I spent hours trying to find more compact code for reproducing it, but I just
couldn't...
The loop, and the addition (i + j) seems to be needed to trigger this.

[Bug c++/93674] GCC eliminates conditions it should not, when strict-enums is on

2020-02-11 Thread gbuella at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93674

--- Comment #2 from Gábor Buella  ---
(In reply to Andrew Pinski from comment #1)
> -fstrict-enums
> Allow the compiler to optimize using the assumption that a value of
> enumerated type can only be one of the values of the enumeration (as defined
> in the C++ standard; basically, a value that can be represented in the
> minimum number of bits needed to represent all the enumerators). This
> assumption may not be valid if the program uses a cast to convert an
> arbitrary integer value to the enumerated type.
> 
>  CUT ---
> I assume that means since there is no negative numbers in the enum, that
> means you might get undefined behavior if j is negative ...

Look at the code, j is never negative when the cast is applied.
You can easily verify it by printing the e.g. values.

[Bug c++/93674] GCC eliminates conditions it should not, when strict-enums is on

2020-02-11 Thread gbuella at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93674

--- Comment #3 from Gábor Buella  ---
In case anyone would still get confused about the what values get casted to
enum, here is another way to write that example:

enum some_enum { x0,  x1,  x2,  x3,  x4,  x5,  x6,  x7,
 x8,  x9,  xa,  xb,  xc,  xd,  xe,  xf,
x00, x01, x02, x03, x04, x05, x06, x07,
x08, x09, x0a, x0b, x0c, x0d, x0e, x0f };

void sink(some_enum);

void func()
{
for (int i = 0; i < 3; ++i) {
int j = 3;
while (j >= 0) {
// Note: (i + j) is always non-negative here
// Actually, (i + j) is always of the values
// given when defining the type above.
sink((some_enum)(i + j));
--j;
}
}
}

[Bug c++/93674] GCC eliminates conditions it should not, when strict-enums is on

2020-02-11 Thread gbuella at gmail dot com 
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93674

--- Comment #6 from Gábor Buella  ---
(In reply to Jonathan Wakely from comment #4)
> I can't reproduce this with GCC 9, only 8.

$ cat code.cc
enum some_enum { x = 1000 };

void sink(some_enum);

void func()
{
for (int i = 0; i < 3; ++i) {
int j = 3;
while (j >= 0) { // turned into an uncoditional jmp by GCC
// Note: (i + j) is always non-negative here
// Actually, (i + j) is always of the values
// given when defining the type above.
sink((some_enum)(i + j));
--j;
}
}
}

$ g++-9 -Wall -Wextra -pedantic -std=c++14 code.cc   -fstrict-enums  -S -o
/dev/stdout -fno-dwarf2-cfi-asm -O3  | head -20
.file   "code.cc"
.text
.p2align 4
.globl  _Z4funcv
.type   _Z4funcv, @function
_Z4funcv:
.LFB0:
endbr64
pushq   %rbx
.LCFI0:
movl$3, %ebx
.p2align 4,,10
.p2align 3
.L2:
movl%ebx, %edi
subl$1, %ebx
call_Z4sink9some_enum@PLT
jmp .L2
.LFE0:
.size   _Z4funcv, .-_Z4funcv

$ g++-9 --version
g++-9 (Ubuntu 9.2.1-9ubuntu2) 9.2.1 20191008
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.