[Bug tree-optimization/10980] vararg functions are not inlined

[Bug tree-optimization/10980] vararg functions are not inlined

[vanboxem.ruben at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

Ruben Van Boxem <vanboxem.ruben at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |vanboxem.ruben at gmail dot
                   |                            |com

--- Comment #7 from Ruben Van Boxem <vanboxem.ruben at gmail dot com> 2011-09-23 13:34:52 UTC ---
This problem popped up when compiling LLVM's libc++:
M:\Development\Source\libc++\src\locale.cpp:1:0: warning: -fPIC ignored for
target (all code is position independent) [enabled by default]
M:\Development\Source\libc++\include/locale: In function 'int
std::__1::__asprintf_l(char**, _locale_t, const char*, ...)':
M:\Development\Source\libc++\include/locale:372:5: sorry, unimplemented:
function 'int std::__1::__asprintf_l(char**, _locale_t, const char*, ...)' can
never be inlined because it uses variable argument lists
M:\Development\Source\libc++\include/locale: In function 'int
std::__1::__sprintf_l(char*, _locale_t, const char*, ...)':
M:\Development\Source\libc++\include/locale:344:5: sorry, unimplemented:
function 'int std::__1::__sprintf_l(char*, _locale_t, const char*, ...)' can
never be inlined because it uses variable argument lists
M:\Development\Source\libc++\include/locale: In function 'int
std::__1::__snprintf_l(char*, size_t, _locale_t, const char*, ...)':
M:\Development\Source\libc++\include/locale:358:5: sorry, unimplemented:
function 'int std::__1::__snprintf_l(char*, size_t, _locale_t, const char*,
...)' can never be inlined because it uses variable argument lists
M:\Development\Source\libc++\include/locale: In function 'int
std::__1::__sscanf_l(const char*, _locale_t, const char*, ...)':
M:\Development\Source\libc++\include/locale:386:5: sorry, unimplemented:
function 'int std::__1::__sscanf_l(const char*, _locale_t, const char*, ...)'
can never be inlined because it uses variable argument lists

The code can be found here:
http://llvm.org/svn/llvm-project/libcxx/trunk/include/locale

[Bug tree-optimization/10980] vararg functions are not inlined

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #8 from Jan Hubicka <hubicka at ucw dot cz> 2011-09-23 13:40:55 UTC ---
> This problem popped up when compiling LLVM's libc++:
> M:\Development\Source\libc++\src\locale.cpp:1:0: warning: -fPIC ignored for
> target (all code is position independent) [enabled by default]
> M:\Development\Source\libc++\include/locale: In function 'int
> std::__1::__asprintf_l(char**, _locale_t, const char*, ...)':
> M:\Development\Source\libc++\include/locale:372:5: sorry, unimplemented:
> function 'int std::__1::__asprintf_l(char**, _locale_t, const char*, ...)' can
> never be inlined because it uses variable argument lists
> M:\Development\Source\libc++\include/locale: In function 'int
> std::__1::__sprintf_l(char*, _locale_t, const char*, ...)':

It seems wrong to add always_inline attribute on something like this.  I am not
aware
that LLVM would be inlining variadic function either
(even though it is doable in some trivial cases)

Honza

[Bug tree-optimization/10980] vararg functions are not inlined

[vanboxem.ruben at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #9 from Ruben Van Boxem <vanboxem.ruben at gmail dot com> 2011-09-23 13:46:27 UTC ---
Current Clang compiled everything without a hitch. It might ignore the
attribute, I haven't checked. Perhaps this error should be made a warning?

Re: [Bug tree-optimization/10980] vararg functions are not inlined

[Jan Hubicka]

I bet it just ignores the attribute that is not how the attribute is
supposed to work.

We output error in this case intentionally (and warning in at the same
spot if function is declared inline and -Winline is given).
Functions marked always_inlined really ought to be always inlined and
attribute is supposed to be useable in the cases where compilation would
fail otherwise (such as SSE buitins).

Our documentation is however not particularly clear here, perhaps it should
be updated...

Honza

[Bug tree-optimization/10980] vararg functions are not inlined

[hubicka at ucw dot cz]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #10 from Jan Hubicka <hubicka at ucw dot cz> 2011-09-23 14:10:53 UTC ---
I bet it just ignores the attribute that is not how the attribute is
supposed to work.

We output error in this case intentionally (and warning in at the same
spot if function is declared inline and -Winline is given).
Functions marked always_inlined really ought to be always inlined and
attribute is supposed to be useable in the cases where compilation would
fail otherwise (such as SSE buitins).

Our documentation is however not particularly clear here, perhaps it should
be updated...

Honza

[Bug tree-optimization/10980] vararg functions are not inlined

[vanboxem.ruben at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #11 from Ruben Van Boxem <vanboxem.ruben at gmail dot com> 2011-10-17 18:44:39 UTC ---
(In reply to comment #10)
> I bet it just ignores the attribute that is not how the attribute is
> supposed to work.
> 
> We output error in this case intentionally (and warning in at the same
> spot if function is declared inline and -Winline is given).
> Functions marked always_inlined really ought to be always inlined and
> attribute is supposed to be useable in the cases where compilation would
> fail otherwise (such as SSE buitins).
> 
> Our documentation is however not particularly clear here, perhaps it should
> be updated...
> 
> Honza

Seeing that documentation is unclear (no mention of when errors or warnings are
produced), and Qt Creator devs want to know what version of GCC started
errorring on this, could you please let me know what GCC version started
producing errors? Thanks!

[Bug tree-optimization/10980] vararg functions are not inlined

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #12 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-10-18 09:54:03 UTC ---
(In reply to comment #11)
> (In reply to comment #10)
> > I bet it just ignores the attribute that is not how the attribute is
> > supposed to work.
> > 
> > We output error in this case intentionally (and warning in at the same
> > spot if function is declared inline and -Winline is given).
> > Functions marked always_inlined really ought to be always inlined and
> > attribute is supposed to be useable in the cases where compilation would
> > fail otherwise (such as SSE buitins).
> > 
> > Our documentation is however not particularly clear here, perhaps it should
> > be updated...
> > 
> > Honza
> 
> Seeing that documentation is unclear (no mention of when errors or warnings are
> produced), and Qt Creator devs want to know what version of GCC started
> errorring on this, could you please let me know what GCC version started
> producing errors? Thanks!

GCC always errored out on this (previously by sorry(), now with error()).

We can possibly inline varargs functions when we delay lowering the
vararg builtins to the stdarg pass (which would clean up its implementation
a lot ...).

[Bug tree-optimization/10980] vararg functions are not inlined

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=10980

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
We have __builtin_va_arg_pack and __builtin_va_arg_pack_len which I think
solves this problem really.


https://gcc.gnu.org/onlinedocs/gcc-11.2.0/gcc/Constructing-Calls.html#index-_005f_005fbuiltin_005fva_005farg_005fpack