Does __attribute__ ((selectany)) make sense now for GCC?

Does __attribute__ ((selectany)) make sense now for GCC?

[Tim Lewis]

In Visual Studio we have __declspec(selectany) to limit the impact of unused
data.

 

I see that GCC for Windows has __attribute__ ((selectany)).

 

Should we me using this for GLOBAL_REMOVE_IF_UNREFERENCED in
MdePkg\Include\Base.h?

 

Tim

Re: Does __attribute__ ((selectany)) make sense now for GCC?

[Andrew Fish]

Tim,

Looks like that is only available on Windows versions of GCC, and is more about compatible behavior. 

selectany
The selectany attribute causes an initialized global variable to have link-once semantics. When multiple definitions of the variable are encountered by the linker, the first is selected and the remainder are discarded. Following usage by the Microsoft compiler, the linker is told not to warn about size or content differences of the multiple definitions.
Although the primary usage of this attribute is for POD types, the attribute can also be applied to global C++ objects that are initialized by a constructor. In this case, the static initialization and destruction code for the object is emitted in each translation defining the object, but the calls to the constructor and destructor are protected by a link-once guard variable.

The selectany attribute is only available on Microsoft Windows targets. You can use __declspec (selectany) as a synonym for __attribute__ ((selectany)) for compatibility with other compilers.


What I've noticed with clang/LLVM is the unreferenced globals get removed when you enable link time optimizations. 

I'd actually ask the opposite question. Does __declspec(selectany) impact dead code removal on current versions of VC++? 

Thanks,

Andrew Fish

> On May 18, 2018, at 3:22 PM, Tim Lewis <tim.lewis@insyde.com> wrote:
> 
> In Visual Studio we have __declspec(selectany) to limit the impact of unused
> data.
> 
> 
> 
> I see that GCC for Windows has __attribute__ ((selectany)).
> 
> 
> 
> Should we me using this for GLOBAL_REMOVE_IF_UNREFERENCED in
> MdePkg\Include\Base.h?
> 
> 
> 
> Tim
> 
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>

Re: Does __attribute__ ((selectany)) make sense now for GCC?

[Tim Lewis]

Andrew -

 

I know for Visual Studio it does the same thing for data that /Gy does for
code.

 

That is, if the data is unreferenced, then it doesn't get brought it.

 

Maybe link-time code generation has the same effect. That's what I'm curious
about. 

 

Tim

 

From: afish@apple.com <afish@apple.com> 
Sent: Friday, May 18, 2018 3:39 PM
To: Tim Lewis <tim.lewis@insyde.com>
Cc: edk2-devel@lists.01.org
Subject: Re: [edk2] Does __attribute__ ((selectany)) make sense now for GCC?

 

Tim,

 

Looks like that is only available on Windows versions of GCC, and is more
about compatible behavior. 

 

selectany

The selectany attribute causes an initialized global variable to have
link-once semantics. When multiple definitions of the variable are
encountered by the linker, the first is selected and the remainder are
discarded. Following usage by the Microsoft compiler, the linker is told not
to warn about size or content differences of the multiple definitions.

Although the primary usage of this attribute is for POD types, the attribute
can also be applied to global C++ objects that are initialized by a
constructor. In this case, the static initialization and destruction code
for the object is emitted in each translation defining the object, but the
calls to the constructor and destructor are protected by a link-once guard
variable.

The selectany attribute is only available on Microsoft Windows targets. You
can use __declspec (selectany) as a synonym for __attribute__ ((selectany))
for compatibility with other compilers.

 

What I've noticed with clang/LLVM is the unreferenced globals get removed
when you enable link time optimizations. 

 

I'd actually ask the opposite question. Does __declspec(selectany) impact
dead code removal on current versions of VC++? 

 

Thanks,

 

Andrew Fish





On May 18, 2018, at 3:22 PM, Tim Lewis <tim.lewis@insyde.com
<mailto:tim.lewis@insyde.com> > wrote:

 

In Visual Studio we have __declspec(selectany) to limit the impact of unused
data.



I see that GCC for Windows has __attribute__ ((selectany)).



Should we me using this for GLOBAL_REMOVE_IF_UNREFERENCED in
MdePkg\Include\Base.h?



Tim

_______________________________________________
edk2-devel mailing list
 <mailto:edk2-devel@lists.01.org> edk2-devel@lists.01.org
 <https://lists.01.org/mailman/listinfo/edk2-devel>
https://lists.01.org/mailman/listinfo/edk2-devel

 

Re: Does __attribute__ ((selectany)) make sense now for GCC?

[Andrew Fish]

Tim,

 VC++ is a strange beast. See C99 support. VC++ seems to do strange things to support non standard hackyness of years past. So the only reason I can think of that  __attribute__ ((selectany)) is associated with dead stripping, is doing it correctly broke some chunk of code that was required for compatibility. From looking on the VC++ site it seems like COMDAT folding may be required for dead code removal, but again this not required to make it work, but to not break old crufty stuff. 

I guess on Windows GCC they could trying to "copy exact" VC++ behavior, but it is most definitely going to be ignored on the non-Windows versions. 

The answer from Microsoft on C99 is to use clang :)

Thanks,

Andrew Fish

> On May 18, 2018, at 3:43 PM, Tim Lewis <tim.lewis@insyde.com> wrote:
> 
> Andrew -
> 
> 
> 
> I know for Visual Studio it does the same thing for data that /Gy does for
> code.
> 
> 
> 
> That is, if the data is unreferenced, then it doesn't get brought it.
> 
> 
> 
> Maybe link-time code generation has the same effect. That's what I'm curious
> about. 
> 
> 
> 
> Tim
> 
> 
> 
> From: afish@apple.com <mailto:afish@apple.com> <afish@apple.com <mailto:afish@apple.com>> 
> Sent: Friday, May 18, 2018 3:39 PM
> To: Tim Lewis <tim.lewis@insyde.com <mailto:tim.lewis@insyde.com>>
> Cc: edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> Subject: Re: [edk2] Does __attribute__ ((selectany)) make sense now for GCC?
> 
> 
> 
> Tim,
> 
> 
> 
> Looks like that is only available on Windows versions of GCC, and is more
> about compatible behavior. 
> 
> 
> 
> selectany
> 
> The selectany attribute causes an initialized global variable to have
> link-once semantics. When multiple definitions of the variable are
> encountered by the linker, the first is selected and the remainder are
> discarded. Following usage by the Microsoft compiler, the linker is told not
> to warn about size or content differences of the multiple definitions.
> 
> Although the primary usage of this attribute is for POD types, the attribute
> can also be applied to global C++ objects that are initialized by a
> constructor. In this case, the static initialization and destruction code
> for the object is emitted in each translation defining the object, but the
> calls to the constructor and destructor are protected by a link-once guard
> variable.
> 
> The selectany attribute is only available on Microsoft Windows targets. You
> can use __declspec (selectany) as a synonym for __attribute__ ((selectany))
> for compatibility with other compilers.
> 
> 
> 
> What I've noticed with clang/LLVM is the unreferenced globals get removed
> when you enable link time optimizations. 
> 
> 
> 
> I'd actually ask the opposite question. Does __declspec(selectany) impact
> dead code removal on current versions of VC++? 
> 
> 
> 
> Thanks,
> 
> 
> 
> Andrew Fish
> 
> 
> 
> 
> 
> On May 18, 2018, at 3:22 PM, Tim Lewis <tim.lewis@insyde.com
> <mailto:tim.lewis@insyde.com <mailto:tim.lewis@insyde.com>> > wrote:
> 
> 
> 
> In Visual Studio we have __declspec(selectany) to limit the impact of unused
> data.
> 
> 
> 
> I see that GCC for Windows has __attribute__ ((selectany)).
> 
> 
> 
> Should we me using this for GLOBAL_REMOVE_IF_UNREFERENCED in
> MdePkg\Include\Base.h?
> 
> 
> 
> Tim
> 
> _______________________________________________
> edk2-devel mailing list
> <mailto:edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>> edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> <https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>>
> https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>
> 
> 
> 
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>

Re: Does __attribute__ ((selectany)) make sense now for GCC?

[Gao, Liming]

Andrew:
  VS has added /Gw option to Optimize Global Data since VS2013. With /Gw option, __declspec(selectany) is not used any more. Now, __declspec(selectany) is kept for old VS tool chain. So, I suggest GCC uses its native optimization option instead of it. 

Thanks
Liming
> -----Original Message-----
> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of Andrew Fish
> Sent: Saturday, May 19, 2018 7:07 AM
> To: Tim Lewis <tim.lewis@insyde.com>
> Cc: edk2-devel@lists.01.org
> Subject: Re: [edk2] Does __attribute__ ((selectany)) make sense now for GCC?
> 
> Tim,
> 
>  VC++ is a strange beast. See C99 support. VC++ seems to do strange things to support non standard hackyness of years past. So the
> only reason I can think of that  __attribute__ ((selectany)) is associated with dead stripping, is doing it correctly broke some chunk of
> code that was required for compatibility. From looking on the VC++ site it seems like COMDAT folding may be required for dead code
> removal, but again this not required to make it work, but to not break old crufty stuff.
> 
> I guess on Windows GCC they could trying to "copy exact" VC++ behavior, but it is most definitely going to be ignored on the
> non-Windows versions.
> 
> The answer from Microsoft on C99 is to use clang :)
> 
> Thanks,
> 
> Andrew Fish
> 
> > On May 18, 2018, at 3:43 PM, Tim Lewis <tim.lewis@insyde.com> wrote:
> >
> > Andrew -
> >
> >
> >
> > I know for Visual Studio it does the same thing for data that /Gy does for
> > code.
> >
> >
> >
> > That is, if the data is unreferenced, then it doesn't get brought it.
> >
> >
> >
> > Maybe link-time code generation has the same effect. That's what I'm curious
> > about.
> >
> >
> >
> > Tim
> >
> >
> >
> > From: afish@apple.com <mailto:afish@apple.com> <afish@apple.com <mailto:afish@apple.com>>
> > Sent: Friday, May 18, 2018 3:39 PM
> > To: Tim Lewis <tim.lewis@insyde.com <mailto:tim.lewis@insyde.com>>
> > Cc: edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> > Subject: Re: [edk2] Does __attribute__ ((selectany)) make sense now for GCC?
> >
> >
> >
> > Tim,
> >
> >
> >
> > Looks like that is only available on Windows versions of GCC, and is more
> > about compatible behavior.
> >
> >
> >
> > selectany
> >
> > The selectany attribute causes an initialized global variable to have
> > link-once semantics. When multiple definitions of the variable are
> > encountered by the linker, the first is selected and the remainder are
> > discarded. Following usage by the Microsoft compiler, the linker is told not
> > to warn about size or content differences of the multiple definitions.
> >
> > Although the primary usage of this attribute is for POD types, the attribute
> > can also be applied to global C++ objects that are initialized by a
> > constructor. In this case, the static initialization and destruction code
> > for the object is emitted in each translation defining the object, but the
> > calls to the constructor and destructor are protected by a link-once guard
> > variable.
> >
> > The selectany attribute is only available on Microsoft Windows targets. You
> > can use __declspec (selectany) as a synonym for __attribute__ ((selectany))
> > for compatibility with other compilers.
> >
> >
> >
> > What I've noticed with clang/LLVM is the unreferenced globals get removed
> > when you enable link time optimizations.
> >
> >
> >
> > I'd actually ask the opposite question. Does __declspec(selectany) impact
> > dead code removal on current versions of VC++?
> >
> >
> >
> > Thanks,
> >
> >
> >
> > Andrew Fish
> >
> >
> >
> >
> >
> > On May 18, 2018, at 3:22 PM, Tim Lewis <tim.lewis@insyde.com
> > <mailto:tim.lewis@insyde.com <mailto:tim.lewis@insyde.com>> > wrote:
> >
> >
> >
> > In Visual Studio we have __declspec(selectany) to limit the impact of unused
> > data.
> >
> >
> >
> > I see that GCC for Windows has __attribute__ ((selectany)).
> >
> >
> >
> > Should we me using this for GLOBAL_REMOVE_IF_UNREFERENCED in
> > MdePkg\Include\Base.h?
> >
> >
> >
> > Tim
> >
> > _______________________________________________
> > edk2-devel mailing list
> > <mailto:edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>> edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> > <https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>>
> > https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>
> >
> >
> >
> > _______________________________________________
> > edk2-devel mailing list
> > edk2-devel@lists.01.org <mailto:edk2-devel@lists.01.org>
> > https://lists.01.org/mailman/listinfo/edk2-devel <https://lists.01.org/mailman/listinfo/edk2-devel>
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org
> https://lists.01.org/mailman/listinfo/edk2-devel