How is it possible to put some inline bytes in the code like the former inline( $1f , $ef , $1A );
instruction?

Rainer Stratmann wrote on Thu, 09 Aug 2012:

How is it possible to put some inline bytes in the code like the former inline( $1f , $ef , $1A ); instruction?

{$asmmode att}
asm
   .byte $0x1f, $0xef, $0x1a
end;
{$asmmode intel}
asm
   db $1a, $ef, $1a
end;

If this is supposed to be the only code/data inside an assembler routine, make sure to also add "assembler; nostackframe;" to that routine's modifiers to prevent the compiler from setting up a stack frame and/or insert stack alignment code.

Jonas

Hi,

Rainer Stratmann wrote

How is it possible to put some inline bytes in the code like the former inline( $1f , $ef , $1A ); instruction?

Not sure if it is the official way, but the following works:

with {$asmmode att}

asm
.byte 0x1f, 0xef, 0x1a
end;

and with {$asmmode intel}

asm
db $1f, $ef, $1a
end;

Hth,
Thomas

Thanks, the next problem then is: I wanted to do that in an inline procedure but if the compiler detects an asm statement then inline is disabled(!).

Maybe a macro?

Henry

Rainer Stratmann wrote on Thu, 09 Aug 2012:

I made an error here, the "$" should be removed (see Thomas' mail)

That's correct, it is not possible inline assembler statements in FPC. You can try a macro instead:

{$macro on}
{$define IReallyKnowWhatIAmDoing:=asm .byte 0x1f, 0xef, 0x1a end}
begin
   IReallyKnowWhatIAmDoing;
end.

You will have to declare the macro in a common include file or in every main source file if you want to use it in more than one compilation module though.

Jonas

Sounds good!

end.

Same thing: inlining disabled(!).

Rainer Stratmann wrote on Thu, 09 Aug 2012:

The idea is to use the macro instead of a function call, not to use the macro inside a function that you then try to inline. A macro is just text substitution, it will never have a different effect than typing its contents directly at the place where the macro is used.

Jonas

Jonas Maebe wrote

The idea is to use the macro instead of a function call, not to use the macro inside a function that you then try to inline. A macro is just text substitution, it will never have a different effect than typing its contents directly at the place where the macro is used.

To avoid disabling procedure/function inlining when the compiler finds an asm instruction it would be good to have at least the mentioned former

inline( $AB, $CD , $EF , ... );

back.

Is that possible?

Macros have to defined for each unit again so this is not userfriendly if a programmer changes a macro.

Rainer Stratmann wrote on Thu, 09 Aug 2012:

There are no plans to do so, and even if it were added it would be quite possible that this would also disable inlining the containing routine for implementation reasons.

Macros have to defined for each unit again so this is not userfriendly if a programmer changes a macro.

A macro can be defined in a file that is included in all units, so then you only have to change it in one place.

Jonas

Jonas Maebe wrote

if it were added it would be quite possible that this would also disable inlining the containing routine for implementation reasons.

I think the reason for disabling is that in the asm code can be identifyers and jumplabels which are difficult to handle then.

In the inline( $AB , $CD , ... ) instruction is only pure data, so for me there is no reason to disable the inlining then.

Rainer Stratmann wrote on Thu, 09 Aug 2012:

inline() can contain identifiers and labels just like assembler blocks. But again: there are no plans to add support for inline().

Jonas