Gloster said: " void ASSERT_GPS_INTERRUPT(void) /*This can probably be compiled without any function call overhead.*/ "
Dave responded: "[..]
[..] The inline keyword might help, if your compiler supports it."
I have just noticed ( news: snipped-for-privacy@news.xsall.nl ) that inlining is part of the current C standard.
From PDF page 290, hardcopy page 205 of
formatting link
:
"On average every fifth statement is a function call. This is very frustrating for writers of optimizers (long sequences of C code without any function calls provide more opportunities to generate high-quality machine code). The introduction of support for the inline function specifier, in C99, offers one way around this function specifier syntax 1512 problem for time-critical code."
From later in that document:
"6.7.4 Function specifiers
-------------------------- 1512 function specifier syntax function-specifier: inline Commentary Rationale The inline keyword, adapted from C++ . . . The keyword inline is invariably used by languages to specify functions that are to be considered for inlining by a translator.
C90 Support for functionspecifier is new in C99. [..]
Coding Guidelines Those coding guideline documents that argue against the use of the register storage-class specifier may well argue against the use of function specifiers for the same reasons. These coding guidelines do not recommend against this usage for the same reason they did not recommend against the use of the register storage-class specifier.
[..]
1519 Making a function an inline function suggests that calls to the function be as fast as possible.118) Commentary The inline specifier is a hint from the developer to the translator. The 1970s gave us the register keyword and the 1990s have given us the inline keyword. In both eras commercially usable translation technology was not up to automating the necessary functionality and assistance from developers was the solution adopted. Published figures on translators that automatically decide which functions to inline[210, 315, 338] show total program execution time reductions of around 2% to 8%. The latest research on inlining[1384] has not significantly improved on these program execution time reductions, but it does not seem to cause the large increase in executable program size seen in earlier work, and the translation overhead associated with deducing what to inline has been reduced.
[.. paragraph about not caring about function delays from a clearly non-embedded persepective and more importantly non-realtime]
It often comes as a surprise to developers that use of the register storage class can slow a program down. The same is also true of the inline function specifier; its use can slow a program down (although the situations in which this occurs appear to be less frequent than for the register storage class). Degradations in to an increase in page swaps, on hosts with limited storage; or an increase in program size causing the number of cache hits to decrease are the most commonly seen reasons. One published report[267] (Fortran source) found that a lack of sophisticated dependency analysis in the translator meant that it had to make worst-case assumptions in a critical loop, that did not apply in the non-inlined source. Even when inline is used intelligently (based on execution counts of function calls) improvements in performance can vary significantly, depending on the characteristics of the source code and on the architecture of the host processor.[315, 338]
[..]
Coding Guidelines The term type safe macro (because the types of the arguments are checked) or simply safe macro (because the arguments are only evaluated once) are sometimes applied to the use of inline functions. 1520 The extent to which such suggestions are effective is implementation-defined.119) Commentary There is no requirement on an implementation to handle calls to a function defined with the inline function specifier any differently than calls to a function defined without one. This behavior parallels that for the register storage-class specifier.
[..]
Coding Guidelines Drawing parallels with the implementation-defined behavior of the register storage class would suggest that although this behavior is implementation-defined, no recommendation against its use be given. However, there is an argument for recommending the use of inline in some circumstances. Developers sometimes use macros because of a perceived performance advantage. Suggesting that an inline function be used instead may satisfy the perceived need for performance (whether or not the translator used performs any inlining is often not relevant), gaining the actual benefit of argument type checking and a nested scope for any object definitions.
1521 Any function with internal linkage can be an inline function.
Commentary The C Standard explicitly gives this permission because it goes on to list restrictions on inline functions with external linkage. [..]
1522 For a function with external linkage, the following restrictions apply: Commentary The following restrictions define a model that has differences from the one used by C++. Rationale Inlining was added to the Standard in such a way that it can be implemented with existing linker technology, and a subset of C99 inlining is compatible with C++. [..]
1526 Therefore, for example, the expansion of a macro used within the body of the function uses the definition it had at the point the function body appears, and not where the function is called; [..]
1531 An inline definition does not provide an external definition for the function, and does not forbid an external definition definition in another translation unit. Commentary An inline definition is explicitly specified as not being an external definition. The status of an inline function as an inline definition or an external definition does not affect the suggest it provides to a translator. An inline definition is intended for use only within the translation unit in which it occurs. Because it is not an external definition, the constraint requirement that only one external definition for an identifier occur in a program does not apply. However, if a function is used within an expression an external definition for it must exist somewhere in the entire program. The absence, in a function declaration, of inline or the inclusion of extern in a declaration creates such an external definition. Also, because an inline definition does not provide an external definition, another translation unit may have to contain an external definition (to satisfy references from other translation units).
1532 An inline definition provides an alternative to an external definition, which a translator may use to implement any call to the function in the same translation unit. Commentary The inline definition provides all the information (return type and parameter information) needed to call the external definition. Its body can also be used to perform inline substitution. An inline function might be said to have ghostly linkage. It exists if the translator believes in it. Otherwise it does not exist and the external definition is referenced. [..]
1533 It is unspecified whether a call to the function uses the inline definition or the external definition.120) Commentary The C Standard does not require that the sequence of tokens representing an inline definition or external definition, of the same function, be the same. However, the intended implication, to be drawn from the unspecified nature of the choice of the definition used, is that a programs external output shall be the same in both cases (apart from execution time performance). If the definitions of the two functions are such that the external program outputs would not be the same, the behavior is undefined. [..] [..]These extensions do offer some advantages; and programmers who are concerned about compatibility can simply abide by the stricter C++ rules. [..] Commentary Although fahr is an external definition, an implementation may still choose to inline calls to it, from within the definition of convert. [..]"