ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.53
Committed: Wed Dec 5 16:31:27 2012 UTC (11 years, 6 months ago) by root
Branch: MAIN
Changes since 1.52: +21 -0 lines
Log Message:
*** empty log message ***

File Contents

# User Rev Content
1 root 1.14 =head1 LIBECB - e-C-Builtins
2 root 1.3
3 root 1.14 =head2 ABOUT LIBECB
4    
5     Libecb is currently a simple header file that doesn't require any
6     configuration to use or include in your project.
7    
8 sf-exg 1.16 It's part of the e-suite of libraries, other members of which include
9 root 1.14 libev and libeio.
10    
11     Its homepage can be found here:
12    
13     http://software.schmorp.de/pkg/libecb
14    
15     It mainly provides a number of wrappers around GCC built-ins, together
16     with replacement functions for other compilers. In addition to this,
17 sf-exg 1.16 it provides a number of other lowlevel C utilities, such as endianness
18 root 1.14 detection, byte swapping or bit rotations.
19    
20 root 1.24 Or in other words, things that should be built into any standard C system,
21     but aren't, implemented as efficient as possible with GCC, and still
22     correct with other compilers.
23 root 1.17
24 root 1.14 More might come.
25 root 1.3
26     =head2 ABOUT THE HEADER
27    
28 root 1.14 At the moment, all you have to do is copy F<ecb.h> somewhere where your
29     compiler can find it and include it:
30    
31     #include <ecb.h>
32    
33     The header should work fine for both C and C++ compilation, and gives you
34     all of F<inttypes.h> in addition to the ECB symbols.
35    
36 sf-exg 1.16 There are currently no object files to link to - future versions might
37 root 1.14 come with an (optional) object code library to link against, to reduce
38     code size or gain access to additional features.
39    
40     It also currently includes everything from F<inttypes.h>.
41    
42     =head2 ABOUT THIS MANUAL / CONVENTIONS
43    
44     This manual mainly describes each (public) function available after
45     including the F<ecb.h> header. The header might define other symbols than
46     these, but these are not part of the public API, and not supported in any
47     way.
48    
49     When the manual mentions a "function" then this could be defined either as
50     as inline function, a macro, or an external symbol.
51    
52     When functions use a concrete standard type, such as C<int> or
53     C<uint32_t>, then the corresponding function works only with that type. If
54     only a generic name is used (C<expr>, C<cond>, C<value> and so on), then
55     the corresponding function relies on C to implement the correct types, and
56     is usually implemented as a macro. Specifically, a "bool" in this manual
57     refers to any kind of boolean value, not a specific type.
58 root 1.1
59 root 1.40 =head2 TYPES / TYPE SUPPORT
60    
61     ecb.h makes sure that the following types are defined (in the expected way):
62    
63 root 1.42 int8_t uint8_t int16_t uint16_t
64     int32_t uint32_t int64_t uint64_t
65 root 1.49 intptr_t uintptr_t
66 root 1.40
67     The macro C<ECB_PTRSIZE> is defined to the size of a pointer on this
68 root 1.45 platform (currently C<4> or C<8>) and can be used in preprocessor
69     expressions.
70 root 1.40
71 root 1.49 For C<ptrdiff_t> and C<size_t> use C<stddef.h>.
72    
73 root 1.43 =head2 LANGUAGE/COMPILER VERSIONS
74    
75 sf-exg 1.46 All the following symbols expand to an expression that can be tested in
76 root 1.44 preprocessor instructions as well as treated as a boolean (use C<!!> to
77     ensure it's either C<0> or C<1> if you need that).
78    
79 root 1.43 =over 4
80    
81 root 1.44 =item ECB_C
82    
83     True if the implementation defines the C<__STDC__> macro to a true value,
84     which is typically true for both C and C++ compilers.
85    
86 root 1.43 =item ECB_C99
87    
88 root 1.47 True if the implementation claims to be compliant to C99 (ISO/IEC
89     9899:1999) or any later version.
90    
91     Note that later versions (ECB_C11) remove core features again (for
92     example, variable length arrays).
93 root 1.43
94     =item ECB_C11
95    
96 root 1.47 True if the implementation claims to be compliant to C11 (ISO/IEC
97     9899:2011) or any later version.
98 root 1.44
99     =item ECB_CPP
100    
101     True if the implementation defines the C<__cplusplus__> macro to a true
102     value, which is typically true for C++ compilers.
103    
104     =item ECB_CPP11
105    
106     True if the implementation claims to be compliant to ISO/IEC 14882:2011
107 sf-exg 1.46 (C++11) or any later version.
108 root 1.43
109     =item ECB_GCC_VERSION(major,minor)
110    
111     Expands to a true value (suitable for testing in by the preprocessor)
112 sf-exg 1.46 if the compiler used is GNU C and the version is the given version, or
113 root 1.43 higher.
114    
115     This macro tries to return false on compilers that claim to be GCC
116     compatible but aren't.
117    
118 root 1.50 =item ECB_EXTERN_C
119    
120     Expands to C<extern "C"> in C++, and a simple C<extern> in C.
121    
122     This can be used to declare a single external C function:
123    
124     ECB_EXTERN_C int printf (const char *format, ...);
125    
126     =item ECB_EXTERN_C_BEG / ECB_EXTERN_C_END
127    
128     These two macros can be used to wrap multiple C<extern "C"> definitions -
129     they expand to nothing in C.
130    
131     They are most useful in header files:
132    
133     ECB_EXTERN_C_BEG
134    
135     int mycfun1 (int x);
136     int mycfun2 (int x);
137    
138     ECB_EXTERN_C_END
139    
140     =item ECB_STDFP
141    
142     If this evaluates to a true value (suitable for testing in by the
143     preprocessor), then C<float> and C<double> use IEEE 754 single/binary32
144     and double/binary64 representations internally I<and> the endianness of
145     both types match the endianness of C<uint32_t> and C<uint64_t>.
146    
147     This means you can just copy the bits of a C<float> (or C<double>) to an
148     C<uint32_t> (or C<uint64_t>) and get the raw IEEE 754 bit representation
149     without having to think about format or endianness.
150    
151     This is true for basically all modern platforms, although F<ecb.h> might
152     not be able to deduce this correctly everywhere and might err on the safe
153     side.
154    
155 root 1.43 =back
156    
157 root 1.1 =head2 GCC ATTRIBUTES
158    
159 root 1.20 A major part of libecb deals with GCC attributes. These are additional
160 sf-exg 1.26 attributes that you can assign to functions, variables and sometimes even
161 root 1.20 types - much like C<const> or C<volatile> in C.
162    
163     While GCC allows declarations to show up in many surprising places,
164 sf-exg 1.26 but not in many expected places, the safest way is to put attribute
165 root 1.20 declarations before the whole declaration:
166    
167     ecb_const int mysqrt (int a);
168     ecb_unused int i;
169    
170     For variables, it is often nicer to put the attribute after the name, and
171     avoid multiple declarations using commas:
172    
173     int i ecb_unused;
174 root 1.3
175 root 1.1 =over 4
176    
177 root 1.2 =item ecb_attribute ((attrs...))
178 root 1.1
179 root 1.15 A simple wrapper that expands to C<__attribute__((attrs))> on GCC, and to
180     nothing on other compilers, so the effect is that only GCC sees these.
181    
182     Example: use the C<deprecated> attribute on a function.
183    
184     ecb_attribute((__deprecated__)) void
185     do_not_use_me_anymore (void);
186 root 1.2
187 root 1.3 =item ecb_unused
188    
189     Marks a function or a variable as "unused", which simply suppresses a
190     warning by GCC when it detects it as unused. This is useful when you e.g.
191     declare a variable but do not always use it:
192    
193 root 1.15 {
194     int var ecb_unused;
195 root 1.3
196 root 1.15 #ifdef SOMECONDITION
197     var = ...;
198     return var;
199     #else
200     return 0;
201     #endif
202     }
203 root 1.3
204 root 1.31 =item ecb_inline
205 root 1.29
206     This is not actually an attribute, but you use it like one. It expands
207     either to C<static inline> or to just C<static>, if inline isn't
208     supported. It should be used to declare functions that should be inlined,
209     for code size or speed reasons.
210    
211     Example: inline this function, it surely will reduce codesize.
212    
213 root 1.31 ecb_inline int
214 root 1.29 negmul (int a, int b)
215     {
216     return - (a * b);
217     }
218    
219 root 1.2 =item ecb_noinline
220    
221 root 1.9 Prevent a function from being inlined - it might be optimised away, but
222 root 1.3 not inlined into other functions. This is useful if you know your function
223     is rarely called and large enough for inlining not to be helpful.
224    
225 root 1.2 =item ecb_noreturn
226    
227 root 1.17 Marks a function as "not returning, ever". Some typical functions that
228     don't return are C<exit> or C<abort> (which really works hard to not
229     return), and now you can make your own:
230    
231     ecb_noreturn void
232     my_abort (const char *errline)
233     {
234     puts (errline);
235     abort ();
236     }
237    
238 sf-exg 1.19 In this case, the compiler would probably be smart enough to deduce it on
239     its own, so this is mainly useful for declarations.
240 root 1.17
241 root 1.53 =item ecb_restrict
242    
243     Expands to the C<restrict> keyword or equivalent on compilers that support
244     them, and to nothing on others. Must be specified on a pointer type or
245     an array index to indicate that the memory doesn't alias with any other
246     restricted pointer in the same scope.
247    
248     Example: multiply a vector, and allow the compiler to parallelise the
249     loop, because it knows it doesn't overwrite input values.
250    
251     void
252     multiply (float *ecb_restrict src,
253     float *ecb_restrict dst,
254     int len, float factor)
255     {
256     int i;
257    
258     for (i = 0; i < len; ++i)
259     dst [i] = src [i] * factor;
260     }
261    
262 root 1.2 =item ecb_const
263    
264 sf-exg 1.19 Declares that the function only depends on the values of its arguments,
265 root 1.17 much like a mathematical function. It specifically does not read or write
266     any memory any arguments might point to, global variables, or call any
267     non-const functions. It also must not have any side effects.
268    
269     Such a function can be optimised much more aggressively by the compiler -
270     for example, multiple calls with the same arguments can be optimised into
271     a single call, which wouldn't be possible if the compiler would have to
272     expect any side effects.
273    
274     It is best suited for functions in the sense of mathematical functions,
275 sf-exg 1.19 such as a function returning the square root of its input argument.
276 root 1.17
277     Not suited would be a function that calculates the hash of some memory
278     area you pass in, prints some messages or looks at a global variable to
279     decide on rounding.
280    
281     See C<ecb_pure> for a slightly less restrictive class of functions.
282    
283 root 1.2 =item ecb_pure
284    
285 root 1.17 Similar to C<ecb_const>, declares a function that has no side
286     effects. Unlike C<ecb_const>, the function is allowed to examine global
287     variables and any other memory areas (such as the ones passed to it via
288     pointers).
289    
290     While these functions cannot be optimised as aggressively as C<ecb_const>
291     functions, they can still be optimised away in many occasions, and the
292     compiler has more freedom in moving calls to them around.
293    
294     Typical examples for such functions would be C<strlen> or C<memcmp>. A
295     function that calculates the MD5 sum of some input and updates some MD5
296     state passed as argument would I<NOT> be pure, however, as it would modify
297     some memory area that is not the return value.
298    
299 root 1.2 =item ecb_hot
300    
301 root 1.17 This declares a function as "hot" with regards to the cache - the function
302     is used so often, that it is very beneficial to keep it in the cache if
303     possible.
304    
305     The compiler reacts by trying to place hot functions near to each other in
306     memory.
307    
308 sf-exg 1.19 Whether a function is hot or not often depends on the whole program,
309 root 1.17 and less on the function itself. C<ecb_cold> is likely more useful in
310     practise.
311    
312 root 1.2 =item ecb_cold
313    
314 root 1.17 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
315     the cache, or in other words, this function is not called often, or not at
316     speed-critical times, and keeping it in the cache might be a waste of said
317     cache.
318    
319     In addition to placing cold functions together (or at least away from hot
320     functions), this knowledge can be used in other ways, for example, the
321     function will be optimised for size, as opposed to speed, and codepaths
322     leading to calls to those functions can automatically be marked as if
323 root 1.27 C<ecb_expect_false> had been used to reach them.
324 root 1.17
325     Good examples for such functions would be error reporting functions, or
326     functions only called in exceptional or rare cases.
327    
328 root 1.2 =item ecb_artificial
329    
330 root 1.17 Declares the function as "artificial", in this case meaning that this
331 root 1.52 function is not really meant to be a function, but more like an accessor
332 root 1.17 - many methods in C++ classes are mere accessor functions, and having a
333     crash reported in such a method, or single-stepping through them, is not
334     usually so helpful, especially when it's inlined to just a few instructions.
335    
336     Marking them as artificial will instruct the debugger about just this,
337     leading to happier debugging and thus happier lives.
338    
339     Example: in some kind of smart-pointer class, mark the pointer accessor as
340     artificial, so that the whole class acts more like a pointer and less like
341     some C++ abstraction monster.
342    
343     template<typename T>
344     struct my_smart_ptr
345     {
346     T *value;
347    
348     ecb_artificial
349     operator T *()
350     {
351     return value;
352     }
353     };
354    
355 root 1.2 =back
356 root 1.1
357     =head2 OPTIMISATION HINTS
358    
359     =over 4
360    
361 root 1.14 =item bool ecb_is_constant(expr)
362 root 1.1
363 root 1.3 Returns true iff the expression can be deduced to be a compile-time
364     constant, and false otherwise.
365    
366     For example, when you have a C<rndm16> function that returns a 16 bit
367     random number, and you have a function that maps this to a range from
368 root 1.5 0..n-1, then you could use this inline function in a header file:
369 root 1.3
370     ecb_inline uint32_t
371     rndm (uint32_t n)
372     {
373 root 1.6 return (n * (uint32_t)rndm16 ()) >> 16;
374 root 1.3 }
375    
376     However, for powers of two, you could use a normal mask, but that is only
377     worth it if, at compile time, you can detect this case. This is the case
378     when the passed number is a constant and also a power of two (C<n & (n -
379     1) == 0>):
380    
381     ecb_inline uint32_t
382     rndm (uint32_t n)
383     {
384     return is_constant (n) && !(n & (n - 1))
385     ? rndm16 () & (num - 1)
386 root 1.6 : (n * (uint32_t)rndm16 ()) >> 16;
387 root 1.3 }
388    
389 root 1.14 =item bool ecb_expect (expr, value)
390 root 1.1
391 root 1.7 Evaluates C<expr> and returns it. In addition, it tells the compiler that
392     the C<expr> evaluates to C<value> a lot, which can be used for static
393     branch optimisations.
394 root 1.1
395 root 1.27 Usually, you want to use the more intuitive C<ecb_expect_true> and
396     C<ecb_expect_false> functions instead.
397 root 1.1
398 root 1.27 =item bool ecb_expect_true (cond)
399 root 1.1
400 root 1.27 =item bool ecb_expect_false (cond)
401 root 1.1
402 root 1.7 These two functions expect a expression that is true or false and return
403     C<1> or C<0>, respectively, so when used in the condition of an C<if> or
404     other conditional statement, it will not change the program:
405    
406     /* these two do the same thing */
407     if (some_condition) ...;
408 root 1.27 if (ecb_expect_true (some_condition)) ...;
409 root 1.7
410 root 1.27 However, by using C<ecb_expect_true>, you tell the compiler that the
411     condition is likely to be true (and for C<ecb_expect_false>, that it is
412     unlikely to be true).
413 root 1.7
414 root 1.9 For example, when you check for a null pointer and expect this to be a
415 root 1.27 rare, exceptional, case, then use C<ecb_expect_false>:
416 root 1.7
417     void my_free (void *ptr)
418     {
419 root 1.27 if (ecb_expect_false (ptr == 0))
420 root 1.7 return;
421     }
422    
423     Consequent use of these functions to mark away exceptional cases or to
424     tell the compiler what the hot path through a function is can increase
425     performance considerably.
426    
427 root 1.27 You might know these functions under the name C<likely> and C<unlikely>
428     - while these are common aliases, we find that the expect name is easier
429     to understand when quickly skimming code. If you wish, you can use
430     C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
431     C<ecb_expect_false> - these are simply aliases.
432    
433 root 1.7 A very good example is in a function that reserves more space for some
434     memory block (for example, inside an implementation of a string stream) -
435 root 1.9 each time something is added, you have to check for a buffer overrun, but
436 root 1.7 you expect that most checks will turn out to be false:
437    
438     /* make sure we have "size" extra room in our buffer */
439     ecb_inline void
440     reserve (int size)
441     {
442 root 1.27 if (ecb_expect_false (current + size > end))
443 root 1.7 real_reserve_method (size); /* presumably noinline */
444     }
445    
446 root 1.14 =item bool ecb_assume (cond)
447 root 1.7
448     Try to tell the compiler that some condition is true, even if it's not
449     obvious.
450    
451     This can be used to teach the compiler about invariants or other
452     conditions that might improve code generation, but which are impossible to
453     deduce form the code itself.
454    
455 root 1.27 For example, the example reservation function from the C<ecb_expect_false>
456 root 1.7 description could be written thus (only C<ecb_assume> was added):
457    
458     ecb_inline void
459     reserve (int size)
460     {
461 root 1.27 if (ecb_expect_false (current + size > end))
462 root 1.7 real_reserve_method (size); /* presumably noinline */
463    
464     ecb_assume (current + size <= end);
465     }
466    
467     If you then call this function twice, like this:
468    
469     reserve (10);
470     reserve (1);
471    
472     Then the compiler I<might> be able to optimise out the second call
473     completely, as it knows that C<< current + 1 > end >> is false and the
474     call will never be executed.
475    
476     =item bool ecb_unreachable ()
477    
478     This function does nothing itself, except tell the compiler that it will
479 root 1.9 never be executed. Apart from suppressing a warning in some cases, this
480 root 1.7 function can be used to implement C<ecb_assume> or similar functions.
481    
482 root 1.14 =item bool ecb_prefetch (addr, rw, locality)
483 root 1.7
484     Tells the compiler to try to prefetch memory at the given C<addr>ess
485 root 1.10 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
486 root 1.7 C<0> means that there will only be one access later, C<3> means that
487     the data will likely be accessed very often, and values in between mean
488     something... in between. The memory pointed to by the address does not
489     need to be accessible (it could be a null pointer for example), but C<rw>
490     and C<locality> must be compile-time constants.
491    
492     An obvious way to use this is to prefetch some data far away, in a big
493 root 1.9 array you loop over. This prefetches memory some 128 array elements later,
494 root 1.7 in the hope that it will be ready when the CPU arrives at that location.
495    
496     int sum = 0;
497    
498     for (i = 0; i < N; ++i)
499     {
500     sum += arr [i]
501     ecb_prefetch (arr + i + 128, 0, 0);
502     }
503    
504     It's hard to predict how far to prefetch, and most CPUs that can prefetch
505     are often good enough to predict this kind of behaviour themselves. It
506     gets more interesting with linked lists, especially when you do some fair
507     processing on each list element:
508    
509     for (node *n = start; n; n = n->next)
510     {
511     ecb_prefetch (n->next, 0, 0);
512     ... do medium amount of work with *n
513     }
514    
515     After processing the node, (part of) the next node might already be in
516     cache.
517 root 1.1
518 root 1.2 =back
519 root 1.1
520 root 1.36 =head2 BIT FIDDLING / BIT WIZARDRY
521 root 1.1
522 root 1.4 =over 4
523    
524 root 1.3 =item bool ecb_big_endian ()
525    
526     =item bool ecb_little_endian ()
527    
528 sf-exg 1.11 These two functions return true if the byte order is big endian
529     (most-significant byte first) or little endian (least-significant byte
530     first) respectively.
531    
532 root 1.24 On systems that are neither, their return values are unspecified.
533    
534 root 1.3 =item int ecb_ctz32 (uint32_t x)
535    
536 root 1.35 =item int ecb_ctz64 (uint64_t x)
537    
538 sf-exg 1.11 Returns the index of the least significant bit set in C<x> (or
539 root 1.24 equivalently the number of bits set to 0 before the least significant bit
540 root 1.35 set), starting from 0. If C<x> is 0 the result is undefined.
541    
542 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
543    
544 root 1.35 For example:
545 sf-exg 1.11
546 root 1.15 ecb_ctz32 (3) = 0
547     ecb_ctz32 (6) = 1
548 sf-exg 1.11
549 root 1.41 =item bool ecb_is_pot32 (uint32_t x)
550    
551     =item bool ecb_is_pot64 (uint32_t x)
552    
553     Return true iff C<x> is a power of two or C<x == 0>.
554    
555     For smaller types then C<uint32_t> you can safely use C<ecb_is_pot32>.
556    
557 root 1.35 =item int ecb_ld32 (uint32_t x)
558    
559     =item int ecb_ld64 (uint64_t x)
560    
561     Returns the index of the most significant bit set in C<x>, or the number
562     of digits the number requires in binary (so that C<< 2**ld <= x <
563     2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
564     to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
565     example to see how many bits a certain number requires to be encoded.
566    
567     This function is similar to the "count leading zero bits" function, except
568     that that one returns how many zero bits are "in front" of the number (in
569     the given data type), while C<ecb_ld> returns how many bits the number
570     itself requires.
571    
572 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
573    
574 root 1.3 =item int ecb_popcount32 (uint32_t x)
575    
576 root 1.35 =item int ecb_popcount64 (uint64_t x)
577    
578 root 1.36 Returns the number of bits set to 1 in C<x>.
579    
580     For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
581    
582     For example:
583 sf-exg 1.11
584 root 1.15 ecb_popcount32 (7) = 3
585     ecb_popcount32 (255) = 8
586 sf-exg 1.11
587 root 1.39 =item uint8_t ecb_bitrev8 (uint8_t x)
588    
589     =item uint16_t ecb_bitrev16 (uint16_t x)
590    
591     =item uint32_t ecb_bitrev32 (uint32_t x)
592    
593     Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
594     and so on.
595    
596     Example:
597    
598     ecb_bitrev8 (0xa7) = 0xea
599     ecb_bitrev32 (0xffcc4411) = 0x882233ff
600    
601 root 1.8 =item uint32_t ecb_bswap16 (uint32_t x)
602    
603 root 1.3 =item uint32_t ecb_bswap32 (uint32_t x)
604    
605 root 1.34 =item uint64_t ecb_bswap64 (uint64_t x)
606 sf-exg 1.13
607 root 1.34 These functions return the value of the 16-bit (32-bit, 64-bit) value
608     C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
609     C<ecb_bswap32>).
610    
611     =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
612    
613     =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
614 root 1.3
615     =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
616    
617 root 1.34 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
618    
619     =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
620    
621     =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
622    
623     =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
624    
625 root 1.33 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
626    
627 root 1.34 These two families of functions return the value of C<x> after rotating
628     all the bits by C<count> positions to the right (C<ecb_rotr>) or left
629     (C<ecb_rotl>).
630 sf-exg 1.11
631 root 1.20 Current GCC versions understand these functions and usually compile them
632 root 1.34 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
633     x86).
634 root 1.20
635 root 1.3 =back
636 root 1.1
637 root 1.50 =head2 FLOATING POINT FIDDLING
638    
639     =over 4
640    
641     =item uint32_t ecb_float_to_binary32 (float x) [-UECB_NO_LIBM]
642    
643     =item uint64_t ecb_double_to_binary64 (double x) [-UECB_NO_LIBM]
644    
645     These functions each take an argument in the native C<float> or C<double>
646     type and return the IEEE 754 bit representation of it.
647    
648     The bit representation is just as IEEE 754 defines it, i.e. the sign bit
649     will be the most significant bit, followed by exponent and mantissa.
650    
651     This function should work even when the native floating point format isn't
652     IEEE compliant, of course at a speed and code size penalty, and of course
653     also within reasonable limits (it tries to convert NaNs, infinities and
654     denormals, but will likely convert negative zero to positive zero).
655    
656     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
657     be able to optimise away this function completely.
658    
659     These functions can be helpful when serialising floats to the network - you
660     can serialise the return value like a normal uint32_t/uint64_t.
661    
662     Another use for these functions is to manipulate floating point values
663     directly.
664    
665     Silly example: toggle the sign bit of a float.
666    
667     /* On gcc-4.7 on amd64, */
668     /* this results in a single add instruction to toggle the bit, and 4 extra */
669     /* instructions to move the float value to an integer register and back. */
670    
671     x = ecb_binary32_to_float (ecb_float_to_binary32 (x) ^ 0x80000000U)
672    
673     =item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM]
674    
675     =item double ecb_binary32_to_double (uint64_t x) [-UECB_NO_LIBM]
676    
677     The reverse operation of the previos function - takes the bit representation
678     of an IEEE binary32 or binary64 number and converts it to the native C<float>
679     or C<double> format.
680    
681     This function should work even when the native floating point format isn't
682     IEEE compliant, of course at a speed and code size penalty, and of course
683     also within reasonable limits (it tries to convert normals and denormals,
684     and might be lucky for infinities, and with extraordinary luck, also for
685     negative zero).
686    
687     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
688     be able to optimise away this function completely.
689    
690     =back
691    
692 root 1.1 =head2 ARITHMETIC
693    
694 root 1.3 =over 4
695    
696 root 1.14 =item x = ecb_mod (m, n)
697 root 1.3
698 root 1.25 Returns C<m> modulo C<n>, which is the same as the positive remainder
699     of the division operation between C<m> and C<n>, using floored
700     division. Unlike the C remainder operator C<%>, this function ensures that
701     the return value is always positive and that the two numbers I<m> and
702     I<m' = m + i * n> result in the same value modulo I<n> - in other words,
703     C<ecb_mod> implements the mathematical modulo operation, which is missing
704     in the language.
705 root 1.14
706 sf-exg 1.23 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
707 root 1.14 negatable, that is, both C<m> and C<-m> must be representable in its
708 root 1.30 type (this typically excludes the minimum signed integer value, the same
709 root 1.25 limitation as for C</> and C<%> in C).
710 sf-exg 1.11
711 root 1.24 Current GCC versions compile this into an efficient branchless sequence on
712 root 1.28 almost all CPUs.
713 root 1.24
714     For example, when you want to rotate forward through the members of an
715     array for increasing C<m> (which might be negative), then you should use
716     C<ecb_mod>, as the C<%> operator might give either negative results, or
717     change direction for negative values:
718    
719     for (m = -100; m <= 100; ++m)
720     int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
721    
722 sf-exg 1.37 =item x = ecb_div_rd (val, div)
723    
724     =item x = ecb_div_ru (val, div)
725    
726     Returns C<val> divided by C<div> rounded down or up, respectively.
727     C<val> and C<div> must have integer types and C<div> must be strictly
728 sf-exg 1.38 positive. Note that these functions are implemented with macros in C
729     and with function templates in C++.
730 sf-exg 1.37
731 root 1.3 =back
732 root 1.1
733     =head2 UTILITY
734    
735 root 1.3 =over 4
736    
737 sf-exg 1.23 =item element_count = ecb_array_length (name)
738 root 1.3
739 sf-exg 1.13 Returns the number of elements in the array C<name>. For example:
740    
741     int primes[] = { 2, 3, 5, 7, 11 };
742     int sum = 0;
743    
744     for (i = 0; i < ecb_array_length (primes); i++)
745     sum += primes [i];
746    
747 root 1.3 =back
748 root 1.1
749 root 1.43 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
750    
751     These symbols need to be defined before including F<ecb.h> the first time.
752    
753     =over 4
754    
755 root 1.51 =item ECB_NO_THREADS
756 root 1.43
757     If F<ecb.h> is never used from multiple threads, then this symbol can
758     be defined, in which case memory fences (and similar constructs) are
759     completely removed, leading to more efficient code and fewer dependencies.
760    
761     Setting this symbol to a true value implies C<ECB_NO_SMP>.
762    
763     =item ECB_NO_SMP
764    
765     The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
766     multiple threads, but never concurrently (e.g. if the system the program
767     runs on has only a single CPU with a single core, no hyperthreading and so
768     on), then this symbol can be defined, leading to more efficient code and
769     fewer dependencies.
770    
771 root 1.50 =item ECB_NO_LIBM
772    
773     When defined to C<1>, do not export any functions that might introduce
774     dependencies on the math library (usually called F<-lm>) - these are
775     marked with [-UECB_NO_LIBM].
776    
777 root 1.43 =back
778    
779 root 1.1