ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.63
Committed: Wed Feb 18 20:29:38 2015 UTC (9 years, 4 months ago) by root
Branch: MAIN
Changes since 1.62: +1 -1 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 1.62 =head2 LANGUAGE/ENVIRONMENT/COMPILER VERSIONS
74 root 1.43
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 root 1.55 while not claiming to be C++.
85 root 1.44
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 root 1.55 9899:1999) or any later version, while not claiming to be C++.
90 root 1.47
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 root 1.55 9899:2011) or any later version, while not claiming to be C++.
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 root 1.57 =item ECB_GCC_VERSION (major, minor)
110 root 1.43
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 1.54 =item ECB_AMD64, ECB_AMD64_X32
156    
157     These two macros are defined to C<1> on the x86_64/amd64 ABI and the X32
158     ABI, respectively, and undefined elsewhere.
159    
160     The designers of the new X32 ABI for some inexplicable reason decided to
161     make it look exactly like amd64, even though it's completely incompatible
162     to that ABI, breaking about every piece of software that assumed that
163     C<__x86_64> stands for, well, the x86-64 ABI, making these macros
164     necessary.
165    
166 root 1.43 =back
167    
168 root 1.62 =head2 MACRO TRICKERY
169    
170     =over 4
171    
172     =item ECB_CONCAT (a, b)
173    
174     Expands any macros in C<a> and C<b>, then concatenates the result to form
175     a single token. This is mainly useful to form identifiers from components,
176     e.g.:
177    
178     #define S1 str
179     #define S2 cpy
180    
181     ECB_CONCAT (S1, S2)(dst, src); // == strcpy (dst, src);
182    
183     =item ECB_STRINGIFY (arg)
184    
185     Expands any macros in C<arg> and returns the stringified version of
186     it. This is mainly useful to get the contents of a macro in string form,
187     e.g.:
188    
189     #define SQL_LIMIT 100
190     sql_exec ("select * from table limit " ECB_STRINGIFY (SQL_LIMIT));
191    
192     =back
193    
194 sf-exg 1.60 =head2 ATTRIBUTES
195 root 1.1
196 sf-exg 1.60 A major part of libecb deals with additional attributes that can be
197     assigned to functions, variables and sometimes even types - much like
198     C<const> or C<volatile> in C. They are implemented using either GCC
199     attributes or other compiler/language specific features. Attributes
200     declarations must be put before the whole declaration:
201 root 1.20
202     ecb_const int mysqrt (int a);
203     ecb_unused int i;
204    
205 root 1.1 =over 4
206    
207 root 1.3 =item ecb_unused
208    
209     Marks a function or a variable as "unused", which simply suppresses a
210     warning by GCC when it detects it as unused. This is useful when you e.g.
211     declare a variable but do not always use it:
212    
213 root 1.15 {
214 sf-exg 1.61 ecb_unused int var;
215 root 1.3
216 root 1.15 #ifdef SOMECONDITION
217     var = ...;
218     return var;
219     #else
220     return 0;
221     #endif
222     }
223 root 1.3
224 root 1.56 =item ecb_deprecated
225    
226     Similar to C<ecb_unused>, but marks a function, variable or type as
227     deprecated. This makes some compilers warn when the type is used.
228    
229 root 1.62 =item ecb_deprecated_message (message)
230    
231     Same as C<ecb_deprecated>, but if possible, supply a diagnostic that is
232     used instead of a generic depreciation message when the object is being
233     used.
234    
235 root 1.31 =item ecb_inline
236 root 1.29
237 sf-exg 1.60 Expands either to C<static inline> or to just C<static>, if inline
238     isn't supported. It should be used to declare functions that should be
239     inlined, for code size or speed reasons.
240 root 1.29
241     Example: inline this function, it surely will reduce codesize.
242    
243 root 1.31 ecb_inline int
244 root 1.29 negmul (int a, int b)
245     {
246     return - (a * b);
247     }
248    
249 root 1.2 =item ecb_noinline
250    
251 root 1.9 Prevent a function from being inlined - it might be optimised away, but
252 root 1.3 not inlined into other functions. This is useful if you know your function
253     is rarely called and large enough for inlining not to be helpful.
254    
255 root 1.2 =item ecb_noreturn
256    
257 root 1.17 Marks a function as "not returning, ever". Some typical functions that
258     don't return are C<exit> or C<abort> (which really works hard to not
259     return), and now you can make your own:
260    
261     ecb_noreturn void
262     my_abort (const char *errline)
263     {
264     puts (errline);
265     abort ();
266     }
267    
268 sf-exg 1.19 In this case, the compiler would probably be smart enough to deduce it on
269     its own, so this is mainly useful for declarations.
270 root 1.17
271 root 1.53 =item ecb_restrict
272    
273     Expands to the C<restrict> keyword or equivalent on compilers that support
274     them, and to nothing on others. Must be specified on a pointer type or
275     an array index to indicate that the memory doesn't alias with any other
276     restricted pointer in the same scope.
277    
278     Example: multiply a vector, and allow the compiler to parallelise the
279     loop, because it knows it doesn't overwrite input values.
280    
281     void
282 sf-exg 1.61 multiply (ecb_restrict float *src,
283     ecb_restrict float *dst,
284 root 1.53 int len, float factor)
285     {
286     int i;
287    
288     for (i = 0; i < len; ++i)
289     dst [i] = src [i] * factor;
290     }
291    
292 root 1.2 =item ecb_const
293    
294 sf-exg 1.19 Declares that the function only depends on the values of its arguments,
295 root 1.17 much like a mathematical function. It specifically does not read or write
296     any memory any arguments might point to, global variables, or call any
297     non-const functions. It also must not have any side effects.
298    
299     Such a function can be optimised much more aggressively by the compiler -
300     for example, multiple calls with the same arguments can be optimised into
301     a single call, which wouldn't be possible if the compiler would have to
302     expect any side effects.
303    
304     It is best suited for functions in the sense of mathematical functions,
305 sf-exg 1.19 such as a function returning the square root of its input argument.
306 root 1.17
307     Not suited would be a function that calculates the hash of some memory
308     area you pass in, prints some messages or looks at a global variable to
309     decide on rounding.
310    
311     See C<ecb_pure> for a slightly less restrictive class of functions.
312    
313 root 1.2 =item ecb_pure
314    
315 root 1.17 Similar to C<ecb_const>, declares a function that has no side
316     effects. Unlike C<ecb_const>, the function is allowed to examine global
317     variables and any other memory areas (such as the ones passed to it via
318     pointers).
319    
320     While these functions cannot be optimised as aggressively as C<ecb_const>
321     functions, they can still be optimised away in many occasions, and the
322     compiler has more freedom in moving calls to them around.
323    
324     Typical examples for such functions would be C<strlen> or C<memcmp>. A
325     function that calculates the MD5 sum of some input and updates some MD5
326     state passed as argument would I<NOT> be pure, however, as it would modify
327     some memory area that is not the return value.
328    
329 root 1.2 =item ecb_hot
330    
331 root 1.17 This declares a function as "hot" with regards to the cache - the function
332     is used so often, that it is very beneficial to keep it in the cache if
333     possible.
334    
335     The compiler reacts by trying to place hot functions near to each other in
336     memory.
337    
338 sf-exg 1.19 Whether a function is hot or not often depends on the whole program,
339 root 1.17 and less on the function itself. C<ecb_cold> is likely more useful in
340     practise.
341    
342 root 1.2 =item ecb_cold
343    
344 root 1.17 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
345     the cache, or in other words, this function is not called often, or not at
346     speed-critical times, and keeping it in the cache might be a waste of said
347     cache.
348    
349     In addition to placing cold functions together (or at least away from hot
350     functions), this knowledge can be used in other ways, for example, the
351     function will be optimised for size, as opposed to speed, and codepaths
352     leading to calls to those functions can automatically be marked as if
353 root 1.27 C<ecb_expect_false> had been used to reach them.
354 root 1.17
355     Good examples for such functions would be error reporting functions, or
356     functions only called in exceptional or rare cases.
357    
358 root 1.2 =item ecb_artificial
359    
360 root 1.17 Declares the function as "artificial", in this case meaning that this
361 root 1.52 function is not really meant to be a function, but more like an accessor
362 root 1.17 - many methods in C++ classes are mere accessor functions, and having a
363     crash reported in such a method, or single-stepping through them, is not
364     usually so helpful, especially when it's inlined to just a few instructions.
365    
366     Marking them as artificial will instruct the debugger about just this,
367     leading to happier debugging and thus happier lives.
368    
369     Example: in some kind of smart-pointer class, mark the pointer accessor as
370     artificial, so that the whole class acts more like a pointer and less like
371     some C++ abstraction monster.
372    
373     template<typename T>
374     struct my_smart_ptr
375     {
376     T *value;
377    
378     ecb_artificial
379     operator T *()
380     {
381     return value;
382     }
383     };
384    
385 root 1.2 =back
386 root 1.1
387     =head2 OPTIMISATION HINTS
388    
389     =over 4
390    
391 root 1.58 =item bool ecb_is_constant (expr)
392 root 1.1
393 root 1.3 Returns true iff the expression can be deduced to be a compile-time
394     constant, and false otherwise.
395    
396     For example, when you have a C<rndm16> function that returns a 16 bit
397     random number, and you have a function that maps this to a range from
398 root 1.5 0..n-1, then you could use this inline function in a header file:
399 root 1.3
400     ecb_inline uint32_t
401     rndm (uint32_t n)
402     {
403 root 1.6 return (n * (uint32_t)rndm16 ()) >> 16;
404 root 1.3 }
405    
406     However, for powers of two, you could use a normal mask, but that is only
407     worth it if, at compile time, you can detect this case. This is the case
408     when the passed number is a constant and also a power of two (C<n & (n -
409     1) == 0>):
410    
411     ecb_inline uint32_t
412     rndm (uint32_t n)
413     {
414     return is_constant (n) && !(n & (n - 1))
415     ? rndm16 () & (num - 1)
416 root 1.6 : (n * (uint32_t)rndm16 ()) >> 16;
417 root 1.3 }
418    
419 root 1.62 =item ecb_expect (expr, value)
420 root 1.1
421 root 1.7 Evaluates C<expr> and returns it. In addition, it tells the compiler that
422     the C<expr> evaluates to C<value> a lot, which can be used for static
423     branch optimisations.
424 root 1.1
425 root 1.27 Usually, you want to use the more intuitive C<ecb_expect_true> and
426     C<ecb_expect_false> functions instead.
427 root 1.1
428 root 1.27 =item bool ecb_expect_true (cond)
429 root 1.1
430 root 1.27 =item bool ecb_expect_false (cond)
431 root 1.1
432 root 1.7 These two functions expect a expression that is true or false and return
433     C<1> or C<0>, respectively, so when used in the condition of an C<if> or
434     other conditional statement, it will not change the program:
435    
436     /* these two do the same thing */
437     if (some_condition) ...;
438 root 1.27 if (ecb_expect_true (some_condition)) ...;
439 root 1.7
440 root 1.27 However, by using C<ecb_expect_true>, you tell the compiler that the
441     condition is likely to be true (and for C<ecb_expect_false>, that it is
442     unlikely to be true).
443 root 1.7
444 root 1.9 For example, when you check for a null pointer and expect this to be a
445 root 1.27 rare, exceptional, case, then use C<ecb_expect_false>:
446 root 1.7
447     void my_free (void *ptr)
448     {
449 root 1.27 if (ecb_expect_false (ptr == 0))
450 root 1.7 return;
451     }
452    
453     Consequent use of these functions to mark away exceptional cases or to
454     tell the compiler what the hot path through a function is can increase
455     performance considerably.
456    
457 root 1.27 You might know these functions under the name C<likely> and C<unlikely>
458     - while these are common aliases, we find that the expect name is easier
459     to understand when quickly skimming code. If you wish, you can use
460     C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
461     C<ecb_expect_false> - these are simply aliases.
462    
463 root 1.7 A very good example is in a function that reserves more space for some
464     memory block (for example, inside an implementation of a string stream) -
465 root 1.9 each time something is added, you have to check for a buffer overrun, but
466 root 1.7 you expect that most checks will turn out to be false:
467    
468     /* make sure we have "size" extra room in our buffer */
469     ecb_inline void
470     reserve (int size)
471     {
472 root 1.27 if (ecb_expect_false (current + size > end))
473 root 1.7 real_reserve_method (size); /* presumably noinline */
474     }
475    
476 root 1.62 =item ecb_assume (cond)
477 root 1.7
478     Try to tell the compiler that some condition is true, even if it's not
479     obvious.
480    
481     This can be used to teach the compiler about invariants or other
482     conditions that might improve code generation, but which are impossible to
483     deduce form the code itself.
484    
485 root 1.27 For example, the example reservation function from the C<ecb_expect_false>
486 root 1.7 description could be written thus (only C<ecb_assume> was added):
487    
488     ecb_inline void
489     reserve (int size)
490     {
491 root 1.27 if (ecb_expect_false (current + size > end))
492 root 1.7 real_reserve_method (size); /* presumably noinline */
493    
494     ecb_assume (current + size <= end);
495     }
496    
497     If you then call this function twice, like this:
498    
499     reserve (10);
500     reserve (1);
501    
502     Then the compiler I<might> be able to optimise out the second call
503     completely, as it knows that C<< current + 1 > end >> is false and the
504     call will never be executed.
505    
506 root 1.62 =item ecb_unreachable ()
507 root 1.7
508     This function does nothing itself, except tell the compiler that it will
509 root 1.9 never be executed. Apart from suppressing a warning in some cases, this
510 root 1.7 function can be used to implement C<ecb_assume> or similar functions.
511    
512 root 1.62 =item ecb_prefetch (addr, rw, locality)
513 root 1.7
514     Tells the compiler to try to prefetch memory at the given C<addr>ess
515 root 1.10 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
516 root 1.7 C<0> means that there will only be one access later, C<3> means that
517     the data will likely be accessed very often, and values in between mean
518     something... in between. The memory pointed to by the address does not
519     need to be accessible (it could be a null pointer for example), but C<rw>
520     and C<locality> must be compile-time constants.
521    
522     An obvious way to use this is to prefetch some data far away, in a big
523 root 1.9 array you loop over. This prefetches memory some 128 array elements later,
524 root 1.7 in the hope that it will be ready when the CPU arrives at that location.
525    
526     int sum = 0;
527    
528     for (i = 0; i < N; ++i)
529     {
530     sum += arr [i]
531     ecb_prefetch (arr + i + 128, 0, 0);
532     }
533    
534     It's hard to predict how far to prefetch, and most CPUs that can prefetch
535     are often good enough to predict this kind of behaviour themselves. It
536     gets more interesting with linked lists, especially when you do some fair
537     processing on each list element:
538    
539     for (node *n = start; n; n = n->next)
540     {
541     ecb_prefetch (n->next, 0, 0);
542     ... do medium amount of work with *n
543     }
544    
545     After processing the node, (part of) the next node might already be in
546     cache.
547 root 1.1
548 root 1.2 =back
549 root 1.1
550 root 1.36 =head2 BIT FIDDLING / BIT WIZARDRY
551 root 1.1
552 root 1.4 =over 4
553    
554 root 1.3 =item bool ecb_big_endian ()
555    
556     =item bool ecb_little_endian ()
557    
558 sf-exg 1.11 These two functions return true if the byte order is big endian
559     (most-significant byte first) or little endian (least-significant byte
560     first) respectively.
561    
562 root 1.24 On systems that are neither, their return values are unspecified.
563    
564 root 1.3 =item int ecb_ctz32 (uint32_t x)
565    
566 root 1.35 =item int ecb_ctz64 (uint64_t x)
567    
568 sf-exg 1.11 Returns the index of the least significant bit set in C<x> (or
569 root 1.24 equivalently the number of bits set to 0 before the least significant bit
570 root 1.35 set), starting from 0. If C<x> is 0 the result is undefined.
571    
572 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
573    
574 root 1.35 For example:
575 sf-exg 1.11
576 root 1.15 ecb_ctz32 (3) = 0
577     ecb_ctz32 (6) = 1
578 sf-exg 1.11
579 root 1.41 =item bool ecb_is_pot32 (uint32_t x)
580    
581     =item bool ecb_is_pot64 (uint32_t x)
582    
583     Return true iff C<x> is a power of two or C<x == 0>.
584    
585     For smaller types then C<uint32_t> you can safely use C<ecb_is_pot32>.
586    
587 root 1.35 =item int ecb_ld32 (uint32_t x)
588    
589     =item int ecb_ld64 (uint64_t x)
590    
591     Returns the index of the most significant bit set in C<x>, or the number
592     of digits the number requires in binary (so that C<< 2**ld <= x <
593     2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
594     to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
595     example to see how many bits a certain number requires to be encoded.
596    
597     This function is similar to the "count leading zero bits" function, except
598     that that one returns how many zero bits are "in front" of the number (in
599     the given data type), while C<ecb_ld> returns how many bits the number
600     itself requires.
601    
602 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
603    
604 root 1.3 =item int ecb_popcount32 (uint32_t x)
605    
606 root 1.35 =item int ecb_popcount64 (uint64_t x)
607    
608 root 1.36 Returns the number of bits set to 1 in C<x>.
609    
610     For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
611    
612     For example:
613 sf-exg 1.11
614 root 1.15 ecb_popcount32 (7) = 3
615     ecb_popcount32 (255) = 8
616 sf-exg 1.11
617 root 1.39 =item uint8_t ecb_bitrev8 (uint8_t x)
618    
619     =item uint16_t ecb_bitrev16 (uint16_t x)
620    
621     =item uint32_t ecb_bitrev32 (uint32_t x)
622    
623     Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
624     and so on.
625    
626     Example:
627    
628     ecb_bitrev8 (0xa7) = 0xea
629     ecb_bitrev32 (0xffcc4411) = 0x882233ff
630    
631 root 1.8 =item uint32_t ecb_bswap16 (uint32_t x)
632    
633 root 1.3 =item uint32_t ecb_bswap32 (uint32_t x)
634    
635 root 1.34 =item uint64_t ecb_bswap64 (uint64_t x)
636 sf-exg 1.13
637 root 1.34 These functions return the value of the 16-bit (32-bit, 64-bit) value
638     C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
639     C<ecb_bswap32>).
640    
641     =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
642    
643     =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
644 root 1.3
645     =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
646    
647 root 1.34 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
648    
649     =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
650    
651     =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
652    
653     =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
654    
655 root 1.33 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
656    
657 root 1.34 These two families of functions return the value of C<x> after rotating
658     all the bits by C<count> positions to the right (C<ecb_rotr>) or left
659     (C<ecb_rotl>).
660 sf-exg 1.11
661 root 1.20 Current GCC versions understand these functions and usually compile them
662 root 1.34 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
663     x86).
664 root 1.20
665 root 1.3 =back
666 root 1.1
667 root 1.50 =head2 FLOATING POINT FIDDLING
668    
669     =over 4
670    
671 root 1.62 =item ECB_INFINITY
672    
673     Evaluates to positive infinity if supported by the platform, otherwise to
674     a truly huge number.
675    
676 root 1.63 =item ECB_NAN
677 root 1.62
678     Evaluates to a quiet NAN if supported by the platform, otherwise to
679     C<ECB_INFINITY>.
680    
681     =item float ecb_ldexpf (float x, int exp)
682    
683     Same as C<ldexpf>, but always available.
684    
685 root 1.50 =item uint32_t ecb_float_to_binary32 (float x) [-UECB_NO_LIBM]
686    
687     =item uint64_t ecb_double_to_binary64 (double x) [-UECB_NO_LIBM]
688    
689     These functions each take an argument in the native C<float> or C<double>
690     type and return the IEEE 754 bit representation of it.
691    
692     The bit representation is just as IEEE 754 defines it, i.e. the sign bit
693     will be the most significant bit, followed by exponent and mantissa.
694    
695     This function should work even when the native floating point format isn't
696     IEEE compliant, of course at a speed and code size penalty, and of course
697     also within reasonable limits (it tries to convert NaNs, infinities and
698     denormals, but will likely convert negative zero to positive zero).
699    
700     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
701     be able to optimise away this function completely.
702    
703     These functions can be helpful when serialising floats to the network - you
704     can serialise the return value like a normal uint32_t/uint64_t.
705    
706     Another use for these functions is to manipulate floating point values
707     directly.
708    
709     Silly example: toggle the sign bit of a float.
710    
711     /* On gcc-4.7 on amd64, */
712     /* this results in a single add instruction to toggle the bit, and 4 extra */
713     /* instructions to move the float value to an integer register and back. */
714    
715     x = ecb_binary32_to_float (ecb_float_to_binary32 (x) ^ 0x80000000U)
716    
717 root 1.58 =item float ecb_binary16_to_float (uint16_t x) [-UECB_NO_LIBM]
718    
719 root 1.50 =item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM]
720    
721     =item double ecb_binary32_to_double (uint64_t x) [-UECB_NO_LIBM]
722    
723 sf-exg 1.59 The reverse operation of the previous function - takes the bit
724 root 1.58 representation of an IEEE binary16, binary32 or binary64 number and
725     converts it to the native C<float> or C<double> format.
726 root 1.50
727     This function should work even when the native floating point format isn't
728     IEEE compliant, of course at a speed and code size penalty, and of course
729     also within reasonable limits (it tries to convert normals and denormals,
730     and might be lucky for infinities, and with extraordinary luck, also for
731     negative zero).
732    
733     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
734     be able to optimise away this function completely.
735    
736     =back
737    
738 root 1.1 =head2 ARITHMETIC
739    
740 root 1.3 =over 4
741    
742 root 1.14 =item x = ecb_mod (m, n)
743 root 1.3
744 root 1.25 Returns C<m> modulo C<n>, which is the same as the positive remainder
745     of the division operation between C<m> and C<n>, using floored
746     division. Unlike the C remainder operator C<%>, this function ensures that
747     the return value is always positive and that the two numbers I<m> and
748     I<m' = m + i * n> result in the same value modulo I<n> - in other words,
749     C<ecb_mod> implements the mathematical modulo operation, which is missing
750     in the language.
751 root 1.14
752 sf-exg 1.23 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
753 root 1.14 negatable, that is, both C<m> and C<-m> must be representable in its
754 root 1.30 type (this typically excludes the minimum signed integer value, the same
755 root 1.25 limitation as for C</> and C<%> in C).
756 sf-exg 1.11
757 root 1.24 Current GCC versions compile this into an efficient branchless sequence on
758 root 1.28 almost all CPUs.
759 root 1.24
760     For example, when you want to rotate forward through the members of an
761     array for increasing C<m> (which might be negative), then you should use
762     C<ecb_mod>, as the C<%> operator might give either negative results, or
763     change direction for negative values:
764    
765     for (m = -100; m <= 100; ++m)
766     int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
767    
768 sf-exg 1.37 =item x = ecb_div_rd (val, div)
769    
770     =item x = ecb_div_ru (val, div)
771    
772     Returns C<val> divided by C<div> rounded down or up, respectively.
773     C<val> and C<div> must have integer types and C<div> must be strictly
774 sf-exg 1.38 positive. Note that these functions are implemented with macros in C
775     and with function templates in C++.
776 sf-exg 1.37
777 root 1.3 =back
778 root 1.1
779     =head2 UTILITY
780    
781 root 1.3 =over 4
782    
783 sf-exg 1.23 =item element_count = ecb_array_length (name)
784 root 1.3
785 sf-exg 1.13 Returns the number of elements in the array C<name>. For example:
786    
787     int primes[] = { 2, 3, 5, 7, 11 };
788     int sum = 0;
789    
790     for (i = 0; i < ecb_array_length (primes); i++)
791     sum += primes [i];
792    
793 root 1.3 =back
794 root 1.1
795 root 1.43 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
796    
797     These symbols need to be defined before including F<ecb.h> the first time.
798    
799     =over 4
800    
801 root 1.51 =item ECB_NO_THREADS
802 root 1.43
803     If F<ecb.h> is never used from multiple threads, then this symbol can
804     be defined, in which case memory fences (and similar constructs) are
805     completely removed, leading to more efficient code and fewer dependencies.
806    
807     Setting this symbol to a true value implies C<ECB_NO_SMP>.
808    
809     =item ECB_NO_SMP
810    
811     The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
812     multiple threads, but never concurrently (e.g. if the system the program
813     runs on has only a single CPU with a single core, no hyperthreading and so
814     on), then this symbol can be defined, leading to more efficient code and
815     fewer dependencies.
816    
817 root 1.50 =item ECB_NO_LIBM
818    
819     When defined to C<1>, do not export any functions that might introduce
820     dependencies on the math library (usually called F<-lm>) - these are
821     marked with [-UECB_NO_LIBM].
822    
823 root 1.43 =back
824    
825 root 1.1