ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.84
Committed: Mon Jan 20 21:10:16 2020 UTC (4 years, 4 months ago) by root
Branch: MAIN
Changes since 1.83: +3 -4 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 1.76 int8_t uint8_
64     int16_t uint16_t
65     int32_t uint32_
66     int64_t uint64_t
67     int_fast8_t uint_fast8_t
68     int_fast16_t uint_fast16_t
69     int_fast32_t uint_fast32_t
70     int_fast64_t uint_fast64_t
71     intptr_t uintptr_t
72 root 1.40
73     The macro C<ECB_PTRSIZE> is defined to the size of a pointer on this
74 root 1.45 platform (currently C<4> or C<8>) and can be used in preprocessor
75     expressions.
76 root 1.40
77 root 1.74 For C<ptrdiff_t> and C<size_t> use C<stddef.h>/C<cstddef>.
78 root 1.49
79 root 1.62 =head2 LANGUAGE/ENVIRONMENT/COMPILER VERSIONS
80 root 1.43
81 sf-exg 1.46 All the following symbols expand to an expression that can be tested in
82 root 1.44 preprocessor instructions as well as treated as a boolean (use C<!!> to
83     ensure it's either C<0> or C<1> if you need that).
84    
85 root 1.43 =over 4
86    
87 root 1.44 =item ECB_C
88    
89     True if the implementation defines the C<__STDC__> macro to a true value,
90 root 1.82 while not claiming to be C++, i..e C, but not C++.
91 root 1.44
92 root 1.43 =item ECB_C99
93    
94 root 1.47 True if the implementation claims to be compliant to C99 (ISO/IEC
95 root 1.55 9899:1999) or any later version, while not claiming to be C++.
96 root 1.47
97     Note that later versions (ECB_C11) remove core features again (for
98     example, variable length arrays).
99 root 1.43
100 root 1.74 =item ECB_C11, ECB_C17
101 root 1.43
102 root 1.74 True if the implementation claims to be compliant to C11/C17 (ISO/IEC
103     9899:2011, :20187) or any later version, while not claiming to be C++.
104 root 1.44
105     =item ECB_CPP
106    
107     True if the implementation defines the C<__cplusplus__> macro to a true
108     value, which is typically true for C++ compilers.
109    
110 root 1.74 =item ECB_CPP11, ECB_CPP14, ECB_CPP17
111 root 1.44
112 root 1.74 True if the implementation claims to be compliant to C++11/C++14/C++17
113     (ISO/IEC 14882:2011, :2014, :2017) or any later version.
114 root 1.43
115 root 1.83 Note that many C++20 features will likely have their own feature test
116     macros (see e.g. L<http://eel.is/c++draft/cpp.predefined#1.8>).
117    
118 root 1.81 =item ECB_OPTIMIZE_SIZE
119    
120     Is C<1> when the compiler optimizes for size, C<0> otherwise. This symbol
121     can also be defined before including F<ecb.h>, in which case it will be
122     unchanged.
123    
124 root 1.57 =item ECB_GCC_VERSION (major, minor)
125 root 1.43
126 root 1.84 Expands to a true value (suitable for testing by the preprocessor) if the
127     compiler used is GNU C and the version is the given version, or higher.
128 root 1.43
129     This macro tries to return false on compilers that claim to be GCC
130     compatible but aren't.
131    
132 root 1.50 =item ECB_EXTERN_C
133    
134     Expands to C<extern "C"> in C++, and a simple C<extern> in C.
135    
136     This can be used to declare a single external C function:
137    
138     ECB_EXTERN_C int printf (const char *format, ...);
139    
140     =item ECB_EXTERN_C_BEG / ECB_EXTERN_C_END
141    
142     These two macros can be used to wrap multiple C<extern "C"> definitions -
143     they expand to nothing in C.
144    
145     They are most useful in header files:
146    
147     ECB_EXTERN_C_BEG
148    
149     int mycfun1 (int x);
150     int mycfun2 (int x);
151    
152     ECB_EXTERN_C_END
153    
154     =item ECB_STDFP
155    
156 root 1.84 If this evaluates to a true value (suitable for testing by the
157 root 1.50 preprocessor), then C<float> and C<double> use IEEE 754 single/binary32
158     and double/binary64 representations internally I<and> the endianness of
159     both types match the endianness of C<uint32_t> and C<uint64_t>.
160    
161     This means you can just copy the bits of a C<float> (or C<double>) to an
162     C<uint32_t> (or C<uint64_t>) and get the raw IEEE 754 bit representation
163     without having to think about format or endianness.
164    
165     This is true for basically all modern platforms, although F<ecb.h> might
166     not be able to deduce this correctly everywhere and might err on the safe
167     side.
168    
169 root 1.54 =item ECB_AMD64, ECB_AMD64_X32
170    
171     These two macros are defined to C<1> on the x86_64/amd64 ABI and the X32
172     ABI, respectively, and undefined elsewhere.
173    
174     The designers of the new X32 ABI for some inexplicable reason decided to
175     make it look exactly like amd64, even though it's completely incompatible
176     to that ABI, breaking about every piece of software that assumed that
177     C<__x86_64> stands for, well, the x86-64 ABI, making these macros
178     necessary.
179    
180 root 1.43 =back
181    
182 root 1.62 =head2 MACRO TRICKERY
183    
184     =over 4
185    
186     =item ECB_CONCAT (a, b)
187    
188     Expands any macros in C<a> and C<b>, then concatenates the result to form
189     a single token. This is mainly useful to form identifiers from components,
190     e.g.:
191    
192     #define S1 str
193     #define S2 cpy
194    
195     ECB_CONCAT (S1, S2)(dst, src); // == strcpy (dst, src);
196    
197     =item ECB_STRINGIFY (arg)
198    
199     Expands any macros in C<arg> and returns the stringified version of
200     it. This is mainly useful to get the contents of a macro in string form,
201     e.g.:
202    
203     #define SQL_LIMIT 100
204     sql_exec ("select * from table limit " ECB_STRINGIFY (SQL_LIMIT));
205    
206 root 1.64 =item ECB_STRINGIFY_EXPR (expr)
207    
208     Like C<ECB_STRINGIFY>, but additionally evaluates C<expr> to make sure it
209     is a valid expression. This is useful to catch typos or cases where the
210     macro isn't available:
211    
212     #include <errno.h>
213    
214     ECB_STRINGIFY (EDOM); // "33" (on my system at least)
215     ECB_STRINGIFY_EXPR (EDOM); // "33"
216    
217     // now imagine we had a typo:
218    
219     ECB_STRINGIFY (EDAM); // "EDAM"
220     ECB_STRINGIFY_EXPR (EDAM); // error: EDAM undefined
221    
222 root 1.62 =back
223    
224 sf-exg 1.60 =head2 ATTRIBUTES
225 root 1.1
226 sf-exg 1.60 A major part of libecb deals with additional attributes that can be
227     assigned to functions, variables and sometimes even types - much like
228     C<const> or C<volatile> in C. They are implemented using either GCC
229     attributes or other compiler/language specific features. Attributes
230     declarations must be put before the whole declaration:
231 root 1.20
232     ecb_const int mysqrt (int a);
233     ecb_unused int i;
234    
235 root 1.1 =over 4
236    
237 root 1.3 =item ecb_unused
238    
239     Marks a function or a variable as "unused", which simply suppresses a
240     warning by GCC when it detects it as unused. This is useful when you e.g.
241     declare a variable but do not always use it:
242    
243 root 1.15 {
244 sf-exg 1.61 ecb_unused int var;
245 root 1.3
246 root 1.15 #ifdef SOMECONDITION
247     var = ...;
248     return var;
249     #else
250     return 0;
251     #endif
252     }
253 root 1.3
254 root 1.56 =item ecb_deprecated
255    
256     Similar to C<ecb_unused>, but marks a function, variable or type as
257     deprecated. This makes some compilers warn when the type is used.
258    
259 root 1.62 =item ecb_deprecated_message (message)
260    
261 root 1.67 Same as C<ecb_deprecated>, but if possible, the specified diagnostic is
262 root 1.62 used instead of a generic depreciation message when the object is being
263     used.
264    
265 root 1.31 =item ecb_inline
266 root 1.29
267 root 1.73 Expands either to (a compiler-specific equivalent of) C<static inline> or
268     to just C<static>, if inline isn't supported. It should be used to declare
269     functions that should be inlined, for code size or speed reasons.
270 root 1.29
271     Example: inline this function, it surely will reduce codesize.
272    
273 root 1.31 ecb_inline int
274 root 1.29 negmul (int a, int b)
275     {
276     return - (a * b);
277     }
278    
279 root 1.2 =item ecb_noinline
280    
281 sf-exg 1.66 Prevents a function from being inlined - it might be optimised away, but
282 root 1.3 not inlined into other functions. This is useful if you know your function
283     is rarely called and large enough for inlining not to be helpful.
284    
285 root 1.2 =item ecb_noreturn
286    
287 root 1.17 Marks a function as "not returning, ever". Some typical functions that
288     don't return are C<exit> or C<abort> (which really works hard to not
289     return), and now you can make your own:
290    
291     ecb_noreturn void
292     my_abort (const char *errline)
293     {
294     puts (errline);
295     abort ();
296     }
297    
298 sf-exg 1.19 In this case, the compiler would probably be smart enough to deduce it on
299     its own, so this is mainly useful for declarations.
300 root 1.17
301 root 1.53 =item ecb_restrict
302    
303     Expands to the C<restrict> keyword or equivalent on compilers that support
304     them, and to nothing on others. Must be specified on a pointer type or
305     an array index to indicate that the memory doesn't alias with any other
306     restricted pointer in the same scope.
307    
308     Example: multiply a vector, and allow the compiler to parallelise the
309     loop, because it knows it doesn't overwrite input values.
310    
311     void
312 sf-exg 1.61 multiply (ecb_restrict float *src,
313     ecb_restrict float *dst,
314 root 1.53 int len, float factor)
315     {
316     int i;
317    
318     for (i = 0; i < len; ++i)
319     dst [i] = src [i] * factor;
320     }
321    
322 root 1.2 =item ecb_const
323    
324 sf-exg 1.19 Declares that the function only depends on the values of its arguments,
325 root 1.17 much like a mathematical function. It specifically does not read or write
326     any memory any arguments might point to, global variables, or call any
327     non-const functions. It also must not have any side effects.
328    
329     Such a function can be optimised much more aggressively by the compiler -
330     for example, multiple calls with the same arguments can be optimised into
331     a single call, which wouldn't be possible if the compiler would have to
332     expect any side effects.
333    
334     It is best suited for functions in the sense of mathematical functions,
335 sf-exg 1.19 such as a function returning the square root of its input argument.
336 root 1.17
337     Not suited would be a function that calculates the hash of some memory
338     area you pass in, prints some messages or looks at a global variable to
339     decide on rounding.
340    
341     See C<ecb_pure> for a slightly less restrictive class of functions.
342    
343 root 1.2 =item ecb_pure
344    
345 root 1.17 Similar to C<ecb_const>, declares a function that has no side
346     effects. Unlike C<ecb_const>, the function is allowed to examine global
347     variables and any other memory areas (such as the ones passed to it via
348     pointers).
349    
350     While these functions cannot be optimised as aggressively as C<ecb_const>
351     functions, they can still be optimised away in many occasions, and the
352     compiler has more freedom in moving calls to them around.
353    
354     Typical examples for such functions would be C<strlen> or C<memcmp>. A
355     function that calculates the MD5 sum of some input and updates some MD5
356     state passed as argument would I<NOT> be pure, however, as it would modify
357     some memory area that is not the return value.
358    
359 root 1.2 =item ecb_hot
360    
361 root 1.17 This declares a function as "hot" with regards to the cache - the function
362     is used so often, that it is very beneficial to keep it in the cache if
363     possible.
364    
365     The compiler reacts by trying to place hot functions near to each other in
366     memory.
367    
368 sf-exg 1.19 Whether a function is hot or not often depends on the whole program,
369 root 1.17 and less on the function itself. C<ecb_cold> is likely more useful in
370     practise.
371    
372 root 1.2 =item ecb_cold
373    
374 root 1.17 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
375     the cache, or in other words, this function is not called often, or not at
376     speed-critical times, and keeping it in the cache might be a waste of said
377     cache.
378    
379     In addition to placing cold functions together (or at least away from hot
380     functions), this knowledge can be used in other ways, for example, the
381     function will be optimised for size, as opposed to speed, and codepaths
382     leading to calls to those functions can automatically be marked as if
383 root 1.27 C<ecb_expect_false> had been used to reach them.
384 root 1.17
385     Good examples for such functions would be error reporting functions, or
386     functions only called in exceptional or rare cases.
387    
388 root 1.2 =item ecb_artificial
389    
390 root 1.17 Declares the function as "artificial", in this case meaning that this
391 root 1.52 function is not really meant to be a function, but more like an accessor
392 root 1.17 - many methods in C++ classes are mere accessor functions, and having a
393     crash reported in such a method, or single-stepping through them, is not
394     usually so helpful, especially when it's inlined to just a few instructions.
395    
396     Marking them as artificial will instruct the debugger about just this,
397     leading to happier debugging and thus happier lives.
398    
399     Example: in some kind of smart-pointer class, mark the pointer accessor as
400     artificial, so that the whole class acts more like a pointer and less like
401     some C++ abstraction monster.
402    
403     template<typename T>
404     struct my_smart_ptr
405     {
406     T *value;
407    
408     ecb_artificial
409     operator T *()
410     {
411     return value;
412     }
413     };
414    
415 root 1.2 =back
416 root 1.1
417     =head2 OPTIMISATION HINTS
418    
419     =over 4
420    
421 root 1.58 =item bool ecb_is_constant (expr)
422 root 1.1
423 root 1.3 Returns true iff the expression can be deduced to be a compile-time
424     constant, and false otherwise.
425    
426     For example, when you have a C<rndm16> function that returns a 16 bit
427     random number, and you have a function that maps this to a range from
428 root 1.5 0..n-1, then you could use this inline function in a header file:
429 root 1.3
430     ecb_inline uint32_t
431     rndm (uint32_t n)
432     {
433 root 1.6 return (n * (uint32_t)rndm16 ()) >> 16;
434 root 1.3 }
435    
436     However, for powers of two, you could use a normal mask, but that is only
437     worth it if, at compile time, you can detect this case. This is the case
438     when the passed number is a constant and also a power of two (C<n & (n -
439     1) == 0>):
440    
441     ecb_inline uint32_t
442     rndm (uint32_t n)
443     {
444     return is_constant (n) && !(n & (n - 1))
445     ? rndm16 () & (num - 1)
446 root 1.6 : (n * (uint32_t)rndm16 ()) >> 16;
447 root 1.3 }
448    
449 root 1.62 =item ecb_expect (expr, value)
450 root 1.1
451 root 1.7 Evaluates C<expr> and returns it. In addition, it tells the compiler that
452     the C<expr> evaluates to C<value> a lot, which can be used for static
453     branch optimisations.
454 root 1.1
455 root 1.27 Usually, you want to use the more intuitive C<ecb_expect_true> and
456     C<ecb_expect_false> functions instead.
457 root 1.1
458 root 1.27 =item bool ecb_expect_true (cond)
459 root 1.1
460 root 1.27 =item bool ecb_expect_false (cond)
461 root 1.1
462 root 1.7 These two functions expect a expression that is true or false and return
463     C<1> or C<0>, respectively, so when used in the condition of an C<if> or
464     other conditional statement, it will not change the program:
465    
466     /* these two do the same thing */
467     if (some_condition) ...;
468 root 1.27 if (ecb_expect_true (some_condition)) ...;
469 root 1.7
470 root 1.27 However, by using C<ecb_expect_true>, you tell the compiler that the
471     condition is likely to be true (and for C<ecb_expect_false>, that it is
472     unlikely to be true).
473 root 1.7
474 root 1.9 For example, when you check for a null pointer and expect this to be a
475 root 1.27 rare, exceptional, case, then use C<ecb_expect_false>:
476 root 1.7
477     void my_free (void *ptr)
478     {
479 root 1.27 if (ecb_expect_false (ptr == 0))
480 root 1.7 return;
481     }
482    
483     Consequent use of these functions to mark away exceptional cases or to
484     tell the compiler what the hot path through a function is can increase
485     performance considerably.
486    
487 root 1.27 You might know these functions under the name C<likely> and C<unlikely>
488     - while these are common aliases, we find that the expect name is easier
489     to understand when quickly skimming code. If you wish, you can use
490     C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
491     C<ecb_expect_false> - these are simply aliases.
492    
493 root 1.7 A very good example is in a function that reserves more space for some
494     memory block (for example, inside an implementation of a string stream) -
495 root 1.9 each time something is added, you have to check for a buffer overrun, but
496 root 1.7 you expect that most checks will turn out to be false:
497    
498     /* make sure we have "size" extra room in our buffer */
499     ecb_inline void
500     reserve (int size)
501     {
502 root 1.27 if (ecb_expect_false (current + size > end))
503 root 1.7 real_reserve_method (size); /* presumably noinline */
504     }
505    
506 root 1.62 =item ecb_assume (cond)
507 root 1.7
508 sf-exg 1.66 Tries to tell the compiler that some condition is true, even if it's not
509 root 1.65 obvious. This is not a function, but a statement: it cannot be used in
510     another expression.
511 root 1.7
512     This can be used to teach the compiler about invariants or other
513     conditions that might improve code generation, but which are impossible to
514     deduce form the code itself.
515    
516 root 1.27 For example, the example reservation function from the C<ecb_expect_false>
517 root 1.7 description could be written thus (only C<ecb_assume> was added):
518    
519     ecb_inline void
520     reserve (int size)
521     {
522 root 1.27 if (ecb_expect_false (current + size > end))
523 root 1.7 real_reserve_method (size); /* presumably noinline */
524    
525     ecb_assume (current + size <= end);
526     }
527    
528     If you then call this function twice, like this:
529    
530     reserve (10);
531     reserve (1);
532    
533     Then the compiler I<might> be able to optimise out the second call
534     completely, as it knows that C<< current + 1 > end >> is false and the
535     call will never be executed.
536    
537 root 1.62 =item ecb_unreachable ()
538 root 1.7
539     This function does nothing itself, except tell the compiler that it will
540 root 1.9 never be executed. Apart from suppressing a warning in some cases, this
541 root 1.65 function can be used to implement C<ecb_assume> or similar functionality.
542 root 1.7
543 root 1.62 =item ecb_prefetch (addr, rw, locality)
544 root 1.7
545     Tells the compiler to try to prefetch memory at the given C<addr>ess
546 root 1.10 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
547 root 1.7 C<0> means that there will only be one access later, C<3> means that
548     the data will likely be accessed very often, and values in between mean
549     something... in between. The memory pointed to by the address does not
550     need to be accessible (it could be a null pointer for example), but C<rw>
551     and C<locality> must be compile-time constants.
552    
553 root 1.65 This is a statement, not a function: you cannot use it as part of an
554     expression.
555    
556 root 1.7 An obvious way to use this is to prefetch some data far away, in a big
557 root 1.9 array you loop over. This prefetches memory some 128 array elements later,
558 root 1.7 in the hope that it will be ready when the CPU arrives at that location.
559    
560     int sum = 0;
561    
562     for (i = 0; i < N; ++i)
563     {
564     sum += arr [i]
565     ecb_prefetch (arr + i + 128, 0, 0);
566     }
567    
568     It's hard to predict how far to prefetch, and most CPUs that can prefetch
569     are often good enough to predict this kind of behaviour themselves. It
570     gets more interesting with linked lists, especially when you do some fair
571     processing on each list element:
572    
573     for (node *n = start; n; n = n->next)
574     {
575     ecb_prefetch (n->next, 0, 0);
576     ... do medium amount of work with *n
577     }
578    
579     After processing the node, (part of) the next node might already be in
580     cache.
581 root 1.1
582 root 1.2 =back
583 root 1.1
584 root 1.36 =head2 BIT FIDDLING / BIT WIZARDRY
585 root 1.1
586 root 1.4 =over 4
587    
588 root 1.3 =item bool ecb_big_endian ()
589    
590     =item bool ecb_little_endian ()
591    
592 sf-exg 1.11 These two functions return true if the byte order is big endian
593     (most-significant byte first) or little endian (least-significant byte
594     first) respectively.
595    
596 root 1.24 On systems that are neither, their return values are unspecified.
597    
598 root 1.3 =item int ecb_ctz32 (uint32_t x)
599    
600 root 1.35 =item int ecb_ctz64 (uint64_t x)
601    
602 root 1.77 =item int ecb_ctz (T x) [C++]
603    
604 sf-exg 1.11 Returns the index of the least significant bit set in C<x> (or
605 root 1.24 equivalently the number of bits set to 0 before the least significant bit
606 root 1.35 set), starting from 0. If C<x> is 0 the result is undefined.
607    
608 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
609    
610 root 1.77 The overloaded C++ C<ecb_ctz> function supports C<uint8_t>, C<uint16_t>,
611     C<uint32_t> and C<uint64_t> types.
612    
613 root 1.35 For example:
614 sf-exg 1.11
615 root 1.15 ecb_ctz32 (3) = 0
616     ecb_ctz32 (6) = 1
617 sf-exg 1.11
618 root 1.41 =item bool ecb_is_pot32 (uint32_t x)
619    
620     =item bool ecb_is_pot64 (uint32_t x)
621    
622 root 1.77 =item bool ecb_is_pot (T x) [C++]
623    
624 sf-exg 1.66 Returns true iff C<x> is a power of two or C<x == 0>.
625 root 1.41
626 sf-exg 1.66 For smaller types than C<uint32_t> you can safely use C<ecb_is_pot32>.
627 root 1.41
628 root 1.77 The overloaded C++ C<ecb_is_pot> function supports C<uint8_t>, C<uint16_t>,
629     C<uint32_t> and C<uint64_t> types.
630    
631 root 1.35 =item int ecb_ld32 (uint32_t x)
632    
633     =item int ecb_ld64 (uint64_t x)
634    
635 root 1.77 =item int ecb_ld64 (T x) [C++]
636    
637 root 1.35 Returns the index of the most significant bit set in C<x>, or the number
638     of digits the number requires in binary (so that C<< 2**ld <= x <
639     2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
640     to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
641     example to see how many bits a certain number requires to be encoded.
642    
643     This function is similar to the "count leading zero bits" function, except
644     that that one returns how many zero bits are "in front" of the number (in
645     the given data type), while C<ecb_ld> returns how many bits the number
646     itself requires.
647    
648 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
649    
650 root 1.77 The overloaded C++ C<ecb_ld> function supports C<uint8_t>, C<uint16_t>,
651     C<uint32_t> and C<uint64_t> types.
652    
653 root 1.3 =item int ecb_popcount32 (uint32_t x)
654    
655 root 1.35 =item int ecb_popcount64 (uint64_t x)
656    
657 root 1.77 =item int ecb_popcount (T x) [C++]
658    
659 root 1.36 Returns the number of bits set to 1 in C<x>.
660    
661     For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
662    
663 root 1.77 The overloaded C++ C<ecb_popcount> function supports C<uint8_t>, C<uint16_t>,
664     C<uint32_t> and C<uint64_t> types.
665    
666 root 1.36 For example:
667 sf-exg 1.11
668 root 1.15 ecb_popcount32 (7) = 3
669     ecb_popcount32 (255) = 8
670 sf-exg 1.11
671 root 1.39 =item uint8_t ecb_bitrev8 (uint8_t x)
672    
673     =item uint16_t ecb_bitrev16 (uint16_t x)
674    
675     =item uint32_t ecb_bitrev32 (uint32_t x)
676    
677 root 1.77 =item T ecb_bitrev (T x) [C++]
678    
679 root 1.39 Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
680     and so on.
681    
682 root 1.77 The overloaded C++ C<ecb_bitrev> function supports C<uint8_t>, C<uint16_t> and C<uint32_t> types.
683    
684 root 1.39 Example:
685    
686     ecb_bitrev8 (0xa7) = 0xea
687     ecb_bitrev32 (0xffcc4411) = 0x882233ff
688    
689 root 1.77 =item T ecb_bitrev (T x) [C++]
690    
691     Overloaded C++ bitrev function.
692    
693     C<T> must be one of C<uint8_t>, C<uint16_t> or C<uint32_t>.
694    
695 root 1.8 =item uint32_t ecb_bswap16 (uint32_t x)
696    
697 root 1.3 =item uint32_t ecb_bswap32 (uint32_t x)
698    
699 root 1.34 =item uint64_t ecb_bswap64 (uint64_t x)
700 sf-exg 1.13
701 root 1.78 =item T ecb_bswap (T x)
702    
703 root 1.34 These functions return the value of the 16-bit (32-bit, 64-bit) value
704     C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
705     C<ecb_bswap32>).
706    
707 root 1.77 The overloaded C++ C<ecb_bswap> function supports C<uint8_t>, C<uint16_t>,
708     C<uint32_t> and C<uint64_t> types.
709 root 1.76
710 root 1.34 =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
711    
712     =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
713 root 1.3
714     =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
715    
716 root 1.34 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
717    
718     =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
719    
720     =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
721    
722     =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
723    
724 root 1.33 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
725    
726 root 1.34 These two families of functions return the value of C<x> after rotating
727     all the bits by C<count> positions to the right (C<ecb_rotr>) or left
728     (C<ecb_rotl>).
729 sf-exg 1.11
730 root 1.20 Current GCC versions understand these functions and usually compile them
731 root 1.34 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
732     x86).
733 root 1.20
734 root 1.77 =item T ecb_rotl (T x, unsigned int count) [C++]
735    
736     =item T ecb_rotr (T x, unsigned int count) [C++]
737    
738     Overloaded C++ rotl/rotr functions.
739    
740     C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>.
741    
742 root 1.3 =back
743 root 1.1
744 root 1.76 =head2 HOST ENDIANNESS CONVERSION
745    
746     =over 4
747    
748     =item uint_fast16_t ecb_be_u16_to_host (uint_fast16_t v)
749    
750     =item uint_fast32_t ecb_be_u32_to_host (uint_fast32_t v)
751    
752     =item uint_fast64_t ecb_be_u64_to_host (uint_fast64_t v)
753    
754     =item uint_fast16_t ecb_le_u16_to_host (uint_fast16_t v)
755    
756     =item uint_fast32_t ecb_le_u32_to_host (uint_fast32_t v)
757    
758     =item uint_fast64_t ecb_le_u64_to_host (uint_fast64_t v)
759    
760     Convert an unsigned 16, 32 or 64 bit value from big or little endian to host byte order.
761    
762     The naming convention is C<ecb_>(C<be>|C<le>)C<_u>C<16|32|64>C<_to_host>,
763 root 1.79 where C<be> and C<le> stand for big endian and little endian, respectively.
764 root 1.76
765     =item uint_fast16_t ecb_host_to_be_u16 (uint_fast16_t v)
766    
767     =item uint_fast32_t ecb_host_to_be_u32 (uint_fast32_t v)
768    
769     =item uint_fast64_t ecb_host_to_be_u64 (uint_fast64_t v)
770    
771     =item uint_fast16_t ecb_host_to_le_u16 (uint_fast16_t v)
772    
773     =item uint_fast32_t ecb_host_to_le_u32 (uint_fast32_t v)
774    
775     =item uint_fast64_t ecb_host_to_le_u64 (uint_fast64_t v)
776    
777     Like above, but converts I<from> host byte order to the specified
778     endianness.
779    
780     =back
781    
782 root 1.77 In C++ the following additional template functions are supported:
783 root 1.76
784     =over 4
785    
786     =item T ecb_be_to_host (T v)
787    
788     =item T ecb_le_to_host (T v)
789    
790     =item T ecb_host_to_be (T v)
791    
792     =item T ecb_host_to_le (T v)
793    
794 root 1.77 These functions work like their C counterparts, above, but use templates,
795     which make them useful in generic code.
796 root 1.76
797     C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>
798     (so unlike their C counterparts, there is a version for C<uint8_t>, which
799     again can be useful in generic code).
800    
801     =head2 UNALIGNED LOAD/STORE
802    
803     These function load or store unaligned multi-byte values.
804    
805     =over 4
806    
807     =item uint_fast16_t ecb_peek_u16_u (const void *ptr)
808    
809     =item uint_fast32_t ecb_peek_u32_u (const void *ptr)
810    
811     =item uint_fast64_t ecb_peek_u64_u (const void *ptr)
812    
813     These functions load an unaligned, unsigned 16, 32 or 64 bit value from
814     memory.
815    
816     =item uint_fast16_t ecb_peek_be_u16_u (const void *ptr)
817    
818     =item uint_fast32_t ecb_peek_be_u32_u (const void *ptr)
819    
820     =item uint_fast64_t ecb_peek_be_u64_u (const void *ptr)
821    
822     =item uint_fast16_t ecb_peek_le_u16_u (const void *ptr)
823    
824     =item uint_fast32_t ecb_peek_le_u32_u (const void *ptr)
825    
826     =item uint_fast64_t ecb_peek_le_u64_u (const void *ptr)
827    
828     Like above, but additionally convert from big endian (C<be>) or little
829     endian (C<le>) byte order to host byte order while doing so.
830    
831     =item ecb_poke_u16_u (void *ptr, uint16_t v)
832    
833     =item ecb_poke_u32_u (void *ptr, uint32_t v)
834    
835     =item ecb_poke_u64_u (void *ptr, uint64_t v)
836    
837     These functions store an unaligned, unsigned 16, 32 or 64 bit value to
838     memory.
839    
840     =item ecb_poke_be_u16_u (void *ptr, uint_fast16_t v)
841    
842     =item ecb_poke_be_u32_u (void *ptr, uint_fast32_t v)
843    
844     =item ecb_poke_be_u64_u (void *ptr, uint_fast64_t v)
845    
846     =item ecb_poke_le_u16_u (void *ptr, uint_fast16_t v)
847    
848     =item ecb_poke_le_u32_u (void *ptr, uint_fast32_t v)
849    
850     =item ecb_poke_le_u64_u (void *ptr, uint_fast64_t v)
851    
852     Like above, but additionally convert from host byte order to big endian
853     (C<be>) or little endian (C<le>) byte order while doing so.
854    
855     =back
856    
857 root 1.77 In C++ the following additional template functions are supported:
858 root 1.76
859     =over 4
860    
861 root 1.80 =item T ecb_peek<T> (const void *ptr)
862 root 1.76
863 root 1.80 =item T ecb_peek_be<T> (const void *ptr)
864 root 1.76
865 root 1.80 =item T ecb_peek_le<T> (const void *ptr)
866 root 1.76
867 root 1.80 =item T ecb_peek_u<T> (const void *ptr)
868 root 1.76
869 root 1.80 =item T ecb_peek_be_u<T> (const void *ptr)
870 root 1.76
871 root 1.80 =item T ecb_peek_le_u<T> (const void *ptr)
872 root 1.76
873     Similarly to their C counterparts, these functions load an unsigned 8, 16,
874     32 or 64 bit value from memory, with optional conversion from big/little
875     endian.
876    
877 root 1.80 Since the type cannot be deduced, it has to be specified explicitly, e.g.
878 root 1.76
879     uint_fast16_t v = ecb_peek<uint16_t> (ptr);
880    
881     C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>.
882    
883     Unlike their C counterparts, these functions support 8 bit quantities
884     (C<uint8_t>) and also have an aligned version (without the C<_u> prefix),
885     all of which hopefully makes them more useful in generic code.
886    
887     =item ecb_poke (void *ptr, T v)
888    
889     =item ecb_poke_be (void *ptr, T v)
890    
891     =item ecb_poke_le (void *ptr, T v)
892    
893     =item ecb_poke_u (void *ptr, T v)
894    
895     =item ecb_poke_be_u (void *ptr, T v)
896    
897     =item ecb_poke_le_u (void *ptr, T v)
898    
899     Again, similarly to their C counterparts, these functions store an
900     unsigned 8, 16, 32 or z64 bit value to memory, with optional conversion to
901     big/little endian.
902    
903     C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>.
904    
905     Unlike their C counterparts, these functions support 8 bit quantities
906     (C<uint8_t>) and also have an aligned version (without the C<_u> prefix),
907     all of which hopefully makes them more useful in generic code.
908    
909     =back
910    
911 root 1.50 =head2 FLOATING POINT FIDDLING
912    
913     =over 4
914    
915 root 1.71 =item ECB_INFINITY [-UECB_NO_LIBM]
916 root 1.62
917     Evaluates to positive infinity if supported by the platform, otherwise to
918     a truly huge number.
919    
920 root 1.71 =item ECB_NAN [-UECB_NO_LIBM]
921 root 1.62
922     Evaluates to a quiet NAN if supported by the platform, otherwise to
923     C<ECB_INFINITY>.
924    
925 root 1.71 =item float ecb_ldexpf (float x, int exp) [-UECB_NO_LIBM]
926 root 1.62
927     Same as C<ldexpf>, but always available.
928    
929 root 1.71 =item uint32_t ecb_float_to_binary16 (float x) [-UECB_NO_LIBM]
930    
931 root 1.50 =item uint32_t ecb_float_to_binary32 (float x) [-UECB_NO_LIBM]
932    
933     =item uint64_t ecb_double_to_binary64 (double x) [-UECB_NO_LIBM]
934    
935     These functions each take an argument in the native C<float> or C<double>
936 root 1.71 type and return the IEEE 754 bit representation of it (binary16/half,
937     binary32/single or binary64/double precision).
938 root 1.50
939     The bit representation is just as IEEE 754 defines it, i.e. the sign bit
940     will be the most significant bit, followed by exponent and mantissa.
941    
942     This function should work even when the native floating point format isn't
943     IEEE compliant, of course at a speed and code size penalty, and of course
944     also within reasonable limits (it tries to convert NaNs, infinities and
945     denormals, but will likely convert negative zero to positive zero).
946    
947     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
948     be able to optimise away this function completely.
949    
950     These functions can be helpful when serialising floats to the network - you
951 root 1.71 can serialise the return value like a normal uint16_t/uint32_t/uint64_t.
952 root 1.50
953     Another use for these functions is to manipulate floating point values
954     directly.
955    
956     Silly example: toggle the sign bit of a float.
957    
958     /* On gcc-4.7 on amd64, */
959     /* this results in a single add instruction to toggle the bit, and 4 extra */
960     /* instructions to move the float value to an integer register and back. */
961    
962     x = ecb_binary32_to_float (ecb_float_to_binary32 (x) ^ 0x80000000U)
963    
964 root 1.58 =item float ecb_binary16_to_float (uint16_t x) [-UECB_NO_LIBM]
965    
966 root 1.50 =item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM]
967    
968 root 1.70 =item double ecb_binary64_to_double (uint64_t x) [-UECB_NO_LIBM]
969 root 1.50
970 sf-exg 1.59 The reverse operation of the previous function - takes the bit
971 root 1.71 representation of an IEEE binary16, binary32 or binary64 number (half,
972     single or double precision) and converts it to the native C<float> or
973     C<double> format.
974 root 1.50
975     This function should work even when the native floating point format isn't
976     IEEE compliant, of course at a speed and code size penalty, and of course
977     also within reasonable limits (it tries to convert normals and denormals,
978     and might be lucky for infinities, and with extraordinary luck, also for
979     negative zero).
980    
981     On all modern platforms (where C<ECB_STDFP> is true), the compiler should
982     be able to optimise away this function completely.
983    
984 root 1.71 =item uint16_t ecb_binary32_to_binary16 (uint32_t x)
985    
986     =item uint32_t ecb_binary16_to_binary32 (uint16_t x)
987    
988     Convert a IEEE binary32/single precision to binary16/half format, and vice
989 root 1.72 versa, handling all details (round-to-nearest-even, subnormals, infinity
990     and NaNs) correctly.
991 root 1.71
992     These are functions are available under C<-DECB_NO_LIBM>, since
993     they do not rely on the platform floating point format. The
994     C<ecb_float_to_binary16> and C<ecb_binary16_to_float> functions are
995     usually what you want.
996    
997 root 1.50 =back
998    
999 root 1.1 =head2 ARITHMETIC
1000    
1001 root 1.3 =over 4
1002    
1003 root 1.14 =item x = ecb_mod (m, n)
1004 root 1.3
1005 root 1.25 Returns C<m> modulo C<n>, which is the same as the positive remainder
1006     of the division operation between C<m> and C<n>, using floored
1007     division. Unlike the C remainder operator C<%>, this function ensures that
1008     the return value is always positive and that the two numbers I<m> and
1009     I<m' = m + i * n> result in the same value modulo I<n> - in other words,
1010     C<ecb_mod> implements the mathematical modulo operation, which is missing
1011     in the language.
1012 root 1.14
1013 sf-exg 1.23 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
1014 root 1.14 negatable, that is, both C<m> and C<-m> must be representable in its
1015 root 1.30 type (this typically excludes the minimum signed integer value, the same
1016 root 1.25 limitation as for C</> and C<%> in C).
1017 sf-exg 1.11
1018 root 1.24 Current GCC versions compile this into an efficient branchless sequence on
1019 root 1.28 almost all CPUs.
1020 root 1.24
1021     For example, when you want to rotate forward through the members of an
1022     array for increasing C<m> (which might be negative), then you should use
1023     C<ecb_mod>, as the C<%> operator might give either negative results, or
1024     change direction for negative values:
1025    
1026     for (m = -100; m <= 100; ++m)
1027     int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
1028    
1029 sf-exg 1.37 =item x = ecb_div_rd (val, div)
1030    
1031     =item x = ecb_div_ru (val, div)
1032    
1033     Returns C<val> divided by C<div> rounded down or up, respectively.
1034     C<val> and C<div> must have integer types and C<div> must be strictly
1035 sf-exg 1.38 positive. Note that these functions are implemented with macros in C
1036     and with function templates in C++.
1037 sf-exg 1.37
1038 root 1.3 =back
1039 root 1.1
1040     =head2 UTILITY
1041    
1042 root 1.3 =over 4
1043    
1044 sf-exg 1.23 =item element_count = ecb_array_length (name)
1045 root 1.3
1046 sf-exg 1.13 Returns the number of elements in the array C<name>. For example:
1047    
1048     int primes[] = { 2, 3, 5, 7, 11 };
1049     int sum = 0;
1050    
1051     for (i = 0; i < ecb_array_length (primes); i++)
1052     sum += primes [i];
1053    
1054 root 1.3 =back
1055 root 1.1
1056 root 1.43 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
1057    
1058     These symbols need to be defined before including F<ecb.h> the first time.
1059    
1060     =over 4
1061    
1062 root 1.51 =item ECB_NO_THREADS
1063 root 1.43
1064     If F<ecb.h> is never used from multiple threads, then this symbol can
1065     be defined, in which case memory fences (and similar constructs) are
1066     completely removed, leading to more efficient code and fewer dependencies.
1067    
1068     Setting this symbol to a true value implies C<ECB_NO_SMP>.
1069    
1070     =item ECB_NO_SMP
1071    
1072     The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
1073     multiple threads, but never concurrently (e.g. if the system the program
1074     runs on has only a single CPU with a single core, no hyperthreading and so
1075     on), then this symbol can be defined, leading to more efficient code and
1076     fewer dependencies.
1077    
1078 root 1.50 =item ECB_NO_LIBM
1079    
1080     When defined to C<1>, do not export any functions that might introduce
1081     dependencies on the math library (usually called F<-lm>) - these are
1082     marked with [-UECB_NO_LIBM].
1083    
1084 sf-exg 1.69 =back
1085    
1086 root 1.68 =head1 UNDOCUMENTED FUNCTIONALITY
1087    
1088     F<ecb.h> is full of undocumented functionality as well, some of which is
1089     intended to be internal-use only, some of which we forgot to document, and
1090     some of which we hide because we are not sure we will keep the interface
1091     stable.
1092    
1093     While you are welcome to rummage around and use whatever you find useful
1094     (we can't stop you), keep in mind that we will change undocumented
1095     functionality in incompatible ways without thinking twice, while we are
1096     considerably more conservative with documented things.
1097    
1098     =head1 AUTHORS
1099    
1100     C<libecb> is designed and maintained by:
1101    
1102     Emanuele Giaquinta <e.giaquinta@glauco.it>
1103     Marc Alexander Lehmann <schmorp@schmorp.de>
1104    
1105 root 1.1