1 |
=head1 LIBECB - e-C-Builtins |
2 |
|
3 |
=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 |
It's part of the e-suite of libraries, other members of which include |
9 |
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 many compiler built-ins, |
16 |
together with replacement functions for other compilers. In addition |
17 |
to this, it provides a number of other lowlevel C utilities, such as |
18 |
endianness detection, byte swapping or bit rotations. |
19 |
|
20 |
Or in other words, things that should be built into any standard C |
21 |
system, but aren't, implemented as efficient as possible with GCC (clang, |
22 |
msvc...), and still correct with other compilers. |
23 |
|
24 |
More might come. |
25 |
|
26 |
=head2 ABOUT THE HEADER |
27 |
|
28 |
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 |
There are currently no object files to link to - future versions might |
37 |
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 |
|
59 |
=head2 TYPES / TYPE SUPPORT |
60 |
|
61 |
ecb.h makes sure that the following types are defined (in the expected way): |
62 |
|
63 |
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 |
|
73 |
The macro C<ECB_PTRSIZE> is defined to the size of a pointer on this |
74 |
platform (currently C<4> or C<8>) and can be used in preprocessor |
75 |
expressions. |
76 |
|
77 |
For C<ptrdiff_t> and C<size_t> use C<stddef.h>/C<cstddef>. |
78 |
|
79 |
=head2 LANGUAGE/ENVIRONMENT/COMPILER VERSIONS |
80 |
|
81 |
All the following symbols expand to an expression that can be tested in |
82 |
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 |
=over 4 |
86 |
|
87 |
=item ECB_C |
88 |
|
89 |
True if the implementation defines the C<__STDC__> macro to a true value, |
90 |
while not claiming to be C++, i..e C, but not C++. |
91 |
|
92 |
=item ECB_C99 |
93 |
|
94 |
True if the implementation claims to be compliant to C99 (ISO/IEC |
95 |
9899:1999) or any later version, while not claiming to be C++. |
96 |
|
97 |
Note that later versions (ECB_C11) remove core features again (for |
98 |
example, variable length arrays). |
99 |
|
100 |
=item ECB_C11, ECB_C17 |
101 |
|
102 |
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 |
|
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 |
=item ECB_CPP11, ECB_CPP14, ECB_CPP17 |
111 |
|
112 |
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 |
|
115 |
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 |
=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 |
=item ECB_GCC_VERSION (major, minor) |
125 |
|
126 |
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 |
|
129 |
This macro tries to return false on compilers that claim to be GCC |
130 |
compatible but aren't. |
131 |
|
132 |
=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 |
If this evaluates to a true value (suitable for testing by the |
157 |
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 |
=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 |
=back |
181 |
|
182 |
=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 |
=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 |
=back |
223 |
|
224 |
=head2 ATTRIBUTES |
225 |
|
226 |
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 |
|
232 |
ecb_const int mysqrt (int a); |
233 |
ecb_unused int i; |
234 |
|
235 |
=over 4 |
236 |
|
237 |
=item ecb_unused |
238 |
|
239 |
Marks a function or a variable as "unused", which simply suppresses a |
240 |
warning by the compiler when it detects it as unused. This is useful when |
241 |
you e.g. declare a variable but do not always use it: |
242 |
|
243 |
{ |
244 |
ecb_unused int var; |
245 |
|
246 |
#ifdef SOMECONDITION |
247 |
var = ...; |
248 |
return var; |
249 |
#else |
250 |
return 0; |
251 |
#endif |
252 |
} |
253 |
|
254 |
=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 |
=item ecb_deprecated_message (message) |
260 |
|
261 |
Same as C<ecb_deprecated>, but if possible, the specified diagnostic is |
262 |
used instead of a generic depreciation message when the object is being |
263 |
used. |
264 |
|
265 |
=item ecb_inline |
266 |
|
267 |
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 |
|
271 |
Example: inline this function, it surely will reduce codesize. |
272 |
|
273 |
ecb_inline int |
274 |
negmul (int a, int b) |
275 |
{ |
276 |
return - (a * b); |
277 |
} |
278 |
|
279 |
=item ecb_noinline |
280 |
|
281 |
Prevents a function from being inlined - it might be optimised away, but |
282 |
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 |
=item ecb_noreturn |
286 |
|
287 |
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 |
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 |
|
301 |
=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 |
multiply (ecb_restrict float *src, |
313 |
ecb_restrict float *dst, |
314 |
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 |
=item ecb_const |
323 |
|
324 |
Declares that the function only depends on the values of its arguments, |
325 |
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 |
such as a function returning the square root of its input argument. |
336 |
|
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 |
=item ecb_pure |
344 |
|
345 |
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 |
=item ecb_hot |
360 |
|
361 |
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 |
Whether a function is hot or not often depends on the whole program, |
369 |
and less on the function itself. C<ecb_cold> is likely more useful in |
370 |
practise. |
371 |
|
372 |
=item ecb_cold |
373 |
|
374 |
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 |
C<ecb_expect_false> had been used to reach them. |
384 |
|
385 |
Good examples for such functions would be error reporting functions, or |
386 |
functions only called in exceptional or rare cases. |
387 |
|
388 |
=item ecb_artificial |
389 |
|
390 |
Declares the function as "artificial", in this case meaning that this |
391 |
function is not really meant to be a function, but more like an accessor |
392 |
- 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 |
=back |
416 |
|
417 |
=head2 OPTIMISATION HINTS |
418 |
|
419 |
=over 4 |
420 |
|
421 |
=item bool ecb_is_constant (expr) |
422 |
|
423 |
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 |
0..n-1, then you could use this inline function in a header file: |
429 |
|
430 |
ecb_inline uint32_t |
431 |
rndm (uint32_t n) |
432 |
{ |
433 |
return (n * (uint32_t)rndm16 ()) >> 16; |
434 |
} |
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 |
: (n * (uint32_t)rndm16 ()) >> 16; |
447 |
} |
448 |
|
449 |
=item ecb_expect (expr, value) |
450 |
|
451 |
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 |
|
455 |
Usually, you want to use the more intuitive C<ecb_expect_true> and |
456 |
C<ecb_expect_false> functions instead. |
457 |
|
458 |
=item bool ecb_expect_true (cond) |
459 |
|
460 |
=item bool ecb_expect_false (cond) |
461 |
|
462 |
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 |
if (ecb_expect_true (some_condition)) ...; |
469 |
|
470 |
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 |
|
474 |
For example, when you check for a null pointer and expect this to be a |
475 |
rare, exceptional, case, then use C<ecb_expect_false>: |
476 |
|
477 |
void my_free (void *ptr) |
478 |
{ |
479 |
if (ecb_expect_false (ptr == 0)) |
480 |
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 |
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 |
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 |
each time something is added, you have to check for a buffer overrun, but |
496 |
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 |
if (ecb_expect_false (current + size > end)) |
503 |
real_reserve_method (size); /* presumably noinline */ |
504 |
} |
505 |
|
506 |
=item ecb_assume (cond) |
507 |
|
508 |
Tries to tell the compiler that some condition is true, even if it's not |
509 |
obvious. This is not a function, but a statement: it cannot be used in |
510 |
another expression. |
511 |
|
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 |
For example, the example reservation function from the C<ecb_expect_false> |
517 |
description could be written thus (only C<ecb_assume> was added): |
518 |
|
519 |
ecb_inline void |
520 |
reserve (int size) |
521 |
{ |
522 |
if (ecb_expect_false (current + size > end)) |
523 |
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 |
=item ecb_unreachable () |
538 |
|
539 |
This function does nothing itself, except tell the compiler that it will |
540 |
never be executed. Apart from suppressing a warning in some cases, this |
541 |
function can be used to implement C<ecb_assume> or similar functionality. |
542 |
|
543 |
=item ecb_prefetch (addr, rw, locality) |
544 |
|
545 |
Tells the compiler to try to prefetch memory at the given C<addr>ess |
546 |
for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of |
547 |
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 |
This is a statement, not a function: you cannot use it as part of an |
554 |
expression. |
555 |
|
556 |
An obvious way to use this is to prefetch some data far away, in a big |
557 |
array you loop over. This prefetches memory some 128 array elements later, |
558 |
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 |
|
582 |
=back |
583 |
|
584 |
=head2 BIT FIDDLING / BIT WIZARDRY |
585 |
|
586 |
=over 4 |
587 |
|
588 |
=item bool ecb_big_endian () |
589 |
|
590 |
=item bool ecb_little_endian () |
591 |
|
592 |
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 |
On systems that are neither, their return values are unspecified. |
597 |
|
598 |
=item int ecb_ctz32 (uint32_t x) |
599 |
|
600 |
=item int ecb_ctz64 (uint64_t x) |
601 |
|
602 |
=item int ecb_ctz (T x) [C++] |
603 |
|
604 |
Returns the index of the least significant bit set in C<x> (or |
605 |
equivalently the number of bits set to 0 before the least significant bit |
606 |
set), starting from 0. If C<x> is 0 the result is undefined. |
607 |
|
608 |
For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>. |
609 |
|
610 |
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 |
For example: |
614 |
|
615 |
ecb_ctz32 (3) = 0 |
616 |
ecb_ctz32 (6) = 1 |
617 |
|
618 |
=item bool ecb_is_pot32 (uint32_t x) |
619 |
|
620 |
=item bool ecb_is_pot64 (uint32_t x) |
621 |
|
622 |
=item bool ecb_is_pot (T x) [C++] |
623 |
|
624 |
Returns true iff C<x> is a power of two or C<x == 0>. |
625 |
|
626 |
For smaller types than C<uint32_t> you can safely use C<ecb_is_pot32>. |
627 |
|
628 |
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 |
=item int ecb_ld32 (uint32_t x) |
632 |
|
633 |
=item int ecb_ld64 (uint64_t x) |
634 |
|
635 |
=item int ecb_ld64 (T x) [C++] |
636 |
|
637 |
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 |
For smaller types than C<uint32_t> you can safely use C<ecb_ld32>. |
649 |
|
650 |
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 |
=item int ecb_popcount32 (uint32_t x) |
654 |
|
655 |
=item int ecb_popcount64 (uint64_t x) |
656 |
|
657 |
=item int ecb_popcount (T x) [C++] |
658 |
|
659 |
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 |
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 |
For example: |
667 |
|
668 |
ecb_popcount32 (7) = 3 |
669 |
ecb_popcount32 (255) = 8 |
670 |
|
671 |
=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 |
=item T ecb_bitrev (T x) [C++] |
678 |
|
679 |
Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1 |
680 |
and so on. |
681 |
|
682 |
The overloaded C++ C<ecb_bitrev> function supports C<uint8_t>, C<uint16_t> and C<uint32_t> types. |
683 |
|
684 |
Example: |
685 |
|
686 |
ecb_bitrev8 (0xa7) = 0xea |
687 |
ecb_bitrev32 (0xffcc4411) = 0x882233ff |
688 |
|
689 |
=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 |
=item uint32_t ecb_bswap16 (uint32_t x) |
696 |
|
697 |
=item uint32_t ecb_bswap32 (uint32_t x) |
698 |
|
699 |
=item uint64_t ecb_bswap64 (uint64_t x) |
700 |
|
701 |
=item T ecb_bswap (T x) |
702 |
|
703 |
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 |
The overloaded C++ C<ecb_bswap> function supports C<uint8_t>, C<uint16_t>, |
708 |
C<uint32_t> and C<uint64_t> types. |
709 |
|
710 |
=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 |
|
714 |
=item uint32_t ecb_rotl32 (uint32_t x, unsigned int count) |
715 |
|
716 |
=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 |
=item uint64_t ecb_rotr64 (uint64_t x, unsigned int count) |
725 |
|
726 |
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 |
|
730 |
Current GCC/clang versions understand these functions and usually compile |
731 |
them to "optimal" code (e.g. a single C<rol> or a combination of C<shld> |
732 |
on x86). |
733 |
|
734 |
=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 |
=back |
743 |
|
744 |
=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 |
where C<be> and C<le> stand for big endian and little endian, respectively. |
764 |
|
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 |
In C++ the following additional template functions are supported: |
783 |
|
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 |
These functions work like their C counterparts, above, but use templates, |
795 |
which make them useful in generic code. |
796 |
|
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 |
In C++ the following additional template functions are supported: |
858 |
|
859 |
=over 4 |
860 |
|
861 |
=item T ecb_peek<T> (const void *ptr) |
862 |
|
863 |
=item T ecb_peek_be<T> (const void *ptr) |
864 |
|
865 |
=item T ecb_peek_le<T> (const void *ptr) |
866 |
|
867 |
=item T ecb_peek_u<T> (const void *ptr) |
868 |
|
869 |
=item T ecb_peek_be_u<T> (const void *ptr) |
870 |
|
871 |
=item T ecb_peek_le_u<T> (const void *ptr) |
872 |
|
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 |
Since the type cannot be deduced, it has to be specified explicitly, e.g. |
878 |
|
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 |
=head2 FLOATING POINT FIDDLING |
912 |
|
913 |
=over 4 |
914 |
|
915 |
=item ECB_INFINITY [-UECB_NO_LIBM] |
916 |
|
917 |
Evaluates to positive infinity if supported by the platform, otherwise to |
918 |
a truly huge number. |
919 |
|
920 |
=item ECB_NAN [-UECB_NO_LIBM] |
921 |
|
922 |
Evaluates to a quiet NAN if supported by the platform, otherwise to |
923 |
C<ECB_INFINITY>. |
924 |
|
925 |
=item float ecb_ldexpf (float x, int exp) [-UECB_NO_LIBM] |
926 |
|
927 |
Same as C<ldexpf>, but always available. |
928 |
|
929 |
=item uint32_t ecb_float_to_binary16 (float x) [-UECB_NO_LIBM] |
930 |
|
931 |
=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 |
type and return the IEEE 754 bit representation of it (binary16/half, |
937 |
binary32/single or binary64/double precision). |
938 |
|
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 |
can serialise the return value like a normal uint16_t/uint32_t/uint64_t. |
952 |
|
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 |
=item float ecb_binary16_to_float (uint16_t x) [-UECB_NO_LIBM] |
965 |
|
966 |
=item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM] |
967 |
|
968 |
=item double ecb_binary64_to_double (uint64_t x) [-UECB_NO_LIBM] |
969 |
|
970 |
The reverse operation of the previous function - takes the bit |
971 |
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 |
|
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 |
=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 |
versa, handling all details (round-to-nearest-even, subnormals, infinity |
990 |
and NaNs) correctly. |
991 |
|
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 |
=back |
998 |
|
999 |
=head2 ARITHMETIC |
1000 |
|
1001 |
=over 4 |
1002 |
|
1003 |
=item x = ecb_mod (m, n) |
1004 |
|
1005 |
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 |
|
1013 |
C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be |
1014 |
negatable, that is, both C<m> and C<-m> must be representable in its |
1015 |
type (this typically excludes the minimum signed integer value, the same |
1016 |
limitation as for C</> and C<%> in C). |
1017 |
|
1018 |
Current GCC/clang versions compile this into an efficient branchless |
1019 |
sequence on almost all CPUs. |
1020 |
|
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 |
=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 |
positive. Note that these functions are implemented with macros in C |
1036 |
and with function templates in C++. |
1037 |
|
1038 |
=back |
1039 |
|
1040 |
=head2 UTILITY |
1041 |
|
1042 |
=over 4 |
1043 |
|
1044 |
=item element_count = ecb_array_length (name) |
1045 |
|
1046 |
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 |
=back |
1055 |
|
1056 |
=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 |
=item ECB_NO_THREADS |
1063 |
|
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 |
=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 |
=back |
1085 |
|
1086 |
=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 |
|