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 GCC built-ins, together |
16 |
with replacement functions for other compilers. In addition to this, |
17 |
it provides a number of other lowlevel C utilities, such as endianness |
18 |
detection, byte swapping or bit rotations. |
19 |
|
20 |
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 |
|
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 in by the preprocessor) |
127 |
if the compiler used is GNU C and the version is the given version, or |
128 |
higher. |
129 |
|
130 |
This macro tries to return false on compilers that claim to be GCC |
131 |
compatible but aren't. |
132 |
|
133 |
=item ECB_EXTERN_C |
134 |
|
135 |
Expands to C<extern "C"> in C++, and a simple C<extern> in C. |
136 |
|
137 |
This can be used to declare a single external C function: |
138 |
|
139 |
ECB_EXTERN_C int printf (const char *format, ...); |
140 |
|
141 |
=item ECB_EXTERN_C_BEG / ECB_EXTERN_C_END |
142 |
|
143 |
These two macros can be used to wrap multiple C<extern "C"> definitions - |
144 |
they expand to nothing in C. |
145 |
|
146 |
They are most useful in header files: |
147 |
|
148 |
ECB_EXTERN_C_BEG |
149 |
|
150 |
int mycfun1 (int x); |
151 |
int mycfun2 (int x); |
152 |
|
153 |
ECB_EXTERN_C_END |
154 |
|
155 |
=item ECB_STDFP |
156 |
|
157 |
If this evaluates to a true value (suitable for testing in by the |
158 |
preprocessor), then C<float> and C<double> use IEEE 754 single/binary32 |
159 |
and double/binary64 representations internally I<and> the endianness of |
160 |
both types match the endianness of C<uint32_t> and C<uint64_t>. |
161 |
|
162 |
This means you can just copy the bits of a C<float> (or C<double>) to an |
163 |
C<uint32_t> (or C<uint64_t>) and get the raw IEEE 754 bit representation |
164 |
without having to think about format or endianness. |
165 |
|
166 |
This is true for basically all modern platforms, although F<ecb.h> might |
167 |
not be able to deduce this correctly everywhere and might err on the safe |
168 |
side. |
169 |
|
170 |
=item ECB_AMD64, ECB_AMD64_X32 |
171 |
|
172 |
These two macros are defined to C<1> on the x86_64/amd64 ABI and the X32 |
173 |
ABI, respectively, and undefined elsewhere. |
174 |
|
175 |
The designers of the new X32 ABI for some inexplicable reason decided to |
176 |
make it look exactly like amd64, even though it's completely incompatible |
177 |
to that ABI, breaking about every piece of software that assumed that |
178 |
C<__x86_64> stands for, well, the x86-64 ABI, making these macros |
179 |
necessary. |
180 |
|
181 |
=back |
182 |
|
183 |
=head2 MACRO TRICKERY |
184 |
|
185 |
=over 4 |
186 |
|
187 |
=item ECB_CONCAT (a, b) |
188 |
|
189 |
Expands any macros in C<a> and C<b>, then concatenates the result to form |
190 |
a single token. This is mainly useful to form identifiers from components, |
191 |
e.g.: |
192 |
|
193 |
#define S1 str |
194 |
#define S2 cpy |
195 |
|
196 |
ECB_CONCAT (S1, S2)(dst, src); // == strcpy (dst, src); |
197 |
|
198 |
=item ECB_STRINGIFY (arg) |
199 |
|
200 |
Expands any macros in C<arg> and returns the stringified version of |
201 |
it. This is mainly useful to get the contents of a macro in string form, |
202 |
e.g.: |
203 |
|
204 |
#define SQL_LIMIT 100 |
205 |
sql_exec ("select * from table limit " ECB_STRINGIFY (SQL_LIMIT)); |
206 |
|
207 |
=item ECB_STRINGIFY_EXPR (expr) |
208 |
|
209 |
Like C<ECB_STRINGIFY>, but additionally evaluates C<expr> to make sure it |
210 |
is a valid expression. This is useful to catch typos or cases where the |
211 |
macro isn't available: |
212 |
|
213 |
#include <errno.h> |
214 |
|
215 |
ECB_STRINGIFY (EDOM); // "33" (on my system at least) |
216 |
ECB_STRINGIFY_EXPR (EDOM); // "33" |
217 |
|
218 |
// now imagine we had a typo: |
219 |
|
220 |
ECB_STRINGIFY (EDAM); // "EDAM" |
221 |
ECB_STRINGIFY_EXPR (EDAM); // error: EDAM undefined |
222 |
|
223 |
=back |
224 |
|
225 |
=head2 ATTRIBUTES |
226 |
|
227 |
A major part of libecb deals with additional attributes that can be |
228 |
assigned to functions, variables and sometimes even types - much like |
229 |
C<const> or C<volatile> in C. They are implemented using either GCC |
230 |
attributes or other compiler/language specific features. Attributes |
231 |
declarations must be put before the whole declaration: |
232 |
|
233 |
ecb_const int mysqrt (int a); |
234 |
ecb_unused int i; |
235 |
|
236 |
=over 4 |
237 |
|
238 |
=item ecb_unused |
239 |
|
240 |
Marks a function or a variable as "unused", which simply suppresses a |
241 |
warning by GCC when it detects it as unused. This is useful when you e.g. |
242 |
declare a variable but do not always use it: |
243 |
|
244 |
{ |
245 |
ecb_unused int var; |
246 |
|
247 |
#ifdef SOMECONDITION |
248 |
var = ...; |
249 |
return var; |
250 |
#else |
251 |
return 0; |
252 |
#endif |
253 |
} |
254 |
|
255 |
=item ecb_deprecated |
256 |
|
257 |
Similar to C<ecb_unused>, but marks a function, variable or type as |
258 |
deprecated. This makes some compilers warn when the type is used. |
259 |
|
260 |
=item ecb_deprecated_message (message) |
261 |
|
262 |
Same as C<ecb_deprecated>, but if possible, the specified diagnostic is |
263 |
used instead of a generic depreciation message when the object is being |
264 |
used. |
265 |
|
266 |
=item ecb_inline |
267 |
|
268 |
Expands either to (a compiler-specific equivalent of) C<static inline> or |
269 |
to just C<static>, if inline isn't supported. It should be used to declare |
270 |
functions that should be inlined, for code size or speed reasons. |
271 |
|
272 |
Example: inline this function, it surely will reduce codesize. |
273 |
|
274 |
ecb_inline int |
275 |
negmul (int a, int b) |
276 |
{ |
277 |
return - (a * b); |
278 |
} |
279 |
|
280 |
=item ecb_noinline |
281 |
|
282 |
Prevents a function from being inlined - it might be optimised away, but |
283 |
not inlined into other functions. This is useful if you know your function |
284 |
is rarely called and large enough for inlining not to be helpful. |
285 |
|
286 |
=item ecb_noreturn |
287 |
|
288 |
Marks a function as "not returning, ever". Some typical functions that |
289 |
don't return are C<exit> or C<abort> (which really works hard to not |
290 |
return), and now you can make your own: |
291 |
|
292 |
ecb_noreturn void |
293 |
my_abort (const char *errline) |
294 |
{ |
295 |
puts (errline); |
296 |
abort (); |
297 |
} |
298 |
|
299 |
In this case, the compiler would probably be smart enough to deduce it on |
300 |
its own, so this is mainly useful for declarations. |
301 |
|
302 |
=item ecb_restrict |
303 |
|
304 |
Expands to the C<restrict> keyword or equivalent on compilers that support |
305 |
them, and to nothing on others. Must be specified on a pointer type or |
306 |
an array index to indicate that the memory doesn't alias with any other |
307 |
restricted pointer in the same scope. |
308 |
|
309 |
Example: multiply a vector, and allow the compiler to parallelise the |
310 |
loop, because it knows it doesn't overwrite input values. |
311 |
|
312 |
void |
313 |
multiply (ecb_restrict float *src, |
314 |
ecb_restrict float *dst, |
315 |
int len, float factor) |
316 |
{ |
317 |
int i; |
318 |
|
319 |
for (i = 0; i < len; ++i) |
320 |
dst [i] = src [i] * factor; |
321 |
} |
322 |
|
323 |
=item ecb_const |
324 |
|
325 |
Declares that the function only depends on the values of its arguments, |
326 |
much like a mathematical function. It specifically does not read or write |
327 |
any memory any arguments might point to, global variables, or call any |
328 |
non-const functions. It also must not have any side effects. |
329 |
|
330 |
Such a function can be optimised much more aggressively by the compiler - |
331 |
for example, multiple calls with the same arguments can be optimised into |
332 |
a single call, which wouldn't be possible if the compiler would have to |
333 |
expect any side effects. |
334 |
|
335 |
It is best suited for functions in the sense of mathematical functions, |
336 |
such as a function returning the square root of its input argument. |
337 |
|
338 |
Not suited would be a function that calculates the hash of some memory |
339 |
area you pass in, prints some messages or looks at a global variable to |
340 |
decide on rounding. |
341 |
|
342 |
See C<ecb_pure> for a slightly less restrictive class of functions. |
343 |
|
344 |
=item ecb_pure |
345 |
|
346 |
Similar to C<ecb_const>, declares a function that has no side |
347 |
effects. Unlike C<ecb_const>, the function is allowed to examine global |
348 |
variables and any other memory areas (such as the ones passed to it via |
349 |
pointers). |
350 |
|
351 |
While these functions cannot be optimised as aggressively as C<ecb_const> |
352 |
functions, they can still be optimised away in many occasions, and the |
353 |
compiler has more freedom in moving calls to them around. |
354 |
|
355 |
Typical examples for such functions would be C<strlen> or C<memcmp>. A |
356 |
function that calculates the MD5 sum of some input and updates some MD5 |
357 |
state passed as argument would I<NOT> be pure, however, as it would modify |
358 |
some memory area that is not the return value. |
359 |
|
360 |
=item ecb_hot |
361 |
|
362 |
This declares a function as "hot" with regards to the cache - the function |
363 |
is used so often, that it is very beneficial to keep it in the cache if |
364 |
possible. |
365 |
|
366 |
The compiler reacts by trying to place hot functions near to each other in |
367 |
memory. |
368 |
|
369 |
Whether a function is hot or not often depends on the whole program, |
370 |
and less on the function itself. C<ecb_cold> is likely more useful in |
371 |
practise. |
372 |
|
373 |
=item ecb_cold |
374 |
|
375 |
The opposite of C<ecb_hot> - declares a function as "cold" with regards to |
376 |
the cache, or in other words, this function is not called often, or not at |
377 |
speed-critical times, and keeping it in the cache might be a waste of said |
378 |
cache. |
379 |
|
380 |
In addition to placing cold functions together (or at least away from hot |
381 |
functions), this knowledge can be used in other ways, for example, the |
382 |
function will be optimised for size, as opposed to speed, and codepaths |
383 |
leading to calls to those functions can automatically be marked as if |
384 |
C<ecb_expect_false> had been used to reach them. |
385 |
|
386 |
Good examples for such functions would be error reporting functions, or |
387 |
functions only called in exceptional or rare cases. |
388 |
|
389 |
=item ecb_artificial |
390 |
|
391 |
Declares the function as "artificial", in this case meaning that this |
392 |
function is not really meant to be a function, but more like an accessor |
393 |
- many methods in C++ classes are mere accessor functions, and having a |
394 |
crash reported in such a method, or single-stepping through them, is not |
395 |
usually so helpful, especially when it's inlined to just a few instructions. |
396 |
|
397 |
Marking them as artificial will instruct the debugger about just this, |
398 |
leading to happier debugging and thus happier lives. |
399 |
|
400 |
Example: in some kind of smart-pointer class, mark the pointer accessor as |
401 |
artificial, so that the whole class acts more like a pointer and less like |
402 |
some C++ abstraction monster. |
403 |
|
404 |
template<typename T> |
405 |
struct my_smart_ptr |
406 |
{ |
407 |
T *value; |
408 |
|
409 |
ecb_artificial |
410 |
operator T *() |
411 |
{ |
412 |
return value; |
413 |
} |
414 |
}; |
415 |
|
416 |
=back |
417 |
|
418 |
=head2 OPTIMISATION HINTS |
419 |
|
420 |
=over 4 |
421 |
|
422 |
=item bool ecb_is_constant (expr) |
423 |
|
424 |
Returns true iff the expression can be deduced to be a compile-time |
425 |
constant, and false otherwise. |
426 |
|
427 |
For example, when you have a C<rndm16> function that returns a 16 bit |
428 |
random number, and you have a function that maps this to a range from |
429 |
0..n-1, then you could use this inline function in a header file: |
430 |
|
431 |
ecb_inline uint32_t |
432 |
rndm (uint32_t n) |
433 |
{ |
434 |
return (n * (uint32_t)rndm16 ()) >> 16; |
435 |
} |
436 |
|
437 |
However, for powers of two, you could use a normal mask, but that is only |
438 |
worth it if, at compile time, you can detect this case. This is the case |
439 |
when the passed number is a constant and also a power of two (C<n & (n - |
440 |
1) == 0>): |
441 |
|
442 |
ecb_inline uint32_t |
443 |
rndm (uint32_t n) |
444 |
{ |
445 |
return is_constant (n) && !(n & (n - 1)) |
446 |
? rndm16 () & (num - 1) |
447 |
: (n * (uint32_t)rndm16 ()) >> 16; |
448 |
} |
449 |
|
450 |
=item ecb_expect (expr, value) |
451 |
|
452 |
Evaluates C<expr> and returns it. In addition, it tells the compiler that |
453 |
the C<expr> evaluates to C<value> a lot, which can be used for static |
454 |
branch optimisations. |
455 |
|
456 |
Usually, you want to use the more intuitive C<ecb_expect_true> and |
457 |
C<ecb_expect_false> functions instead. |
458 |
|
459 |
=item bool ecb_expect_true (cond) |
460 |
|
461 |
=item bool ecb_expect_false (cond) |
462 |
|
463 |
These two functions expect a expression that is true or false and return |
464 |
C<1> or C<0>, respectively, so when used in the condition of an C<if> or |
465 |
other conditional statement, it will not change the program: |
466 |
|
467 |
/* these two do the same thing */ |
468 |
if (some_condition) ...; |
469 |
if (ecb_expect_true (some_condition)) ...; |
470 |
|
471 |
However, by using C<ecb_expect_true>, you tell the compiler that the |
472 |
condition is likely to be true (and for C<ecb_expect_false>, that it is |
473 |
unlikely to be true). |
474 |
|
475 |
For example, when you check for a null pointer and expect this to be a |
476 |
rare, exceptional, case, then use C<ecb_expect_false>: |
477 |
|
478 |
void my_free (void *ptr) |
479 |
{ |
480 |
if (ecb_expect_false (ptr == 0)) |
481 |
return; |
482 |
} |
483 |
|
484 |
Consequent use of these functions to mark away exceptional cases or to |
485 |
tell the compiler what the hot path through a function is can increase |
486 |
performance considerably. |
487 |
|
488 |
You might know these functions under the name C<likely> and C<unlikely> |
489 |
- while these are common aliases, we find that the expect name is easier |
490 |
to understand when quickly skimming code. If you wish, you can use |
491 |
C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of |
492 |
C<ecb_expect_false> - these are simply aliases. |
493 |
|
494 |
A very good example is in a function that reserves more space for some |
495 |
memory block (for example, inside an implementation of a string stream) - |
496 |
each time something is added, you have to check for a buffer overrun, but |
497 |
you expect that most checks will turn out to be false: |
498 |
|
499 |
/* make sure we have "size" extra room in our buffer */ |
500 |
ecb_inline void |
501 |
reserve (int size) |
502 |
{ |
503 |
if (ecb_expect_false (current + size > end)) |
504 |
real_reserve_method (size); /* presumably noinline */ |
505 |
} |
506 |
|
507 |
=item ecb_assume (cond) |
508 |
|
509 |
Tries to tell the compiler that some condition is true, even if it's not |
510 |
obvious. This is not a function, but a statement: it cannot be used in |
511 |
another expression. |
512 |
|
513 |
This can be used to teach the compiler about invariants or other |
514 |
conditions that might improve code generation, but which are impossible to |
515 |
deduce form the code itself. |
516 |
|
517 |
For example, the example reservation function from the C<ecb_expect_false> |
518 |
description could be written thus (only C<ecb_assume> was added): |
519 |
|
520 |
ecb_inline void |
521 |
reserve (int size) |
522 |
{ |
523 |
if (ecb_expect_false (current + size > end)) |
524 |
real_reserve_method (size); /* presumably noinline */ |
525 |
|
526 |
ecb_assume (current + size <= end); |
527 |
} |
528 |
|
529 |
If you then call this function twice, like this: |
530 |
|
531 |
reserve (10); |
532 |
reserve (1); |
533 |
|
534 |
Then the compiler I<might> be able to optimise out the second call |
535 |
completely, as it knows that C<< current + 1 > end >> is false and the |
536 |
call will never be executed. |
537 |
|
538 |
=item ecb_unreachable () |
539 |
|
540 |
This function does nothing itself, except tell the compiler that it will |
541 |
never be executed. Apart from suppressing a warning in some cases, this |
542 |
function can be used to implement C<ecb_assume> or similar functionality. |
543 |
|
544 |
=item ecb_prefetch (addr, rw, locality) |
545 |
|
546 |
Tells the compiler to try to prefetch memory at the given C<addr>ess |
547 |
for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of |
548 |
C<0> means that there will only be one access later, C<3> means that |
549 |
the data will likely be accessed very often, and values in between mean |
550 |
something... in between. The memory pointed to by the address does not |
551 |
need to be accessible (it could be a null pointer for example), but C<rw> |
552 |
and C<locality> must be compile-time constants. |
553 |
|
554 |
This is a statement, not a function: you cannot use it as part of an |
555 |
expression. |
556 |
|
557 |
An obvious way to use this is to prefetch some data far away, in a big |
558 |
array you loop over. This prefetches memory some 128 array elements later, |
559 |
in the hope that it will be ready when the CPU arrives at that location. |
560 |
|
561 |
int sum = 0; |
562 |
|
563 |
for (i = 0; i < N; ++i) |
564 |
{ |
565 |
sum += arr [i] |
566 |
ecb_prefetch (arr + i + 128, 0, 0); |
567 |
} |
568 |
|
569 |
It's hard to predict how far to prefetch, and most CPUs that can prefetch |
570 |
are often good enough to predict this kind of behaviour themselves. It |
571 |
gets more interesting with linked lists, especially when you do some fair |
572 |
processing on each list element: |
573 |
|
574 |
for (node *n = start; n; n = n->next) |
575 |
{ |
576 |
ecb_prefetch (n->next, 0, 0); |
577 |
... do medium amount of work with *n |
578 |
} |
579 |
|
580 |
After processing the node, (part of) the next node might already be in |
581 |
cache. |
582 |
|
583 |
=back |
584 |
|
585 |
=head2 BIT FIDDLING / BIT WIZARDRY |
586 |
|
587 |
=over 4 |
588 |
|
589 |
=item bool ecb_big_endian () |
590 |
|
591 |
=item bool ecb_little_endian () |
592 |
|
593 |
These two functions return true if the byte order is big endian |
594 |
(most-significant byte first) or little endian (least-significant byte |
595 |
first) respectively. |
596 |
|
597 |
On systems that are neither, their return values are unspecified. |
598 |
|
599 |
=item int ecb_ctz32 (uint32_t x) |
600 |
|
601 |
=item int ecb_ctz64 (uint64_t x) |
602 |
|
603 |
=item int ecb_ctz (T x) [C++] |
604 |
|
605 |
Returns the index of the least significant bit set in C<x> (or |
606 |
equivalently the number of bits set to 0 before the least significant bit |
607 |
set), starting from 0. If C<x> is 0 the result is undefined. |
608 |
|
609 |
For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>. |
610 |
|
611 |
The overloaded C++ C<ecb_ctz> function supports C<uint8_t>, C<uint16_t>, |
612 |
C<uint32_t> and C<uint64_t> types. |
613 |
|
614 |
For example: |
615 |
|
616 |
ecb_ctz32 (3) = 0 |
617 |
ecb_ctz32 (6) = 1 |
618 |
|
619 |
=item bool ecb_is_pot32 (uint32_t x) |
620 |
|
621 |
=item bool ecb_is_pot64 (uint32_t x) |
622 |
|
623 |
=item bool ecb_is_pot (T x) [C++] |
624 |
|
625 |
Returns true iff C<x> is a power of two or C<x == 0>. |
626 |
|
627 |
For smaller types than C<uint32_t> you can safely use C<ecb_is_pot32>. |
628 |
|
629 |
The overloaded C++ C<ecb_is_pot> function supports C<uint8_t>, C<uint16_t>, |
630 |
C<uint32_t> and C<uint64_t> types. |
631 |
|
632 |
=item int ecb_ld32 (uint32_t x) |
633 |
|
634 |
=item int ecb_ld64 (uint64_t x) |
635 |
|
636 |
=item int ecb_ld64 (T x) [C++] |
637 |
|
638 |
Returns the index of the most significant bit set in C<x>, or the number |
639 |
of digits the number requires in binary (so that C<< 2**ld <= x < |
640 |
2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is |
641 |
to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for |
642 |
example to see how many bits a certain number requires to be encoded. |
643 |
|
644 |
This function is similar to the "count leading zero bits" function, except |
645 |
that that one returns how many zero bits are "in front" of the number (in |
646 |
the given data type), while C<ecb_ld> returns how many bits the number |
647 |
itself requires. |
648 |
|
649 |
For smaller types than C<uint32_t> you can safely use C<ecb_ld32>. |
650 |
|
651 |
The overloaded C++ C<ecb_ld> function supports C<uint8_t>, C<uint16_t>, |
652 |
C<uint32_t> and C<uint64_t> types. |
653 |
|
654 |
=item int ecb_popcount32 (uint32_t x) |
655 |
|
656 |
=item int ecb_popcount64 (uint64_t x) |
657 |
|
658 |
=item int ecb_popcount (T x) [C++] |
659 |
|
660 |
Returns the number of bits set to 1 in C<x>. |
661 |
|
662 |
For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>. |
663 |
|
664 |
The overloaded C++ C<ecb_popcount> function supports C<uint8_t>, C<uint16_t>, |
665 |
C<uint32_t> and C<uint64_t> types. |
666 |
|
667 |
For example: |
668 |
|
669 |
ecb_popcount32 (7) = 3 |
670 |
ecb_popcount32 (255) = 8 |
671 |
|
672 |
=item uint8_t ecb_bitrev8 (uint8_t x) |
673 |
|
674 |
=item uint16_t ecb_bitrev16 (uint16_t x) |
675 |
|
676 |
=item uint32_t ecb_bitrev32 (uint32_t x) |
677 |
|
678 |
=item T ecb_bitrev (T x) [C++] |
679 |
|
680 |
Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1 |
681 |
and so on. |
682 |
|
683 |
The overloaded C++ C<ecb_bitrev> function supports C<uint8_t>, C<uint16_t> and C<uint32_t> types. |
684 |
|
685 |
Example: |
686 |
|
687 |
ecb_bitrev8 (0xa7) = 0xea |
688 |
ecb_bitrev32 (0xffcc4411) = 0x882233ff |
689 |
|
690 |
=item T ecb_bitrev (T x) [C++] |
691 |
|
692 |
Overloaded C++ bitrev function. |
693 |
|
694 |
C<T> must be one of C<uint8_t>, C<uint16_t> or C<uint32_t>. |
695 |
|
696 |
=item uint32_t ecb_bswap16 (uint32_t x) |
697 |
|
698 |
=item uint32_t ecb_bswap32 (uint32_t x) |
699 |
|
700 |
=item uint64_t ecb_bswap64 (uint64_t x) |
701 |
|
702 |
=item T ecb_bswap (T x) |
703 |
|
704 |
These functions return the value of the 16-bit (32-bit, 64-bit) value |
705 |
C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in |
706 |
C<ecb_bswap32>). |
707 |
|
708 |
The overloaded C++ C<ecb_bswap> function supports C<uint8_t>, C<uint16_t>, |
709 |
C<uint32_t> and C<uint64_t> types. |
710 |
|
711 |
=item uint8_t ecb_rotl8 (uint8_t x, unsigned int count) |
712 |
|
713 |
=item uint16_t ecb_rotl16 (uint16_t x, unsigned int count) |
714 |
|
715 |
=item uint32_t ecb_rotl32 (uint32_t x, unsigned int count) |
716 |
|
717 |
=item uint64_t ecb_rotl64 (uint64_t x, unsigned int count) |
718 |
|
719 |
=item uint8_t ecb_rotr8 (uint8_t x, unsigned int count) |
720 |
|
721 |
=item uint16_t ecb_rotr16 (uint16_t x, unsigned int count) |
722 |
|
723 |
=item uint32_t ecb_rotr32 (uint32_t x, unsigned int count) |
724 |
|
725 |
=item uint64_t ecb_rotr64 (uint64_t x, unsigned int count) |
726 |
|
727 |
These two families of functions return the value of C<x> after rotating |
728 |
all the bits by C<count> positions to the right (C<ecb_rotr>) or left |
729 |
(C<ecb_rotl>). |
730 |
|
731 |
Current GCC versions understand these functions and usually compile them |
732 |
to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on |
733 |
x86). |
734 |
|
735 |
=item T ecb_rotl (T x, unsigned int count) [C++] |
736 |
|
737 |
=item T ecb_rotr (T x, unsigned int count) [C++] |
738 |
|
739 |
Overloaded C++ rotl/rotr functions. |
740 |
|
741 |
C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>. |
742 |
|
743 |
=back |
744 |
|
745 |
=head2 HOST ENDIANNESS CONVERSION |
746 |
|
747 |
=over 4 |
748 |
|
749 |
=item uint_fast16_t ecb_be_u16_to_host (uint_fast16_t v) |
750 |
|
751 |
=item uint_fast32_t ecb_be_u32_to_host (uint_fast32_t v) |
752 |
|
753 |
=item uint_fast64_t ecb_be_u64_to_host (uint_fast64_t v) |
754 |
|
755 |
=item uint_fast16_t ecb_le_u16_to_host (uint_fast16_t v) |
756 |
|
757 |
=item uint_fast32_t ecb_le_u32_to_host (uint_fast32_t v) |
758 |
|
759 |
=item uint_fast64_t ecb_le_u64_to_host (uint_fast64_t v) |
760 |
|
761 |
Convert an unsigned 16, 32 or 64 bit value from big or little endian to host byte order. |
762 |
|
763 |
The naming convention is C<ecb_>(C<be>|C<le>)C<_u>C<16|32|64>C<_to_host>, |
764 |
where C<be> and C<le> stand for big endian and little endian, respectively. |
765 |
|
766 |
=item uint_fast16_t ecb_host_to_be_u16 (uint_fast16_t v) |
767 |
|
768 |
=item uint_fast32_t ecb_host_to_be_u32 (uint_fast32_t v) |
769 |
|
770 |
=item uint_fast64_t ecb_host_to_be_u64 (uint_fast64_t v) |
771 |
|
772 |
=item uint_fast16_t ecb_host_to_le_u16 (uint_fast16_t v) |
773 |
|
774 |
=item uint_fast32_t ecb_host_to_le_u32 (uint_fast32_t v) |
775 |
|
776 |
=item uint_fast64_t ecb_host_to_le_u64 (uint_fast64_t v) |
777 |
|
778 |
Like above, but converts I<from> host byte order to the specified |
779 |
endianness. |
780 |
|
781 |
=back |
782 |
|
783 |
In C++ the following additional template functions are supported: |
784 |
|
785 |
=over 4 |
786 |
|
787 |
=item T ecb_be_to_host (T v) |
788 |
|
789 |
=item T ecb_le_to_host (T v) |
790 |
|
791 |
=item T ecb_host_to_be (T v) |
792 |
|
793 |
=item T ecb_host_to_le (T v) |
794 |
|
795 |
These functions work like their C counterparts, above, but use templates, |
796 |
which make them useful in generic code. |
797 |
|
798 |
C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t> |
799 |
(so unlike their C counterparts, there is a version for C<uint8_t>, which |
800 |
again can be useful in generic code). |
801 |
|
802 |
=head2 UNALIGNED LOAD/STORE |
803 |
|
804 |
These function load or store unaligned multi-byte values. |
805 |
|
806 |
=over 4 |
807 |
|
808 |
=item uint_fast16_t ecb_peek_u16_u (const void *ptr) |
809 |
|
810 |
=item uint_fast32_t ecb_peek_u32_u (const void *ptr) |
811 |
|
812 |
=item uint_fast64_t ecb_peek_u64_u (const void *ptr) |
813 |
|
814 |
These functions load an unaligned, unsigned 16, 32 or 64 bit value from |
815 |
memory. |
816 |
|
817 |
=item uint_fast16_t ecb_peek_be_u16_u (const void *ptr) |
818 |
|
819 |
=item uint_fast32_t ecb_peek_be_u32_u (const void *ptr) |
820 |
|
821 |
=item uint_fast64_t ecb_peek_be_u64_u (const void *ptr) |
822 |
|
823 |
=item uint_fast16_t ecb_peek_le_u16_u (const void *ptr) |
824 |
|
825 |
=item uint_fast32_t ecb_peek_le_u32_u (const void *ptr) |
826 |
|
827 |
=item uint_fast64_t ecb_peek_le_u64_u (const void *ptr) |
828 |
|
829 |
Like above, but additionally convert from big endian (C<be>) or little |
830 |
endian (C<le>) byte order to host byte order while doing so. |
831 |
|
832 |
=item ecb_poke_u16_u (void *ptr, uint16_t v) |
833 |
|
834 |
=item ecb_poke_u32_u (void *ptr, uint32_t v) |
835 |
|
836 |
=item ecb_poke_u64_u (void *ptr, uint64_t v) |
837 |
|
838 |
These functions store an unaligned, unsigned 16, 32 or 64 bit value to |
839 |
memory. |
840 |
|
841 |
=item ecb_poke_be_u16_u (void *ptr, uint_fast16_t v) |
842 |
|
843 |
=item ecb_poke_be_u32_u (void *ptr, uint_fast32_t v) |
844 |
|
845 |
=item ecb_poke_be_u64_u (void *ptr, uint_fast64_t v) |
846 |
|
847 |
=item ecb_poke_le_u16_u (void *ptr, uint_fast16_t v) |
848 |
|
849 |
=item ecb_poke_le_u32_u (void *ptr, uint_fast32_t v) |
850 |
|
851 |
=item ecb_poke_le_u64_u (void *ptr, uint_fast64_t v) |
852 |
|
853 |
Like above, but additionally convert from host byte order to big endian |
854 |
(C<be>) or little endian (C<le>) byte order while doing so. |
855 |
|
856 |
=back |
857 |
|
858 |
In C++ the following additional template functions are supported: |
859 |
|
860 |
=over 4 |
861 |
|
862 |
=item T ecb_peek<T> (const void *ptr) |
863 |
|
864 |
=item T ecb_peek_be<T> (const void *ptr) |
865 |
|
866 |
=item T ecb_peek_le<T> (const void *ptr) |
867 |
|
868 |
=item T ecb_peek_u<T> (const void *ptr) |
869 |
|
870 |
=item T ecb_peek_be_u<T> (const void *ptr) |
871 |
|
872 |
=item T ecb_peek_le_u<T> (const void *ptr) |
873 |
|
874 |
Similarly to their C counterparts, these functions load an unsigned 8, 16, |
875 |
32 or 64 bit value from memory, with optional conversion from big/little |
876 |
endian. |
877 |
|
878 |
Since the type cannot be deduced, it has to be specified explicitly, e.g. |
879 |
|
880 |
uint_fast16_t v = ecb_peek<uint16_t> (ptr); |
881 |
|
882 |
C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>. |
883 |
|
884 |
Unlike their C counterparts, these functions support 8 bit quantities |
885 |
(C<uint8_t>) and also have an aligned version (without the C<_u> prefix), |
886 |
all of which hopefully makes them more useful in generic code. |
887 |
|
888 |
=item ecb_poke (void *ptr, T v) |
889 |
|
890 |
=item ecb_poke_be (void *ptr, T v) |
891 |
|
892 |
=item ecb_poke_le (void *ptr, T v) |
893 |
|
894 |
=item ecb_poke_u (void *ptr, T v) |
895 |
|
896 |
=item ecb_poke_be_u (void *ptr, T v) |
897 |
|
898 |
=item ecb_poke_le_u (void *ptr, T v) |
899 |
|
900 |
Again, similarly to their C counterparts, these functions store an |
901 |
unsigned 8, 16, 32 or z64 bit value to memory, with optional conversion to |
902 |
big/little endian. |
903 |
|
904 |
C<T> must be one of C<uint8_t>, C<uint16_t>, C<uint32_t> or C<uint64_t>. |
905 |
|
906 |
Unlike their C counterparts, these functions support 8 bit quantities |
907 |
(C<uint8_t>) and also have an aligned version (without the C<_u> prefix), |
908 |
all of which hopefully makes them more useful in generic code. |
909 |
|
910 |
=back |
911 |
|
912 |
=head2 FLOATING POINT FIDDLING |
913 |
|
914 |
=over 4 |
915 |
|
916 |
=item ECB_INFINITY [-UECB_NO_LIBM] |
917 |
|
918 |
Evaluates to positive infinity if supported by the platform, otherwise to |
919 |
a truly huge number. |
920 |
|
921 |
=item ECB_NAN [-UECB_NO_LIBM] |
922 |
|
923 |
Evaluates to a quiet NAN if supported by the platform, otherwise to |
924 |
C<ECB_INFINITY>. |
925 |
|
926 |
=item float ecb_ldexpf (float x, int exp) [-UECB_NO_LIBM] |
927 |
|
928 |
Same as C<ldexpf>, but always available. |
929 |
|
930 |
=item uint32_t ecb_float_to_binary16 (float x) [-UECB_NO_LIBM] |
931 |
|
932 |
=item uint32_t ecb_float_to_binary32 (float x) [-UECB_NO_LIBM] |
933 |
|
934 |
=item uint64_t ecb_double_to_binary64 (double x) [-UECB_NO_LIBM] |
935 |
|
936 |
These functions each take an argument in the native C<float> or C<double> |
937 |
type and return the IEEE 754 bit representation of it (binary16/half, |
938 |
binary32/single or binary64/double precision). |
939 |
|
940 |
The bit representation is just as IEEE 754 defines it, i.e. the sign bit |
941 |
will be the most significant bit, followed by exponent and mantissa. |
942 |
|
943 |
This function should work even when the native floating point format isn't |
944 |
IEEE compliant, of course at a speed and code size penalty, and of course |
945 |
also within reasonable limits (it tries to convert NaNs, infinities and |
946 |
denormals, but will likely convert negative zero to positive zero). |
947 |
|
948 |
On all modern platforms (where C<ECB_STDFP> is true), the compiler should |
949 |
be able to optimise away this function completely. |
950 |
|
951 |
These functions can be helpful when serialising floats to the network - you |
952 |
can serialise the return value like a normal uint16_t/uint32_t/uint64_t. |
953 |
|
954 |
Another use for these functions is to manipulate floating point values |
955 |
directly. |
956 |
|
957 |
Silly example: toggle the sign bit of a float. |
958 |
|
959 |
/* On gcc-4.7 on amd64, */ |
960 |
/* this results in a single add instruction to toggle the bit, and 4 extra */ |
961 |
/* instructions to move the float value to an integer register and back. */ |
962 |
|
963 |
x = ecb_binary32_to_float (ecb_float_to_binary32 (x) ^ 0x80000000U) |
964 |
|
965 |
=item float ecb_binary16_to_float (uint16_t x) [-UECB_NO_LIBM] |
966 |
|
967 |
=item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM] |
968 |
|
969 |
=item double ecb_binary64_to_double (uint64_t x) [-UECB_NO_LIBM] |
970 |
|
971 |
The reverse operation of the previous function - takes the bit |
972 |
representation of an IEEE binary16, binary32 or binary64 number (half, |
973 |
single or double precision) and converts it to the native C<float> or |
974 |
C<double> format. |
975 |
|
976 |
This function should work even when the native floating point format isn't |
977 |
IEEE compliant, of course at a speed and code size penalty, and of course |
978 |
also within reasonable limits (it tries to convert normals and denormals, |
979 |
and might be lucky for infinities, and with extraordinary luck, also for |
980 |
negative zero). |
981 |
|
982 |
On all modern platforms (where C<ECB_STDFP> is true), the compiler should |
983 |
be able to optimise away this function completely. |
984 |
|
985 |
=item uint16_t ecb_binary32_to_binary16 (uint32_t x) |
986 |
|
987 |
=item uint32_t ecb_binary16_to_binary32 (uint16_t x) |
988 |
|
989 |
Convert a IEEE binary32/single precision to binary16/half format, and vice |
990 |
versa, handling all details (round-to-nearest-even, subnormals, infinity |
991 |
and NaNs) correctly. |
992 |
|
993 |
These are functions are available under C<-DECB_NO_LIBM>, since |
994 |
they do not rely on the platform floating point format. The |
995 |
C<ecb_float_to_binary16> and C<ecb_binary16_to_float> functions are |
996 |
usually what you want. |
997 |
|
998 |
=back |
999 |
|
1000 |
=head2 ARITHMETIC |
1001 |
|
1002 |
=over 4 |
1003 |
|
1004 |
=item x = ecb_mod (m, n) |
1005 |
|
1006 |
Returns C<m> modulo C<n>, which is the same as the positive remainder |
1007 |
of the division operation between C<m> and C<n>, using floored |
1008 |
division. Unlike the C remainder operator C<%>, this function ensures that |
1009 |
the return value is always positive and that the two numbers I<m> and |
1010 |
I<m' = m + i * n> result in the same value modulo I<n> - in other words, |
1011 |
C<ecb_mod> implements the mathematical modulo operation, which is missing |
1012 |
in the language. |
1013 |
|
1014 |
C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be |
1015 |
negatable, that is, both C<m> and C<-m> must be representable in its |
1016 |
type (this typically excludes the minimum signed integer value, the same |
1017 |
limitation as for C</> and C<%> in C). |
1018 |
|
1019 |
Current GCC versions compile this into an efficient branchless sequence on |
1020 |
almost all CPUs. |
1021 |
|
1022 |
For example, when you want to rotate forward through the members of an |
1023 |
array for increasing C<m> (which might be negative), then you should use |
1024 |
C<ecb_mod>, as the C<%> operator might give either negative results, or |
1025 |
change direction for negative values: |
1026 |
|
1027 |
for (m = -100; m <= 100; ++m) |
1028 |
int elem = myarray [ecb_mod (m, ecb_array_length (myarray))]; |
1029 |
|
1030 |
=item x = ecb_div_rd (val, div) |
1031 |
|
1032 |
=item x = ecb_div_ru (val, div) |
1033 |
|
1034 |
Returns C<val> divided by C<div> rounded down or up, respectively. |
1035 |
C<val> and C<div> must have integer types and C<div> must be strictly |
1036 |
positive. Note that these functions are implemented with macros in C |
1037 |
and with function templates in C++. |
1038 |
|
1039 |
=back |
1040 |
|
1041 |
=head2 UTILITY |
1042 |
|
1043 |
=over 4 |
1044 |
|
1045 |
=item element_count = ecb_array_length (name) |
1046 |
|
1047 |
Returns the number of elements in the array C<name>. For example: |
1048 |
|
1049 |
int primes[] = { 2, 3, 5, 7, 11 }; |
1050 |
int sum = 0; |
1051 |
|
1052 |
for (i = 0; i < ecb_array_length (primes); i++) |
1053 |
sum += primes [i]; |
1054 |
|
1055 |
=back |
1056 |
|
1057 |
=head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF |
1058 |
|
1059 |
These symbols need to be defined before including F<ecb.h> the first time. |
1060 |
|
1061 |
=over 4 |
1062 |
|
1063 |
=item ECB_NO_THREADS |
1064 |
|
1065 |
If F<ecb.h> is never used from multiple threads, then this symbol can |
1066 |
be defined, in which case memory fences (and similar constructs) are |
1067 |
completely removed, leading to more efficient code and fewer dependencies. |
1068 |
|
1069 |
Setting this symbol to a true value implies C<ECB_NO_SMP>. |
1070 |
|
1071 |
=item ECB_NO_SMP |
1072 |
|
1073 |
The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from |
1074 |
multiple threads, but never concurrently (e.g. if the system the program |
1075 |
runs on has only a single CPU with a single core, no hyperthreading and so |
1076 |
on), then this symbol can be defined, leading to more efficient code and |
1077 |
fewer dependencies. |
1078 |
|
1079 |
=item ECB_NO_LIBM |
1080 |
|
1081 |
When defined to C<1>, do not export any functions that might introduce |
1082 |
dependencies on the math library (usually called F<-lm>) - these are |
1083 |
marked with [-UECB_NO_LIBM]. |
1084 |
|
1085 |
=back |
1086 |
|
1087 |
=head1 UNDOCUMENTED FUNCTIONALITY |
1088 |
|
1089 |
F<ecb.h> is full of undocumented functionality as well, some of which is |
1090 |
intended to be internal-use only, some of which we forgot to document, and |
1091 |
some of which we hide because we are not sure we will keep the interface |
1092 |
stable. |
1093 |
|
1094 |
While you are welcome to rummage around and use whatever you find useful |
1095 |
(we can't stop you), keep in mind that we will change undocumented |
1096 |
functionality in incompatible ways without thinking twice, while we are |
1097 |
considerably more conservative with documented things. |
1098 |
|
1099 |
=head1 AUTHORS |
1100 |
|
1101 |
C<libecb> is designed and maintained by: |
1102 |
|
1103 |
Emanuele Giaquinta <e.giaquinta@glauco.it> |
1104 |
Marc Alexander Lehmann <schmorp@schmorp.de> |
1105 |
|
1106 |
|