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 GCC ATTRIBUTES |
60 |
|
61 |
A major part of libecb deals with GCC attributes. These are additional |
62 |
attributes that you cna assign to functions, variables and sometimes even |
63 |
types - much like C<const> or C<volatile> in C. |
64 |
|
65 |
While GCC allows declarations to show up in many surprising places, |
66 |
but not in many expeted places, the safest way is to put attribute |
67 |
declarations before the whole declaration: |
68 |
|
69 |
ecb_const int mysqrt (int a); |
70 |
ecb_unused int i; |
71 |
|
72 |
For variables, it is often nicer to put the attribute after the name, and |
73 |
avoid multiple declarations using commas: |
74 |
|
75 |
int i ecb_unused; |
76 |
|
77 |
=over 4 |
78 |
|
79 |
=item ecb_attribute ((attrs...)) |
80 |
|
81 |
A simple wrapper that expands to C<__attribute__((attrs))> on GCC, and to |
82 |
nothing on other compilers, so the effect is that only GCC sees these. |
83 |
|
84 |
Example: use the C<deprecated> attribute on a function. |
85 |
|
86 |
ecb_attribute((__deprecated__)) void |
87 |
do_not_use_me_anymore (void); |
88 |
|
89 |
=item ecb_unused |
90 |
|
91 |
Marks a function or a variable as "unused", which simply suppresses a |
92 |
warning by GCC when it detects it as unused. This is useful when you e.g. |
93 |
declare a variable but do not always use it: |
94 |
|
95 |
{ |
96 |
int var ecb_unused; |
97 |
|
98 |
#ifdef SOMECONDITION |
99 |
var = ...; |
100 |
return var; |
101 |
#else |
102 |
return 0; |
103 |
#endif |
104 |
} |
105 |
|
106 |
=item ecb_noinline |
107 |
|
108 |
Prevent a function from being inlined - it might be optimised away, but |
109 |
not inlined into other functions. This is useful if you know your function |
110 |
is rarely called and large enough for inlining not to be helpful. |
111 |
|
112 |
=item ecb_noreturn |
113 |
|
114 |
Marks a function as "not returning, ever". Some typical functions that |
115 |
don't return are C<exit> or C<abort> (which really works hard to not |
116 |
return), and now you can make your own: |
117 |
|
118 |
ecb_noreturn void |
119 |
my_abort (const char *errline) |
120 |
{ |
121 |
puts (errline); |
122 |
abort (); |
123 |
} |
124 |
|
125 |
In this case, the compiler would probably be smart enough to deduce it on |
126 |
its own, so this is mainly useful for declarations. |
127 |
|
128 |
=item ecb_const |
129 |
|
130 |
Declares that the function only depends on the values of its arguments, |
131 |
much like a mathematical function. It specifically does not read or write |
132 |
any memory any arguments might point to, global variables, or call any |
133 |
non-const functions. It also must not have any side effects. |
134 |
|
135 |
Such a function can be optimised much more aggressively by the compiler - |
136 |
for example, multiple calls with the same arguments can be optimised into |
137 |
a single call, which wouldn't be possible if the compiler would have to |
138 |
expect any side effects. |
139 |
|
140 |
It is best suited for functions in the sense of mathematical functions, |
141 |
such as a function returning the square root of its input argument. |
142 |
|
143 |
Not suited would be a function that calculates the hash of some memory |
144 |
area you pass in, prints some messages or looks at a global variable to |
145 |
decide on rounding. |
146 |
|
147 |
See C<ecb_pure> for a slightly less restrictive class of functions. |
148 |
|
149 |
=item ecb_pure |
150 |
|
151 |
Similar to C<ecb_const>, declares a function that has no side |
152 |
effects. Unlike C<ecb_const>, the function is allowed to examine global |
153 |
variables and any other memory areas (such as the ones passed to it via |
154 |
pointers). |
155 |
|
156 |
While these functions cannot be optimised as aggressively as C<ecb_const> |
157 |
functions, they can still be optimised away in many occasions, and the |
158 |
compiler has more freedom in moving calls to them around. |
159 |
|
160 |
Typical examples for such functions would be C<strlen> or C<memcmp>. A |
161 |
function that calculates the MD5 sum of some input and updates some MD5 |
162 |
state passed as argument would I<NOT> be pure, however, as it would modify |
163 |
some memory area that is not the return value. |
164 |
|
165 |
=item ecb_hot |
166 |
|
167 |
This declares a function as "hot" with regards to the cache - the function |
168 |
is used so often, that it is very beneficial to keep it in the cache if |
169 |
possible. |
170 |
|
171 |
The compiler reacts by trying to place hot functions near to each other in |
172 |
memory. |
173 |
|
174 |
Whether a function is hot or not often depends on the whole program, |
175 |
and less on the function itself. C<ecb_cold> is likely more useful in |
176 |
practise. |
177 |
|
178 |
=item ecb_cold |
179 |
|
180 |
The opposite of C<ecb_hot> - declares a function as "cold" with regards to |
181 |
the cache, or in other words, this function is not called often, or not at |
182 |
speed-critical times, and keeping it in the cache might be a waste of said |
183 |
cache. |
184 |
|
185 |
In addition to placing cold functions together (or at least away from hot |
186 |
functions), this knowledge can be used in other ways, for example, the |
187 |
function will be optimised for size, as opposed to speed, and codepaths |
188 |
leading to calls to those functions can automatically be marked as if |
189 |
C<ecb_unlikely> had been used to reach them. |
190 |
|
191 |
Good examples for such functions would be error reporting functions, or |
192 |
functions only called in exceptional or rare cases. |
193 |
|
194 |
=item ecb_artificial |
195 |
|
196 |
Declares the function as "artificial", in this case meaning that this |
197 |
function is not really mean to be a function, but more like an accessor |
198 |
- many methods in C++ classes are mere accessor functions, and having a |
199 |
crash reported in such a method, or single-stepping through them, is not |
200 |
usually so helpful, especially when it's inlined to just a few instructions. |
201 |
|
202 |
Marking them as artificial will instruct the debugger about just this, |
203 |
leading to happier debugging and thus happier lives. |
204 |
|
205 |
Example: in some kind of smart-pointer class, mark the pointer accessor as |
206 |
artificial, so that the whole class acts more like a pointer and less like |
207 |
some C++ abstraction monster. |
208 |
|
209 |
template<typename T> |
210 |
struct my_smart_ptr |
211 |
{ |
212 |
T *value; |
213 |
|
214 |
ecb_artificial |
215 |
operator T *() |
216 |
{ |
217 |
return value; |
218 |
} |
219 |
}; |
220 |
|
221 |
=back |
222 |
|
223 |
=head2 OPTIMISATION HINTS |
224 |
|
225 |
=over 4 |
226 |
|
227 |
=item bool ecb_is_constant(expr) |
228 |
|
229 |
Returns true iff the expression can be deduced to be a compile-time |
230 |
constant, and false otherwise. |
231 |
|
232 |
For example, when you have a C<rndm16> function that returns a 16 bit |
233 |
random number, and you have a function that maps this to a range from |
234 |
0..n-1, then you could use this inline function in a header file: |
235 |
|
236 |
ecb_inline uint32_t |
237 |
rndm (uint32_t n) |
238 |
{ |
239 |
return (n * (uint32_t)rndm16 ()) >> 16; |
240 |
} |
241 |
|
242 |
However, for powers of two, you could use a normal mask, but that is only |
243 |
worth it if, at compile time, you can detect this case. This is the case |
244 |
when the passed number is a constant and also a power of two (C<n & (n - |
245 |
1) == 0>): |
246 |
|
247 |
ecb_inline uint32_t |
248 |
rndm (uint32_t n) |
249 |
{ |
250 |
return is_constant (n) && !(n & (n - 1)) |
251 |
? rndm16 () & (num - 1) |
252 |
: (n * (uint32_t)rndm16 ()) >> 16; |
253 |
} |
254 |
|
255 |
=item bool ecb_expect (expr, value) |
256 |
|
257 |
Evaluates C<expr> and returns it. In addition, it tells the compiler that |
258 |
the C<expr> evaluates to C<value> a lot, which can be used for static |
259 |
branch optimisations. |
260 |
|
261 |
Usually, you want to use the more intuitive C<ecb_likely> and |
262 |
C<ecb_unlikely> functions instead. |
263 |
|
264 |
=item bool ecb_likely (cond) |
265 |
|
266 |
=item bool ecb_unlikely (cond) |
267 |
|
268 |
These two functions expect a expression that is true or false and return |
269 |
C<1> or C<0>, respectively, so when used in the condition of an C<if> or |
270 |
other conditional statement, it will not change the program: |
271 |
|
272 |
/* these two do the same thing */ |
273 |
if (some_condition) ...; |
274 |
if (ecb_likely (some_condition)) ...; |
275 |
|
276 |
However, by using C<ecb_likely>, you tell the compiler that the condition |
277 |
is likely to be true (and for C<ecb_unlikely>, that it is unlikely to be |
278 |
true). |
279 |
|
280 |
For example, when you check for a null pointer and expect this to be a |
281 |
rare, exceptional, case, then use C<ecb_unlikely>: |
282 |
|
283 |
void my_free (void *ptr) |
284 |
{ |
285 |
if (ecb_unlikely (ptr == 0)) |
286 |
return; |
287 |
} |
288 |
|
289 |
Consequent use of these functions to mark away exceptional cases or to |
290 |
tell the compiler what the hot path through a function is can increase |
291 |
performance considerably. |
292 |
|
293 |
A very good example is in a function that reserves more space for some |
294 |
memory block (for example, inside an implementation of a string stream) - |
295 |
each time something is added, you have to check for a buffer overrun, but |
296 |
you expect that most checks will turn out to be false: |
297 |
|
298 |
/* make sure we have "size" extra room in our buffer */ |
299 |
ecb_inline void |
300 |
reserve (int size) |
301 |
{ |
302 |
if (ecb_unlikely (current + size > end)) |
303 |
real_reserve_method (size); /* presumably noinline */ |
304 |
} |
305 |
|
306 |
=item bool ecb_assume (cond) |
307 |
|
308 |
Try to tell the compiler that some condition is true, even if it's not |
309 |
obvious. |
310 |
|
311 |
This can be used to teach the compiler about invariants or other |
312 |
conditions that might improve code generation, but which are impossible to |
313 |
deduce form the code itself. |
314 |
|
315 |
For example, the example reservation function from the C<ecb_unlikely> |
316 |
description could be written thus (only C<ecb_assume> was added): |
317 |
|
318 |
ecb_inline void |
319 |
reserve (int size) |
320 |
{ |
321 |
if (ecb_unlikely (current + size > end)) |
322 |
real_reserve_method (size); /* presumably noinline */ |
323 |
|
324 |
ecb_assume (current + size <= end); |
325 |
} |
326 |
|
327 |
If you then call this function twice, like this: |
328 |
|
329 |
reserve (10); |
330 |
reserve (1); |
331 |
|
332 |
Then the compiler I<might> be able to optimise out the second call |
333 |
completely, as it knows that C<< current + 1 > end >> is false and the |
334 |
call will never be executed. |
335 |
|
336 |
=item bool ecb_unreachable () |
337 |
|
338 |
This function does nothing itself, except tell the compiler that it will |
339 |
never be executed. Apart from suppressing a warning in some cases, this |
340 |
function can be used to implement C<ecb_assume> or similar functions. |
341 |
|
342 |
=item bool ecb_prefetch (addr, rw, locality) |
343 |
|
344 |
Tells the compiler to try to prefetch memory at the given C<addr>ess |
345 |
for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of |
346 |
C<0> means that there will only be one access later, C<3> means that |
347 |
the data will likely be accessed very often, and values in between mean |
348 |
something... in between. The memory pointed to by the address does not |
349 |
need to be accessible (it could be a null pointer for example), but C<rw> |
350 |
and C<locality> must be compile-time constants. |
351 |
|
352 |
An obvious way to use this is to prefetch some data far away, in a big |
353 |
array you loop over. This prefetches memory some 128 array elements later, |
354 |
in the hope that it will be ready when the CPU arrives at that location. |
355 |
|
356 |
int sum = 0; |
357 |
|
358 |
for (i = 0; i < N; ++i) |
359 |
{ |
360 |
sum += arr [i] |
361 |
ecb_prefetch (arr + i + 128, 0, 0); |
362 |
} |
363 |
|
364 |
It's hard to predict how far to prefetch, and most CPUs that can prefetch |
365 |
are often good enough to predict this kind of behaviour themselves. It |
366 |
gets more interesting with linked lists, especially when you do some fair |
367 |
processing on each list element: |
368 |
|
369 |
for (node *n = start; n; n = n->next) |
370 |
{ |
371 |
ecb_prefetch (n->next, 0, 0); |
372 |
... do medium amount of work with *n |
373 |
} |
374 |
|
375 |
After processing the node, (part of) the next node might already be in |
376 |
cache. |
377 |
|
378 |
=back |
379 |
|
380 |
=head2 BIT FIDDLING / BITSTUFFS |
381 |
|
382 |
=over 4 |
383 |
|
384 |
=item bool ecb_big_endian () |
385 |
|
386 |
=item bool ecb_little_endian () |
387 |
|
388 |
These two functions return true if the byte order is big endian |
389 |
(most-significant byte first) or little endian (least-significant byte |
390 |
first) respectively. |
391 |
|
392 |
On systems that are neither, their return values are unspecified. |
393 |
|
394 |
=item int ecb_ctz32 (uint32_t x) |
395 |
|
396 |
Returns the index of the least significant bit set in C<x> (or |
397 |
equivalently the number of bits set to 0 before the least significant bit |
398 |
set), starting from 0. If C<x> is 0 the result is undefined. A common use |
399 |
case is to compute the integer binary logarithm, i.e., C<floor (log2 |
400 |
(n))>. For example: |
401 |
|
402 |
ecb_ctz32 (3) = 0 |
403 |
ecb_ctz32 (6) = 1 |
404 |
|
405 |
=item int ecb_popcount32 (uint32_t x) |
406 |
|
407 |
Returns the number of bits set to 1 in C<x>. For example: |
408 |
|
409 |
ecb_popcount32 (7) = 3 |
410 |
ecb_popcount32 (255) = 8 |
411 |
|
412 |
=item uint32_t ecb_bswap16 (uint32_t x) |
413 |
|
414 |
=item uint32_t ecb_bswap32 (uint32_t x) |
415 |
|
416 |
These two functions return the value of the 16-bit (32-bit) value C<x> |
417 |
after reversing the order of bytes (0x11223344 becomes 0x44332211). |
418 |
|
419 |
=item uint32_t ecb_rotr32 (uint32_t x, unsigned int count) |
420 |
|
421 |
=item uint32_t ecb_rotl32 (uint32_t x, unsigned int count) |
422 |
|
423 |
These two functions return the value of C<x> after rotating all the bits |
424 |
by C<count> positions to the right or left respectively. |
425 |
|
426 |
Current GCC versions understand these functions and usually compile them |
427 |
to "optimal" code (e.g. a single C<roll> on x86). |
428 |
|
429 |
=back |
430 |
|
431 |
=head2 ARITHMETIC |
432 |
|
433 |
=over 4 |
434 |
|
435 |
=item x = ecb_mod (m, n) |
436 |
|
437 |
Returns C<m> modulo C<n>, which is the same as the positive remainder |
438 |
of the division operation between C<m> and C<n>, using floored |
439 |
division. Unlike the C remainder operator C<%>, this function ensures that |
440 |
the return value is always positive and that the two numbers I<m> and |
441 |
I<m' = m + i * n> result in the same value modulo I<n> - in other words, |
442 |
C<ecb_mod> implements the mathematical modulo operation, which is missing |
443 |
in the language. |
444 |
|
445 |
C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be |
446 |
negatable, that is, both C<m> and C<-m> must be representable in its |
447 |
type (this typically includes the minimum signed integer value, the same |
448 |
limitation as for C</> and C<%> in C). |
449 |
|
450 |
Current GCC versions compile this into an efficient branchless sequence on |
451 |
many systems. |
452 |
|
453 |
For example, when you want to rotate forward through the members of an |
454 |
array for increasing C<m> (which might be negative), then you should use |
455 |
C<ecb_mod>, as the C<%> operator might give either negative results, or |
456 |
change direction for negative values: |
457 |
|
458 |
for (m = -100; m <= 100; ++m) |
459 |
int elem = myarray [ecb_mod (m, ecb_array_length (myarray))]; |
460 |
|
461 |
=back |
462 |
|
463 |
=head2 UTILITY |
464 |
|
465 |
=over 4 |
466 |
|
467 |
=item element_count = ecb_array_length (name) |
468 |
|
469 |
Returns the number of elements in the array C<name>. For example: |
470 |
|
471 |
int primes[] = { 2, 3, 5, 7, 11 }; |
472 |
int sum = 0; |
473 |
|
474 |
for (i = 0; i < ecb_array_length (primes); i++) |
475 |
sum += primes [i]; |
476 |
|
477 |
=back |
478 |
|
479 |
|