ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.52
Committed: Sun Sep 23 22:32:33 2012 UTC (11 years, 8 months ago) by root
Branch: MAIN
Changes since 1.51: +1 -1 lines
Log Message:
*** empty log message ***

File Contents

# Content
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_t int16_t uint16_t
64 int32_t uint32_t int64_t uint64_t
65 intptr_t uintptr_t
66
67 The macro C<ECB_PTRSIZE> is defined to the size of a pointer on this
68 platform (currently C<4> or C<8>) and can be used in preprocessor
69 expressions.
70
71 For C<ptrdiff_t> and C<size_t> use C<stddef.h>.
72
73 =head2 LANGUAGE/COMPILER VERSIONS
74
75 All the following symbols expand to an expression that can be tested in
76 preprocessor instructions as well as treated as a boolean (use C<!!> to
77 ensure it's either C<0> or C<1> if you need that).
78
79 =over 4
80
81 =item ECB_C
82
83 True if the implementation defines the C<__STDC__> macro to a true value,
84 which is typically true for both C and C++ compilers.
85
86 =item ECB_C99
87
88 True if the implementation claims to be compliant to C99 (ISO/IEC
89 9899:1999) or any later version.
90
91 Note that later versions (ECB_C11) remove core features again (for
92 example, variable length arrays).
93
94 =item ECB_C11
95
96 True if the implementation claims to be compliant to C11 (ISO/IEC
97 9899:2011) or any later version.
98
99 =item ECB_CPP
100
101 True if the implementation defines the C<__cplusplus__> macro to a true
102 value, which is typically true for C++ compilers.
103
104 =item ECB_CPP11
105
106 True if the implementation claims to be compliant to ISO/IEC 14882:2011
107 (C++11) or any later version.
108
109 =item ECB_GCC_VERSION(major,minor)
110
111 Expands to a true value (suitable for testing in by the preprocessor)
112 if the compiler used is GNU C and the version is the given version, or
113 higher.
114
115 This macro tries to return false on compilers that claim to be GCC
116 compatible but aren't.
117
118 =item ECB_EXTERN_C
119
120 Expands to C<extern "C"> in C++, and a simple C<extern> in C.
121
122 This can be used to declare a single external C function:
123
124 ECB_EXTERN_C int printf (const char *format, ...);
125
126 =item ECB_EXTERN_C_BEG / ECB_EXTERN_C_END
127
128 These two macros can be used to wrap multiple C<extern "C"> definitions -
129 they expand to nothing in C.
130
131 They are most useful in header files:
132
133 ECB_EXTERN_C_BEG
134
135 int mycfun1 (int x);
136 int mycfun2 (int x);
137
138 ECB_EXTERN_C_END
139
140 =item ECB_STDFP
141
142 If this evaluates to a true value (suitable for testing in by the
143 preprocessor), then C<float> and C<double> use IEEE 754 single/binary32
144 and double/binary64 representations internally I<and> the endianness of
145 both types match the endianness of C<uint32_t> and C<uint64_t>.
146
147 This means you can just copy the bits of a C<float> (or C<double>) to an
148 C<uint32_t> (or C<uint64_t>) and get the raw IEEE 754 bit representation
149 without having to think about format or endianness.
150
151 This is true for basically all modern platforms, although F<ecb.h> might
152 not be able to deduce this correctly everywhere and might err on the safe
153 side.
154
155 =back
156
157 =head2 GCC ATTRIBUTES
158
159 A major part of libecb deals with GCC attributes. These are additional
160 attributes that you can assign to functions, variables and sometimes even
161 types - much like C<const> or C<volatile> in C.
162
163 While GCC allows declarations to show up in many surprising places,
164 but not in many expected places, the safest way is to put attribute
165 declarations before the whole declaration:
166
167 ecb_const int mysqrt (int a);
168 ecb_unused int i;
169
170 For variables, it is often nicer to put the attribute after the name, and
171 avoid multiple declarations using commas:
172
173 int i ecb_unused;
174
175 =over 4
176
177 =item ecb_attribute ((attrs...))
178
179 A simple wrapper that expands to C<__attribute__((attrs))> on GCC, and to
180 nothing on other compilers, so the effect is that only GCC sees these.
181
182 Example: use the C<deprecated> attribute on a function.
183
184 ecb_attribute((__deprecated__)) void
185 do_not_use_me_anymore (void);
186
187 =item ecb_unused
188
189 Marks a function or a variable as "unused", which simply suppresses a
190 warning by GCC when it detects it as unused. This is useful when you e.g.
191 declare a variable but do not always use it:
192
193 {
194 int var ecb_unused;
195
196 #ifdef SOMECONDITION
197 var = ...;
198 return var;
199 #else
200 return 0;
201 #endif
202 }
203
204 =item ecb_inline
205
206 This is not actually an attribute, but you use it like one. It expands
207 either to C<static inline> or to just C<static>, if inline isn't
208 supported. It should be used to declare functions that should be inlined,
209 for code size or speed reasons.
210
211 Example: inline this function, it surely will reduce codesize.
212
213 ecb_inline int
214 negmul (int a, int b)
215 {
216 return - (a * b);
217 }
218
219 =item ecb_noinline
220
221 Prevent a function from being inlined - it might be optimised away, but
222 not inlined into other functions. This is useful if you know your function
223 is rarely called and large enough for inlining not to be helpful.
224
225 =item ecb_noreturn
226
227 Marks a function as "not returning, ever". Some typical functions that
228 don't return are C<exit> or C<abort> (which really works hard to not
229 return), and now you can make your own:
230
231 ecb_noreturn void
232 my_abort (const char *errline)
233 {
234 puts (errline);
235 abort ();
236 }
237
238 In this case, the compiler would probably be smart enough to deduce it on
239 its own, so this is mainly useful for declarations.
240
241 =item ecb_const
242
243 Declares that the function only depends on the values of its arguments,
244 much like a mathematical function. It specifically does not read or write
245 any memory any arguments might point to, global variables, or call any
246 non-const functions. It also must not have any side effects.
247
248 Such a function can be optimised much more aggressively by the compiler -
249 for example, multiple calls with the same arguments can be optimised into
250 a single call, which wouldn't be possible if the compiler would have to
251 expect any side effects.
252
253 It is best suited for functions in the sense of mathematical functions,
254 such as a function returning the square root of its input argument.
255
256 Not suited would be a function that calculates the hash of some memory
257 area you pass in, prints some messages or looks at a global variable to
258 decide on rounding.
259
260 See C<ecb_pure> for a slightly less restrictive class of functions.
261
262 =item ecb_pure
263
264 Similar to C<ecb_const>, declares a function that has no side
265 effects. Unlike C<ecb_const>, the function is allowed to examine global
266 variables and any other memory areas (such as the ones passed to it via
267 pointers).
268
269 While these functions cannot be optimised as aggressively as C<ecb_const>
270 functions, they can still be optimised away in many occasions, and the
271 compiler has more freedom in moving calls to them around.
272
273 Typical examples for such functions would be C<strlen> or C<memcmp>. A
274 function that calculates the MD5 sum of some input and updates some MD5
275 state passed as argument would I<NOT> be pure, however, as it would modify
276 some memory area that is not the return value.
277
278 =item ecb_hot
279
280 This declares a function as "hot" with regards to the cache - the function
281 is used so often, that it is very beneficial to keep it in the cache if
282 possible.
283
284 The compiler reacts by trying to place hot functions near to each other in
285 memory.
286
287 Whether a function is hot or not often depends on the whole program,
288 and less on the function itself. C<ecb_cold> is likely more useful in
289 practise.
290
291 =item ecb_cold
292
293 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
294 the cache, or in other words, this function is not called often, or not at
295 speed-critical times, and keeping it in the cache might be a waste of said
296 cache.
297
298 In addition to placing cold functions together (or at least away from hot
299 functions), this knowledge can be used in other ways, for example, the
300 function will be optimised for size, as opposed to speed, and codepaths
301 leading to calls to those functions can automatically be marked as if
302 C<ecb_expect_false> had been used to reach them.
303
304 Good examples for such functions would be error reporting functions, or
305 functions only called in exceptional or rare cases.
306
307 =item ecb_artificial
308
309 Declares the function as "artificial", in this case meaning that this
310 function is not really meant to be a function, but more like an accessor
311 - many methods in C++ classes are mere accessor functions, and having a
312 crash reported in such a method, or single-stepping through them, is not
313 usually so helpful, especially when it's inlined to just a few instructions.
314
315 Marking them as artificial will instruct the debugger about just this,
316 leading to happier debugging and thus happier lives.
317
318 Example: in some kind of smart-pointer class, mark the pointer accessor as
319 artificial, so that the whole class acts more like a pointer and less like
320 some C++ abstraction monster.
321
322 template<typename T>
323 struct my_smart_ptr
324 {
325 T *value;
326
327 ecb_artificial
328 operator T *()
329 {
330 return value;
331 }
332 };
333
334 =back
335
336 =head2 OPTIMISATION HINTS
337
338 =over 4
339
340 =item bool ecb_is_constant(expr)
341
342 Returns true iff the expression can be deduced to be a compile-time
343 constant, and false otherwise.
344
345 For example, when you have a C<rndm16> function that returns a 16 bit
346 random number, and you have a function that maps this to a range from
347 0..n-1, then you could use this inline function in a header file:
348
349 ecb_inline uint32_t
350 rndm (uint32_t n)
351 {
352 return (n * (uint32_t)rndm16 ()) >> 16;
353 }
354
355 However, for powers of two, you could use a normal mask, but that is only
356 worth it if, at compile time, you can detect this case. This is the case
357 when the passed number is a constant and also a power of two (C<n & (n -
358 1) == 0>):
359
360 ecb_inline uint32_t
361 rndm (uint32_t n)
362 {
363 return is_constant (n) && !(n & (n - 1))
364 ? rndm16 () & (num - 1)
365 : (n * (uint32_t)rndm16 ()) >> 16;
366 }
367
368 =item bool ecb_expect (expr, value)
369
370 Evaluates C<expr> and returns it. In addition, it tells the compiler that
371 the C<expr> evaluates to C<value> a lot, which can be used for static
372 branch optimisations.
373
374 Usually, you want to use the more intuitive C<ecb_expect_true> and
375 C<ecb_expect_false> functions instead.
376
377 =item bool ecb_expect_true (cond)
378
379 =item bool ecb_expect_false (cond)
380
381 These two functions expect a expression that is true or false and return
382 C<1> or C<0>, respectively, so when used in the condition of an C<if> or
383 other conditional statement, it will not change the program:
384
385 /* these two do the same thing */
386 if (some_condition) ...;
387 if (ecb_expect_true (some_condition)) ...;
388
389 However, by using C<ecb_expect_true>, you tell the compiler that the
390 condition is likely to be true (and for C<ecb_expect_false>, that it is
391 unlikely to be true).
392
393 For example, when you check for a null pointer and expect this to be a
394 rare, exceptional, case, then use C<ecb_expect_false>:
395
396 void my_free (void *ptr)
397 {
398 if (ecb_expect_false (ptr == 0))
399 return;
400 }
401
402 Consequent use of these functions to mark away exceptional cases or to
403 tell the compiler what the hot path through a function is can increase
404 performance considerably.
405
406 You might know these functions under the name C<likely> and C<unlikely>
407 - while these are common aliases, we find that the expect name is easier
408 to understand when quickly skimming code. If you wish, you can use
409 C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
410 C<ecb_expect_false> - these are simply aliases.
411
412 A very good example is in a function that reserves more space for some
413 memory block (for example, inside an implementation of a string stream) -
414 each time something is added, you have to check for a buffer overrun, but
415 you expect that most checks will turn out to be false:
416
417 /* make sure we have "size" extra room in our buffer */
418 ecb_inline void
419 reserve (int size)
420 {
421 if (ecb_expect_false (current + size > end))
422 real_reserve_method (size); /* presumably noinline */
423 }
424
425 =item bool ecb_assume (cond)
426
427 Try to tell the compiler that some condition is true, even if it's not
428 obvious.
429
430 This can be used to teach the compiler about invariants or other
431 conditions that might improve code generation, but which are impossible to
432 deduce form the code itself.
433
434 For example, the example reservation function from the C<ecb_expect_false>
435 description could be written thus (only C<ecb_assume> was added):
436
437 ecb_inline void
438 reserve (int size)
439 {
440 if (ecb_expect_false (current + size > end))
441 real_reserve_method (size); /* presumably noinline */
442
443 ecb_assume (current + size <= end);
444 }
445
446 If you then call this function twice, like this:
447
448 reserve (10);
449 reserve (1);
450
451 Then the compiler I<might> be able to optimise out the second call
452 completely, as it knows that C<< current + 1 > end >> is false and the
453 call will never be executed.
454
455 =item bool ecb_unreachable ()
456
457 This function does nothing itself, except tell the compiler that it will
458 never be executed. Apart from suppressing a warning in some cases, this
459 function can be used to implement C<ecb_assume> or similar functions.
460
461 =item bool ecb_prefetch (addr, rw, locality)
462
463 Tells the compiler to try to prefetch memory at the given C<addr>ess
464 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
465 C<0> means that there will only be one access later, C<3> means that
466 the data will likely be accessed very often, and values in between mean
467 something... in between. The memory pointed to by the address does not
468 need to be accessible (it could be a null pointer for example), but C<rw>
469 and C<locality> must be compile-time constants.
470
471 An obvious way to use this is to prefetch some data far away, in a big
472 array you loop over. This prefetches memory some 128 array elements later,
473 in the hope that it will be ready when the CPU arrives at that location.
474
475 int sum = 0;
476
477 for (i = 0; i < N; ++i)
478 {
479 sum += arr [i]
480 ecb_prefetch (arr + i + 128, 0, 0);
481 }
482
483 It's hard to predict how far to prefetch, and most CPUs that can prefetch
484 are often good enough to predict this kind of behaviour themselves. It
485 gets more interesting with linked lists, especially when you do some fair
486 processing on each list element:
487
488 for (node *n = start; n; n = n->next)
489 {
490 ecb_prefetch (n->next, 0, 0);
491 ... do medium amount of work with *n
492 }
493
494 After processing the node, (part of) the next node might already be in
495 cache.
496
497 =back
498
499 =head2 BIT FIDDLING / BIT WIZARDRY
500
501 =over 4
502
503 =item bool ecb_big_endian ()
504
505 =item bool ecb_little_endian ()
506
507 These two functions return true if the byte order is big endian
508 (most-significant byte first) or little endian (least-significant byte
509 first) respectively.
510
511 On systems that are neither, their return values are unspecified.
512
513 =item int ecb_ctz32 (uint32_t x)
514
515 =item int ecb_ctz64 (uint64_t x)
516
517 Returns the index of the least significant bit set in C<x> (or
518 equivalently the number of bits set to 0 before the least significant bit
519 set), starting from 0. If C<x> is 0 the result is undefined.
520
521 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
522
523 For example:
524
525 ecb_ctz32 (3) = 0
526 ecb_ctz32 (6) = 1
527
528 =item bool ecb_is_pot32 (uint32_t x)
529
530 =item bool ecb_is_pot64 (uint32_t x)
531
532 Return true iff C<x> is a power of two or C<x == 0>.
533
534 For smaller types then C<uint32_t> you can safely use C<ecb_is_pot32>.
535
536 =item int ecb_ld32 (uint32_t x)
537
538 =item int ecb_ld64 (uint64_t x)
539
540 Returns the index of the most significant bit set in C<x>, or the number
541 of digits the number requires in binary (so that C<< 2**ld <= x <
542 2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
543 to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
544 example to see how many bits a certain number requires to be encoded.
545
546 This function is similar to the "count leading zero bits" function, except
547 that that one returns how many zero bits are "in front" of the number (in
548 the given data type), while C<ecb_ld> returns how many bits the number
549 itself requires.
550
551 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
552
553 =item int ecb_popcount32 (uint32_t x)
554
555 =item int ecb_popcount64 (uint64_t x)
556
557 Returns the number of bits set to 1 in C<x>.
558
559 For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
560
561 For example:
562
563 ecb_popcount32 (7) = 3
564 ecb_popcount32 (255) = 8
565
566 =item uint8_t ecb_bitrev8 (uint8_t x)
567
568 =item uint16_t ecb_bitrev16 (uint16_t x)
569
570 =item uint32_t ecb_bitrev32 (uint32_t x)
571
572 Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
573 and so on.
574
575 Example:
576
577 ecb_bitrev8 (0xa7) = 0xea
578 ecb_bitrev32 (0xffcc4411) = 0x882233ff
579
580 =item uint32_t ecb_bswap16 (uint32_t x)
581
582 =item uint32_t ecb_bswap32 (uint32_t x)
583
584 =item uint64_t ecb_bswap64 (uint64_t x)
585
586 These functions return the value of the 16-bit (32-bit, 64-bit) value
587 C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
588 C<ecb_bswap32>).
589
590 =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
591
592 =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
593
594 =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
595
596 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
597
598 =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
599
600 =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
601
602 =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
603
604 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
605
606 These two families of functions return the value of C<x> after rotating
607 all the bits by C<count> positions to the right (C<ecb_rotr>) or left
608 (C<ecb_rotl>).
609
610 Current GCC versions understand these functions and usually compile them
611 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
612 x86).
613
614 =back
615
616 =head2 FLOATING POINT FIDDLING
617
618 =over 4
619
620 =item uint32_t ecb_float_to_binary32 (float x) [-UECB_NO_LIBM]
621
622 =item uint64_t ecb_double_to_binary64 (double x) [-UECB_NO_LIBM]
623
624 These functions each take an argument in the native C<float> or C<double>
625 type and return the IEEE 754 bit representation of it.
626
627 The bit representation is just as IEEE 754 defines it, i.e. the sign bit
628 will be the most significant bit, followed by exponent and mantissa.
629
630 This function should work even when the native floating point format isn't
631 IEEE compliant, of course at a speed and code size penalty, and of course
632 also within reasonable limits (it tries to convert NaNs, infinities and
633 denormals, but will likely convert negative zero to positive zero).
634
635 On all modern platforms (where C<ECB_STDFP> is true), the compiler should
636 be able to optimise away this function completely.
637
638 These functions can be helpful when serialising floats to the network - you
639 can serialise the return value like a normal uint32_t/uint64_t.
640
641 Another use for these functions is to manipulate floating point values
642 directly.
643
644 Silly example: toggle the sign bit of a float.
645
646 /* On gcc-4.7 on amd64, */
647 /* this results in a single add instruction to toggle the bit, and 4 extra */
648 /* instructions to move the float value to an integer register and back. */
649
650 x = ecb_binary32_to_float (ecb_float_to_binary32 (x) ^ 0x80000000U)
651
652 =item float ecb_binary32_to_float (uint32_t x) [-UECB_NO_LIBM]
653
654 =item double ecb_binary32_to_double (uint64_t x) [-UECB_NO_LIBM]
655
656 The reverse operation of the previos function - takes the bit representation
657 of an IEEE binary32 or binary64 number and converts it to the native C<float>
658 or C<double> format.
659
660 This function should work even when the native floating point format isn't
661 IEEE compliant, of course at a speed and code size penalty, and of course
662 also within reasonable limits (it tries to convert normals and denormals,
663 and might be lucky for infinities, and with extraordinary luck, also for
664 negative zero).
665
666 On all modern platforms (where C<ECB_STDFP> is true), the compiler should
667 be able to optimise away this function completely.
668
669 =back
670
671 =head2 ARITHMETIC
672
673 =over 4
674
675 =item x = ecb_mod (m, n)
676
677 Returns C<m> modulo C<n>, which is the same as the positive remainder
678 of the division operation between C<m> and C<n>, using floored
679 division. Unlike the C remainder operator C<%>, this function ensures that
680 the return value is always positive and that the two numbers I<m> and
681 I<m' = m + i * n> result in the same value modulo I<n> - in other words,
682 C<ecb_mod> implements the mathematical modulo operation, which is missing
683 in the language.
684
685 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
686 negatable, that is, both C<m> and C<-m> must be representable in its
687 type (this typically excludes the minimum signed integer value, the same
688 limitation as for C</> and C<%> in C).
689
690 Current GCC versions compile this into an efficient branchless sequence on
691 almost all CPUs.
692
693 For example, when you want to rotate forward through the members of an
694 array for increasing C<m> (which might be negative), then you should use
695 C<ecb_mod>, as the C<%> operator might give either negative results, or
696 change direction for negative values:
697
698 for (m = -100; m <= 100; ++m)
699 int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
700
701 =item x = ecb_div_rd (val, div)
702
703 =item x = ecb_div_ru (val, div)
704
705 Returns C<val> divided by C<div> rounded down or up, respectively.
706 C<val> and C<div> must have integer types and C<div> must be strictly
707 positive. Note that these functions are implemented with macros in C
708 and with function templates in C++.
709
710 =back
711
712 =head2 UTILITY
713
714 =over 4
715
716 =item element_count = ecb_array_length (name)
717
718 Returns the number of elements in the array C<name>. For example:
719
720 int primes[] = { 2, 3, 5, 7, 11 };
721 int sum = 0;
722
723 for (i = 0; i < ecb_array_length (primes); i++)
724 sum += primes [i];
725
726 =back
727
728 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
729
730 These symbols need to be defined before including F<ecb.h> the first time.
731
732 =over 4
733
734 =item ECB_NO_THREADS
735
736 If F<ecb.h> is never used from multiple threads, then this symbol can
737 be defined, in which case memory fences (and similar constructs) are
738 completely removed, leading to more efficient code and fewer dependencies.
739
740 Setting this symbol to a true value implies C<ECB_NO_SMP>.
741
742 =item ECB_NO_SMP
743
744 The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
745 multiple threads, but never concurrently (e.g. if the system the program
746 runs on has only a single CPU with a single core, no hyperthreading and so
747 on), then this symbol can be defined, leading to more efficient code and
748 fewer dependencies.
749
750 =item ECB_NO_LIBM
751
752 When defined to C<1>, do not export any functions that might introduce
753 dependencies on the math library (usually called F<-lm>) - these are
754 marked with [-UECB_NO_LIBM].
755
756 =back
757
758