ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.45
Committed: Tue May 29 14:35:43 2012 UTC (12 years ago) by root
Branch: MAIN
Changes since 1.44: +2 -1 lines
Log Message:
*** empty log message ***

File Contents

# User Rev Content
1 root 1.14 =head1 LIBECB - e-C-Builtins
2 root 1.3
3 root 1.14 =head2 ABOUT LIBECB
4    
5     Libecb is currently a simple header file that doesn't require any
6     configuration to use or include in your project.
7    
8 sf-exg 1.16 It's part of the e-suite of libraries, other members of which include
9 root 1.14 libev and libeio.
10    
11     Its homepage can be found here:
12    
13     http://software.schmorp.de/pkg/libecb
14    
15     It mainly provides a number of wrappers around GCC built-ins, together
16     with replacement functions for other compilers. In addition to this,
17 sf-exg 1.16 it provides a number of other lowlevel C utilities, such as endianness
18 root 1.14 detection, byte swapping or bit rotations.
19    
20 root 1.24 Or in other words, things that should be built into any standard C system,
21     but aren't, implemented as efficient as possible with GCC, and still
22     correct with other compilers.
23 root 1.17
24 root 1.14 More might come.
25 root 1.3
26     =head2 ABOUT THE HEADER
27    
28 root 1.14 At the moment, all you have to do is copy F<ecb.h> somewhere where your
29     compiler can find it and include it:
30    
31     #include <ecb.h>
32    
33     The header should work fine for both C and C++ compilation, and gives you
34     all of F<inttypes.h> in addition to the ECB symbols.
35    
36 sf-exg 1.16 There are currently no object files to link to - future versions might
37 root 1.14 come with an (optional) object code library to link against, to reduce
38     code size or gain access to additional features.
39    
40     It also currently includes everything from F<inttypes.h>.
41    
42     =head2 ABOUT THIS MANUAL / CONVENTIONS
43    
44     This manual mainly describes each (public) function available after
45     including the F<ecb.h> header. The header might define other symbols than
46     these, but these are not part of the public API, and not supported in any
47     way.
48    
49     When the manual mentions a "function" then this could be defined either as
50     as inline function, a macro, or an external symbol.
51    
52     When functions use a concrete standard type, such as C<int> or
53     C<uint32_t>, then the corresponding function works only with that type. If
54     only a generic name is used (C<expr>, C<cond>, C<value> and so on), then
55     the corresponding function relies on C to implement the correct types, and
56     is usually implemented as a macro. Specifically, a "bool" in this manual
57     refers to any kind of boolean value, not a specific type.
58 root 1.1
59 root 1.40 =head2 TYPES / TYPE SUPPORT
60    
61     ecb.h makes sure that the following types are defined (in the expected way):
62    
63 root 1.42 int8_t uint8_t int16_t uint16_t
64     int32_t uint32_t int64_t uint64_t
65     intptr_t uintptr_t ptrdiff_t
66 root 1.40
67     The macro C<ECB_PTRSIZE> is defined to the size of a pointer on this
68 root 1.45 platform (currently C<4> or C<8>) and can be used in preprocessor
69     expressions.
70 root 1.40
71 root 1.43 =head2 LANGUAGE/COMPILER VERSIONS
72    
73 root 1.44 All the following symbols expand to an expressionb that cna be tested in
74     preprocessor instructions as well as treated as a boolean (use C<!!> to
75     ensure it's either C<0> or C<1> if you need that).
76    
77 root 1.43 =over 4
78    
79 root 1.44 =item ECB_C
80    
81     True if the implementation defines the C<__STDC__> macro to a true value,
82     which is typically true for both C and C++ compilers.
83    
84 root 1.43 =item ECB_C99
85    
86 root 1.44 True if the implementation claims to be C99 compliant.
87 root 1.43
88     =item ECB_C11
89    
90 root 1.44 True if the implementation claims to be C11 compliant.
91    
92     =item ECB_CPP
93    
94     True if the implementation defines the C<__cplusplus__> macro to a true
95     value, which is typically true for C++ compilers.
96    
97     =item ECB_CPP98
98    
99     True if the implementation claims to be compliant to ISO/IEC 14882:1998
100     (the first C++ ISO standard) or any later vwersion. Typically true for all
101     C++ compilers.
102    
103     =item ECB_CPP11
104    
105     True if the implementation claims to be compliant to ISO/IEC 14882:2011
106     (C++11) or any later vwersion.
107 root 1.43
108     =item ECB_GCC_VERSION(major,minor)
109    
110     Expands to a true value (suitable for testing in by the preprocessor)
111     if the compiler used is GNU C and the version is the givne version, or
112     higher.
113    
114     This macro tries to return false on compilers that claim to be GCC
115     compatible but aren't.
116    
117     =back
118    
119 root 1.1 =head2 GCC ATTRIBUTES
120    
121 root 1.20 A major part of libecb deals with GCC attributes. These are additional
122 sf-exg 1.26 attributes that you can assign to functions, variables and sometimes even
123 root 1.20 types - much like C<const> or C<volatile> in C.
124    
125     While GCC allows declarations to show up in many surprising places,
126 sf-exg 1.26 but not in many expected places, the safest way is to put attribute
127 root 1.20 declarations before the whole declaration:
128    
129     ecb_const int mysqrt (int a);
130     ecb_unused int i;
131    
132     For variables, it is often nicer to put the attribute after the name, and
133     avoid multiple declarations using commas:
134    
135     int i ecb_unused;
136 root 1.3
137 root 1.1 =over 4
138    
139 root 1.2 =item ecb_attribute ((attrs...))
140 root 1.1
141 root 1.15 A simple wrapper that expands to C<__attribute__((attrs))> on GCC, and to
142     nothing on other compilers, so the effect is that only GCC sees these.
143    
144     Example: use the C<deprecated> attribute on a function.
145    
146     ecb_attribute((__deprecated__)) void
147     do_not_use_me_anymore (void);
148 root 1.2
149 root 1.3 =item ecb_unused
150    
151     Marks a function or a variable as "unused", which simply suppresses a
152     warning by GCC when it detects it as unused. This is useful when you e.g.
153     declare a variable but do not always use it:
154    
155 root 1.15 {
156     int var ecb_unused;
157 root 1.3
158 root 1.15 #ifdef SOMECONDITION
159     var = ...;
160     return var;
161     #else
162     return 0;
163     #endif
164     }
165 root 1.3
166 root 1.31 =item ecb_inline
167 root 1.29
168     This is not actually an attribute, but you use it like one. It expands
169     either to C<static inline> or to just C<static>, if inline isn't
170     supported. It should be used to declare functions that should be inlined,
171     for code size or speed reasons.
172    
173     Example: inline this function, it surely will reduce codesize.
174    
175 root 1.31 ecb_inline int
176 root 1.29 negmul (int a, int b)
177     {
178     return - (a * b);
179     }
180    
181 root 1.2 =item ecb_noinline
182    
183 root 1.9 Prevent a function from being inlined - it might be optimised away, but
184 root 1.3 not inlined into other functions. This is useful if you know your function
185     is rarely called and large enough for inlining not to be helpful.
186    
187 root 1.2 =item ecb_noreturn
188    
189 root 1.17 Marks a function as "not returning, ever". Some typical functions that
190     don't return are C<exit> or C<abort> (which really works hard to not
191     return), and now you can make your own:
192    
193     ecb_noreturn void
194     my_abort (const char *errline)
195     {
196     puts (errline);
197     abort ();
198     }
199    
200 sf-exg 1.19 In this case, the compiler would probably be smart enough to deduce it on
201     its own, so this is mainly useful for declarations.
202 root 1.17
203 root 1.2 =item ecb_const
204    
205 sf-exg 1.19 Declares that the function only depends on the values of its arguments,
206 root 1.17 much like a mathematical function. It specifically does not read or write
207     any memory any arguments might point to, global variables, or call any
208     non-const functions. It also must not have any side effects.
209    
210     Such a function can be optimised much more aggressively by the compiler -
211     for example, multiple calls with the same arguments can be optimised into
212     a single call, which wouldn't be possible if the compiler would have to
213     expect any side effects.
214    
215     It is best suited for functions in the sense of mathematical functions,
216 sf-exg 1.19 such as a function returning the square root of its input argument.
217 root 1.17
218     Not suited would be a function that calculates the hash of some memory
219     area you pass in, prints some messages or looks at a global variable to
220     decide on rounding.
221    
222     See C<ecb_pure> for a slightly less restrictive class of functions.
223    
224 root 1.2 =item ecb_pure
225    
226 root 1.17 Similar to C<ecb_const>, declares a function that has no side
227     effects. Unlike C<ecb_const>, the function is allowed to examine global
228     variables and any other memory areas (such as the ones passed to it via
229     pointers).
230    
231     While these functions cannot be optimised as aggressively as C<ecb_const>
232     functions, they can still be optimised away in many occasions, and the
233     compiler has more freedom in moving calls to them around.
234    
235     Typical examples for such functions would be C<strlen> or C<memcmp>. A
236     function that calculates the MD5 sum of some input and updates some MD5
237     state passed as argument would I<NOT> be pure, however, as it would modify
238     some memory area that is not the return value.
239    
240 root 1.2 =item ecb_hot
241    
242 root 1.17 This declares a function as "hot" with regards to the cache - the function
243     is used so often, that it is very beneficial to keep it in the cache if
244     possible.
245    
246     The compiler reacts by trying to place hot functions near to each other in
247     memory.
248    
249 sf-exg 1.19 Whether a function is hot or not often depends on the whole program,
250 root 1.17 and less on the function itself. C<ecb_cold> is likely more useful in
251     practise.
252    
253 root 1.2 =item ecb_cold
254    
255 root 1.17 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
256     the cache, or in other words, this function is not called often, or not at
257     speed-critical times, and keeping it in the cache might be a waste of said
258     cache.
259    
260     In addition to placing cold functions together (or at least away from hot
261     functions), this knowledge can be used in other ways, for example, the
262     function will be optimised for size, as opposed to speed, and codepaths
263     leading to calls to those functions can automatically be marked as if
264 root 1.27 C<ecb_expect_false> had been used to reach them.
265 root 1.17
266     Good examples for such functions would be error reporting functions, or
267     functions only called in exceptional or rare cases.
268    
269 root 1.2 =item ecb_artificial
270    
271 root 1.17 Declares the function as "artificial", in this case meaning that this
272     function is not really mean to be a function, but more like an accessor
273     - many methods in C++ classes are mere accessor functions, and having a
274     crash reported in such a method, or single-stepping through them, is not
275     usually so helpful, especially when it's inlined to just a few instructions.
276    
277     Marking them as artificial will instruct the debugger about just this,
278     leading to happier debugging and thus happier lives.
279    
280     Example: in some kind of smart-pointer class, mark the pointer accessor as
281     artificial, so that the whole class acts more like a pointer and less like
282     some C++ abstraction monster.
283    
284     template<typename T>
285     struct my_smart_ptr
286     {
287     T *value;
288    
289     ecb_artificial
290     operator T *()
291     {
292     return value;
293     }
294     };
295    
296 root 1.2 =back
297 root 1.1
298     =head2 OPTIMISATION HINTS
299    
300     =over 4
301    
302 root 1.14 =item bool ecb_is_constant(expr)
303 root 1.1
304 root 1.3 Returns true iff the expression can be deduced to be a compile-time
305     constant, and false otherwise.
306    
307     For example, when you have a C<rndm16> function that returns a 16 bit
308     random number, and you have a function that maps this to a range from
309 root 1.5 0..n-1, then you could use this inline function in a header file:
310 root 1.3
311     ecb_inline uint32_t
312     rndm (uint32_t n)
313     {
314 root 1.6 return (n * (uint32_t)rndm16 ()) >> 16;
315 root 1.3 }
316    
317     However, for powers of two, you could use a normal mask, but that is only
318     worth it if, at compile time, you can detect this case. This is the case
319     when the passed number is a constant and also a power of two (C<n & (n -
320     1) == 0>):
321    
322     ecb_inline uint32_t
323     rndm (uint32_t n)
324     {
325     return is_constant (n) && !(n & (n - 1))
326     ? rndm16 () & (num - 1)
327 root 1.6 : (n * (uint32_t)rndm16 ()) >> 16;
328 root 1.3 }
329    
330 root 1.14 =item bool ecb_expect (expr, value)
331 root 1.1
332 root 1.7 Evaluates C<expr> and returns it. In addition, it tells the compiler that
333     the C<expr> evaluates to C<value> a lot, which can be used for static
334     branch optimisations.
335 root 1.1
336 root 1.27 Usually, you want to use the more intuitive C<ecb_expect_true> and
337     C<ecb_expect_false> functions instead.
338 root 1.1
339 root 1.27 =item bool ecb_expect_true (cond)
340 root 1.1
341 root 1.27 =item bool ecb_expect_false (cond)
342 root 1.1
343 root 1.7 These two functions expect a expression that is true or false and return
344     C<1> or C<0>, respectively, so when used in the condition of an C<if> or
345     other conditional statement, it will not change the program:
346    
347     /* these two do the same thing */
348     if (some_condition) ...;
349 root 1.27 if (ecb_expect_true (some_condition)) ...;
350 root 1.7
351 root 1.27 However, by using C<ecb_expect_true>, you tell the compiler that the
352     condition is likely to be true (and for C<ecb_expect_false>, that it is
353     unlikely to be true).
354 root 1.7
355 root 1.9 For example, when you check for a null pointer and expect this to be a
356 root 1.27 rare, exceptional, case, then use C<ecb_expect_false>:
357 root 1.7
358     void my_free (void *ptr)
359     {
360 root 1.27 if (ecb_expect_false (ptr == 0))
361 root 1.7 return;
362     }
363    
364     Consequent use of these functions to mark away exceptional cases or to
365     tell the compiler what the hot path through a function is can increase
366     performance considerably.
367    
368 root 1.27 You might know these functions under the name C<likely> and C<unlikely>
369     - while these are common aliases, we find that the expect name is easier
370     to understand when quickly skimming code. If you wish, you can use
371     C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
372     C<ecb_expect_false> - these are simply aliases.
373    
374 root 1.7 A very good example is in a function that reserves more space for some
375     memory block (for example, inside an implementation of a string stream) -
376 root 1.9 each time something is added, you have to check for a buffer overrun, but
377 root 1.7 you expect that most checks will turn out to be false:
378    
379     /* make sure we have "size" extra room in our buffer */
380     ecb_inline void
381     reserve (int size)
382     {
383 root 1.27 if (ecb_expect_false (current + size > end))
384 root 1.7 real_reserve_method (size); /* presumably noinline */
385     }
386    
387 root 1.14 =item bool ecb_assume (cond)
388 root 1.7
389     Try to tell the compiler that some condition is true, even if it's not
390     obvious.
391    
392     This can be used to teach the compiler about invariants or other
393     conditions that might improve code generation, but which are impossible to
394     deduce form the code itself.
395    
396 root 1.27 For example, the example reservation function from the C<ecb_expect_false>
397 root 1.7 description could be written thus (only C<ecb_assume> was added):
398    
399     ecb_inline void
400     reserve (int size)
401     {
402 root 1.27 if (ecb_expect_false (current + size > end))
403 root 1.7 real_reserve_method (size); /* presumably noinline */
404    
405     ecb_assume (current + size <= end);
406     }
407    
408     If you then call this function twice, like this:
409    
410     reserve (10);
411     reserve (1);
412    
413     Then the compiler I<might> be able to optimise out the second call
414     completely, as it knows that C<< current + 1 > end >> is false and the
415     call will never be executed.
416    
417     =item bool ecb_unreachable ()
418    
419     This function does nothing itself, except tell the compiler that it will
420 root 1.9 never be executed. Apart from suppressing a warning in some cases, this
421 root 1.7 function can be used to implement C<ecb_assume> or similar functions.
422    
423 root 1.14 =item bool ecb_prefetch (addr, rw, locality)
424 root 1.7
425     Tells the compiler to try to prefetch memory at the given C<addr>ess
426 root 1.10 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
427 root 1.7 C<0> means that there will only be one access later, C<3> means that
428     the data will likely be accessed very often, and values in between mean
429     something... in between. The memory pointed to by the address does not
430     need to be accessible (it could be a null pointer for example), but C<rw>
431     and C<locality> must be compile-time constants.
432    
433     An obvious way to use this is to prefetch some data far away, in a big
434 root 1.9 array you loop over. This prefetches memory some 128 array elements later,
435 root 1.7 in the hope that it will be ready when the CPU arrives at that location.
436    
437     int sum = 0;
438    
439     for (i = 0; i < N; ++i)
440     {
441     sum += arr [i]
442     ecb_prefetch (arr + i + 128, 0, 0);
443     }
444    
445     It's hard to predict how far to prefetch, and most CPUs that can prefetch
446     are often good enough to predict this kind of behaviour themselves. It
447     gets more interesting with linked lists, especially when you do some fair
448     processing on each list element:
449    
450     for (node *n = start; n; n = n->next)
451     {
452     ecb_prefetch (n->next, 0, 0);
453     ... do medium amount of work with *n
454     }
455    
456     After processing the node, (part of) the next node might already be in
457     cache.
458 root 1.1
459 root 1.2 =back
460 root 1.1
461 root 1.36 =head2 BIT FIDDLING / BIT WIZARDRY
462 root 1.1
463 root 1.4 =over 4
464    
465 root 1.3 =item bool ecb_big_endian ()
466    
467     =item bool ecb_little_endian ()
468    
469 sf-exg 1.11 These two functions return true if the byte order is big endian
470     (most-significant byte first) or little endian (least-significant byte
471     first) respectively.
472    
473 root 1.24 On systems that are neither, their return values are unspecified.
474    
475 root 1.3 =item int ecb_ctz32 (uint32_t x)
476    
477 root 1.35 =item int ecb_ctz64 (uint64_t x)
478    
479 sf-exg 1.11 Returns the index of the least significant bit set in C<x> (or
480 root 1.24 equivalently the number of bits set to 0 before the least significant bit
481 root 1.35 set), starting from 0. If C<x> is 0 the result is undefined.
482    
483 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
484    
485 root 1.35 For example:
486 sf-exg 1.11
487 root 1.15 ecb_ctz32 (3) = 0
488     ecb_ctz32 (6) = 1
489 sf-exg 1.11
490 root 1.41 =item bool ecb_is_pot32 (uint32_t x)
491    
492     =item bool ecb_is_pot64 (uint32_t x)
493    
494     Return true iff C<x> is a power of two or C<x == 0>.
495    
496     For smaller types then C<uint32_t> you can safely use C<ecb_is_pot32>.
497    
498 root 1.35 =item int ecb_ld32 (uint32_t x)
499    
500     =item int ecb_ld64 (uint64_t x)
501    
502     Returns the index of the most significant bit set in C<x>, or the number
503     of digits the number requires in binary (so that C<< 2**ld <= x <
504     2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
505     to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
506     example to see how many bits a certain number requires to be encoded.
507    
508     This function is similar to the "count leading zero bits" function, except
509     that that one returns how many zero bits are "in front" of the number (in
510     the given data type), while C<ecb_ld> returns how many bits the number
511     itself requires.
512    
513 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
514    
515 root 1.3 =item int ecb_popcount32 (uint32_t x)
516    
517 root 1.35 =item int ecb_popcount64 (uint64_t x)
518    
519 root 1.36 Returns the number of bits set to 1 in C<x>.
520    
521     For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
522    
523     For example:
524 sf-exg 1.11
525 root 1.15 ecb_popcount32 (7) = 3
526     ecb_popcount32 (255) = 8
527 sf-exg 1.11
528 root 1.39 =item uint8_t ecb_bitrev8 (uint8_t x)
529    
530     =item uint16_t ecb_bitrev16 (uint16_t x)
531    
532     =item uint32_t ecb_bitrev32 (uint32_t x)
533    
534     Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
535     and so on.
536    
537     Example:
538    
539     ecb_bitrev8 (0xa7) = 0xea
540     ecb_bitrev32 (0xffcc4411) = 0x882233ff
541    
542 root 1.8 =item uint32_t ecb_bswap16 (uint32_t x)
543    
544 root 1.3 =item uint32_t ecb_bswap32 (uint32_t x)
545    
546 root 1.34 =item uint64_t ecb_bswap64 (uint64_t x)
547 sf-exg 1.13
548 root 1.34 These functions return the value of the 16-bit (32-bit, 64-bit) value
549     C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
550     C<ecb_bswap32>).
551    
552     =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
553    
554     =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
555 root 1.3
556     =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
557    
558 root 1.34 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
559    
560     =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
561    
562     =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
563    
564     =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
565    
566 root 1.33 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
567    
568 root 1.34 These two families of functions return the value of C<x> after rotating
569     all the bits by C<count> positions to the right (C<ecb_rotr>) or left
570     (C<ecb_rotl>).
571 sf-exg 1.11
572 root 1.20 Current GCC versions understand these functions and usually compile them
573 root 1.34 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
574     x86).
575 root 1.20
576 root 1.3 =back
577 root 1.1
578     =head2 ARITHMETIC
579    
580 root 1.3 =over 4
581    
582 root 1.14 =item x = ecb_mod (m, n)
583 root 1.3
584 root 1.25 Returns C<m> modulo C<n>, which is the same as the positive remainder
585     of the division operation between C<m> and C<n>, using floored
586     division. Unlike the C remainder operator C<%>, this function ensures that
587     the return value is always positive and that the two numbers I<m> and
588     I<m' = m + i * n> result in the same value modulo I<n> - in other words,
589     C<ecb_mod> implements the mathematical modulo operation, which is missing
590     in the language.
591 root 1.14
592 sf-exg 1.23 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
593 root 1.14 negatable, that is, both C<m> and C<-m> must be representable in its
594 root 1.30 type (this typically excludes the minimum signed integer value, the same
595 root 1.25 limitation as for C</> and C<%> in C).
596 sf-exg 1.11
597 root 1.24 Current GCC versions compile this into an efficient branchless sequence on
598 root 1.28 almost all CPUs.
599 root 1.24
600     For example, when you want to rotate forward through the members of an
601     array for increasing C<m> (which might be negative), then you should use
602     C<ecb_mod>, as the C<%> operator might give either negative results, or
603     change direction for negative values:
604    
605     for (m = -100; m <= 100; ++m)
606     int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
607    
608 sf-exg 1.37 =item x = ecb_div_rd (val, div)
609    
610     =item x = ecb_div_ru (val, div)
611    
612     Returns C<val> divided by C<div> rounded down or up, respectively.
613     C<val> and C<div> must have integer types and C<div> must be strictly
614 sf-exg 1.38 positive. Note that these functions are implemented with macros in C
615     and with function templates in C++.
616 sf-exg 1.37
617 root 1.3 =back
618 root 1.1
619     =head2 UTILITY
620    
621 root 1.3 =over 4
622    
623 sf-exg 1.23 =item element_count = ecb_array_length (name)
624 root 1.3
625 sf-exg 1.13 Returns the number of elements in the array C<name>. For example:
626    
627     int primes[] = { 2, 3, 5, 7, 11 };
628     int sum = 0;
629    
630     for (i = 0; i < ecb_array_length (primes); i++)
631     sum += primes [i];
632    
633 root 1.3 =back
634 root 1.1
635 root 1.43 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
636    
637     These symbols need to be defined before including F<ecb.h> the first time.
638    
639     =over 4
640    
641     =item ECB_NO_THRADS
642    
643     If F<ecb.h> is never used from multiple threads, then this symbol can
644     be defined, in which case memory fences (and similar constructs) are
645     completely removed, leading to more efficient code and fewer dependencies.
646    
647     Setting this symbol to a true value implies C<ECB_NO_SMP>.
648    
649     =item ECB_NO_SMP
650    
651     The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
652     multiple threads, but never concurrently (e.g. if the system the program
653     runs on has only a single CPU with a single core, no hyperthreading and so
654     on), then this symbol can be defined, leading to more efficient code and
655     fewer dependencies.
656    
657     =back
658    
659 root 1.1