ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libecb/ecb.pod
Revision: 1.44
Committed: Tue May 29 14:30:44 2012 UTC (12 years ago) by root
Branch: MAIN
Changes since 1.43: +27 -4 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     platform (currently C<4> or C<8>).
69    
70 root 1.43 =head2 LANGUAGE/COMPILER VERSIONS
71    
72 root 1.44 All the following symbols expand to an expressionb that cna be tested in
73     preprocessor instructions as well as treated as a boolean (use C<!!> to
74     ensure it's either C<0> or C<1> if you need that).
75    
76 root 1.43 =over 4
77    
78 root 1.44 =item ECB_C
79    
80     True if the implementation defines the C<__STDC__> macro to a true value,
81     which is typically true for both C and C++ compilers.
82    
83 root 1.43 =item ECB_C99
84    
85 root 1.44 True if the implementation claims to be C99 compliant.
86 root 1.43
87     =item ECB_C11
88    
89 root 1.44 True if the implementation claims to be C11 compliant.
90    
91     =item ECB_CPP
92    
93     True if the implementation defines the C<__cplusplus__> macro to a true
94     value, which is typically true for C++ compilers.
95    
96     =item ECB_CPP98
97    
98     True if the implementation claims to be compliant to ISO/IEC 14882:1998
99     (the first C++ ISO standard) or any later vwersion. Typically true for all
100     C++ compilers.
101    
102     =item ECB_CPP11
103    
104     True if the implementation claims to be compliant to ISO/IEC 14882:2011
105     (C++11) or any later vwersion.
106 root 1.43
107     =item ECB_GCC_VERSION(major,minor)
108    
109     Expands to a true value (suitable for testing in by the preprocessor)
110     if the compiler used is GNU C and the version is the givne version, or
111     higher.
112    
113     This macro tries to return false on compilers that claim to be GCC
114     compatible but aren't.
115    
116     =back
117    
118 root 1.1 =head2 GCC ATTRIBUTES
119    
120 root 1.20 A major part of libecb deals with GCC attributes. These are additional
121 sf-exg 1.26 attributes that you can assign to functions, variables and sometimes even
122 root 1.20 types - much like C<const> or C<volatile> in C.
123    
124     While GCC allows declarations to show up in many surprising places,
125 sf-exg 1.26 but not in many expected places, the safest way is to put attribute
126 root 1.20 declarations before the whole declaration:
127    
128     ecb_const int mysqrt (int a);
129     ecb_unused int i;
130    
131     For variables, it is often nicer to put the attribute after the name, and
132     avoid multiple declarations using commas:
133    
134     int i ecb_unused;
135 root 1.3
136 root 1.1 =over 4
137    
138 root 1.2 =item ecb_attribute ((attrs...))
139 root 1.1
140 root 1.15 A simple wrapper that expands to C<__attribute__((attrs))> on GCC, and to
141     nothing on other compilers, so the effect is that only GCC sees these.
142    
143     Example: use the C<deprecated> attribute on a function.
144    
145     ecb_attribute((__deprecated__)) void
146     do_not_use_me_anymore (void);
147 root 1.2
148 root 1.3 =item ecb_unused
149    
150     Marks a function or a variable as "unused", which simply suppresses a
151     warning by GCC when it detects it as unused. This is useful when you e.g.
152     declare a variable but do not always use it:
153    
154 root 1.15 {
155     int var ecb_unused;
156 root 1.3
157 root 1.15 #ifdef SOMECONDITION
158     var = ...;
159     return var;
160     #else
161     return 0;
162     #endif
163     }
164 root 1.3
165 root 1.31 =item ecb_inline
166 root 1.29
167     This is not actually an attribute, but you use it like one. It expands
168     either to C<static inline> or to just C<static>, if inline isn't
169     supported. It should be used to declare functions that should be inlined,
170     for code size or speed reasons.
171    
172     Example: inline this function, it surely will reduce codesize.
173    
174 root 1.31 ecb_inline int
175 root 1.29 negmul (int a, int b)
176     {
177     return - (a * b);
178     }
179    
180 root 1.2 =item ecb_noinline
181    
182 root 1.9 Prevent a function from being inlined - it might be optimised away, but
183 root 1.3 not inlined into other functions. This is useful if you know your function
184     is rarely called and large enough for inlining not to be helpful.
185    
186 root 1.2 =item ecb_noreturn
187    
188 root 1.17 Marks a function as "not returning, ever". Some typical functions that
189     don't return are C<exit> or C<abort> (which really works hard to not
190     return), and now you can make your own:
191    
192     ecb_noreturn void
193     my_abort (const char *errline)
194     {
195     puts (errline);
196     abort ();
197     }
198    
199 sf-exg 1.19 In this case, the compiler would probably be smart enough to deduce it on
200     its own, so this is mainly useful for declarations.
201 root 1.17
202 root 1.2 =item ecb_const
203    
204 sf-exg 1.19 Declares that the function only depends on the values of its arguments,
205 root 1.17 much like a mathematical function. It specifically does not read or write
206     any memory any arguments might point to, global variables, or call any
207     non-const functions. It also must not have any side effects.
208    
209     Such a function can be optimised much more aggressively by the compiler -
210     for example, multiple calls with the same arguments can be optimised into
211     a single call, which wouldn't be possible if the compiler would have to
212     expect any side effects.
213    
214     It is best suited for functions in the sense of mathematical functions,
215 sf-exg 1.19 such as a function returning the square root of its input argument.
216 root 1.17
217     Not suited would be a function that calculates the hash of some memory
218     area you pass in, prints some messages or looks at a global variable to
219     decide on rounding.
220    
221     See C<ecb_pure> for a slightly less restrictive class of functions.
222    
223 root 1.2 =item ecb_pure
224    
225 root 1.17 Similar to C<ecb_const>, declares a function that has no side
226     effects. Unlike C<ecb_const>, the function is allowed to examine global
227     variables and any other memory areas (such as the ones passed to it via
228     pointers).
229    
230     While these functions cannot be optimised as aggressively as C<ecb_const>
231     functions, they can still be optimised away in many occasions, and the
232     compiler has more freedom in moving calls to them around.
233    
234     Typical examples for such functions would be C<strlen> or C<memcmp>. A
235     function that calculates the MD5 sum of some input and updates some MD5
236     state passed as argument would I<NOT> be pure, however, as it would modify
237     some memory area that is not the return value.
238    
239 root 1.2 =item ecb_hot
240    
241 root 1.17 This declares a function as "hot" with regards to the cache - the function
242     is used so often, that it is very beneficial to keep it in the cache if
243     possible.
244    
245     The compiler reacts by trying to place hot functions near to each other in
246     memory.
247    
248 sf-exg 1.19 Whether a function is hot or not often depends on the whole program,
249 root 1.17 and less on the function itself. C<ecb_cold> is likely more useful in
250     practise.
251    
252 root 1.2 =item ecb_cold
253    
254 root 1.17 The opposite of C<ecb_hot> - declares a function as "cold" with regards to
255     the cache, or in other words, this function is not called often, or not at
256     speed-critical times, and keeping it in the cache might be a waste of said
257     cache.
258    
259     In addition to placing cold functions together (or at least away from hot
260     functions), this knowledge can be used in other ways, for example, the
261     function will be optimised for size, as opposed to speed, and codepaths
262     leading to calls to those functions can automatically be marked as if
263 root 1.27 C<ecb_expect_false> had been used to reach them.
264 root 1.17
265     Good examples for such functions would be error reporting functions, or
266     functions only called in exceptional or rare cases.
267    
268 root 1.2 =item ecb_artificial
269    
270 root 1.17 Declares the function as "artificial", in this case meaning that this
271     function is not really mean to be a function, but more like an accessor
272     - many methods in C++ classes are mere accessor functions, and having a
273     crash reported in such a method, or single-stepping through them, is not
274     usually so helpful, especially when it's inlined to just a few instructions.
275    
276     Marking them as artificial will instruct the debugger about just this,
277     leading to happier debugging and thus happier lives.
278    
279     Example: in some kind of smart-pointer class, mark the pointer accessor as
280     artificial, so that the whole class acts more like a pointer and less like
281     some C++ abstraction monster.
282    
283     template<typename T>
284     struct my_smart_ptr
285     {
286     T *value;
287    
288     ecb_artificial
289     operator T *()
290     {
291     return value;
292     }
293     };
294    
295 root 1.2 =back
296 root 1.1
297     =head2 OPTIMISATION HINTS
298    
299     =over 4
300    
301 root 1.14 =item bool ecb_is_constant(expr)
302 root 1.1
303 root 1.3 Returns true iff the expression can be deduced to be a compile-time
304     constant, and false otherwise.
305    
306     For example, when you have a C<rndm16> function that returns a 16 bit
307     random number, and you have a function that maps this to a range from
308 root 1.5 0..n-1, then you could use this inline function in a header file:
309 root 1.3
310     ecb_inline uint32_t
311     rndm (uint32_t n)
312     {
313 root 1.6 return (n * (uint32_t)rndm16 ()) >> 16;
314 root 1.3 }
315    
316     However, for powers of two, you could use a normal mask, but that is only
317     worth it if, at compile time, you can detect this case. This is the case
318     when the passed number is a constant and also a power of two (C<n & (n -
319     1) == 0>):
320    
321     ecb_inline uint32_t
322     rndm (uint32_t n)
323     {
324     return is_constant (n) && !(n & (n - 1))
325     ? rndm16 () & (num - 1)
326 root 1.6 : (n * (uint32_t)rndm16 ()) >> 16;
327 root 1.3 }
328    
329 root 1.14 =item bool ecb_expect (expr, value)
330 root 1.1
331 root 1.7 Evaluates C<expr> and returns it. In addition, it tells the compiler that
332     the C<expr> evaluates to C<value> a lot, which can be used for static
333     branch optimisations.
334 root 1.1
335 root 1.27 Usually, you want to use the more intuitive C<ecb_expect_true> and
336     C<ecb_expect_false> functions instead.
337 root 1.1
338 root 1.27 =item bool ecb_expect_true (cond)
339 root 1.1
340 root 1.27 =item bool ecb_expect_false (cond)
341 root 1.1
342 root 1.7 These two functions expect a expression that is true or false and return
343     C<1> or C<0>, respectively, so when used in the condition of an C<if> or
344     other conditional statement, it will not change the program:
345    
346     /* these two do the same thing */
347     if (some_condition) ...;
348 root 1.27 if (ecb_expect_true (some_condition)) ...;
349 root 1.7
350 root 1.27 However, by using C<ecb_expect_true>, you tell the compiler that the
351     condition is likely to be true (and for C<ecb_expect_false>, that it is
352     unlikely to be true).
353 root 1.7
354 root 1.9 For example, when you check for a null pointer and expect this to be a
355 root 1.27 rare, exceptional, case, then use C<ecb_expect_false>:
356 root 1.7
357     void my_free (void *ptr)
358     {
359 root 1.27 if (ecb_expect_false (ptr == 0))
360 root 1.7 return;
361     }
362    
363     Consequent use of these functions to mark away exceptional cases or to
364     tell the compiler what the hot path through a function is can increase
365     performance considerably.
366    
367 root 1.27 You might know these functions under the name C<likely> and C<unlikely>
368     - while these are common aliases, we find that the expect name is easier
369     to understand when quickly skimming code. If you wish, you can use
370     C<ecb_likely> instead of C<ecb_expect_true> and C<ecb_unlikely> instead of
371     C<ecb_expect_false> - these are simply aliases.
372    
373 root 1.7 A very good example is in a function that reserves more space for some
374     memory block (for example, inside an implementation of a string stream) -
375 root 1.9 each time something is added, you have to check for a buffer overrun, but
376 root 1.7 you expect that most checks will turn out to be false:
377    
378     /* make sure we have "size" extra room in our buffer */
379     ecb_inline void
380     reserve (int size)
381     {
382 root 1.27 if (ecb_expect_false (current + size > end))
383 root 1.7 real_reserve_method (size); /* presumably noinline */
384     }
385    
386 root 1.14 =item bool ecb_assume (cond)
387 root 1.7
388     Try to tell the compiler that some condition is true, even if it's not
389     obvious.
390    
391     This can be used to teach the compiler about invariants or other
392     conditions that might improve code generation, but which are impossible to
393     deduce form the code itself.
394    
395 root 1.27 For example, the example reservation function from the C<ecb_expect_false>
396 root 1.7 description could be written thus (only C<ecb_assume> was added):
397    
398     ecb_inline void
399     reserve (int size)
400     {
401 root 1.27 if (ecb_expect_false (current + size > end))
402 root 1.7 real_reserve_method (size); /* presumably noinline */
403    
404     ecb_assume (current + size <= end);
405     }
406    
407     If you then call this function twice, like this:
408    
409     reserve (10);
410     reserve (1);
411    
412     Then the compiler I<might> be able to optimise out the second call
413     completely, as it knows that C<< current + 1 > end >> is false and the
414     call will never be executed.
415    
416     =item bool ecb_unreachable ()
417    
418     This function does nothing itself, except tell the compiler that it will
419 root 1.9 never be executed. Apart from suppressing a warning in some cases, this
420 root 1.7 function can be used to implement C<ecb_assume> or similar functions.
421    
422 root 1.14 =item bool ecb_prefetch (addr, rw, locality)
423 root 1.7
424     Tells the compiler to try to prefetch memory at the given C<addr>ess
425 root 1.10 for either reading (C<rw> = 0) or writing (C<rw> = 1). A C<locality> of
426 root 1.7 C<0> means that there will only be one access later, C<3> means that
427     the data will likely be accessed very often, and values in between mean
428     something... in between. The memory pointed to by the address does not
429     need to be accessible (it could be a null pointer for example), but C<rw>
430     and C<locality> must be compile-time constants.
431    
432     An obvious way to use this is to prefetch some data far away, in a big
433 root 1.9 array you loop over. This prefetches memory some 128 array elements later,
434 root 1.7 in the hope that it will be ready when the CPU arrives at that location.
435    
436     int sum = 0;
437    
438     for (i = 0; i < N; ++i)
439     {
440     sum += arr [i]
441     ecb_prefetch (arr + i + 128, 0, 0);
442     }
443    
444     It's hard to predict how far to prefetch, and most CPUs that can prefetch
445     are often good enough to predict this kind of behaviour themselves. It
446     gets more interesting with linked lists, especially when you do some fair
447     processing on each list element:
448    
449     for (node *n = start; n; n = n->next)
450     {
451     ecb_prefetch (n->next, 0, 0);
452     ... do medium amount of work with *n
453     }
454    
455     After processing the node, (part of) the next node might already be in
456     cache.
457 root 1.1
458 root 1.2 =back
459 root 1.1
460 root 1.36 =head2 BIT FIDDLING / BIT WIZARDRY
461 root 1.1
462 root 1.4 =over 4
463    
464 root 1.3 =item bool ecb_big_endian ()
465    
466     =item bool ecb_little_endian ()
467    
468 sf-exg 1.11 These two functions return true if the byte order is big endian
469     (most-significant byte first) or little endian (least-significant byte
470     first) respectively.
471    
472 root 1.24 On systems that are neither, their return values are unspecified.
473    
474 root 1.3 =item int ecb_ctz32 (uint32_t x)
475    
476 root 1.35 =item int ecb_ctz64 (uint64_t x)
477    
478 sf-exg 1.11 Returns the index of the least significant bit set in C<x> (or
479 root 1.24 equivalently the number of bits set to 0 before the least significant bit
480 root 1.35 set), starting from 0. If C<x> is 0 the result is undefined.
481    
482 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ctz32>.
483    
484 root 1.35 For example:
485 sf-exg 1.11
486 root 1.15 ecb_ctz32 (3) = 0
487     ecb_ctz32 (6) = 1
488 sf-exg 1.11
489 root 1.41 =item bool ecb_is_pot32 (uint32_t x)
490    
491     =item bool ecb_is_pot64 (uint32_t x)
492    
493     Return true iff C<x> is a power of two or C<x == 0>.
494    
495     For smaller types then C<uint32_t> you can safely use C<ecb_is_pot32>.
496    
497 root 1.35 =item int ecb_ld32 (uint32_t x)
498    
499     =item int ecb_ld64 (uint64_t x)
500    
501     Returns the index of the most significant bit set in C<x>, or the number
502     of digits the number requires in binary (so that C<< 2**ld <= x <
503     2**(ld+1) >>). If C<x> is 0 the result is undefined. A common use case is
504     to compute the integer binary logarithm, i.e. C<floor (log2 (n))>, for
505     example to see how many bits a certain number requires to be encoded.
506    
507     This function is similar to the "count leading zero bits" function, except
508     that that one returns how many zero bits are "in front" of the number (in
509     the given data type), while C<ecb_ld> returns how many bits the number
510     itself requires.
511    
512 root 1.36 For smaller types than C<uint32_t> you can safely use C<ecb_ld32>.
513    
514 root 1.3 =item int ecb_popcount32 (uint32_t x)
515    
516 root 1.35 =item int ecb_popcount64 (uint64_t x)
517    
518 root 1.36 Returns the number of bits set to 1 in C<x>.
519    
520     For smaller types than C<uint32_t> you can safely use C<ecb_popcount32>.
521    
522     For example:
523 sf-exg 1.11
524 root 1.15 ecb_popcount32 (7) = 3
525     ecb_popcount32 (255) = 8
526 sf-exg 1.11
527 root 1.39 =item uint8_t ecb_bitrev8 (uint8_t x)
528    
529     =item uint16_t ecb_bitrev16 (uint16_t x)
530    
531     =item uint32_t ecb_bitrev32 (uint32_t x)
532    
533     Reverses the bits in x, i.e. the MSB becomes the LSB, MSB-1 becomes LSB+1
534     and so on.
535    
536     Example:
537    
538     ecb_bitrev8 (0xa7) = 0xea
539     ecb_bitrev32 (0xffcc4411) = 0x882233ff
540    
541 root 1.8 =item uint32_t ecb_bswap16 (uint32_t x)
542    
543 root 1.3 =item uint32_t ecb_bswap32 (uint32_t x)
544    
545 root 1.34 =item uint64_t ecb_bswap64 (uint64_t x)
546 sf-exg 1.13
547 root 1.34 These functions return the value of the 16-bit (32-bit, 64-bit) value
548     C<x> after reversing the order of bytes (0x11223344 becomes 0x44332211 in
549     C<ecb_bswap32>).
550    
551     =item uint8_t ecb_rotl8 (uint8_t x, unsigned int count)
552    
553     =item uint16_t ecb_rotl16 (uint16_t x, unsigned int count)
554 root 1.3
555     =item uint32_t ecb_rotl32 (uint32_t x, unsigned int count)
556    
557 root 1.34 =item uint64_t ecb_rotl64 (uint64_t x, unsigned int count)
558    
559     =item uint8_t ecb_rotr8 (uint8_t x, unsigned int count)
560    
561     =item uint16_t ecb_rotr16 (uint16_t x, unsigned int count)
562    
563     =item uint32_t ecb_rotr32 (uint32_t x, unsigned int count)
564    
565 root 1.33 =item uint64_t ecb_rotr64 (uint64_t x, unsigned int count)
566    
567 root 1.34 These two families of functions return the value of C<x> after rotating
568     all the bits by C<count> positions to the right (C<ecb_rotr>) or left
569     (C<ecb_rotl>).
570 sf-exg 1.11
571 root 1.20 Current GCC versions understand these functions and usually compile them
572 root 1.34 to "optimal" code (e.g. a single C<rol> or a combination of C<shld> on
573     x86).
574 root 1.20
575 root 1.3 =back
576 root 1.1
577     =head2 ARITHMETIC
578    
579 root 1.3 =over 4
580    
581 root 1.14 =item x = ecb_mod (m, n)
582 root 1.3
583 root 1.25 Returns C<m> modulo C<n>, which is the same as the positive remainder
584     of the division operation between C<m> and C<n>, using floored
585     division. Unlike the C remainder operator C<%>, this function ensures that
586     the return value is always positive and that the two numbers I<m> and
587     I<m' = m + i * n> result in the same value modulo I<n> - in other words,
588     C<ecb_mod> implements the mathematical modulo operation, which is missing
589     in the language.
590 root 1.14
591 sf-exg 1.23 C<n> must be strictly positive (i.e. C<< >= 1 >>), while C<m> must be
592 root 1.14 negatable, that is, both C<m> and C<-m> must be representable in its
593 root 1.30 type (this typically excludes the minimum signed integer value, the same
594 root 1.25 limitation as for C</> and C<%> in C).
595 sf-exg 1.11
596 root 1.24 Current GCC versions compile this into an efficient branchless sequence on
597 root 1.28 almost all CPUs.
598 root 1.24
599     For example, when you want to rotate forward through the members of an
600     array for increasing C<m> (which might be negative), then you should use
601     C<ecb_mod>, as the C<%> operator might give either negative results, or
602     change direction for negative values:
603    
604     for (m = -100; m <= 100; ++m)
605     int elem = myarray [ecb_mod (m, ecb_array_length (myarray))];
606    
607 sf-exg 1.37 =item x = ecb_div_rd (val, div)
608    
609     =item x = ecb_div_ru (val, div)
610    
611     Returns C<val> divided by C<div> rounded down or up, respectively.
612     C<val> and C<div> must have integer types and C<div> must be strictly
613 sf-exg 1.38 positive. Note that these functions are implemented with macros in C
614     and with function templates in C++.
615 sf-exg 1.37
616 root 1.3 =back
617 root 1.1
618     =head2 UTILITY
619    
620 root 1.3 =over 4
621    
622 sf-exg 1.23 =item element_count = ecb_array_length (name)
623 root 1.3
624 sf-exg 1.13 Returns the number of elements in the array C<name>. For example:
625    
626     int primes[] = { 2, 3, 5, 7, 11 };
627     int sum = 0;
628    
629     for (i = 0; i < ecb_array_length (primes); i++)
630     sum += primes [i];
631    
632 root 1.3 =back
633 root 1.1
634 root 1.43 =head2 SYMBOLS GOVERNING COMPILATION OF ECB.H ITSELF
635    
636     These symbols need to be defined before including F<ecb.h> the first time.
637    
638     =over 4
639    
640     =item ECB_NO_THRADS
641    
642     If F<ecb.h> is never used from multiple threads, then this symbol can
643     be defined, in which case memory fences (and similar constructs) are
644     completely removed, leading to more efficient code and fewer dependencies.
645    
646     Setting this symbol to a true value implies C<ECB_NO_SMP>.
647    
648     =item ECB_NO_SMP
649    
650     The weaker version of C<ECB_NO_THREADS> - if F<ecb.h> is used from
651     multiple threads, but never concurrently (e.g. if the system the program
652     runs on has only a single CPU with a single core, no hyperthreading and so
653     on), then this symbol can be defined, leading to more efficient code and
654     fewer dependencies.
655    
656     =back
657    
658 root 1.1