先分析fwrite，直接找到PHP源代码：
static size_t _php_stream_write_buffer(php_stream *stream, const char *buf, size_t count TSRMLS_DC)
{vv
size_t didwrite = , towrite, justwrote; /* if we have a seekable stream we need to ensure that data is written at the
* current stream->position. This means invalidating the read buffer and then
* performing a low-level seek */
if (stream->ops->seek && (stream->flags & PHP_STREAM_FLAG_NO_SEEK) == && stream->readpos != stream->writepos) {
stream->readpos = stream->writepos = ; stream->ops->seek(stream, stream->position, SEEK_SET, &stream->position TSRMLS_CC);
} while (count > ) {
towrite = count;
//chunk_size是8K
if (towrite > stream->chunk_size)
towrite = stream->chunk_size;
//分多次调用
justwrote = stream->ops->write(stream, buf, towrite TSRMLS_CC); /* convert justwrote to an integer, since normally it is unsigned */
if ((int)justwrote > ) {
buf += justwrote;
count -= justwrote;
didwrite += justwrote; /* Only screw with the buffer if we can seek, otherwise we lose data
* buffered from fifos and sockets */
if (stream->ops->seek && (stream->flags & PHP_STREAM_FLAG_NO_SEEK) == ) {
stream->position += justwrote;
}
} else {
break;
}
}
return didwrite; }
上图中红色注释是我加的。可以看出，fwrite会判断写入的内容长度，如果大于8K，会分多次write。每次write是直接调用了系统的write，这个是原子的。因为有多次write，就会有被中断的可能，就可能会出现错乱。
结论：写入小于8K，是原子性操作，不用加锁。反之需要。
error_log和file_put_contents函数
看这两个函数的源代码，发现最终也会调用_php_stream_write_buffer，所以结果是一样的。
程序验证：
先构造一个脚本，每次写10000字节（超过8K）
$a=str_repeat($argv[1],10000);
$i=0;
while($i++<10000)
file_put_contents('temp',$a."\n",FILE_APPEND);
用shell模拟并发
for((j=;j<;j++));do
php test.php a&
php test.php b&
done
结果可以明显看出来，出现了大量的错乱。
wc -l temp
temp
grep -P '(ab|ba)' temp|wc -l
如果把每次10000字节，改成8000如何呢，下面结果很明显了？
wc -l temp
temp
grep -P '(ab|ba)' temp|wc -l
TIP:怎么快速找到源代码呢？参考我的一篇博客中的方法：http://www.cnblogs.com/hxdoit/p/5193598.html