With new timer algorithm, timer should be initialized during startup. So
add them to the bsps. Use these commands to get which bsp is missing
calling the function:

    % git grep rt_system_timer_init bsp|sed -n 's|bsp/\([^/]*\).*|\1|p' | sort | uniq > have_tm_init
    % ls -1 bsp |sed -n 's|\([^/]*\).*|\1|p' | sort > all_bsp
    % comm -3 all_bsp have_tm_init
    beaglebone
    lpc176x
    lpc178x
    ls1bdev
    mb9bf506r
    stm32f10x
    xplorer4330

kernel: timer: fix compiling bug in d59aa279

Finsh ex

We split the history handling form the key handling. So we could handle
the direction key even if the history is disabled. As a "side effect", I
also remove the unnecessary "use_history" bit.

kernel: use skip list to implement timer list

make sure the device is in STREAM mode when used by rt_kprintf.

UART DMA tx_complete() with buffer information

simulator/jffs2: remove _TIME_T_DEFINED when built by VC++

_TIME_T_DEFINED was pre-defined in order to build jffs2 with VC.
While it may cause many side effects.

simulator: add createdef.py to generate def file for VS

kernel: mutex could only be released in thread context

Now we can get rid off Mingw to create def file. To create def file, one
should run:

    scons --def

in the bsp/simulator.

Mutex has the idea of ownership, only the thread which owns the mutex
can release it. So rt_mutex_release could only be called in thread
context.  Add a debug guard to it.

Skip list is a "random" data structure that in high possibilities it
would get O(log(N)) time complexity in inserting while the old list get
O(N). Forthermore, when set RT_TIMER_SKIP_LIST_LEVEL to 1, it will just
the same as the old double linked list, both in time and space
complexity.

Benchmarks shows that when RT_TIMER_SKIP_LIST_LEVEL is 3, the average
time of random insertion of new timer is about 2 times faster than the
old timer when there are 100 timers and 3 times faster when there are
200 timers.

However, it restores the deprecated funcion rt_system_timer_init. BSPs
must invoke it upon system startup.

kernel: fix compiling error when RT_DEBUG is not defined

Define RT_DEBUG_IN_THREAD_CONTEXT when RT_DEBUG is not defined.

kernel: add RT_DEBUG_IN_THREAD_CONTEXT

In thread context means: 1) the scheduler has been started; 2) not in
interrupt context. It is more stronger than RT_DEBUG_NOT_IN_INTERRUPT.
With this commit, you will catch the error on situations like taking
mutex before scheduling instead of crashing on NULL pointer reference.

DFS: add dfs_file_flush prototype in dfs_file.h

log_trace: fix the memmove use

The first parameter of memmove is the dst, not src. Besides, fix the
wrong move size.

components/utilities: add ymodem

Real-YModem implemented a flexible YModem support. It use callback-based
structure to let the user application to deal with the data. It contains
3 examples:

    1. echo.c: write the data recieved on YModem to an other device
    2. null.c: discard the YModem data
    3. tofile.c: write the data to the file system

Currently, it does not support batch file transmission.

scons script: support to generate vs2012 project xml

rename template.vcproj to template_vs2005.vcproj
add template_vs2012.vcxproj
NOTE: the vs2012.py is ugly, just can work.

DFS/jffs2: fix file mode init bug in jffs2_open/opendir

found by haitao5198
clear mount parition device talbe when jffs2 mount falied

found by haitao52198
URL:http://www.rt-thread.org/phpBB3/viewtopic.php?f=3&t=3112&p=17153#p17153

Compiling fix

*buffer) with buffer pointer