Rm48x50

You need to turn on RT_VFP_LAZY_STACKING in rtconfig.h. By default, RTT
will turn on VFP for all threads and stack all the VFP registers. When
doing lazy stacking, VFP will only be turned on for the thread who
issued VFP instructions.

Currently, if a thread turned on VFP, it cannot be turned off. RTT will
never know what time the thread doesn't need VFP any more. The thread
might could turn off the VFP for it self in proper time.

VFP instructions only have IA(Increment After)/DB(Decrement Before)
mode. To keep consistency, just change STM/LDM to DB/IA accordingly.

This support Common VFPv2 sub-architecture.

For pull request

use __rt_ffs instead of ffs

Use condition flag in the ORR. This could eliminate a BEQ.

When saving thread registers in context_switch_interrupt_to, we don't
change them, just move them. So there is no need to always r0-r3 from
stack to the real r0-r3. So just use the intermediate registers and
eliminate 2 MOV.

It does not satisfy AAPCS.

The code is GPLv2 so I think we could use it for free(both free beer and
free speech).

Substitude STMFD, MOV, ADD with STMFD, SUB. It reduce one instruction.
Tested on board and it works like a charm.

use __rt_ffs instead of ffs

usbdevice/core: add reset function to support hotplug

When the USB got RESET packet from the host and the address is setup,
all the classes will got reset. The reset is done by class stop and than
class run. So the classes should reset their internal state in
class_{run,stop}.

Besides, the USB device driver could post a USB_MSG_RESET message on
every RESET packet.

cdc_vcom: reset the vcom state in proper cases

Reset the vcom state in class_{run,stop} and when the PC is not
receiving data from me.

We currently only support building with CCS and SCons is not using.
bsp/rm48x50/HALCoGen/HALCoGen.{hcg,dil} is the HALCoGen project file.
You may need to regenerate the source file as you like, providing that:

    1, IRQ is in Dispatch Mode and the table entry is IRQ_Handler. The
    channel 5 in enabled and connected to IRQ.

    2, RTI driver is enabled and compare3 source is selected to counter1
    and the compare3 will generate tick in the period of 10ms. This
    value is coresponding with RT_TICK_PER_SECOND in rtconfig.h.

In CCS, you need to create a new CCS project and create link folders
pointing at bsp/rm48x50, libcpu/arm/rm48x50 and src/, include/. Remember
to add the include path to the Build Properties.

For pull request