CCS/AM3354: the different between assembler 'LDMIA SP!, {R0-R12,LR,PC} ' and ' LDMIA SP!, {R0-R12,LR,PC}^ '

Tool/software: Code Composer Studio

Hello 

I porting an RTOS to AM3354,  after the program execute assembler :

    LDMIA   SP!, {R0}                          ; Pickup saved CPSR value
    MSR     SPSR_cxsf, R0                      ; Setup new context SPSR value
    LDMIA   SP!, {R0-R12,LR,PC}               ; Pop new tasks context 

the  program jump into the entry point in the task,  and the PC Register value is 0x40302B1C,

then I press the button 'step into', the program execute one sentence code and the 

PC value become 0x40302B22( the value is not 4 byte align ). Then the program jump into 

the following code in document memset32.asm .

_word_aligned:
	AND	R1, R1, #255		; be safe since prototype has value as
					; as an int rather than unsigned char

	ORR	R1, R1, R1, LSL	#8      ; replicate byte in 2nd byte of
					; register

	CMP	R2,#4			; are at least 4 bytes being set
	BCC	_memset3

Then the hard fault happened.

But, if I change the code:

LDMIA   SP!, {R0-R12,LR,PC}^

into 

LDMIA   SP!, {R0-R12,LR,PC}

everything is fun, why this happen?

Best Regards

Jack