the coprocessor save area in function pxPortInitialiseStack would not be same address in function vPortStoreTaskMPUSettings

[frog7259]

Describe the bug
1.
In function pxPortInitialiseStack(StackType_t *pxTopOfStack....)
p = (uint32_t *)(((uint32_t) pxTopOfStack - XT_CP_SIZE) & ~0xf);

In function prvInitialiseNewTask (file: task.c)
pxTopOfStack =( pxStack + (ulStackDepth - 1) ) &( ~portBYTE_ALIGNMENT_MASK)

so we know co-processor stack is at
p = (uint32_t *)(((uint32_t) (( pxStack + (ulStackDepth - 1) ) &( ~portBYTE_ALIGNMENT_MASK))- XT_CP_SIZE) & ~0xf);

In function vPortStoreTaskMPUSettings( .... , StackType_t pxBottomOfStack ...)
xMPUSettings->coproc_area = (StackType_t)((((uint32_t)(pxBottomOfStack + usStackDepth - 1)) - XT_CP_SIZE ) & ~0xf);

pxBottomOfStack = pxStack

=> xMPUSettings->coproc_area = (StackType_t*)((((uint32_t)(pxStack+ ulStackDepth - 1)) - XT_CP_SIZE ) & ~0xf);

combine 1 & 2, p should equal to coproc_area.
when portBYTE_ALIGNMENT_MASK set 128 bytes.

(uint32_t )(((uint32_t) (( pxStack + (ulStackDepth - 1) ) &( ~portBYTE_ALIGNMENT_MASK))- XT_CP_SIZE) & ~0xf)
not equal to
(StackType_t)((((uint32_t)(pxStack+ usStackDepth - 1)) - XT_CP_SIZE ) & ~0xf);

And it would cause co-processor context switch error, because the stack save area is not correct

btw, because the cache line is 128 alignment, so we need to set the portBYTE_ALIGNMENT_MASK = 0x007F

System information

• Which hardware board or part numbers? Hifi3
• IDE used : RI-2018.0-linux
• Operating System : Linux
• Version of Amazon FreeRTOS (run git describe --tags to find it) : V10.0.1
• Project [Custom Application]
• If your project is a Custom Application, please add the relevant code snippet in the section Code to reproduce the bug.
  (file: portable.h)
  #if portBYTE_ALIGNMENT == 128
  #define portBYTE_ALIGNMENT_MASK ( 0x007FU )
  #endif

Expected behavior
co-processor context switch is normal, not trigger exception

Screenshots or console output
no

To reproduce
Steps to reproduce the behavior:

1. ...
2. ...

Code to reproduce the bug

If the code is longer than 30 lines, GIST is preferred.

Additional context
Add any other context about the problem here.

Thank you!