arm/cxd56: use chip specific vectors to allow smpcall update regs

[jasonbu]

Summary

When sig dispatch do up_schedule_sigaction, need to make a new frame to run arm_sigdeliver. But the exception_direct cannot handle xcp.regs as we are using c-function exception handler.
Need to use exception_common to handle SMP call.

issue from #13606 (comment)
@masayuki2009 please review the update, and see it at your environments also work as expected.

use chip specific vector cxd56_vectors to make CXD56_IRQ_SMP_CALL not use c-call version exception handler.

Impact

can pass the self-test in Spresense CPU=3 case.
for self-test, the _vector still in .map, maybe we should add option to make unique between cxd56_vectors and _vectors.

Testing

CI-test.

./tools/configure.sh -l spresense/smp 
kconfig-tweak --set-val CONFIG_SMP_NCPUS 3
make -j
./tools/flash_writer.py -s -c /dev/ttyUSB0 -d -b 115200 -n nuttx.spk
minicom -D /dev/ttyUSB0 -b 115200 -C log/ttysprense-$(date +"%Y-%m-%d-%H-%M-%S").log
ostest

[jasonbu]

ramvectors should allso can solve the problem, but not work in my board.

[masayuki2009]

@jasonbu

When sig dispatch do up_schedule_sigaction, need to make a new frame to run arm_sigdeliver. But the exception_direct cannot handle xcp.regs as we are using c-function exception handler.
Need to use exception_common to handle SMP call.

I think a similar issue would happen with rp2040 (dual Cortex-M0+), lc823450 (dual Cortex-M3) that use exception_direct.
So they should be fixed as well.

[jasonbu]

@jasonbu

When sig dispatch do up_schedule_sigaction, need to make a new frame to run arm_sigdeliver. But the exception_direct cannot handle xcp.regs as we are using c-function exception handler.
Need to use exception_common to handle SMP call.

I think a similar issue would happen with rp2040 (dual Cortex-M0+), lc823450 (dual Cortex-M3) that use exception_direct. So they should be fixed as well.

let's make sure problem already converged, then talk about how to update correctly. for example if always one smp call have to use exception_common, we should make it a optional as a support in arm_vectors.c.
please give me a feedback if this patch can solve your problem.

[GUIDINGLI]

@masayuki2009

Cause we don't have enough environment to test the code, it leads to regression issues.
Can you provide us with a copy of your testing script so that we can prevent such problems from happening?

As this PR said:
#14007

[masayuki2009]

let's make sure problem already converged, then talk about how to update correctly. for example if always one smp call have to use exception_common, we should make it a optional as a support in arm_vectors.c.
please give me a feedback if this patch can solve your problem.

@jasonbu
I confirmed that this PR fixes #13606 (comment)

[xiaoxiang781216]

let's fix other arm-m SMP chip.