feat(dbltrp): add support for critical-error

[lewislzh]

This pr raise critical-error in difftest when XiangShan issues a critical-error signal.
The critical-error state is defined in the smdbltrp extension. This extension specifies that when the critical-error signal is asserted, the CPU should halt execution. This PR implements a method for stopping execution in simulation mode, where difftest halts by triggering an abort.
In the future, an option to enter debug mode rather than directly halting the CPU will be introduced for handling critical-error. This incoming change won’t require modifications to the difftest framework; Xiangshan will only need to control it through an valid signal.

[poemonsense]

what's the difference between this and trap?

[lewislzh]

what's the difference between this and trap?
the critical-error state is defined in smdbltrp extension. critical-error
Xiangshan will send a critical-error signal to the top level and difftest. At this point, the core should stop running: during simulation comparison, difftest will stop by triggering an abort; in other cases, after Xiangshan generates the critical-error signal, the ROB will stop committing instructions.
I think this represents an interruption more severe than exceptions or interrupts, causing the CPU to halt execution directly.

[lewislzh]

In the future, an option will be introduced to enter debug mode in the event of a critical-error, allowing the CPU to handle the critical-error in a way that improves error diagnosis and resolution. Currently, the handling method for critical-error is to directly halt CPU execution.

[poemonsense]

In the future, an option will be introduced to enter debug mode in the event of a critical-error, allowing the CPU to handle the critical-error in a way that improves error diagnosis and resolution. Currently, the handling method for critical-error is to directly halt CPU execution.

Make sense. Look good. Maybe we should add comments to explain its relationship with the double trap extension.

[lewislzh]

In the future, an option will be introduced to enter debug mode in the event of a critical-error, allowing the CPU to handle the critical-error in a way that improves error diagnosis and resolution. Currently, the handling method for critical-error is to directly halt CPU execution.

Make sense. Look good. Maybe we should add comments to explain its relationship with the double trap extension.

Sure, I have updated the comment to clarify the relationship.

[poemonsense]

Thanks. Merged.