In VCS simulation, multi-core simulation of some harts ended prematurely due to incorrect execution of SEQZ instruction

[meiqin0]

Before start

• [X] I have read the RISC-V ISA Manual and this is not a RISC-V ISA question. 我已经阅读过 RISC-V 指令集手册，这不是一个指令集本身的问题。
• [X] I have read the XiangShan Documents. 我已经阅读过香山文档。
• [X] I have searched the previous issues and did not find anything relevant. 我已经搜索过之前的 issue，并没有找到相关的。
• [X] I have reviewed the commit messages from the relevant commit history. 我已经浏览过相关的提交历史和提交信息。

Describe the bug

我在仿真多hart的时候，发现四个hart有部分hart没有等到同步点而是提前结束了，然后找了一下问题，发现好像是伪指令seqz没有正确。

[TRANSLATION]

During the simulation of multiple harts, I noticed that some of the four harts ended prematurely without waiting for the synchronization point. Upon further investigation, it seems that the pseudo-instruction SEQZ was not executed correctly.

后面80003a8a这条分支指令的src0/1都是0， a4寄存器没有取反，应该是80003a66指令没有执行：

[TRANSLATION]

The source registers src0/1 for the branch instruction at 80003a8a are both zero, and register a4 has not been inverted, which suggests that the instruction at 80003a66 was not executed:

测试用例基于am->amtest, 问题发生点位于mpe.c的_barrier函数:

[TRANSLATION]

The test case is based on am->amtest, and the point of issue is located in the _barrier function of mpe.c:

Expected behavior

应该等到4个hart都到达同步点后再打印结果（正确）：

[TRANSLATION]

The results should be printed only after all four harts have reached the synchronization point (correct):

而不是部分hart提前结束（此错误）： [TRANSLATION] Instead of some harts ending prematurely (this is the error):

To Reproduce

elf,bin,dasm amtest-riscv64-xs-dual.zip

Environment

• XiangShan branch: master
• XiangShan commit id: commit 5adc4829471a0ea417766f3b0e57679ab3feb696
• NEMU commit id:
• SPIKE commit id:

Additional context

注：需要使用vcs仿真，verilator发现不了这个错误。

[TRANSLATION]

Note: VCS simulation is required, as Verilator cannot detect this error.

[Tang-Haojin]

感谢您的反馈，多核的测试尚不充分，我们后续会进一步调查这一问题。

[TRANSLATION]

Thank you for your feedback. The multi-core testing is still insufficient, and we will further investigate this issue in the follow-up.