On Tue, Jan 14, 2020 at 10:32 AM Richard Henderson <address@hidden
On 1/9/20 11:05 PM, address@hidden wrote:
> Because ctx->mstatus_fs changes dynamically during runtime, we should
> remove the mstatus_fs check at the translation stage.
This change is incorrect.
The actual bug is in cpu_get_tb_cpu_state(), as I just noticed during review:
"[PATCH 2/3] RISC-V: use FIELD macro to define tb flags"
I didn't understand the purpose of mstatus_fs, but now I get it.
Thanks for pointing me to.
I will resend my patches.