On 10/2/19 9:47 AM, Richard Henderson wrote:
> There is still the special case of EXECUTE of MVCL, which I suspect must have
> some failure mode that we're not considering -- the setting and clearing of
> ex_value can't help. I have a suspicion that we need to special case that
> within helper_ex, just so that ex_value doesn't enter into it.
I had a walk and a think. I now believe that we're ok:
(1) TB with EXECUTE runs, at address Ae
- env->psw_addr stored with Ae.
- helper_ex() runs, memory address Am computed
from D2a(X2a,B2a) or from psw.addr+RI2.
- env->ex_value stored with memory value modified by R1a
(2) TB of executee runs,
- env->ex_value stored with 0.
- helper_mvcl() runs, using and updating R1b, R1b+1, R2b, R2b+1.
(3a) helper_mvcl() completes,
- TB of executee continues, psw.addr += ilen.
- Next instruction is the one following EXECUTE.
(3b) helper_mvcl() exits to main loop,
- cpu_loop_exit_restore() unwinds psw.addr = Ae.
- Next instruction is the EXECUTE itself...
- goto 1.
If we can agree that the result is undefined if registers R1a, X2a, B2a overlap
R1b, R1b+1, R2b, R2b+1, or if the memory address Am is modified by the
interrupted MVCL, then we're ok.
r~
