Re: [PATCH] target/s390x: Move trans_exc_code update to do_program_interrupt

[Ilya Leoshkevich <iii@linux.ibm.com>]

On Mon, 2023-07-31 at 07:55 -0700, Richard Henderson wrote:
> On 7/31/23 01:26, Ilya Leoshkevich wrote:
> > > > +    case PGM_PROTECTION:
> > > > +    case PGM_TRANS_SPEC:
> > > > +        assert(env->int_pgm_code == env->tlb_fill_exc);
> > > > +        set_trans_exc_code = true;
> > > > +        advance = true;
> > > > +        break;
> > > 
> > > These exceptions came from seeing an early kernel fault, grepping
> > > for
> > > the set of
> > > exceptions raised in mmu_helper.c, and eliminating PGM_ADDRESSING
> > > per
> > > the first hunk.
> > 
> > Does POp specify that the CPU stores Translation-Exception
> > Identification on Translation-Specification Exceptions
> > (PGM_TRANS_SPEC)? I re-read the 0xA8 documentation a few times, but
> > could not find it.
> 
> Neither could I.
> 
> > It's also interesting what the kernel was attempting when it got
> > PGM_TRANS_SPEC and recovered from it. Maybe something else is
> > wrong?
> 
> I think the kernel was testing PGM_PROTECTION, to see if LowCore
> protection worked.
> 
> But since that one wasn't listed for 0xA8 either, I preserved
> previous behaviour in 
> setting trans_exc_code for all mmu exceptions except for
> PGM_ADDRESSING.
> 
> 
> r~

Actually, PGM_PROTECTION is there. If you scroll a few pages down, you
will see: "Translation-Exception Identification for Protection
Exceptions". But nothing like this for PGM_TRANS_SPEC.