Re: [Qemu-devel] [Qemu-ppc] [PATCH] 4/4] target-ppc: Implement bcdctz. instruction

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 3138 bytes --]

On Thu, Oct 27, 2016 at 06:08:20PM -0200, joserz@linux.vnet.ibm.com wrote:
> On Thu, Oct 27, 2016 at 12:47:34PM +1100, David Gibson wrote:
> > On Wed, Oct 26, 2016 at 11:18:58AM -0200, Jose Ricardo Ziviani wrote:
> > > bcdctz. converts from BCD to Zoned numeric format. Zoned format uses
> > > a byte to represent a digit where the most significant nibble is 0x3
> > > or 0xf, depending on the preferred signal.
> > > 
> > > Signed-off-by: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com>
> > > ---
> > >  target-ppc/helper.h                 |  1 +
> > >  target-ppc/int_helper.c             | 54 +++++++++++++++++++++++++++++++++++++
> > >  target-ppc/translate/vmx-impl.inc.c |  7 +++--
> > >  3 files changed, 60 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/target-ppc/helper.h b/target-ppc/helper.h
> > > index f460635..3b928b8 100644
> > > --- a/target-ppc/helper.h
> > > +++ b/target-ppc/helper.h
> > > @@ -372,6 +372,7 @@ DEF_HELPER_4(bcdsub, i32, avr, avr, avr, i32)
> > >  DEF_HELPER_3(bcdcfn, i32, avr, avr, i32)
> > >  DEF_HELPER_2(bcdctn, i32, avr, avr)
> > >  DEF_HELPER_3(bcdcfz, i32, avr, avr, i32)
> > > +DEF_HELPER_3(bcdctz, i32, avr, avr, i32)
> > >  
> > >  DEF_HELPER_2(xsadddp, void, env, i32)
> > >  DEF_HELPER_2(xssubdp, void, env, i32)
> > > diff --git a/target-ppc/int_helper.c b/target-ppc/int_helper.c
> > > index 8cbbdfc..e0a84bb 100644
> > > --- a/target-ppc/int_helper.c
> > > +++ b/target-ppc/int_helper.c
> > > @@ -2768,6 +2768,60 @@ uint32_t helper_bcdcfz(ppc_avr_t *r, ppc_avr_t *b, uint32_t ps)
> > >      *r = ret;
> > >      return cr;
> > >  }
> > > +
> > > +uint32_t helper_bcdctz(ppc_avr_t *r, ppc_avr_t *b, uint32_t ps)
> > > +{
> > > +    int i;
> > > +    int j;
> > > +    int cr = 0;
> > > +    int invalid = 0;
> > > +    uint8_t digit = 0;
> > > +    int sgnb = bcd_get_sgn(b);
> > > +    int zone_lead = (ps) ? 0xF0 : 0x30;
> > > +    ppc_avr_t ret = { .u64 = { 0, 0 } };
> > > +
> > > +    int eq_flag = (b->u64[HI_IDX] == 0) && ((b->u64[LO_IDX] >> 4) == 0);
> > > +    int ox_flag = (b->u64[HI_IDX] != 0);
> > 
> > This doesn't look right.  The zoned format can accomodate 16 digits.
> > The low word of the packed format can accomodate 15 digits + sign.  So
> > it should be possible to convert a packed for with a value in the low
> > nibble of the upper word.
> 
> David, it seems that zoned format can accommodate 15 digits + sign
> at most. Suppose I have this 16 bytes zoned number:
> f9f9f9f9f9f9f9f9f9f9f9f9f9f9f9xx. Converting it to packed bcd will
> give 999999999999999x. Do you agree? can I keep that ox_flag logic?

Uh.. no.  AFAICT from the description of the instruction the last byte
of a zoned number contains *both* a sign nibble *and* the least
significant digit of the number.  So zoned format can represent 16
digits + sign.

But IIRC you changed this in v2, so maybe you realised that already.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]