[Mesa-dev] [PATCH 59/59] i965/fs: fix MOV_INDIRECT exec_size for doubles

[Jordan Justen]

On 2016-05-02 00:31:41, Iago Toral wrote:
> On Sun, 2016-05-01 at 20:04 -0700, Jordan Justen wrote:
> > On 2016-04-29 04:29:56, Samuel Iglesias Gonsálvez wrote:
> > > In that case, the writes need two times the size of a 32-bit value.
> > > We need to adjust the exec_size, so it is not breaking any hardware
> > > rule.
> > > 
> > > Signed-off-by: Samuel Iglesias Gonsálvez <siglesias at igalia.com>
> > > ---
> > >  src/mesa/drivers/dri/i965/brw_fs.cpp | 8 +++++++-
> > >  1 file changed, 7 insertions(+), 1 deletion(-)
> > > 
> > > diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
> > > index 059edc6..85d430c 100644
> > > --- a/src/mesa/drivers/dri/i965/brw_fs.cpp
> > > +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
> > > @@ -4655,7 +4655,13 @@ get_lowered_simd_width(const struct brw_device_info *devinfo,
> > >  
> > >     case SHADER_OPCODE_MOV_INDIRECT:
> > >        /* Prior to Broadwell, we only have 8 address subregisters */
> > > -      return devinfo->gen < 8 ? 8 : MIN2(inst->exec_size, 16);
> > > +      if (devinfo->gen < 8)
> > > +         return 8;
> > > +
> > > +      if (inst->exec_size < 16)
> > > +         return inst->exec_size;
> > > +      else
> > > +         return MIN2(inst->exec_size / (type_sz(inst->dst.type) / 4), 16);
> > 
> > Can't type_sz return < 4, which could cause a divide by 0? If that
> > can't happen along this code path, would an assert be a good idea?
> 
> Looking at the places where the driver inserts MOV INDIRECT opcodes it
> seems that it always uses region sizes that are multiples of 4 and dst
> types are usually UD (except in load uniform where I don't think we can
> see any types that are not at least 4 bytes, since we implement bool as
> UD too), so it seems to me that we shouldn't be seeing types with sizes
> < 4 here.
> 
> That said, I am adding Jason in the CC just in case I am missing
> something.
> 
> In any case, adding the assert for extra safety sounds good to me.
> 

Ok, with the assert: Reviewed-by: Jordan Justen <jordan.l.justen at intel.com>