[Mesa-dev] [PATCH 07/17] i965/fs: Move offset() and half() to the fs_builder

Jason Ekstrand jason at jlekstrand.net
Wed Jun 24 06:38:41 PDT 2015


On Jun 24, 2015 6:29 AM, "Francisco Jerez" <currojerez at riseup.net> wrote:
>
> Jason Ekstrand <jason at jlekstrand.net> writes:
>
> > On Jun 24, 2015 4:29 AM, "Francisco Jerez" <currojerez at riseup.net>
wrote:
> >>
> >> Jason Ekstrand <jason at jlekstrand.net> writes:
> >>
> >> > On Tue, Jun 23, 2015 at 9:22 AM, Francisco Jerez <
currojerez at riseup.net>
> > wrote:
> >> >> Jason Ekstrand <jason at jlekstrand.net> writes:
> >> >>
> >> >>> We want to move these into the builder so that they know the
current
> >> >>> builder's dispatch width.  This will be needed by a later commit.
> >> >>
> >> >> I very much like the idea of this series, but, why do you need to
move
> >> >> these register manipulators into the builder?  The builder is an
object
> >> >> you can use to:
> >> >>  - Manipulate and query parameters affecting code generation.
> >> >>  - Create instructions into the program (::emit and friends).
> >> >>  - Allocate virtual registers from the program (::vgrf and friends).
> >> >>
> >> >> offset() and half() logically perform an action on a given register
> >> >> object (or rather, compute a function of a given register object),
not
> >> >> on a builder object, the builder is only required as an auxiliary
> >> >> parameter -- Any reason you didn't just pass it as a third
parameter?
> >> >
> >> > What's required as a third parameter is the current execution size.
I
> >> > could have passed that directly, but I figured that, especially for
> >> > half(), it would get messed up.  I could pass the builder in but I
> >> > don't see a whole lot of difference between that and what I'm doing
> >> > right now.
> >>
> >> Assembly-wise there's no difference, but it seems inconsistent with
both
> >> the remaining register manipulators and remaining builder methods, and
> >> IMHO it's kind of an anti-pattern to make something a method that
> >> doesn't need access to any internal details of the object.
> >>
> >> > As is, it's not entirely obvious whether you should call
> >> > half(reg) on the half-width or full-width builder.  I'm not 100% sure
> >> > what to do about that.
> >> >
> >> Actually, does half() really need to know about the builder?  AFAICT it
> >> only needs it because of dispatch_width(), and before doing anything
> >> useful with it it asserts that it's equal to 16, what points at the
> >> parameter being redundant.  By convention a "half" is a group of 8
> >> channels (we may want to revise this convention when we implement
SIMD32
> >> -- E.g. make half a group of 16 channels and quarter a group of 8
> >> channels), so 'half(reg)' could simply be implemented as
> >> "horiz_offset(reg, 8 * i)" without any dependency on the builder.  As
> >> additional paranoia to catch half() being called on a non-16-aligned
> >> register you could assert that either 'stride == 0' or 16 divides
> >> '(REG_SIZE * reg_offset + subreg_offset) / (stride * type_size)' (why
> >> don't we have a reg_offset already in bytes again?) -- That would also
> >> catch cases in which the register and builder "widths" get out of sync,
> >> e.g. if half is called in an already halved register but the builder
> >> used happens to be of the correct exec_size.
> >
> > OK, fine, we can pull half() back out.  Should offset() stay in the
> > builder? If not, where should it get its dispatch width.
> >
> I'm for leaving it as a stand-alone function (like all other register
> manipulators), and add a third argument to pass the 'fs_builder' it can
> take the dispatch width from?

I'm not a big fan.  However, in the interest of keeping the builder clean,
I'm willing to go with that.
--Jason

> >> >> As offset() and half() don't require access to any private details
of
> >> >> the builder, that would actually improve encapsulation, and would
avoid
> >> >> the dubious overloading of fs_builder::half() with two methods with
> >> >> completely different semantics.
> >> >
> >> > Yeah, I don't really like that either.  I just couldn't come up with
> >> > anything better at the time.
> >> >
> >> > Suggestions are very much welcome.  But I would like to settle on
> >> > whatever we do fairly quickly so as to limit the amount of
> >> > refactoring.
> >> > --Jason
> >> >
> >> >> Thanks.
> >> >>
> >> >>> ---
> >> >>>  src/mesa/drivers/dri/i965/brw_fs.cpp         |  52 ++++++----
> >> >>>  src/mesa/drivers/dri/i965/brw_fs_builder.h   |  46 +++++++++
> >> >>>  src/mesa/drivers/dri/i965/brw_fs_cse.cpp     |   2 +-
> >> >>>  src/mesa/drivers/dri/i965/brw_fs_nir.cpp     |  60 +++++------
> >> >>>  src/mesa/drivers/dri/i965/brw_fs_visitor.cpp | 149
> > ++++++++++++++-------------
> >> >>>  src/mesa/drivers/dri/i965/brw_ir_fs.h        |  51 ---------
> >> >>>  6 files changed, 182 insertions(+), 178 deletions(-)
> >> >>>
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp
> > b/src/mesa/drivers/dri/i965/brw_fs.cpp
> >> >>> index 4f98d63..c13ac7d 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_fs.cpp
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
> >> >>> @@ -267,7 +267,7 @@ fs_visitor::VARYING_PULL_CONSTANT_LOAD(const
> > fs_builder &bld,
> >> >>>           inst->mlen = 1 + dispatch_width / 8;
> >> >>>     }
> >> >>>
> >> >>> -   bld.MOV(dst, offset(vec4_result, (const_offset & 3) * scale));
> >> >>> +   bld.MOV(dst, bld.offset(vec4_result, (const_offset & 3) *
scale));
> >> >>>  }
> >> >>>
> >> >>>  /**
> >> >>> @@ -361,7 +361,12 @@ fs_inst::is_copy_payload(const
> > brw::simple_allocator &grf_alloc) const
> >> >>>        reg.width = this->src[i].width;
> >> >>>        if (!this->src[i].equals(reg))
> >> >>>           return false;
> >> >>> -      reg = ::offset(reg, 1);
> >> >>> +
> >> >>> +      if (i < this->header_size) {
> >> >>> +         reg.reg_offset += 1;
> >> >>> +      } else {
> >> >>> +         reg.reg_offset += this->exec_size / 8;
> >> >>> +      }
> >> >>>     }
> >> >>>
> >> >>>     return true;
> >> >>> @@ -963,7 +968,7 @@ fs_visitor::emit_fragcoord_interpolation(bool
> > pixel_center_integer,
> >> >>>     } else {
> >> >>>        bld.ADD(wpos, this->pixel_x, fs_reg(0.5f));
> >> >>>     }
> >> >>> -   wpos = offset(wpos, 1);
> >> >>> +   wpos = bld.offset(wpos, 1);
> >> >>>
> >> >>>     /* gl_FragCoord.y */
> >> >>>     if (!flip && pixel_center_integer) {
> >> >>> @@ -979,7 +984,7 @@ fs_visitor::emit_fragcoord_interpolation(bool
> > pixel_center_integer,
> >> >>>
> >> >>>        bld.ADD(wpos, pixel_y, fs_reg(offset));
> >> >>>     }
> >> >>> -   wpos = offset(wpos, 1);
> >> >>> +   wpos = bld.offset(wpos, 1);
> >> >>>
> >> >>>     /* gl_FragCoord.z */
> >> >>>     if (devinfo->gen >= 6) {
> >> >>> @@ -989,7 +994,7 @@ fs_visitor::emit_fragcoord_interpolation(bool
> > pixel_center_integer,
> >> >>>             this->delta_xy[BRW_WM_PERSPECTIVE_PIXEL_BARYCENTRIC],
> >> >>>             interp_reg(VARYING_SLOT_POS, 2));
> >> >>>     }
> >> >>> -   wpos = offset(wpos, 1);
> >> >>> +   wpos = bld.offset(wpos, 1);
> >> >>>
> >> >>>     /* gl_FragCoord.w: Already set up in emit_interpolation */
> >> >>>     bld.MOV(wpos, this->wpos_w);
> >> >>> @@ -1072,7 +1077,7 @@ fs_visitor::emit_general_interpolation(fs_reg
> > attr, const char *name,
> >> >>>           /* If there's no incoming setup data for this slot, don't
> >> >>>            * emit interpolation for it.
> >> >>>            */
> >> >>> -         attr = offset(attr, type->vector_elements);
> >> >>> +         attr = bld.offset(attr, type->vector_elements);
> >> >>>           location++;
> >> >>>           continue;
> >> >>>        }
> >> >>> @@ -1087,7 +1092,7 @@ fs_visitor::emit_general_interpolation(fs_reg
> > attr, const char *name,
> >> >>>              interp = suboffset(interp, 3);
> >> >>>                 interp.type = attr.type;
> >> >>>                 bld.emit(FS_OPCODE_CINTERP, attr, fs_reg(interp));
> >> >>> -            attr = offset(attr, 1);
> >> >>> +            attr = bld.offset(attr, 1);
> >> >>>           }
> >> >>>        } else {
> >> >>>           /* Smooth/noperspective interpolation case. */
> >> >>> @@ -1125,7 +1130,7 @@ fs_visitor::emit_general_interpolation(fs_reg
> > attr, const char *name,
> >> >>>                 if (devinfo->gen < 6 && interpolation_mode ==
> > INTERP_QUALIFIER_SMOOTH) {
> >> >>>                    bld.MUL(attr, attr, this->pixel_w);
> >> >>>                 }
> >> >>> -            attr = offset(attr, 1);
> >> >>> +            attr = bld.offset(attr, 1);
> >> >>>           }
> >> >>>
> >> >>>        }
> >> >>> @@ -1227,19 +1232,19 @@ fs_visitor::emit_samplepos_setup()
> >> >>>     if (dispatch_width == 8) {
> >> >>>        abld.MOV(int_sample_x, fs_reg(sample_pos_reg));
> >> >>>     } else {
> >> >>> -      abld.half(0).MOV(half(int_sample_x, 0),
> > fs_reg(sample_pos_reg));
> >> >>> -      abld.half(1).MOV(half(int_sample_x, 1),
> >> >>> +      abld.half(0).MOV(abld.half(int_sample_x, 0),
> > fs_reg(sample_pos_reg));
> >> >>> +      abld.half(1).MOV(abld.half(int_sample_x, 1),
> >> >>>                         fs_reg(suboffset(sample_pos_reg, 16)));
> >> >>>     }
> >> >>>     /* Compute gl_SamplePosition.x */
> >> >>>     compute_sample_position(pos, int_sample_x);
> >> >>> -   pos = offset(pos, 1);
> >> >>> +   pos = abld.offset(pos, 1);
> >> >>>     if (dispatch_width == 8) {
> >> >>>        abld.MOV(int_sample_y, fs_reg(suboffset(sample_pos_reg,
1)));
> >> >>>     } else {
> >> >>> -      abld.half(0).MOV(half(int_sample_y, 0),
> >> >>> +      abld.half(0).MOV(abld.half(int_sample_y, 0),
> >> >>>                         fs_reg(suboffset(sample_pos_reg, 1)));
> >> >>> -      abld.half(1).MOV(half(int_sample_y, 1),
> >> >>> +      abld.half(1).MOV(abld.half(int_sample_y, 1),
> >> >>>                         fs_reg(suboffset(sample_pos_reg, 17)));
> >> >>>     }
> >> >>>     /* Compute gl_SamplePosition.y */
> >> >>> @@ -3018,10 +3023,6 @@ fs_visitor::lower_load_payload()
> >> >>>
> >> >>>        assert(inst->dst.file == MRF || inst->dst.file == GRF);
> >> >>>        assert(inst->saturate == false);
> >> >>> -
> >> >>> -      const fs_builder ibld = bld.group(inst->exec_size,
> > inst->force_sechalf)
> >> >>> -
 .exec_all(inst->force_writemask_all)
> >> >>> -                                 .at(block, inst);
> >> >>>        fs_reg dst = inst->dst;
> >> >>>
> >> >>>        /* Get rid of COMPR4.  We'll add it back in if we need it */
> >> >>> @@ -3029,17 +3030,23 @@ fs_visitor::lower_load_payload()
> >> >>>           dst.reg = dst.reg & ~BRW_MRF_COMPR4;
> >> >>>
> >> >>>        dst.width = 8;
> >> >>> +      const fs_builder hbld = bld.group(8, 0).exec_all().at(block,
> > inst);
> >> >>> +
> >> >>>        for (uint8_t i = 0; i < inst->header_size; i++) {
> >> >>>           if (inst->src[i].file != BAD_FILE) {
> >> >>>              fs_reg mov_dst = retype(dst, BRW_REGISTER_TYPE_UD);
> >> >>>              fs_reg mov_src = retype(inst->src[i],
> > BRW_REGISTER_TYPE_UD);
> >> >>>              mov_src.width = 8;
> >> >>> -            ibld.exec_all().MOV(mov_dst, mov_src);
> >> >>> +            hbld.MOV(mov_dst, mov_src);
> >> >>>           }
> >> >>> -         dst = offset(dst, 1);
> >> >>> +         dst = hbld.offset(dst, 1);
> >> >>>        }
> >> >>>
> >> >>>        dst.width = inst->exec_size;
> >> >>> +      const fs_builder ibld = bld.group(inst->exec_size,
> > inst->force_sechalf)
> >> >>> +
 .exec_all(inst->force_writemask_all)
> >> >>> +                                 .at(block, inst);
> >> >>> +
> >> >>>        if (inst->dst.file == MRF && (inst->dst.reg &
BRW_MRF_COMPR4)
> > &&
> >> >>>            inst->exec_size > 8) {
> >> >>>           /* In this case, the payload portion of the LOAD_PAYLOAD
> > isn't
> >> >>> @@ -3070,8 +3077,9 @@ fs_visitor::lower_load_payload()
> >> >>>                    /* Platform doesn't have COMPR4.  We have to
fake
> > it */
> >> >>>                    fs_reg mov_dst = retype(dst, inst->src[i].type);
> >> >>>                    mov_dst.width = 8;
> >> >>> -                  ibld.half(0).MOV(mov_dst, half(inst->src[i],
0));
> >> >>> -                  ibld.half(1).MOV(offset(mov_dst, 4),
> > half(inst->src[i], 1));
> >> >>> +                  ibld.half(0).MOV(mov_dst,
ibld.half(inst->src[i],
> > 0));
> >> >>> +                  mov_dst.reg += 4;
> >> >>> +                  ibld.half(1).MOV(mov_dst,
ibld.half(inst->src[i],
> > 1));
> >> >>>                 }
> >> >>>              }
> >> >>>
> >> >>> @@ -3096,7 +3104,7 @@ fs_visitor::lower_load_payload()
> >> >>>        for (uint8_t i = inst->header_size; i < inst->sources; i++)
{
> >> >>>           if (inst->src[i].file != BAD_FILE)
> >> >>>              ibld.MOV(retype(dst, inst->src[i].type),
inst->src[i]);
> >> >>> -         dst = offset(dst, 1);
> >> >>> +         dst = ibld.offset(dst, 1);
> >> >>>        }
> >> >>>
> >> >>>        inst->remove(block);
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_fs_builder.h
> > b/src/mesa/drivers/dri/i965/brw_fs_builder.h
> >> >>> index 58ac598..594e252 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_fs_builder.h
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_fs_builder.h
> >> >>> @@ -151,6 +151,52 @@ namespace brw {
> >> >>>           return _dispatch_width;
> >> >>>        }
> >> >>>
> >> >>> +      src_reg
> >> >>> +      offset(src_reg reg, unsigned delta) const
> >> >>> +      {
> >> >>> +         switch (reg.file) {
> >> >>> +         case BAD_FILE:
> >> >>> +            break;
> >> >>> +         case GRF:
> >> >>> +         case MRF:
> >> >>> +         case ATTR:
> >> >>> +            return byte_offset(reg,
> >> >>> +                               delta * MAX2(reg.width *
reg.stride,
> > 1) *
> >> >>> +                               type_sz(reg.type));
> >> >>> +         case UNIFORM:
> >> >>> +            reg.reg_offset += delta;
> >> >>> +            break;
> >> >>> +         default:
> >> >>> +            assert(delta == 0);
> >> >>> +         }
> >> >>> +         return reg;
> >> >>> +      }
> >> >>> +
> >> >>> +      fs_reg
> >> >>> +      half(fs_reg reg, unsigned idx) const
> >> >>> +      {
> >> >>> +         assert(idx < 2);
> >> >>> +
> >> >>> +         switch (reg.file) {
> >> >>> +         case BAD_FILE:
> >> >>> +         case UNIFORM:
> >> >>> +         case IMM:
> >> >>> +            return reg;
> >> >>> +
> >> >>> +         case GRF:
> >> >>> +         case MRF:
> >> >>> +            assert(reg.width == 16);
> >> >>> +            reg.width = 8;
> >> >>> +            return horiz_offset(reg, 8 * idx);
> >> >>> +
> >> >>> +         case ATTR:
> >> >>> +         case HW_REG:
> >> >>> +         default:
> >> >>> +            unreachable("Cannot take half of this register type");
> >> >>> +         }
> >> >>> +         return reg;
> >> >>> +      }
> >> >>> +
> >> >>>        /**
> >> >>>         * Allocate a virtual register of natural vector size (one
for
> > this IR)
> >> >>>         * and SIMD width.  \p n gives the amount of space to
allocate
> > in
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
> > b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
> >> >>> index 70f0217..5ea66c5 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_fs_cse.cpp
> >> >>> @@ -205,7 +205,7 @@ create_copy_instr(const fs_builder &bld,
fs_inst
> > *inst, fs_reg src, bool negate)
> >> >>>        }
> >> >>>        for (int i = header_size; i < sources; i++) {
> >> >>>           payload[i] = src;
> >> >>> -         src = offset(src, 1);
> >> >>> +         src = ubld.offset(src, 1);
> >> >>>        }
> >> >>>        copy = ubld.LOAD_PAYLOAD(inst->dst, payload, sources,
> > header_size);
> >> >>>     } else {
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> > b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> >> >>> index 59081ea..0ede634 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> >> >>> @@ -76,7 +76,7 @@ fs_visitor::nir_setup_inputs(nir_shader *shader)
> >> >>>  {
> >> >>>     foreach_list_typed(nir_variable, var, node, &shader->inputs) {
> >> >>>        enum brw_reg_type type = brw_type_for_base_type(var->type);
> >> >>> -      fs_reg input = offset(nir_inputs,
var->data.driver_location);
> >> >>> +      fs_reg input = bld.offset(nir_inputs,
> > var->data.driver_location);
> >> >>>
> >> >>>        fs_reg reg;
> >> >>>        switch (stage) {
> >> >>> @@ -95,8 +95,8 @@ fs_visitor::nir_setup_inputs(nir_shader *shader)
> >> >>>           unsigned array_length = var->type->is_array() ?
> > var->type->length : 1;
> >> >>>           for (unsigned i = 0; i < array_length; i++) {
> >> >>>              for (unsigned j = 0; j < components; j++) {
> >> >>> -               bld.MOV(retype(offset(input, components * i + j),
> > type),
> >> >>> -                       offset(fs_reg(ATTR, var->data.location + i,
> > type), j));
> >> >>> +               bld.MOV(retype(bld.offset(input, components * i +
j),
> > type),
> >> >>> +                       bld.offset(fs_reg(ATTR, var->data.location
+
> > i, type), j));
> >> >>>              }
> >> >>>           }
> >> >>>           break;
> >> >>> @@ -127,7 +127,7 @@ fs_visitor::nir_setup_outputs(nir_shader
*shader)
> >> >>>     brw_wm_prog_key *key = (brw_wm_prog_key*) this->key;
> >> >>>
> >> >>>     foreach_list_typed(nir_variable, var, node, &shader->outputs) {
> >> >>> -      fs_reg reg = offset(nir_outputs, var->data.driver_location);
> >> >>> +      fs_reg reg = bld.offset(nir_outputs,
> > var->data.driver_location);
> >> >>>
> >> >>>        int vector_elements =
> >> >>>           var->type->is_array() ?
> > var->type->fields.array->vector_elements
> >> >>> @@ -136,7 +136,7 @@ fs_visitor::nir_setup_outputs(nir_shader
*shader)
> >> >>>        if (stage == MESA_SHADER_VERTEX) {
> >> >>>           for (int i = 0; i < ALIGN(type_size(var->type), 4) / 4;
> > i++) {
> >> >>>              int output = var->data.location + i;
> >> >>> -            this->outputs[output] = offset(reg, 4 * i);
> >> >>> +            this->outputs[output] = bld.offset(reg, 4 * i);
> >> >>>              this->output_components[output] = vector_elements;
> >> >>>           }
> >> >>>        } else if (var->data.index > 0) {
> >> >>> @@ -162,7 +162,7 @@ fs_visitor::nir_setup_outputs(nir_shader
*shader)
> >> >>>           /* General color output. */
> >> >>>           for (unsigned int i = 0; i < MAX2(1, var->type->length);
> > i++) {
> >> >>>              int output = var->data.location - FRAG_RESULT_DATA0 +
i;
> >> >>> -            this->outputs[output] = offset(reg, vector_elements *
i);
> >> >>> +            this->outputs[output] = bld.offset(reg,
vector_elements
> > * i);
> >> >>>              this->output_components[output] = vector_elements;
> >> >>>           }
> >> >>>        }
> >> >>> @@ -618,11 +618,11 @@ fs_visitor::nir_emit_alu(const fs_builder
&bld,
> > nir_alu_instr *instr)
> >> >>>              continue;
> >> >>>
> >> >>>           if (instr->op == nir_op_imov || instr->op ==
nir_op_fmov) {
> >> >>> -            inst = bld.MOV(offset(temp, i),
> >> >>> -                           offset(op[0],
instr->src[0].swizzle[i]));
> >> >>> +            inst = bld.MOV(bld.offset(temp, i),
> >> >>> +                           bld.offset(op[0],
> > instr->src[0].swizzle[i]));
> >> >>>           } else {
> >> >>> -            inst = bld.MOV(offset(temp, i),
> >> >>> -                           offset(op[i],
instr->src[i].swizzle[0]));
> >> >>> +            inst = bld.MOV(bld.offset(temp, i),
> >> >>> +                           bld.offset(op[i],
> > instr->src[i].swizzle[0]));
> >> >>>           }
> >> >>>           inst->saturate = instr->dest.saturate;
> >> >>>        }
> >> >>> @@ -636,7 +636,7 @@ fs_visitor::nir_emit_alu(const fs_builder &bld,
> > nir_alu_instr *instr)
> >> >>>              if (!(instr->dest.write_mask & (1 << i)))
> >> >>>                 continue;
> >> >>>
> >> >>> -            bld.MOV(offset(result, i), offset(temp, i));
> >> >>> +            bld.MOV(bld.offset(result, i), bld.offset(temp, i));
> >> >>>           }
> >> >>>        }
> >> >>>        return;
> >> >>> @@ -657,12 +657,12 @@ fs_visitor::nir_emit_alu(const fs_builder
&bld,
> > nir_alu_instr *instr)
> >> >>>        assert(_mesa_bitcount(instr->dest.write_mask) == 1);
> >> >>>        channel = ffs(instr->dest.write_mask) - 1;
> >> >>>
> >> >>> -      result = offset(result, channel);
> >> >>> +      result = bld.offset(result, channel);
> >> >>>     }
> >> >>>
> >> >>>     for (unsigned i = 0; i < nir_op_infos[instr->op].num_inputs;
i++)
> > {
> >> >>>        assert(nir_op_infos[instr->op].input_sizes[i] < 2);
> >> >>> -      op[i] = offset(op[i], instr->src[i].swizzle[channel]);
> >> >>> +      op[i] = bld.offset(op[i], instr->src[i].swizzle[channel]);
> >> >>>     }
> >> >>>
> >> >>>     switch (instr->op) {
> >> >>> @@ -1156,7 +1156,7 @@ fs_reg_for_nir_reg(fs_visitor *v,
nir_register
> > *nir_reg,
> >> >>>     else
> >> >>>        reg = v->nir_locals[nir_reg->index];
> >> >>>
> >> >>> -   reg = offset(reg, base_offset * nir_reg->num_components);
> >> >>> +   reg = v->bld.offset(reg, base_offset *
nir_reg->num_components);
> >> >>>     if (indirect) {
> >> >>>        int multiplier = nir_reg->num_components *
(v->dispatch_width
> > / 8);
> >> >>>
> >> >>> @@ -1177,7 +1177,7 @@ fs_visitor::get_nir_src(nir_src src)
> >> >>>        fs_reg reg = bld.vgrf(BRW_REGISTER_TYPE_D,
> > src.ssa->num_components);
> >> >>>
> >> >>>        for (unsigned i = 0; i < src.ssa->num_components; ++i)
> >> >>> -         bld.MOV(offset(reg, i), fs_reg(load->value.i[i]));
> >> >>> +         bld.MOV(bld.offset(reg, i), fs_reg(load->value.i[i]));
> >> >>>
> >> >>>        return reg;
> >> >>>     } else {
> >> >>> @@ -1208,10 +1208,10 @@ fs_visitor::emit_percomp(const fs_builder
> > &bld, const fs_inst &inst,
> >> >>>           continue;
> >> >>>
> >> >>>        fs_inst *new_inst = new(mem_ctx) fs_inst(inst);
> >> >>> -      new_inst->dst = offset(new_inst->dst, i);
> >> >>> +      new_inst->dst = bld.offset(new_inst->dst, i);
> >> >>>        for (unsigned j = 0; j < new_inst->sources; j++)
> >> >>>           if (new_inst->src[j].file == GRF)
> >> >>> -            new_inst->src[j] = offset(new_inst->src[j], i);
> >> >>> +            new_inst->src[j] = bld.offset(new_inst->src[j], i);
> >> >>>
> >> >>>        bld.emit(new_inst);
> >> >>>     }
> >> >>> @@ -1322,7 +1322,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>        assert(sample_pos.file != BAD_FILE);
> >> >>>        dest.type = sample_pos.type;
> >> >>>        bld.MOV(dest, sample_pos);
> >> >>> -      bld.MOV(offset(dest, 1), offset(sample_pos, 1));
> >> >>> +      bld.MOV(bld.offset(dest, 1), bld.offset(sample_pos, 1));
> >> >>>        break;
> >> >>>     }
> >> >>>
> >> >>> @@ -1349,13 +1349,13 @@ fs_visitor::nir_emit_intrinsic(const
> > fs_builder &bld, nir_intrinsic_instr *instr
> >> >>>        }
> >> >>>
> >> >>>        for (unsigned j = 0; j < instr->num_components; j++) {
> >> >>> -         fs_reg src = offset(retype(uniform_reg, dest.type),
index);
> >> >>> +         fs_reg src = bld.offset(retype(uniform_reg, dest.type),
> > index);
> >> >>>           if (has_indirect)
> >> >>>              src.reladdr = new(mem_ctx)
> > fs_reg(get_nir_src(instr->src[0]));
> >> >>>           index++;
> >> >>>
> >> >>>           bld.MOV(dest, src);
> >> >>> -         dest = offset(dest, 1);
> >> >>> +         dest = bld.offset(dest, 1);
> >> >>>        }
> >> >>>        break;
> >> >>>     }
> >> >>> @@ -1397,7 +1397,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>
> >> >>>           unsigned vec4_offset = instr->const_index[0] / 4;
> >> >>>           for (int i = 0; i < instr->num_components; i++)
> >> >>> -            VARYING_PULL_CONSTANT_LOAD(bld, offset(dest, i),
> > surf_index,
> >> >>> +            VARYING_PULL_CONSTANT_LOAD(bld, bld.offset(dest, i),
> > surf_index,
> >> >>>                                         base_offset, vec4_offset +
i);
> >> >>>        } else {
> >> >>>           fs_reg packed_consts = vgrf(glsl_type::float_type);
> >> >>> @@ -1416,7 +1416,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>              assert(packed_consts.subreg_offset < 32);
> >> >>>
> >> >>>              bld.MOV(dest, packed_consts);
> >> >>> -            dest = offset(dest, 1);
> >> >>> +            dest = bld.offset(dest, 1);
> >> >>>           }
> >> >>>        }
> >> >>>        break;
> >> >>> @@ -1428,14 +1428,14 @@ fs_visitor::nir_emit_intrinsic(const
> > fs_builder &bld, nir_intrinsic_instr *instr
> >> >>>     case nir_intrinsic_load_input: {
> >> >>>        unsigned index = 0;
> >> >>>        for (unsigned j = 0; j < instr->num_components; j++) {
> >> >>> -         fs_reg src = offset(retype(nir_inputs, dest.type),
> >> >>> +         fs_reg src = bld.offset(retype(nir_inputs, dest.type),
> >> >>>                               instr->const_index[0] + index);
> >> >>>           if (has_indirect)
> >> >>>              src.reladdr = new(mem_ctx)
> > fs_reg(get_nir_src(instr->src[0]));
> >> >>>           index++;
> >> >>>
> >> >>>           bld.MOV(dest, src);
> >> >>> -         dest = offset(dest, 1);
> >> >>> +         dest = bld.offset(dest, 1);
> >> >>>        }
> >> >>>        break;
> >> >>>     }
> >> >>> @@ -1508,7 +1508,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>                                         BRW_REGISTER_TYPE_F);
> >> >>>              for (int i = 0; i < 2; i++) {
> >> >>>                 fs_reg temp = vgrf(glsl_type::float_type);
> >> >>> -               bld.MUL(temp, offset(offset_src, i),
fs_reg(16.0f));
> >> >>> +               bld.MUL(temp, bld.offset(offset_src, i),
> > fs_reg(16.0f));
> >> >>>                 fs_reg itemp = vgrf(glsl_type::int_type);
> >> >>>                 bld.MOV(itemp, temp);  /* float to int */
> >> >>>
> >> >>> @@ -1528,7 +1528,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>                  * FRAGMENT_INTERPOLATION_OFFSET_BITS"
> >> >>>                  */
> >> >>>                 set_condmod(BRW_CONDITIONAL_L,
> >> >>> -                           bld.SEL(offset(src, i), itemp,
> > fs_reg(7)));
> >> >>> +                           bld.SEL(bld.offset(src, i), itemp,
> > fs_reg(7)));
> >> >>>              }
> >> >>>
> >> >>>              mlen = 2;
> >> >>> @@ -1552,7 +1552,7 @@ fs_visitor::nir_emit_intrinsic(const
fs_builder
> > &bld, nir_intrinsic_instr *instr
> >> >>>           src.type = dest.type;
> >> >>>
> >> >>>           bld.emit(FS_OPCODE_LINTERP, dest, dst_xy, src);
> >> >>> -         dest = offset(dest, 1);
> >> >>> +         dest = bld.offset(dest, 1);
> >> >>>        }
> >> >>>        break;
> >> >>>     }
> >> >>> @@ -1564,13 +1564,13 @@ fs_visitor::nir_emit_intrinsic(const
> > fs_builder &bld, nir_intrinsic_instr *instr
> >> >>>        fs_reg src = get_nir_src(instr->src[0]);
> >> >>>        unsigned index = 0;
> >> >>>        for (unsigned j = 0; j < instr->num_components; j++) {
> >> >>> -         fs_reg new_dest = offset(retype(nir_outputs, src.type),
> >> >>> -                                  instr->const_index[0] + index);
> >> >>> +         fs_reg new_dest = bld.offset(retype(nir_outputs,
src.type),
> >> >>> +                                      instr->const_index[0] +
index);
> >> >>>           if (has_indirect)
> >> >>>              src.reladdr = new(mem_ctx)
> > fs_reg(get_nir_src(instr->src[1]));
> >> >>>           index++;
> >> >>>           bld.MOV(new_dest, src);
> >> >>> -         src = offset(src, 1);
> >> >>> +         src = bld.offset(src, 1);
> >> >>>        }
> >> >>>        break;
> >> >>>     }
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> > b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> >> >>> index 8a43ec8..68cd454 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
> >> >>> @@ -95,7 +95,7 @@ fs_visitor::emit_texture_gen4(ir_texture_opcode
op,
> > fs_reg dst,
> >> >>>     if (shadow_c.file != BAD_FILE) {
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen + i), coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        }
> >> >>>
> >> >>>        /* gen4's SIMD8 sampler always has the slots for u,v,r
present.
> >> >>> @@ -124,7 +124,7 @@ fs_visitor::emit_texture_gen4(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>     } else if (op == ir_tex) {
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen + i), coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        }
> >> >>>        /* zero the others. */
> >> >>>        for (int i = coord_components; i<3; i++) {
> >> >>> @@ -137,7 +137,7 @@ fs_visitor::emit_texture_gen4(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen + i), coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        }
> >> >>>        /* the slots for u and v are always present, but r is
optional
> > */
> >> >>>        mlen += MAX2(coord_components, 2);
> >> >>> @@ -158,13 +158,13 @@
fs_visitor::emit_texture_gen4(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>         */
> >> >>>        for (int i = 0; i < grad_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen), dPdx);
> >> >>> -      dPdx = offset(dPdx, 1);
> >> >>> +      dPdx = bld.offset(dPdx, 1);
> >> >>>        }
> >> >>>        mlen += MAX2(grad_components, 2);
> >> >>>
> >> >>>        for (int i = 0; i < grad_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen), dPdy);
> >> >>> -      dPdy = offset(dPdy, 1);
> >> >>> +      dPdy = bld.offset(dPdy, 1);
> >> >>>        }
> >> >>>        mlen += MAX2(grad_components, 2);
> >> >>>     } else if (op == ir_txs) {
> >> >>> @@ -182,7 +182,7 @@ fs_visitor::emit_texture_gen4(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(fs_reg(MRF, base_mrf + mlen + i * 2,
> > coordinate.type),
> >> >>>                   coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        }
> >> >>>
> >> >>>        /* Initialize the rest of u/v/r with 0.0.  Empirically, this
> > seems to
> >> >>> @@ -232,8 +232,8 @@ fs_visitor::emit_texture_gen4(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>     if (simd16) {
> >> >>>        for (int i = 0; i < 4; i++) {
> >> >>>           bld.MOV(orig_dst, dst);
> >> >>> -      orig_dst = offset(orig_dst, 1);
> >> >>> -      dst = offset(dst, 2);
> >> >>> +      orig_dst = bld.offset(orig_dst, 1);
> >> >>> +      dst = bld.offset(dst, 2);
> >> >>>        }
> >> >>>     }
> >> >>>
> >> >>> @@ -257,30 +257,30 @@
> > fs_visitor::emit_texture_gen4_simd16(ir_texture_opcode op, fs_reg dst,
> >> >>>
> >> >>>     /* Copy the coordinates. */
> >> >>>     for (int i = 0; i < vector_elements; i++) {
> >> >>> -      bld.MOV(retype(offset(message, i), coordinate.type),
> > coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      bld.MOV(retype(bld.offset(message, i), coordinate.type),
> > coordinate);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>     }
> >> >>>
> >> >>> -   fs_reg msg_end = offset(message, vector_elements);
> >> >>> +   fs_reg msg_end = bld.offset(message, vector_elements);
> >> >>>
> >> >>>     /* Messages other than sample and ld require all three
components
> > */
> >> >>>     if (has_lod || shadow_c.file != BAD_FILE) {
> >> >>>        for (int i = vector_elements; i < 3; i++) {
> >> >>> -         bld.MOV(offset(message, i), fs_reg(0.0f));
> >> >>> +         bld.MOV(bld.offset(message, i), fs_reg(0.0f));
> >> >>>        }
> >> >>>     }
> >> >>>
> >> >>>     if (has_lod) {
> >> >>> -      fs_reg msg_lod = retype(offset(message, 3), op == ir_txf ?
> >> >>> +      fs_reg msg_lod = retype(bld.offset(message, 3), op ==
ir_txf ?
> >> >>>                                BRW_REGISTER_TYPE_UD :
> > BRW_REGISTER_TYPE_F);
> >> >>>        bld.MOV(msg_lod, lod);
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>     }
> >> >>>
> >> >>>     if (shadow_c.file != BAD_FILE) {
> >> >>> -      fs_reg msg_ref = offset(message, 3 + has_lod);
> >> >>> +      fs_reg msg_ref = bld.offset(message, 3 + has_lod);
> >> >>>        bld.MOV(msg_ref, shadow_c);
> >> >>> -      msg_end = offset(msg_ref, 1);
> >> >>> +      msg_end = bld.offset(msg_ref, 1);
> >> >>>     }
> >> >>>
> >> >>>     enum opcode opcode;
> >> >>> @@ -334,16 +334,16 @@
fs_visitor::emit_texture_gen5(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>     }
> >> >>>
> >> >>>     for (int i = 0; i < vector_elements; i++) {
> >> >>> -      bld.MOV(retype(offset(msg_coords, i), coordinate.type),
> > coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      bld.MOV(retype(bld.offset(msg_coords, i), coordinate.type),
> > coordinate);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>     }
> >> >>> -   fs_reg msg_end = offset(msg_coords, vector_elements);
> >> >>> -   fs_reg msg_lod = offset(msg_coords, 4);
> >> >>> +   fs_reg msg_end = bld.offset(msg_coords, vector_elements);
> >> >>> +   fs_reg msg_lod = bld.offset(msg_coords, 4);
> >> >>>
> >> >>>     if (shadow_c.file != BAD_FILE) {
> >> >>>        fs_reg msg_shadow = msg_lod;
> >> >>>        bld.MOV(msg_shadow, shadow_c);
> >> >>> -      msg_lod = offset(msg_shadow, 1);
> >> >>> +      msg_lod = bld.offset(msg_shadow, 1);
> >> >>>        msg_end = msg_lod;
> >> >>>     }
> >> >>>
> >> >>> @@ -354,13 +354,13 @@
fs_visitor::emit_texture_gen5(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>        break;
> >> >>>     case ir_txb:
> >> >>>        bld.MOV(msg_lod, lod);
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>
> >> >>>        opcode = FS_OPCODE_TXB;
> >> >>>        break;
> >> >>>     case ir_txl:
> >> >>>        bld.MOV(msg_lod, lod);
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXL;
> >> >>>        break;
> >> >>> @@ -377,12 +377,12 @@
fs_visitor::emit_texture_gen5(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>        msg_end = msg_lod;
> >> >>>        for (int i = 0; i < grad_components; i++) {
> >> >>>           bld.MOV(msg_end, lod);
> >> >>> -         lod = offset(lod, 1);
> >> >>> -         msg_end = offset(msg_end, 1);
> >> >>> +         lod = bld.offset(lod, 1);
> >> >>> +         msg_end = bld.offset(msg_end, 1);
> >> >>>
> >> >>>           bld.MOV(msg_end, lod2);
> >> >>> -         lod2 = offset(lod2, 1);
> >> >>> -         msg_end = offset(msg_end, 1);
> >> >>> +         lod2 = bld.offset(lod2, 1);
> >> >>> +         msg_end = bld.offset(msg_end, 1);
> >> >>>        }
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXD;
> >> >>> @@ -391,31 +391,31 @@
fs_visitor::emit_texture_gen5(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>     case ir_txs:
> >> >>>        msg_lod = retype(msg_end, BRW_REGISTER_TYPE_UD);
> >> >>>        bld.MOV(msg_lod, lod);
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXS;
> >> >>>        break;
> >> >>>     case ir_query_levels:
> >> >>>        msg_lod = msg_end;
> >> >>>        bld.MOV(retype(msg_lod, BRW_REGISTER_TYPE_UD), fs_reg(0u));
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXS;
> >> >>>        break;
> >> >>>     case ir_txf:
> >> >>> -      msg_lod = offset(msg_coords, 3);
> >> >>> +      msg_lod = bld.offset(msg_coords, 3);
> >> >>>        bld.MOV(retype(msg_lod, BRW_REGISTER_TYPE_UD), lod);
> >> >>> -      msg_end = offset(msg_lod, 1);
> >> >>> +      msg_end = bld.offset(msg_lod, 1);
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXF;
> >> >>>        break;
> >> >>>     case ir_txf_ms:
> >> >>> -      msg_lod = offset(msg_coords, 3);
> >> >>> +      msg_lod = bld.offset(msg_coords, 3);
> >> >>>        /* lod */
> >> >>>        bld.MOV(retype(msg_lod, BRW_REGISTER_TYPE_UD), fs_reg(0u));
> >> >>>        /* sample index */
> >> >>> -      bld.MOV(retype(offset(msg_lod, 1), BRW_REGISTER_TYPE_UD),
> > sample_index);
> >> >>> -      msg_end = offset(msg_lod, 2);
> >> >>> +      bld.MOV(retype(bld.offset(msg_lod, 1),
BRW_REGISTER_TYPE_UD),
> > sample_index);
> >> >>> +      msg_end = bld.offset(msg_lod, 2);
> >> >>>
> >> >>>        opcode = SHADER_OPCODE_TXF_CMS;
> >> >>>        break;
> >> >>> @@ -525,7 +525,7 @@ fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>         */
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(sources[length], coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        length++;
> >> >>>
> >> >>>           /* For cube map array, the coordinate is (u,v,r,ai) but
> > there are
> >> >>> @@ -533,11 +533,11 @@
fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>            */
> >> >>>           if (i < grad_components) {
> >> >>>              bld.MOV(sources[length], lod);
> >> >>> -            lod = offset(lod, 1);
> >> >>> +            lod = bld.offset(lod, 1);
> >> >>>              length++;
> >> >>>
> >> >>>              bld.MOV(sources[length], lod2);
> >> >>> -            lod2 = offset(lod2, 1);
> >> >>> +            lod2 = bld.offset(lod2, 1);
> >> >>>              length++;
> >> >>>           }
> >> >>>        }
> >> >>> @@ -559,13 +559,13 @@
fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>         */
> >> >>>
> >> >>>        bld.MOV(retype(sources[length], BRW_REGISTER_TYPE_D),
> > coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        length++;
> >> >>>
> >> >>>        if (devinfo->gen >= 9) {
> >> >>>           if (coord_components >= 2) {
> >> >>>              bld.MOV(retype(sources[length], BRW_REGISTER_TYPE_D),
> > coordinate);
> >> >>> -            coordinate = offset(coordinate, 1);
> >> >>> +            coordinate = bld.offset(coordinate, 1);
> >> >>>           }
> >> >>>           length++;
> >> >>>        }
> >> >>> @@ -575,7 +575,7 @@ fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>
> >> >>>        for (int i = devinfo->gen >= 9 ? 2 : 1; i <
coord_components;
> > i++) {
> >> >>>           bld.MOV(retype(sources[length], BRW_REGISTER_TYPE_D),
> > coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>        length++;
> >> >>>        }
> >> >>>
> >> >>> @@ -594,7 +594,7 @@ fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>         */
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(retype(sources[length], BRW_REGISTER_TYPE_D),
> > coordinate);
> >> >>> -         coordinate = offset(coordinate, 1);
> >> >>> +         coordinate = bld.offset(coordinate, 1);
> >> >>>           length++;
> >> >>>        }
> >> >>>
> >> >>> @@ -608,19 +608,19 @@
fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>           /* More crazy intermixing */
> >> >>>           for (int i = 0; i < 2; i++) { /* u, v */
> >> >>>              bld.MOV(sources[length], coordinate);
> >> >>> -            coordinate = offset(coordinate, 1);
> >> >>> +            coordinate = bld.offset(coordinate, 1);
> >> >>>              length++;
> >> >>>           }
> >> >>>
> >> >>>           for (int i = 0; i < 2; i++) { /* offu, offv */
> >> >>>              bld.MOV(retype(sources[length], BRW_REGISTER_TYPE_D),
> > offset_value);
> >> >>> -            offset_value = offset(offset_value, 1);
> >> >>> +            offset_value = bld.offset(offset_value, 1);
> >> >>>              length++;
> >> >>>           }
> >> >>>
> >> >>>           if (coord_components == 3) { /* r if present */
> >> >>>              bld.MOV(sources[length], coordinate);
> >> >>> -            coordinate = offset(coordinate, 1);
> >> >>> +            coordinate = bld.offset(coordinate, 1);
> >> >>>              length++;
> >> >>>           }
> >> >>>
> >> >>> @@ -633,7 +633,7 @@ fs_visitor::emit_texture_gen7(ir_texture_opcode
> > op, fs_reg dst,
> >> >>>     if (!coordinate_done) {
> >> >>>        for (int i = 0; i < coord_components; i++) {
> >> >>>           bld.MOV(sources[length], coordinate);
> >> >>> -         coordinate = offset(coordinate, 1);
> >> >>> +         coordinate = bld.offset(coordinate, 1);
> >> >>>           length++;
> >> >>>        }
> >> >>>     }
> >> >>> @@ -746,8 +746,8 @@ fs_visitor::rescale_texcoord(fs_reg coordinate,
> > int coord_components,
> >> >>>        coordinate = dst;
> >> >>>
> >> >>>        bld.MUL(dst, src, scale_x);
> >> >>> -      dst = offset(dst, 1);
> >> >>> -      src = offset(src, 1);
> >> >>> +      dst = bld.offset(dst, 1);
> >> >>> +      src = bld.offset(src, 1);
> >> >>>        bld.MUL(dst, src, scale_y);
> >> >>>     } else if (is_rect) {
> >> >>>        /* On gen6+, the sampler handles the rectangle coordinates
> >> >>> @@ -760,7 +760,7 @@ fs_visitor::rescale_texcoord(fs_reg coordinate,
> > int coord_components,
> >> >>>        for (int i = 0; i < 2; i++) {
> >> >>>        if (key_tex->gl_clamp_mask[i] & (1 << sampler)) {
> >> >>>           fs_reg chan = coordinate;
> >> >>> -         chan = offset(chan, i);
> >> >>> +         chan = bld.offset(chan, i);
> >> >>>
> >> >>>              set_condmod(BRW_CONDITIONAL_GE,
> >> >>>                          bld.emit(BRW_OPCODE_SEL, chan, chan,
> > fs_reg(0.0f)));
> >> >>> @@ -785,7 +785,7 @@ fs_visitor::rescale_texcoord(fs_reg coordinate,
> > int coord_components,
> >> >>>        for (int i = 0; i < MIN2(coord_components, 3); i++) {
> >> >>>        if (key_tex->gl_clamp_mask[i] & (1 << sampler)) {
> >> >>>           fs_reg chan = coordinate;
> >> >>> -         chan = offset(chan, i);
> >> >>> +         chan = bld.offset(chan, i);
> >> >>>              set_saturate(true, bld.MOV(chan, chan));
> >> >>>        }
> >> >>>        }
> >> >>> @@ -807,7 +807,7 @@ fs_visitor::emit_mcs_fetch(fs_reg coordinate,
int
> > components, fs_reg sampler)
> >> >>>     for (int i = 0; i < components; i++) {
> >> >>>        sources[i] = vgrf(glsl_type::float_type);
> >> >>>        bld.MOV(retype(sources[i], BRW_REGISTER_TYPE_D),
coordinate);
> >> >>> -      coordinate = offset(coordinate, 1);
> >> >>> +      coordinate = bld.offset(coordinate, 1);
> >> >>>     }
> >> >>>
> >> >>>     bld.LOAD_PAYLOAD(payload, sources, components, 0);
> >> >>> @@ -853,7 +853,7 @@ fs_visitor::emit_texture(ir_texture_opcode op,
> >> >>>
> >> >>>           for (int i=0; i<4; i++) {
> >> >>>              bld.MOV(res, fs_reg(swiz == SWIZZLE_ZERO ? 0.0f :
1.0f));
> >> >>> -            res = offset(res, 1);
> >> >>> +            res = bld.offset(res, 1);
> >> >>>           }
> >> >>>           return;
> >> >>>        }
> >> >>> @@ -907,7 +907,7 @@ fs_visitor::emit_texture(ir_texture_opcode op,
> >> >>>
> >> >>>     /* fixup #layers for cube map arrays */
> >> >>>     if (op == ir_txs && is_cube_array) {
> >> >>> -      fs_reg depth = offset(dst, 2);
> >> >>> +      fs_reg depth = bld.offset(dst, 2);
> >> >>>        fs_reg fixed_depth = vgrf(glsl_type::int_type);
> >> >>>        bld.emit(SHADER_OPCODE_INT_QUOTIENT, fixed_depth, depth,
> > fs_reg(6));
> >> >>>
> >> >>> @@ -917,7 +917,7 @@ fs_visitor::emit_texture(ir_texture_opcode op,
> >> >>>           if (i == 2) {
> >> >>>              fixed_payload[i] = fixed_depth;
> >> >>>           } else {
> >> >>> -            fixed_payload[i] = offset(dst, i);
> >> >>> +            fixed_payload[i] = bld.offset(dst, i);
> >> >>>           }
> >> >>>        }
> >> >>>        bld.LOAD_PAYLOAD(dst, fixed_payload, components, 0);
> >> >>> @@ -952,7 +952,7 @@ fs_visitor::emit_gen6_gather_wa(uint8_t wa,
> > fs_reg dst)
> >> >>>           bld.ASR(dst, dst, fs_reg(32 - width));
> >> >>>        }
> >> >>>
> >> >>> -      dst = offset(dst, 1);
> >> >>> +      dst = bld.offset(dst, 1);
> >> >>>     }
> >> >>>  }
> >> >>>
> >> >>> @@ -989,7 +989,7 @@ fs_visitor::swizzle_result(ir_texture_opcode
op,
> > int dest_components,
> >> >>>  {
> >> >>>     if (op == ir_query_levels) {
> >> >>>        /* # levels is in .w */
> >> >>> -      this->result = offset(orig_val, 3);
> >> >>> +      this->result = bld.offset(orig_val, 3);
> >> >>>        return;
> >> >>>     }
> >> >>>
> >> >>> @@ -1010,15 +1010,15 @@
fs_visitor::swizzle_result(ir_texture_opcode
> > op, int dest_components,
> >> >>>        for (int i = 0; i < 4; i++) {
> >> >>>        int swiz = GET_SWZ(key_tex->swizzles[sampler], i);
> >> >>>        fs_reg l = swizzled_result;
> >> >>> -      l = offset(l, i);
> >> >>> +      l = bld.offset(l, i);
> >> >>>
> >> >>>        if (swiz == SWIZZLE_ZERO) {
> >> >>>              bld.MOV(l, fs_reg(0.0f));
> >> >>>        } else if (swiz == SWIZZLE_ONE) {
> >> >>>              bld.MOV(l, fs_reg(1.0f));
> >> >>>        } else {
> >> >>> -            bld.MOV(l, offset(orig_val,
> >> >>> -                              GET_SWZ(key_tex->swizzles[sampler],
> > i)));
> >> >>> +            bld.MOV(l, bld.offset(orig_val,
> >> >>> +
> > GET_SWZ(key_tex->swizzles[sampler], i)));
> >> >>>        }
> >> >>>        }
> >> >>>        this->result = swizzled_result;
> >> >>> @@ -1315,14 +1315,14 @@ fs_visitor::emit_interpolation_setup_gen4()
> >> >>>
> >> >>>     if (devinfo->has_pln && dispatch_width == 16) {
> >> >>>        for (unsigned i = 0; i < 2; i++) {
> >> >>> -         abld.half(i).ADD(half(offset(delta_xy, i), 0),
> >> >>> -                          half(this->pixel_x, i), xstart);
> >> >>> -         abld.half(i).ADD(half(offset(delta_xy, i), 1),
> >> >>> -                          half(this->pixel_y, i), ystart);
> >> >>> +         abld.half(i).ADD(abld.half(abld.offset(delta_xy, i), 0),
> >> >>> +                          abld.half(this->pixel_x, i), xstart);
> >> >>> +         abld.half(i).ADD(abld.half(abld.offset(delta_xy, i), 1),
> >> >>> +                          abld.half(this->pixel_y, i), ystart);
> >> >>>        }
> >> >>>     } else {
> >> >>> -      abld.ADD(offset(delta_xy, 0), this->pixel_x, xstart);
> >> >>> -      abld.ADD(offset(delta_xy, 1), this->pixel_y, ystart);
> >> >>> +      abld.ADD(abld.offset(delta_xy, 0), this->pixel_x, xstart);
> >> >>> +      abld.ADD(abld.offset(delta_xy, 1), this->pixel_y, ystart);
> >> >>>     }
> >> >>>
> >> >>>     abld = bld.annotate("compute pos.w and 1/pos.w");
> >> >>> @@ -1419,7 +1419,7 @@ fs_visitor::setup_color_payload(fs_reg *dst,
> > fs_reg color, unsigned components,
> >> >>>        fs_reg tmp = vgrf(glsl_type::vec4_type);
> >> >>>        assert(color.type == BRW_REGISTER_TYPE_F);
> >> >>>        for (unsigned i = 0; i < components; i++) {
> >> >>> -         inst = bld.MOV(offset(tmp, i), offset(color, i));
> >> >>> +         inst = bld.MOV(bld.offset(tmp, i), bld.offset(color, i));
> >> >>>           inst->saturate = true;
> >> >>>        }
> >> >>>        color = tmp;
> >> >>> @@ -1428,10 +1428,10 @@ fs_visitor::setup_color_payload(fs_reg
*dst,
> > fs_reg color, unsigned components,
> >> >>>     if (exec_size < dispatch_width) {
> >> >>>        unsigned half_idx = use_2nd_half ? 1 : 0;
> >> >>>        for (unsigned i = 0; i < components; i++)
> >> >>> -         dst[i] = half(offset(color, i), half_idx);
> >> >>> +         dst[i] = bld.half(bld.offset(color, i), half_idx);
> >> >>>     } else {
> >> >>>        for (unsigned i = 0; i < components; i++)
> >> >>> -         dst[i] = offset(color, i);
> >> >>> +         dst[i] = bld.offset(color, i);
> >> >>>     }
> >> >>>  }
> >> >>>
> >> >>> @@ -1479,7 +1479,7 @@ fs_visitor::emit_alpha_test()
> >> >>>                       BRW_CONDITIONAL_NEQ);
> >> >>>     } else {
> >> >>>        /* RT0 alpha */
> >> >>> -      fs_reg color = offset(outputs[0], 3);
> >> >>> +      fs_reg color = bld.offset(outputs[0], 3);
> >> >>>
> >> >>>        /* f0.1 &= func(color, ref) */
> >> >>>        cmp = abld.CMP(bld.null_reg_f(), color,
> > fs_reg(key->alpha_test_ref),
> >> >>> @@ -1556,7 +1556,8 @@ fs_visitor::emit_single_fb_write(const
> > fs_builder &bld,
> >> >>>         * alpha-testing, alpha-to-coverage, and so on.
> >> >>>         */
> >> >>>        if (this->outputs[0].file != BAD_FILE)
> >> >>> -         setup_color_payload(&sources[length + 3],
> > offset(this->outputs[0], 3),
> >> >>> +         setup_color_payload(&sources[length + 3],
> >> >>> +                             bld.offset(this->outputs[0], 3),
> >> >>>                               1, exec_size, false);
> >> >>>        length += 4;
> >> >>>     } else if (color1.file == BAD_FILE) {
> >> >>> @@ -1591,7 +1592,7 @@ fs_visitor::emit_single_fb_write(const
> > fs_builder &bld,
> >> >>>        /* Hand over gl_FragDepth. */
> >> >>>        assert(this->frag_depth.file != BAD_FILE);
> >> >>>           if (exec_size < dispatch_width) {
> >> >>> -            sources[length] = half(this->frag_depth,
use_2nd_half);
> >> >>> +            sources[length] = bld.half(this->frag_depth,
> > use_2nd_half);
> >> >>>           } else {
> >> >>>              sources[length] = this->frag_depth;
> >> >>>           }
> >> >>> @@ -1692,7 +1693,7 @@ fs_visitor::emit_fb_writes()
> >> >>>
> >> >>>           fs_reg src0_alpha;
> >> >>>           if (devinfo->gen >= 6 && key->replicate_alpha && target
!=
> > 0)
> >> >>> -            src0_alpha = offset(outputs[0], 3);
> >> >>> +            src0_alpha = bld.offset(outputs[0], 3);
> >> >>>
> >> >>>           inst = emit_single_fb_write(abld, this->outputs[target],
> > reg_undef,
> >> >>>                                       src0_alpha,
> >> >>> @@ -1776,7 +1777,7 @@ void fs_visitor::compute_clip_distance()
> >> >>>        abld.MUL(output, outputs[clip_vertex], u);
> >> >>>        for (int j = 1; j < 4; j++) {
> >> >>>           u.reg = userplane[i].reg + j;
> >> >>> -         abld.MAD(output, output, offset(outputs[clip_vertex], j),
> > u);
> >> >>> +         abld.MAD(output, output, bld.offset(outputs[clip_vertex],
> > j), u);
> >> >>>        }
> >> >>>     }
> >> >>>  }
> >> >>> @@ -1890,13 +1891,13 @@ fs_visitor::emit_urb_writes()
> >> >>>               */
> >> >>>              for (int i = 0; i < 4; i++) {
> >> >>>                 reg = fs_reg(GRF, alloc.allocate(1),
> > outputs[varying].type);
> >> >>> -               src = offset(this->outputs[varying], i);
> >> >>> +               src = bld.offset(this->outputs[varying], i);
> >> >>>                 set_saturate(true, bld.MOV(reg, src));
> >> >>>                 sources[length++] = reg;
> >> >>>              }
> >> >>>           } else {
> >> >>>              for (int i = 0; i < 4; i++)
> >> >>> -               sources[length++] = offset(this->outputs[varying],
i);
> >> >>> +               sources[length++] =
> > bld.offset(this->outputs[varying], i);
> >> >>>           }
> >> >>>           break;
> >> >>>        }
> >> >>> diff --git a/src/mesa/drivers/dri/i965/brw_ir_fs.h
> > b/src/mesa/drivers/dri/i965/brw_ir_fs.h
> >> >>> index 96dc20d..10033ca 100644
> >> >>> --- a/src/mesa/drivers/dri/i965/brw_ir_fs.h
> >> >>> +++ b/src/mesa/drivers/dri/i965/brw_ir_fs.h
> >> >>> @@ -129,27 +129,6 @@ horiz_offset(fs_reg reg, unsigned delta)
> >> >>>  }
> >> >>>
> >> >>>  static inline fs_reg
> >> >>> -offset(fs_reg reg, unsigned delta)
> >> >>> -{
> >> >>> -   switch (reg.file) {
> >> >>> -   case BAD_FILE:
> >> >>> -      break;
> >> >>> -   case GRF:
> >> >>> -   case MRF:
> >> >>> -   case ATTR:
> >> >>> -      return byte_offset(reg,
> >> >>> -                         delta * MAX2(reg.width * reg.stride, 1) *
> >> >>> -                         type_sz(reg.type));
> >> >>> -   case UNIFORM:
> >> >>> -      reg.reg_offset += delta;
> >> >>> -      break;
> >> >>> -   default:
> >> >>> -      assert(delta == 0);
> >> >>> -   }
> >> >>> -   return reg;
> >> >>> -}
> >> >>> -
> >> >>> -static inline fs_reg
> >> >>>  component(fs_reg reg, unsigned idx)
> >> >>>  {
> >> >>>     assert(reg.subreg_offset == 0);
> >> >>> @@ -167,36 +146,6 @@ is_uniform(const fs_reg &reg)
> >> >>>            (!reg.reladdr || is_uniform(*reg.reladdr));
> >> >>>  }
> >> >>>
> >> >>> -/**
> >> >>> - * Get either of the 8-component halves of a 16-component
register.
> >> >>> - *
> >> >>> - * Note: this also works if \c reg represents a SIMD16 pair of
> > registers.
> >> >>> - */
> >> >>> -static inline fs_reg
> >> >>> -half(fs_reg reg, unsigned idx)
> >> >>> -{
> >> >>> -   assert(idx < 2);
> >> >>> -
> >> >>> -   switch (reg.file) {
> >> >>> -   case BAD_FILE:
> >> >>> -   case UNIFORM:
> >> >>> -   case IMM:
> >> >>> -      return reg;
> >> >>> -
> >> >>> -   case GRF:
> >> >>> -   case MRF:
> >> >>> -      assert(reg.width == 16);
> >> >>> -      reg.width = 8;
> >> >>> -      return horiz_offset(reg, 8 * idx);
> >> >>> -
> >> >>> -   case ATTR:
> >> >>> -   case HW_REG:
> >> >>> -   default:
> >> >>> -      unreachable("Cannot take half of this register type");
> >> >>> -   }
> >> >>> -   return reg;
> >> >>> -}
> >> >>> -
> >> >>>  static const fs_reg reg_undef;
> >> >>>
> >> >>>  class fs_inst : public backend_instruction {
> >> >>> --
> >> >>> 2.4.3
> >> >>>
> >> >>> _______________________________________________
> >> >>> mesa-dev mailing list
> >> >>> mesa-dev at lists.freedesktop.org
> >> >>> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20150624/16e8a218/attachment-0001.html>


More information about the mesa-dev mailing list