[Mesa-dev] [PATCH V2 15/22] i965/gen9: Use _mesa_meta_pbo_TexSubImage to write to YF/YS surfaces
Pohjolainen, Topi
topi.pohjolainen at intel.com
Fri Apr 24 17:15:22 PDT 2015
On Fri, Apr 24, 2015 at 05:12:34PM -0700, Anuj Phogat wrote:
> On Fri, Apr 24, 2015 at 12:22 PM, Pohjolainen, Topi
> <topi.pohjolainen at intel.com> wrote:
> > On Fri, Apr 17, 2015 at 04:51:36PM -0700, Anuj Phogat wrote:
> >> No other path currently supports uploading data to these surfaces.
> >>
> >> Signed-off-by: Anuj Phogat <anuj.phogat at gmail.com>
> >> ---
> >> src/mesa/drivers/dri/i965/intel_tex_image.c | 24 ++++++++++++++++++++++--
> >> src/mesa/drivers/dri/i965/intel_tex_subimage.c | 23 +++++++++++++++++++++--
> >> 2 files changed, 43 insertions(+), 4 deletions(-)
> >>
> >> diff --git a/src/mesa/drivers/dri/i965/intel_tex_image.c b/src/mesa/drivers/dri/i965/intel_tex_image.c
> >> index 31cbabe..03db100 100644
> >> --- a/src/mesa/drivers/dri/i965/intel_tex_image.c
> >> +++ b/src/mesa/drivers/dri/i965/intel_tex_image.c
> >> @@ -93,8 +93,10 @@ intelTexImage(struct gl_context * ctx,
> >> const struct gl_pixelstore_attrib *unpack)
> >> {
> >> struct intel_texture_image *intelImage = intel_texture_image(texImage);
> >> + struct brw_context *brw = brw_context(ctx);
> >> bool ok;
> >> -
> >> + bool create_pbo = false;
> >> + uint32_t tr_mode = INTEL_MIPTREE_TRMODE_NONE;
> >> bool tex_busy = intelImage->mt && drm_intel_bo_busy(intelImage->mt->bo);
> >>
> >> DBG("%s mesa_format %s target %s format %s type %s level %d %dx%dx%d\n",
> >> @@ -111,15 +113,33 @@ intelTexImage(struct gl_context * ctx,
> >>
> >> assert(intelImage->mt);
> >>
> >> + if (brw->gen >= 9) {
> >> + tr_mode = intelImage->mt->tr_mode;
> >> +
> >> + /* Set create_pbo = true for surfaces with INTEL_MIPTREE_TRMODE_{YF/YS}.
> >> + * _mesa_meta_pbo_TexSubImage() is the only working path to upload data
> >> + * to such surfaces.
> >> + */
> >> + create_pbo = tex_busy || (intelImage->mt &&
> >> + intelImage->mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE);
> >> + } else {
> >> + create_pbo = tex_busy;
> >> + }
> >> +
> >
> > What would you think about a helper?
> Yes, we can use a helper here.
> >
> > static bool need_to_create_pbo(const struct intel_texture_image *img)
> > {
> > if (!img->mt)
> > return false;
> >
> > const bool tex_busy = drm_intel_bo_busy(img->mt->bo);
> > if (tex_busy || brw->gen < 9)
> > return tex_busy;
> >
> > /* Set create_pbo = true for surfaces with INTEL_MIPTREE_TRMODE_{YF/YS}.
> > * _mesa_meta_pbo_TexSubImage() is the only working path to upload data
> > * to such surfaces.
> > */
> > return img->mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE;
> > }
> >
> >> ok = _mesa_meta_pbo_TexSubImage(ctx, dims, texImage, 0, 0, 0,
> >> texImage->Width, texImage->Height,
> >> texImage->Depth,
> >> format, type, pixels,
> >> false /*allocate_storage*/,
> >> - tex_busy, unpack);
> >> + create_pbo, unpack);
> >> if (ok)
> >> return;
> >>
> >> + /* Currently there are no fallback paths to upload data to surfaces with
> >> + * tr_mode != INTEL_MIPTREE_TRMODE_NONE.
> >> + */
> >> + assert(tr_mode == INTEL_MIPTREE_TRMODE_NONE);
> >
> > And I would put this assertion into _mesa_meta_pbo_TexSubImage() instead
> > of duplicating it for both callers. What do you think?
> >
> Meta functions are not supposed to use any Intel specific code. I'm anyway
> getting rid of these assertions in my later patches. So, I'll just keep them
> here.
Ah, of course, sorry for the noise.
More information about the mesa-dev
mailing list