renderer: add texture view internals (v4)

This adds the renderer support for exposing texture views.

The change adds support to sampler views and surfaces for the backing
object to be a separately allocated textureview instead of the actual
resource. It also adds support for parsing the updated packet where
we put the sampler view target in the top 8 bits of the format,
which should be safe as up until now it's always 0.

This doesn't do the final enable of the extension, but it will
cause us to use views in a few places we didn't before so could have
some other side effects

v2: check for buffers on surface creation, reword comment
to better explain if statement.
v3: fix logic bug (Gert). This uncovered a wierd corner case, possibly
due to how the state tracker allocates textures, but it tries to create
a surface with more layers than there are, I think it then reallocate
the whole texture object in the guest and it works out in the end,
this adds check to catch this, and to avoid creating a texture view
in that case.
v4: realise the problem with v3 was 3D textures and use
util_max_layer to fix the logic properly.

Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
macos/master
Dave Airlie 7 years ago
parent 6b4d7f5cf8
commit 937e724023
  1. 96
      src/vrend_renderer.c

@ -118,6 +118,7 @@ struct global_renderer_state {
bool have_polygon_offset_clamp;
bool have_texture_storage;
bool have_tessellation;
bool have_texture_view;
/* these appeared broken on at least one driver */
bool use_explicit_locations;
@ -230,6 +231,7 @@ struct vrend_sampler_view {
struct pipe_reference reference;
GLuint id;
GLuint format;
GLenum target;
GLuint val0, val1;
GLuint gl_swizzle_r;
GLuint gl_swizzle_g;
@ -424,6 +426,8 @@ static void vrend_apply_sampler_state(struct vrend_context *ctx,
struct vrend_resource *res,
uint32_t shader_type,
int id, int sampler_id, uint32_t srgb_decode);
static GLenum tgsitargettogltarget(const enum pipe_texture_target target, int nr_samples);
void vrend_update_stencil_state(struct vrend_context *ctx);
static struct vrend_format_table tex_conv_table[VIRGL_FORMAT_MAX];
@ -547,6 +551,8 @@ static void __report_gles_missing_func(const char *fname, struct vrend_context *
static void vrend_destroy_surface(struct vrend_surface *surf)
{
if (surf->id != surf->texture->id)
glDeleteTextures(1, &surf->id);
vrend_resource_reference(&surf->texture, NULL);
free(surf);
}
@ -563,6 +569,8 @@ vrend_surface_reference(struct vrend_surface **ptr, struct vrend_surface *surf)
static void vrend_destroy_sampler_view(struct vrend_sampler_view *samp)
{
if (samp->texture->id != samp->id)
glDeleteTextures(1, &samp->id);
vrend_resource_reference(&samp->texture, NULL);
free(samp);
}
@ -1212,6 +1220,31 @@ int vrend_create_surface(struct vrend_context *ctx,
surf->format = format;
surf->val0 = val0;
surf->val1 = val1;
surf->id = res->id;
if (vrend_state.have_texture_view && !res->is_buffer) {
/* We don't need texture views for buffer objects.
* Otherwise we only need a texture view if the
* a) formats differ between the surface and base texture
* b) we need to map a sub range > 1 layer to a surface,
* GL can make a single layer fine without a view, and it
* can map the whole texure fine. In those cases we don't
* create a texture view.
*/
int first_layer = surf->val1 & 0xffff;
int last_layer = (surf->val1 >> 16) & 0xffff;
if ((first_layer != last_layer &&
(first_layer != 0 || (last_layer != util_max_layer(&res->base, surf->val0)))) ||
surf->format != res->base.format) {
GLenum internalformat = tex_conv_table[surf->format].internalformat;
glGenTextures(1, &surf->id);
glTextureView(surf->id, res->target, res->id, internalformat,
0, res->base.last_level + 1,
first_layer, last_layer - first_layer + 1);
}
}
pipe_reference_init(&surf->reference, 1);
vrend_resource_reference(&surf->texture, res);
@ -1416,7 +1449,8 @@ int vrend_create_sampler_view(struct vrend_context *ctx,
return ENOMEM;
pipe_reference_init(&view->reference, 1);
view->format = format;
view->format = format & 0xffffff;
view->target = tgsitargettogltarget((format >> 24) & 0xff, res->base.nr_samples);
view->val0 = val0;
view->val1 = val1;
view->cur_base = -1;
@ -1429,6 +1463,49 @@ int vrend_create_sampler_view(struct vrend_context *ctx,
vrend_resource_reference(&view->texture, res);
view->id = view->texture->id;
if (!view->target)
view->target = view->texture->target;
if (vrend_state.have_texture_view && !view->texture->is_buffer) {
enum pipe_format format;
bool needs_view = false;
/*
* Need to use a texture view if the gallium
* view target is different than the underlying
* texture target.
*/
if (view->target != view->texture->target)
needs_view = true;
/*
* If the formats are different and this isn't
* a DS texture a view is required.
* DS are special as they use different gallium
* formats for DS views into a combined resource.
* GL texture views can't be use for this, stencil
* texturing is used instead. For DS formats
* aways program the underlying DS format as a
* view could be required for layers.
*/
format = view->format;
if (util_format_is_depth_or_stencil(view->texture->base.format))
format = view->texture->base.format;
else if (view->format != view->texture->base.format)
needs_view = true;
if (needs_view) {
glGenTextures(1, &view->id);
GLenum internalformat = tex_conv_table[format].internalformat;
unsigned base_layer = view->val0 & 0xffff;
unsigned max_layer = (view->val0 >> 16) & 0xffff;
view->cur_base = view->val1 & 0xff;
view->cur_max = (view->val1 >> 8) & 0xff;
glTextureView(view->id, view->target, view->texture->id, internalformat,
view->cur_base, (view->cur_max - view->cur_base) + 1,
base_layer, max_layer - base_layer + 1);
}
}
view->srgb_decode = GL_DECODE_EXT;
if (view->format != view->texture->base.format) {
if (util_format_is_srgb(view->texture->base.format) &&
@ -1567,8 +1644,8 @@ static void vrend_hw_set_zsurf_texture(struct vrend_context *ctx)
if (!surf->texture)
return;
vrend_fb_bind_texture(surf->texture, 0, surf->val0,
first_layer != last_layer ? 0xffffffff : first_layer);
vrend_fb_bind_texture_id(surf->texture, surf->id, 0, surf->val0,
first_layer != last_layer ? 0xffffffff : first_layer);
}
}
@ -1585,8 +1662,8 @@ static void vrend_hw_set_color_surface(struct vrend_context *ctx, int index)
uint32_t first_layer = ctx->sub->surf[index]->val1 & 0xffff;
uint32_t last_layer = (ctx->sub->surf[index]->val1 >> 16) & 0xffff;
vrend_fb_bind_texture(surf->texture, index, surf->val0,
first_layer != last_layer ? 0xffffffff : first_layer);
vrend_fb_bind_texture_id(surf->texture, surf->id, index, surf->val0,
first_layer != last_layer ? 0xffffffff : first_layer);
}
}
@ -2084,7 +2161,7 @@ void vrend_set_single_sampler_view(struct vrend_context *ctx,
return;
}
if (!view->texture->is_buffer) {
glBindTexture(view->texture->target, view->texture->id);
glBindTexture(view->target, view->id);
if (util_format_is_depth_or_stencil(view->format)) {
if (vrend_state.use_core_profile == false) {
@ -2985,13 +3062,13 @@ static void vrend_draw_bind_samplers_shader(struct vrend_context *ctx,
if (tview->texture) {
GLuint id;
struct vrend_resource *texture = tview->texture;
GLenum target = texture->target;
GLenum target = tview->target;
if (texture->is_buffer) {
id = texture->tbo_tex_id;
target = GL_TEXTURE_BUFFER;
} else
id = texture->id;
id = tview->id;
glBindTexture(target, id);
if (ctx->sub->views[shader_type].old_ids[i] != id || ctx->sub->sampler_state_dirty) {
@ -4444,6 +4521,9 @@ int vrend_renderer_init(struct vrend_if_cbs *cbs, uint32_t flags)
if (gl_ver >= 42 || epoxy_has_gl_extension("GL_ARB_texture_storage"))
vrend_state.have_texture_storage = true;
if (gl_ver >= 43 || epoxy_has_gl_extension("GL_ARB_texture_view"))
vrend_state.have_texture_view = true;
if (gl_ver >= 46 || epoxy_has_gl_extension("GL_ARB_polygon_offset_clamp"))
vrend_state.have_polygon_offset_clamp = true;

Loading…
Cancel
Save