* SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
*/
+#include <drm_fourcc.h>
#include <errno.h>
#include <time.h>
#include <unistd.h>
-#include "main/glheader.h"
#include "main/context.h"
#include "main/framebuffer.h"
#include "main/renderbuffer.h"
#include "main/version.h"
#include "swrast/s_renderbuffer.h"
#include "util/ralloc.h"
-#include "brw_shader.h"
-#include "glsl/nir/nir.h"
+#include "brw_defines.h"
+#include "compiler/nir/nir.h"
#include "utils.h"
#include "xmlpool.h"
+#ifndef DRM_FORMAT_MOD_INVALID
+#define DRM_FORMAT_MOD_INVALID ((1ULL<<56) - 1)
+#endif
+
+#ifndef DRM_FORMAT_MOD_LINEAR
+#define DRM_FORMAT_MOD_LINEAR 0
+#endif
+
static const __DRIconfigOptionsExtension brw_config_options = {
.base = { __DRI_CONFIG_OPTIONS, 1 },
.xml =
DRI_CONF_ENUM(1, "Enable reuse of all sizes of buffer objects")
DRI_CONF_DESC_END
DRI_CONF_OPT_END
-
- DRI_CONF_OPT_BEGIN_B(hiz, "true")
- DRI_CONF_DESC(en, "Enable Hierarchical Z on gen6+")
- DRI_CONF_OPT_END
DRI_CONF_SECTION_END
DRI_CONF_SECTION_QUALITY
DRI_CONF_FORCE_S3TC_ENABLE("false")
+ DRI_CONF_PRECISE_TRIG("false")
+
DRI_CONF_OPT_BEGIN(clamp_max_samples, int, -1)
DRI_CONF_DESC(en, "Clamp the value of GL_MAX_SAMPLES to the "
"given integer. If negative, then do not clamp.")
DRI_CONF_ALWAYS_FLUSH_CACHE("false")
DRI_CONF_DISABLE_THROTTLING("false")
DRI_CONF_FORCE_GLSL_EXTENSIONS_WARN("false")
+ DRI_CONF_FORCE_GLSL_VERSION(0)
DRI_CONF_DISABLE_GLSL_LINE_CONTINUATIONS("false")
DRI_CONF_DISABLE_BLEND_FUNC_EXTENDED("false")
+ DRI_CONF_DUAL_COLOR_BLEND_BY_LOCATION("false")
DRI_CONF_ALLOW_GLSL_EXTENSION_DIRECTIVE_MIDSHADER("false")
+ DRI_CONF_ALLOW_HIGHER_COMPAT_VERSION("false")
DRI_CONF_OPT_BEGIN_B(shader_precompile, "true")
DRI_CONF_DESC(en, "Perform code generation at shader link time.")
DRI_CONF_OPT_END
DRI_CONF_SECTION_END
+
+ DRI_CONF_SECTION_MISCELLANEOUS
+ DRI_CONF_GLSL_ZERO_INIT("false")
+ DRI_CONF_SECTION_END
DRI_CONF_END
};
return tp.tv_sec + tp.tv_nsec / 1000000000.0;
}
-void
-aub_dump_bmp(struct gl_context *ctx)
-{
- struct gl_framebuffer *fb = ctx->DrawBuffer;
-
- for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
- struct intel_renderbuffer *irb =
- intel_renderbuffer(fb->_ColorDrawBuffers[i]);
-
- if (irb && irb->mt) {
- enum aub_dump_bmp_format format;
-
- switch (irb->Base.Base.Format) {
- case MESA_FORMAT_B8G8R8A8_UNORM:
- case MESA_FORMAT_B8G8R8X8_UNORM:
- format = AUB_DUMP_BMP_FORMAT_ARGB_8888;
- break;
- default:
- continue;
- }
-
- drm_intel_gem_bo_aub_dump_bmp(irb->mt->bo,
- irb->draw_x,
- irb->draw_y,
- irb->Base.Base.Width,
- irb->Base.Base.Height,
- format,
- irb->mt->pitch,
- 0);
- }
- }
-}
-
static const __DRItexBufferExtension intelTexBufferExtension = {
.base = { __DRI_TEX_BUFFER, 3 },
brw->need_flush_throttle = true;
intel_batchbuffer_flush(brw);
-
- if (INTEL_DEBUG & DEBUG_AUB) {
- aub_dump_bmp(ctx);
- }
}
/**
{ __DRI_IMAGE_FOURCC_XBGR8888, __DRI_IMAGE_COMPONENTS_RGB, 1,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_XBGR8888, 4 }, } },
+ { __DRI_IMAGE_FOURCC_ARGB1555, __DRI_IMAGE_COMPONENTS_RGBA, 1,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_ARGB1555, 2 } } },
+
{ __DRI_IMAGE_FOURCC_RGB565, __DRI_IMAGE_COMPONENTS_RGB, 1,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_RGB565, 2 } } },
{ __DRI_IMAGE_FOURCC_R8, __DRI_IMAGE_COMPONENTS_R, 1,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 }, } },
+ { __DRI_IMAGE_FOURCC_R16, __DRI_IMAGE_COMPONENTS_R, 1,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R16, 1 }, } },
+
{ __DRI_IMAGE_FOURCC_GR88, __DRI_IMAGE_COMPONENTS_RG, 1,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_GR88, 2 }, } },
+ { __DRI_IMAGE_FOURCC_GR1616, __DRI_IMAGE_COMPONENTS_RG, 1,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_GR1616, 2 }, } },
+
{ __DRI_IMAGE_FOURCC_YUV410, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
{ 1, 2, 2, __DRI_IMAGE_FORMAT_R8, 1 },
{ 1, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
{ 2, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 } } },
+ { __DRI_IMAGE_FOURCC_YVU410, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 2, 2, 2, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 1, 2, 2, __DRI_IMAGE_FORMAT_R8, 1 } } },
+
+ { __DRI_IMAGE_FOURCC_YVU411, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 2, 2, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 1, 2, 0, __DRI_IMAGE_FORMAT_R8, 1 } } },
+
+ { __DRI_IMAGE_FOURCC_YVU420, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 2, 1, 1, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 1, 1, 1, __DRI_IMAGE_FORMAT_R8, 1 } } },
+
+ { __DRI_IMAGE_FOURCC_YVU422, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 2, 1, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 1, 1, 0, __DRI_IMAGE_FORMAT_R8, 1 } } },
+
+ { __DRI_IMAGE_FOURCC_YVU444, __DRI_IMAGE_COMPONENTS_Y_U_V, 3,
+ { { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 2, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
+ { 1, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 } } },
+
{ __DRI_IMAGE_FOURCC_NV12, __DRI_IMAGE_COMPONENTS_Y_UV, 2,
{ { 0, 0, 0, __DRI_IMAGE_FORMAT_R8, 1 },
{ 1, 1, 1, __DRI_IMAGE_FORMAT_GR88, 2 } } },
}
static __DRIimage *
-intel_allocate_image(int dri_format, void *loaderPrivate)
+intel_allocate_image(struct intel_screen *screen, int dri_format,
+ void *loaderPrivate)
{
__DRIimage *image;
if (image == NULL)
return NULL;
+ image->screen = screen;
image->dri_format = dri_format;
image->offset = 0;
}
static __DRIimage *
-intel_create_image_from_name(__DRIscreen *screen,
+intel_create_image_from_name(__DRIscreen *dri_screen,
int width, int height, int format,
int name, int pitch, void *loaderPrivate)
{
- struct intel_screen *intelScreen = screen->driverPrivate;
+ struct intel_screen *screen = dri_screen->driverPrivate;
__DRIimage *image;
int cpp;
- image = intel_allocate_image(format, loaderPrivate);
+ image = intel_allocate_image(screen, format, loaderPrivate);
if (image == NULL)
return NULL;
image->width = width;
image->height = height;
image->pitch = pitch * cpp;
- image->bo = drm_intel_bo_gem_create_from_name(intelScreen->bufmgr, "image",
+ image->bo = drm_intel_bo_gem_create_from_name(screen->bufmgr, "image",
name);
if (!image->bo) {
free(image);
return NULL;
}
- return image;
+ return image;
}
static __DRIimage *
free(image);
}
+enum modifier_priority {
+ MODIFIER_PRIORITY_INVALID = 0,
+ MODIFIER_PRIORITY_LINEAR,
+ MODIFIER_PRIORITY_X,
+ MODIFIER_PRIORITY_Y,
+};
+
+const uint64_t priority_to_modifier[] = {
+ [MODIFIER_PRIORITY_INVALID] = DRM_FORMAT_MOD_INVALID,
+ [MODIFIER_PRIORITY_LINEAR] = DRM_FORMAT_MOD_LINEAR,
+ [MODIFIER_PRIORITY_X] = I915_FORMAT_MOD_X_TILED,
+ [MODIFIER_PRIORITY_Y] = I915_FORMAT_MOD_Y_TILED,
+};
+
+static uint64_t
+select_best_modifier(struct gen_device_info *devinfo,
+ const uint64_t *modifiers,
+ const unsigned count)
+{
+ enum modifier_priority prio = MODIFIER_PRIORITY_INVALID;
+
+ for (int i = 0; i < count; i++) {
+ switch (modifiers[i]) {
+ case I915_FORMAT_MOD_Y_TILED:
+ prio = MAX2(prio, MODIFIER_PRIORITY_Y);
+ break;
+ case I915_FORMAT_MOD_X_TILED:
+ prio = MAX2(prio, MODIFIER_PRIORITY_X);
+ break;
+ case DRM_FORMAT_MOD_LINEAR:
+ prio = MAX2(prio, MODIFIER_PRIORITY_LINEAR);
+ break;
+ case DRM_FORMAT_MOD_INVALID:
+ default:
+ break;
+ }
+ }
+
+ return priority_to_modifier[prio];
+}
+
static __DRIimage *
-intel_create_image(__DRIscreen *screen,
- int width, int height, int format,
- unsigned int use,
- void *loaderPrivate)
+intel_create_image_common(__DRIscreen *dri_screen,
+ int width, int height, int format,
+ unsigned int use,
+ const uint64_t *modifiers,
+ unsigned count,
+ void *loaderPrivate)
{
__DRIimage *image;
- struct intel_screen *intelScreen = screen->driverPrivate;
- uint32_t tiling;
+ struct intel_screen *screen = dri_screen->driverPrivate;
+ /* Historically, X-tiled was the default, and so lack of modifier means
+ * X-tiled.
+ */
+ uint32_t tiling = I915_TILING_X;
int cpp;
unsigned long pitch;
- tiling = I915_TILING_X;
+ /* Callers of this may specify a modifier, or a dri usage, but not both. The
+ * newer modifier interface deprecates the older usage flags newer modifier
+ * interface deprecates the older usage flags.
+ */
+ assert(!(use && count));
+
+ uint64_t modifier = select_best_modifier(&screen->devinfo, modifiers, count);
+ switch (modifier) {
+ case I915_FORMAT_MOD_X_TILED:
+ assert(tiling == I915_TILING_X);
+ break;
+ case DRM_FORMAT_MOD_LINEAR:
+ tiling = I915_TILING_NONE;
+ break;
+ case I915_FORMAT_MOD_Y_TILED:
+ tiling = I915_TILING_Y;
+ break;
+ case DRM_FORMAT_MOD_INVALID:
+ if (modifiers)
+ return NULL;
+ default:
+ break;
+ }
+
if (use & __DRI_IMAGE_USE_CURSOR) {
if (width != 64 || height != 64)
return NULL;
if (use & __DRI_IMAGE_USE_LINEAR)
tiling = I915_TILING_NONE;
- image = intel_allocate_image(format, loaderPrivate);
+ image = intel_allocate_image(screen, format, loaderPrivate);
if (image == NULL)
return NULL;
-
cpp = _mesa_get_format_bytes(image->format);
- image->bo = drm_intel_bo_alloc_tiled(intelScreen->bufmgr, "image",
+ image->bo = drm_intel_bo_alloc_tiled(screen->bufmgr, "image",
width, height, cpp, &tiling,
&pitch, 0);
if (image->bo == NULL) {
image->width = width;
image->height = height;
image->pitch = pitch;
+ image->modifier = modifier;
return image;
}
+static __DRIimage *
+intel_create_image(__DRIscreen *dri_screen,
+ int width, int height, int format,
+ unsigned int use,
+ void *loaderPrivate)
+{
+ return intel_create_image_common(dri_screen, width, height, format, use, NULL, 0,
+ loaderPrivate);
+}
+
+static __DRIimage *
+intel_create_image_with_modifiers(__DRIscreen *dri_screen,
+ int width, int height, int format,
+ const uint64_t *modifiers,
+ const unsigned count,
+ void *loaderPrivate)
+{
+ return intel_create_image_common(dri_screen, width, height, format, 0,
+ modifiers, count, loaderPrivate);
+}
+
static GLboolean
intel_query_image(__DRIimage *image, int attrib, int *value)
{
*value = image->planar_format->components;
return true;
case __DRI_IMAGE_ATTRIB_FD:
- if (drm_intel_bo_gem_export_to_prime(image->bo, value) == 0)
- return true;
- return false;
+ return !drm_intel_bo_gem_export_to_prime(image->bo, value);
case __DRI_IMAGE_ATTRIB_FOURCC:
- if (intel_lookup_fourcc(image->dri_format, value))
- return true;
- return false;
+ return intel_lookup_fourcc(image->dri_format, value);
case __DRI_IMAGE_ATTRIB_NUM_PLANES:
*value = 1;
return true;
+ case __DRI_IMAGE_ATTRIB_OFFSET:
+ *value = image->offset;
+ return true;
+ case __DRI_IMAGE_ATTRIB_MODIFIER_LOWER:
+ *value = (image->modifier & 0xffffffff);
+ return true;
+ case __DRI_IMAGE_ATTRIB_MODIFIER_UPPER:
+ *value = ((image->modifier >> 32) & 0xffffffff);
+ return true;
default:
return false;
}
static __DRIimage *
-intel_create_image_from_names(__DRIscreen *screen,
+intel_create_image_from_names(__DRIscreen *dri_screen,
int width, int height, int fourcc,
int *names, int num_names,
int *strides, int *offsets,
__DRIimage *image;
int i, index;
- if (screen == NULL || names == NULL || num_names != 1)
+ if (dri_screen == NULL || names == NULL || num_names != 1)
return NULL;
f = intel_image_format_lookup(fourcc);
if (f == NULL)
return NULL;
- image = intel_create_image_from_name(screen, width, height,
+ image = intel_create_image_from_name(dri_screen, width, height,
__DRI_IMAGE_FORMAT_NONE,
names[0], strides[0],
loaderPrivate);
}
static __DRIimage *
-intel_create_image_from_fds(__DRIscreen *screen,
+intel_create_image_from_fds(__DRIscreen *dri_screen,
int width, int height, int fourcc,
int *fds, int num_fds, int *strides, int *offsets,
void *loaderPrivate)
{
- struct intel_screen *intelScreen = screen->driverPrivate;
+ struct intel_screen *screen = dri_screen->driverPrivate;
struct intel_image_format *f;
__DRIimage *image;
int i, index;
- if (fds == NULL || num_fds != 1)
+ if (fds == NULL || num_fds < 1)
return NULL;
+ /* We only support all planes from the same bo */
+ for (i = 0; i < num_fds; i++)
+ if (fds[0] != fds[i])
+ return NULL;
+
f = intel_image_format_lookup(fourcc);
if (f == NULL)
return NULL;
if (f->nplanes == 1)
- image = intel_allocate_image(f->planes[0].dri_format, loaderPrivate);
+ image = intel_allocate_image(screen, f->planes[0].dri_format,
+ loaderPrivate);
else
- image = intel_allocate_image(__DRI_IMAGE_FORMAT_NONE, loaderPrivate);
+ image = intel_allocate_image(screen, __DRI_IMAGE_FORMAT_NONE,
+ loaderPrivate);
if (image == NULL)
return NULL;
- image->bo = drm_intel_bo_gem_create_from_prime(intelScreen->bufmgr,
- fds[0],
- height * strides[0]);
- if (image->bo == NULL) {
- free(image);
- return NULL;
- }
image->width = width;
image->height = height;
image->pitch = strides[0];
image->planar_format = f;
+ int size = 0;
for (i = 0; i < f->nplanes; i++) {
index = f->planes[i].buffer_index;
image->offsets[index] = offsets[index];
image->strides[index] = strides[index];
+
+ const int plane_height = height >> f->planes[i].height_shift;
+ const int end = offsets[index] + plane_height * strides[index];
+ if (size < end)
+ size = end;
+ }
+
+ image->bo = drm_intel_bo_gem_create_from_prime(screen->bufmgr,
+ fds[0], size);
+ if (image->bo == NULL) {
+ free(image);
+ return NULL;
}
if (f->nplanes == 1) {
}
static __DRIimage *
-intel_create_image_from_dma_bufs(__DRIscreen *screen,
+intel_create_image_from_dma_bufs(__DRIscreen *dri_screen,
int width, int height, int fourcc,
int *fds, int num_fds,
int *strides, int *offsets,
__DRIimage *image;
struct intel_image_format *f = intel_image_format_lookup(fourcc);
- /* For now only packed formats that have native sampling are supported. */
- if (!f || f->nplanes != 1) {
+ if (!f) {
*error = __DRI_IMAGE_ERROR_BAD_MATCH;
return NULL;
}
- image = intel_create_image_from_fds(screen, width, height, fourcc, fds,
+ image = intel_create_image_from_fds(dri_screen, width, height, fourcc, fds,
num_fds, strides, offsets,
loaderPrivate);
offset = parent->offsets[index];
stride = parent->strides[index];
- image = intel_allocate_image(dri_format, loaderPrivate);
+ image = intel_allocate_image(parent->screen, dri_format, loaderPrivate);
if (image == NULL)
return NULL;
}
static const __DRIimageExtension intelImageExtension = {
- .base = { __DRI_IMAGE, 11 },
+ .base = { __DRI_IMAGE, 14 },
.createImageFromName = intel_create_image_from_name,
.createImageFromRenderbuffer = intel_create_image_from_renderbuffer,
.createImageFromFds = intel_create_image_from_fds,
.createImageFromDmaBufs = intel_create_image_from_dma_bufs,
.blitImage = NULL,
- .getCapabilities = NULL
+ .getCapabilities = NULL,
+ .mapImage = NULL,
+ .unmapImage = NULL,
+ .createImageWithModifiers = intel_create_image_with_modifiers,
};
static int
-brw_query_renderer_integer(__DRIscreen *psp, int param, unsigned int *value)
+brw_query_renderer_integer(__DRIscreen *dri_screen,
+ int param, unsigned int *value)
{
- const struct intel_screen *const intelScreen =
- (struct intel_screen *) psp->driverPrivate;
+ const struct intel_screen *const screen =
+ (struct intel_screen *) dri_screen->driverPrivate;
switch (param) {
case __DRI2_RENDERER_VENDOR_ID:
value[0] = 0x8086;
return 0;
case __DRI2_RENDERER_DEVICE_ID:
- value[0] = intelScreen->deviceID;
+ value[0] = screen->deviceID;
return 0;
case __DRI2_RENDERER_ACCELERATED:
value[0] = 1;
size_t aper_size;
size_t mappable_size;
- drm_intel_get_aperture_sizes(psp->fd, &mappable_size, &aper_size);
+ drm_intel_get_aperture_sizes(dri_screen->fd, &mappable_size, &aper_size);
const unsigned gpu_mappable_megabytes =
(aper_size / (1024 * 1024)) * 3 / 4;
case __DRI2_RENDERER_UNIFIED_MEMORY_ARCHITECTURE:
value[0] = 1;
return 0;
+ case __DRI2_RENDERER_HAS_TEXTURE_3D:
+ value[0] = 1;
+ return 0;
default:
- return driQueryRendererIntegerCommon(psp, param, value);
+ return driQueryRendererIntegerCommon(dri_screen, param, value);
}
return -1;
}
static int
-brw_query_renderer_string(__DRIscreen *psp, int param, const char **value)
+brw_query_renderer_string(__DRIscreen *dri_screen,
+ int param, const char **value)
{
- const struct intel_screen *intelScreen =
- (struct intel_screen *) psp->driverPrivate;
+ const struct intel_screen *screen =
+ (struct intel_screen *) dri_screen->driverPrivate;
switch (param) {
case __DRI2_RENDERER_VENDOR_ID:
value[0] = brw_vendor_string;
return 0;
case __DRI2_RENDERER_DEVICE_ID:
- value[0] = brw_get_renderer_string(intelScreen->deviceID);
+ value[0] = brw_get_renderer_string(screen);
return 0;
default:
break;
.base = { __DRI2_ROBUSTNESS, 1 }
};
-static const __DRIextension *intelScreenExtensions[] = {
+static const __DRIextension *screenExtensions[] = {
&intelTexBufferExtension.base,
&intelFenceExtension.base,
&intelFlushExtension.base,
NULL
};
-static bool
-intel_get_param(__DRIscreen *psp, int param, int *value)
+static int
+intel_get_param(struct intel_screen *screen, int param, int *value)
{
- int ret;
+ int ret = 0;
struct drm_i915_getparam gp;
memset(&gp, 0, sizeof(gp));
gp.param = param;
gp.value = value;
- ret = drmCommandWriteRead(psp->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
- if (ret) {
+ if (drmIoctl(screen->driScrnPriv->fd, DRM_IOCTL_I915_GETPARAM, &gp) == -1) {
+ ret = -errno;
if (ret != -EINVAL)
- _mesa_warning(NULL, "drm_i915_getparam: %d", ret);
- return false;
+ _mesa_warning(NULL, "drm_i915_getparam: %d", ret);
}
- return true;
+ return ret;
}
static bool
-intel_get_boolean(__DRIscreen *psp, int param)
+intel_get_boolean(struct intel_screen *screen, int param)
{
int value = 0;
- return intel_get_param(psp, param, &value) && value;
+ return (intel_get_param(screen, param, &value) == 0) && value;
+}
+
+static int
+intel_get_integer(struct intel_screen *screen, int param)
+{
+ int value = -1;
+
+ if (intel_get_param(screen, param, &value) == 0)
+ return value;
+
+ return -1;
}
static void
intelDestroyScreen(__DRIscreen * sPriv)
{
- struct intel_screen *intelScreen = sPriv->driverPrivate;
+ struct intel_screen *screen = sPriv->driverPrivate;
- dri_bufmgr_destroy(intelScreen->bufmgr);
- driDestroyOptionInfo(&intelScreen->optionCache);
+ dri_bufmgr_destroy(screen->bufmgr);
+ driDestroyOptionInfo(&screen->optionCache);
- ralloc_free(intelScreen);
+ ralloc_free(screen);
sPriv->driverPrivate = NULL;
}
* This is called when we need to set up GL rendering to a new X window.
*/
static GLboolean
-intelCreateBuffer(__DRIscreen * driScrnPriv,
+intelCreateBuffer(__DRIscreen *dri_screen,
__DRIdrawable * driDrawPriv,
const struct gl_config * mesaVis, GLboolean isPixmap)
{
struct intel_renderbuffer *rb;
- struct intel_screen *screen = (struct intel_screen*) driScrnPriv->driverPrivate;
+ struct intel_screen *screen = (struct intel_screen *)
+ dri_screen->driverPrivate;
mesa_format rgbFormat;
- unsigned num_samples = intel_quantize_num_samples(screen, mesaVis->samples);
+ unsigned num_samples =
+ intel_quantize_num_samples(screen, mesaVis->samples);
struct gl_framebuffer *fb;
if (isPixmap)
fb->Visual.samples = num_samples;
}
- if (mesaVis->redBits == 5)
- rgbFormat = MESA_FORMAT_B5G6R5_UNORM;
- else if (mesaVis->sRGBCapable)
- rgbFormat = MESA_FORMAT_B8G8R8A8_SRGB;
- else if (mesaVis->alphaBits == 0)
- rgbFormat = MESA_FORMAT_B8G8R8X8_UNORM;
- else {
- rgbFormat = MESA_FORMAT_B8G8R8A8_SRGB;
+ if (mesaVis->redBits == 5) {
+ rgbFormat = mesaVis->redMask == 0x1f ? MESA_FORMAT_R5G6B5_UNORM
+ : MESA_FORMAT_B5G6R5_UNORM;
+ } else if (mesaVis->sRGBCapable) {
+ rgbFormat = mesaVis->redMask == 0xff ? MESA_FORMAT_R8G8B8A8_SRGB
+ : MESA_FORMAT_B8G8R8A8_SRGB;
+ } else if (mesaVis->alphaBits == 0) {
+ rgbFormat = mesaVis->redMask == 0xff ? MESA_FORMAT_R8G8B8X8_UNORM
+ : MESA_FORMAT_B8G8R8X8_UNORM;
+ } else {
+ rgbFormat = mesaVis->redMask == 0xff ? MESA_FORMAT_R8G8B8A8_SRGB
+ : MESA_FORMAT_B8G8R8A8_SRGB;
fb->Visual.sRGBCapable = true;
}
if (mesaVis->depthBits == 24) {
assert(mesaVis->stencilBits == 8);
- if (screen->devinfo->has_hiz_and_separate_stencil) {
+ if (screen->devinfo.has_hiz_and_separate_stencil) {
rb = intel_create_private_renderbuffer(MESA_FORMAT_Z24_UNORM_X8_UINT,
num_samples);
_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &rb->Base.Base);
_mesa_reference_framebuffer(&fb, NULL);
}
+static void
+intel_detect_sseu(struct intel_screen *screen)
+{
+ assert(screen->devinfo.gen >= 8);
+ int ret;
+
+ screen->subslice_total = -1;
+ screen->eu_total = -1;
+
+ ret = intel_get_param(screen, I915_PARAM_SUBSLICE_TOTAL,
+ &screen->subslice_total);
+ if (ret < 0 && ret != -EINVAL)
+ goto err_out;
+
+ ret = intel_get_param(screen,
+ I915_PARAM_EU_TOTAL, &screen->eu_total);
+ if (ret < 0 && ret != -EINVAL)
+ goto err_out;
+
+ /* Without this information, we cannot get the right Braswell brandstrings,
+ * and we have to use conservative numbers for GPGPU on many platforms, but
+ * otherwise, things will just work.
+ */
+ if (screen->subslice_total < 1 || screen->eu_total < 1)
+ _mesa_warning(NULL,
+ "Kernel 4.1 required to properly query GPU properties.\n");
+
+ return;
+
+err_out:
+ screen->subslice_total = -1;
+ screen->eu_total = -1;
+ _mesa_warning(NULL, "Failed to query GPU properties (%s).\n", strerror(-ret));
+}
+
static bool
-intel_init_bufmgr(struct intel_screen *intelScreen)
+intel_init_bufmgr(struct intel_screen *screen)
{
- __DRIscreen *spriv = intelScreen->driScrnPriv;
+ __DRIscreen *dri_screen = screen->driScrnPriv;
- intelScreen->no_hw = getenv("INTEL_NO_HW") != NULL;
+ screen->no_hw = getenv("INTEL_NO_HW") != NULL;
- intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
- if (intelScreen->bufmgr == NULL) {
+ screen->bufmgr = intel_bufmgr_gem_init(dri_screen->fd, BATCH_SZ);
+ if (screen->bufmgr == NULL) {
fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
__func__, __LINE__);
return false;
}
- drm_intel_bufmgr_gem_enable_fenced_relocs(intelScreen->bufmgr);
+ drm_intel_bufmgr_gem_enable_fenced_relocs(screen->bufmgr);
- if (!intel_get_boolean(spriv, I915_PARAM_HAS_RELAXED_DELTA)) {
+ if (!intel_get_boolean(screen, I915_PARAM_HAS_RELAXED_DELTA)) {
fprintf(stderr, "[%s: %u] Kernel 2.6.39 required.\n", __func__, __LINE__);
return false;
}
return 0;
}
+ /**
+ * Test if we can use MI_LOAD_REGISTER_MEM from an untrusted batchbuffer.
+ *
+ * Some combinations of hardware and kernel versions allow this feature,
+ * while others don't. Instead of trying to enumerate every case, just
+ * try and write a register and see if works.
+ */
+static bool
+intel_detect_pipelined_register(struct intel_screen *screen,
+ int reg, uint32_t expected_value, bool reset)
+{
+ drm_intel_bo *results, *bo;
+ uint32_t *batch;
+ uint32_t offset = 0;
+ bool success = false;
+
+ /* Create a zero'ed temporary buffer for reading our results */
+ results = drm_intel_bo_alloc(screen->bufmgr, "registers", 4096, 0);
+ if (results == NULL)
+ goto err;
+
+ bo = drm_intel_bo_alloc(screen->bufmgr, "batchbuffer", 4096, 0);
+ if (bo == NULL)
+ goto err_results;
+
+ if (drm_intel_bo_map(bo, 1))
+ goto err_batch;
+
+ batch = bo->virtual;
+
+ /* Write the register. */
+ *batch++ = MI_LOAD_REGISTER_IMM | (3 - 2);
+ *batch++ = reg;
+ *batch++ = expected_value;
+
+ /* Save the register's value back to the buffer. */
+ *batch++ = MI_STORE_REGISTER_MEM | (3 - 2);
+ *batch++ = reg;
+ drm_intel_bo_emit_reloc(bo, (char *)batch -(char *)bo->virtual,
+ results, offset*sizeof(uint32_t),
+ I915_GEM_DOMAIN_INSTRUCTION,
+ I915_GEM_DOMAIN_INSTRUCTION);
+ *batch++ = results->offset + offset*sizeof(uint32_t);
+
+ /* And afterwards clear the register */
+ if (reset) {
+ *batch++ = MI_LOAD_REGISTER_IMM | (3 - 2);
+ *batch++ = reg;
+ *batch++ = 0;
+ }
+
+ *batch++ = MI_BATCH_BUFFER_END;
+
+ drm_intel_bo_mrb_exec(bo, ALIGN((char *)batch - (char *)bo->virtual, 8),
+ NULL, 0, 0,
+ I915_EXEC_RENDER);
+
+ /* Check whether the value got written. */
+ if (drm_intel_bo_map(results, false) == 0) {
+ success = *((uint32_t *)results->virtual + offset) == expected_value;
+ drm_intel_bo_unmap(results);
+ }
+
+err_batch:
+ drm_intel_bo_unreference(bo);
+err_results:
+ drm_intel_bo_unreference(results);
+err:
+ return success;
+}
+
+static bool
+intel_detect_pipelined_so(struct intel_screen *screen)
+{
+ const struct gen_device_info *devinfo = &screen->devinfo;
+
+ /* Supposedly, Broadwell just works. */
+ if (devinfo->gen >= 8)
+ return true;
+
+ if (devinfo->gen <= 6)
+ return false;
+
+ /* See the big explanation about command parser versions below */
+ if (screen->cmd_parser_version >= (devinfo->is_haswell ? 7 : 2))
+ return true;
+
+ /* We use SO_WRITE_OFFSET0 since you're supposed to write it (unlike the
+ * statistics registers), and we already reset it to zero before using it.
+ */
+ return intel_detect_pipelined_register(screen,
+ GEN7_SO_WRITE_OFFSET(0),
+ 0x1337d0d0,
+ false);
+}
+
/**
* Return array of MSAA modes supported by the hardware. The array is
* zero-terminated and sorted in decreasing order.
const int*
intel_supported_msaa_modes(const struct intel_screen *screen)
{
+ static const int gen9_modes[] = {16, 8, 4, 2, 0, -1};
static const int gen8_modes[] = {8, 4, 2, 0, -1};
static const int gen7_modes[] = {8, 4, 0, -1};
static const int gen6_modes[] = {4, 0, -1};
static const int gen4_modes[] = {0, -1};
- if (screen->devinfo->gen >= 8) {
+ if (screen->devinfo.gen >= 9) {
+ return gen9_modes;
+ } else if (screen->devinfo.gen >= 8) {
return gen8_modes;
- } else if (screen->devinfo->gen >= 7) {
+ } else if (screen->devinfo.gen >= 7) {
return gen7_modes;
- } else if (screen->devinfo->gen == 6) {
+ } else if (screen->devinfo.gen == 6) {
return gen6_modes;
} else {
return gen4_modes;
static const uint8_t multisample_samples[2] = {4, 8};
struct intel_screen *screen = dri_screen->driverPrivate;
- const struct brw_device_info *devinfo = screen->devinfo;
+ const struct gen_device_info *devinfo = &screen->devinfo;
uint8_t depth_bits[4], stencil_bits[4];
__DRIconfig **configs = NULL;
num_depth_stencil_bits,
back_buffer_modes, 2,
singlesample_samples, 1,
- false);
+ false, false);
configs = driConcatConfigs(configs, new_configs);
}
depth_bits, stencil_bits, 1,
back_buffer_modes, 1,
singlesample_samples, 1,
- true);
+ true, false);
configs = driConcatConfigs(configs, new_configs);
}
back_buffer_modes, 1,
multisample_samples,
num_msaa_modes,
- false);
+ false, false);
configs = driConcatConfigs(configs, new_configs);
}
static void
set_max_gl_versions(struct intel_screen *screen)
{
- __DRIscreen *psp = screen->driScrnPriv;
+ __DRIscreen *dri_screen = screen->driScrnPriv;
+ const bool has_astc = screen->devinfo.gen >= 9;
- switch (screen->devinfo->gen) {
+ switch (screen->devinfo.gen) {
case 9:
case 8:
+ dri_screen->max_gl_core_version = 45;
+ dri_screen->max_gl_compat_version = 30;
+ dri_screen->max_gl_es1_version = 11;
+ dri_screen->max_gl_es2_version = has_astc ? 32 : 31;
+ break;
case 7:
+ dri_screen->max_gl_core_version = 33;
+ if (screen->devinfo.is_haswell &&
+ can_do_pipelined_register_writes(screen)) {
+ dri_screen->max_gl_core_version = 42;
+ if (can_do_compute_dispatch(screen))
+ dri_screen->max_gl_core_version = 43;
+ if (can_do_mi_math_and_lrr(screen))
+ dri_screen->max_gl_core_version = 45;
+ }
+ dri_screen->max_gl_compat_version = 30;
+ dri_screen->max_gl_es1_version = 11;
+ dri_screen->max_gl_es2_version = screen->devinfo.is_haswell ? 31 : 30;
+ break;
case 6:
- psp->max_gl_core_version = 33;
- psp->max_gl_compat_version = 30;
- psp->max_gl_es1_version = 11;
- psp->max_gl_es2_version = 30;
+ dri_screen->max_gl_core_version = 33;
+ dri_screen->max_gl_compat_version = 30;
+ dri_screen->max_gl_es1_version = 11;
+ dri_screen->max_gl_es2_version = 30;
break;
case 5:
case 4:
- psp->max_gl_core_version = 0;
- psp->max_gl_compat_version = 21;
- psp->max_gl_es1_version = 11;
- psp->max_gl_es2_version = 20;
+ dri_screen->max_gl_core_version = 0;
+ dri_screen->max_gl_compat_version = 21;
+ dri_screen->max_gl_es1_version = 11;
+ dri_screen->max_gl_es2_version = 20;
break;
default:
unreachable("unrecognized intel_screen::gen");
}
}
-static int
+/**
+ * Return the revision (generally the revid field of the PCI header) of the
+ * graphics device.
+ *
+ * XXX: This function is useful to keep around even if it is not currently in
+ * use. It is necessary for new platforms and revision specific workarounds or
+ * features. Please don't remove it so that we know it at least continues to
+ * build.
+ */
+static __attribute__((__unused__)) int
brw_get_revision(int fd)
{
struct drm_i915_getparam gp;
return revision;
}
-/* Drop when RS headers get pulled to libdrm */
-#ifndef I915_PARAM_HAS_RESOURCE_STREAMER
-#define I915_PARAM_HAS_RESOURCE_STREAMER 36
-#endif
+static void
+shader_debug_log_mesa(void *data, const char *fmt, ...)
+{
+ struct brw_context *brw = (struct brw_context *)data;
+ va_list args;
+
+ va_start(args, fmt);
+ GLuint msg_id = 0;
+ _mesa_gl_vdebug(&brw->ctx, &msg_id,
+ MESA_DEBUG_SOURCE_SHADER_COMPILER,
+ MESA_DEBUG_TYPE_OTHER,
+ MESA_DEBUG_SEVERITY_NOTIFICATION, fmt, args);
+ va_end(args);
+}
+
+static void
+shader_perf_log_mesa(void *data, const char *fmt, ...)
+{
+ struct brw_context *brw = (struct brw_context *)data;
+
+ va_list args;
+ va_start(args, fmt);
+
+ if (unlikely(INTEL_DEBUG & DEBUG_PERF)) {
+ va_list args_copy;
+ va_copy(args_copy, args);
+ vfprintf(stderr, fmt, args_copy);
+ va_end(args_copy);
+ }
+
+ if (brw->perf_debug) {
+ GLuint msg_id = 0;
+ _mesa_gl_vdebug(&brw->ctx, &msg_id,
+ MESA_DEBUG_SOURCE_SHADER_COMPILER,
+ MESA_DEBUG_TYPE_PERFORMANCE,
+ MESA_DEBUG_SEVERITY_MEDIUM, fmt, args);
+ }
+ va_end(args);
+}
/**
* This is the driver specific part of the createNewScreen entry point.
* \return the struct gl_config supported by this driver
*/
static const
-__DRIconfig **intelInitScreen2(__DRIscreen *psp)
+__DRIconfig **intelInitScreen2(__DRIscreen *dri_screen)
{
- struct intel_screen *intelScreen;
+ struct intel_screen *screen;
- if (psp->image.loader) {
- } else if (psp->dri2.loader->base.version <= 2 ||
- psp->dri2.loader->getBuffersWithFormat == NULL) {
+ if (dri_screen->image.loader) {
+ } else if (dri_screen->dri2.loader->base.version <= 2 ||
+ dri_screen->dri2.loader->getBuffersWithFormat == NULL) {
fprintf(stderr,
"\nERROR! DRI2 loader with getBuffersWithFormat() "
"support required\n");
- return false;
+ return NULL;
}
/* Allocate the private area */
- intelScreen = rzalloc(NULL, struct intel_screen);
- if (!intelScreen) {
+ screen = rzalloc(NULL, struct intel_screen);
+ if (!screen) {
fprintf(stderr, "\nERROR! Allocating private area failed\n");
- return false;
+ return NULL;
}
/* parse information in __driConfigOptions */
- driParseOptionInfo(&intelScreen->optionCache, brw_config_options.xml);
+ driParseOptionInfo(&screen->optionCache, brw_config_options.xml);
- intelScreen->driScrnPriv = psp;
- psp->driverPrivate = (void *) intelScreen;
+ screen->driScrnPriv = dri_screen;
+ dri_screen->driverPrivate = (void *) screen;
- if (!intel_init_bufmgr(intelScreen))
- return false;
+ if (!intel_init_bufmgr(screen))
+ return NULL;
- intelScreen->deviceID = drm_intel_bufmgr_gem_get_devid(intelScreen->bufmgr);
- intelScreen->devinfo = brw_get_device_info(intelScreen->deviceID,
- brw_get_revision(psp->fd));
- if (!intelScreen->devinfo)
- return false;
+ screen->deviceID = drm_intel_bufmgr_gem_get_devid(screen->bufmgr);
+ if (!gen_get_device_info(screen->deviceID, &screen->devinfo))
+ return NULL;
+
+ const struct gen_device_info *devinfo = &screen->devinfo;
- brw_process_intel_debug_variable(intelScreen);
+ brw_process_intel_debug_variable();
- intelScreen->hw_must_use_separate_stencil = intelScreen->devinfo->gen >= 7;
+ if (INTEL_DEBUG & DEBUG_BUFMGR)
+ dri_bufmgr_set_debug(screen->bufmgr, true);
+
+ if ((INTEL_DEBUG & DEBUG_SHADER_TIME) && devinfo->gen < 7) {
+ fprintf(stderr,
+ "shader_time debugging requires gen7 (Ivybridge) or better.\n");
+ INTEL_DEBUG &= ~DEBUG_SHADER_TIME;
+ }
+
+ if (intel_get_integer(screen, I915_PARAM_MMAP_GTT_VERSION) >= 1) {
+ /* Theorectically unlimited! At least for individual objects...
+ *
+ * Currently the entire (global) address space for all GTT maps is
+ * limited to 64bits. That is all objects on the system that are
+ * setup for GTT mmapping must fit within 64bits. An attempt to use
+ * one that exceeds the limit with fail in drm_intel_bo_map_gtt().
+ *
+ * Long before we hit that limit, we will be practically limited by
+ * that any single object must fit in physical memory (RAM). The upper
+ * limit on the CPU's address space is currently 48bits (Skylake), of
+ * which only 39bits can be physical memory. (The GPU itself also has
+ * a 48bit addressable virtual space.) We can fit over 32 million
+ * objects of the current maximum allocable size before running out
+ * of mmap space.
+ */
+ screen->max_gtt_map_object_size = UINT64_MAX;
+ } else {
+ /* Estimate the size of the mappable aperture into the GTT. There's an
+ * ioctl to get the whole GTT size, but not one to get the mappable subset.
+ * It turns out it's basically always 256MB, though some ancient hardware
+ * was smaller.
+ */
+ uint32_t gtt_size = 256 * 1024 * 1024;
+
+ /* We don't want to map two objects such that a memcpy between them would
+ * just fault one mapping in and then the other over and over forever. So
+ * we would need to divide the GTT size by 2. Additionally, some GTT is
+ * taken up by things like the framebuffer and the ringbuffer and such, so
+ * be more conservative.
+ */
+ screen->max_gtt_map_object_size = gtt_size / 4;
+ }
+
+ screen->hw_has_swizzling = intel_detect_swizzling(screen);
+ screen->hw_has_timestamp = intel_detect_timestamp(screen);
+
+ /* GENs prior to 8 do not support EU/Subslice info */
+ if (devinfo->gen >= 8) {
+ intel_detect_sseu(screen);
+ } else if (devinfo->gen == 7) {
+ screen->subslice_total = 1 << (devinfo->gt - 1);
+ }
+
+ /* Gen7-7.5 kernel requirements / command parser saga:
+ *
+ * - pre-v3.16:
+ * Haswell and Baytrail cannot use any privileged batchbuffer features.
+ *
+ * Ivybridge has aliasing PPGTT on by default, which accidentally marks
+ * all batches secure, allowing them to use any feature with no checking.
+ * This is effectively equivalent to a command parser version of
+ * \infinity - everything is possible.
+ *
+ * The command parser does not exist, and querying the version will
+ * return -EINVAL.
+ *
+ * - v3.16:
+ * The kernel enables the command parser by default, for systems with
+ * aliasing PPGTT enabled (Ivybridge and Haswell). However, the
+ * hardware checker is still enabled, so Haswell and Baytrail cannot
+ * do anything.
+ *
+ * Ivybridge goes from "everything is possible" to "only what the
+ * command parser allows" (if the user boots with i915.cmd_parser=0,
+ * then everything is possible again). We can only safely use features
+ * allowed by the supported command parser version.
+ *
+ * Annoyingly, I915_PARAM_CMD_PARSER_VERSION reports the static version
+ * implemented by the kernel, even if it's turned off. So, checking
+ * for version > 0 does not mean that you can write registers. We have
+ * to try it and see. The version does, however, indicate the age of
+ * the kernel.
+ *
+ * Instead of matching the hardware checker's behavior of converting
+ * privileged commands to MI_NOOP, it makes execbuf2 start returning
+ * -EINVAL, making it dangerous to try and use privileged features.
+ *
+ * Effective command parser versions:
+ * - Haswell: 0 (reporting 1, writes don't work)
+ * - Baytrail: 0 (reporting 1, writes don't work)
+ * - Ivybridge: 1 (enabled) or infinite (disabled)
+ *
+ * - v3.17:
+ * Baytrail aliasing PPGTT is enabled, making it like Ivybridge:
+ * effectively version 1 (enabled) or infinite (disabled).
+ *
+ * - v3.19: f1f55cc0556031c8ee3fe99dae7251e78b9b653b
+ * Command parser v2 supports predicate writes.
+ *
+ * - Haswell: 0 (reporting 1, writes don't work)
+ * - Baytrail: 2 (enabled) or infinite (disabled)
+ * - Ivybridge: 2 (enabled) or infinite (disabled)
+ *
+ * So version >= 2 is enough to know that Ivybridge and Baytrail
+ * will work. Haswell still can't do anything.
+ *
+ * - v4.0: Version 3 happened. Largely not relevant.
+ *
+ * - v4.1: 6702cf16e0ba8b0129f5aa1b6609d4e9c70bc13b
+ * L3 config registers are properly saved and restored as part
+ * of the hardware context. We can approximately detect this point
+ * in time by checking if I915_PARAM_REVISION is recognized - it
+ * landed in a later commit, but in the same release cycle.
+ *
+ * - v4.2: 245054a1fe33c06ad233e0d58a27ec7b64db9284
+ * Command parser finally gains secure batch promotion. On Haswell,
+ * the hardware checker gets disabled, which finally allows it to do
+ * privileged commands.
+ *
+ * I915_PARAM_CMD_PARSER_VERSION reports 3. Effective versions:
+ * - Haswell: 3 (enabled) or 0 (disabled)
+ * - Baytrail: 3 (enabled) or infinite (disabled)
+ * - Ivybridge: 3 (enabled) or infinite (disabled)
+ *
+ * Unfortunately, detecting this point in time is tricky, because
+ * no version bump happened when this important change occurred.
+ * On Haswell, if we can write any register, then the kernel is at
+ * least this new, and we can start trusting the version number.
+ *
+ * - v4.4: 2bbe6bbb0dc94fd4ce287bdac9e1bd184e23057b and
+ * Command parser reaches version 4, allowing access to Haswell
+ * atomic scratch and chicken3 registers. If version >= 4, we know
+ * the kernel is new enough to support privileged features on all
+ * hardware. However, the user might have disabled it...and the
+ * kernel will still report version 4. So we still have to guess
+ * and check.
+ *
+ * - v4.4: 7b9748cb513a6bef4af87b79f0da3ff7e8b56cd8
+ * Command parser v5 whitelists indirect compute shader dispatch
+ * registers, needed for OpenGL 4.3 and later.
+ *
+ * - v4.8:
+ * Command parser v7 lets us use MI_MATH on Haswell.
+ *
+ * Additionally, the kernel begins reporting version 0 when
+ * the command parser is disabled, allowing us to skip the
+ * guess-and-check step on Haswell. Unfortunately, this also
+ * means that we can no longer use it as an indicator of the
+ * age of the kernel.
+ */
+ if (intel_get_param(screen, I915_PARAM_CMD_PARSER_VERSION,
+ &screen->cmd_parser_version) < 0) {
+ /* Command parser does not exist - getparam is unrecognized */
+ screen->cmd_parser_version = 0;
+ }
- intelScreen->hw_has_swizzling = intel_detect_swizzling(intelScreen);
- intelScreen->hw_has_timestamp = intel_detect_timestamp(intelScreen);
+ if (!intel_detect_pipelined_so(screen)) {
+ /* We can't do anything, so the effective version is 0. */
+ screen->cmd_parser_version = 0;
+ } else {
+ screen->kernel_features |= KERNEL_ALLOWS_SOL_OFFSET_WRITES;
+ }
const char *force_msaa = getenv("INTEL_FORCE_MSAA");
if (force_msaa) {
- intelScreen->winsys_msaa_samples_override =
- intel_quantize_num_samples(intelScreen, atoi(force_msaa));
+ screen->winsys_msaa_samples_override =
+ intel_quantize_num_samples(screen, atoi(force_msaa));
printf("Forcing winsys sample count to %d\n",
- intelScreen->winsys_msaa_samples_override);
+ screen->winsys_msaa_samples_override);
} else {
- intelScreen->winsys_msaa_samples_override = -1;
+ screen->winsys_msaa_samples_override = -1;
}
- set_max_gl_versions(intelScreen);
+ set_max_gl_versions(screen);
/* Notification of GPU resets requires hardware contexts and a kernel new
* enough to support DRM_IOCTL_I915_GET_RESET_STATS. If the ioctl is
*
* Don't even try on pre-Gen6, since we don't attempt to use contexts there.
*/
- if (intelScreen->devinfo->gen >= 6) {
+ if (devinfo->gen >= 6) {
struct drm_i915_reset_stats stats;
memset(&stats, 0, sizeof(stats));
- const int ret = drmIoctl(psp->fd, DRM_IOCTL_I915_GET_RESET_STATS, &stats);
+ const int ret = drmIoctl(dri_screen->fd, DRM_IOCTL_I915_GET_RESET_STATS, &stats);
- intelScreen->has_context_reset_notification =
+ screen->has_context_reset_notification =
(ret != -1 || errno != EINVAL);
}
- struct drm_i915_getparam getparam;
- getparam.param = I915_PARAM_CMD_PARSER_VERSION;
- getparam.value = &intelScreen->cmd_parser_version;
- const int ret = drmIoctl(psp->fd, DRM_IOCTL_I915_GETPARAM, &getparam);
- if (ret == -1)
- intelScreen->cmd_parser_version = 0;
+ if (devinfo->gen >= 8 || screen->cmd_parser_version >= 2)
+ screen->kernel_features |= KERNEL_ALLOWS_PREDICATE_WRITES;
- psp->extensions = !intelScreen->has_context_reset_notification
- ? intelScreenExtensions : intelRobustScreenExtensions;
+ /* Haswell requires command parser version 4 in order to have L3
+ * atomic scratch1 and chicken3 bits
+ */
+ if (devinfo->is_haswell && screen->cmd_parser_version >= 4) {
+ screen->kernel_features |=
+ KERNEL_ALLOWS_HSW_SCRATCH1_AND_ROW_CHICKEN3;
+ }
- intelScreen->compiler = brw_compiler_create(intelScreen,
- intelScreen->devinfo);
+ /* Haswell requires command parser version 6 in order to write to the
+ * MI_MATH GPR registers, and version 7 in order to use
+ * MI_LOAD_REGISTER_REG (which all users of MI_MATH use).
+ */
+ if (devinfo->gen >= 8 ||
+ (devinfo->is_haswell && screen->cmd_parser_version >= 7)) {
+ screen->kernel_features |= KERNEL_ALLOWS_MI_MATH_AND_LRR;
+ }
- if (intelScreen->devinfo->has_resource_streamer) {
- int val = -1;
- getparam.param = I915_PARAM_HAS_RESOURCE_STREAMER;
- getparam.value = &val;
+ /* Gen7 needs at least command parser version 5 to support compute */
+ if (devinfo->gen >= 8 || screen->cmd_parser_version >= 5)
+ screen->kernel_features |= KERNEL_ALLOWS_COMPUTE_DISPATCH;
- drmIoctl(psp->fd, DRM_IOCTL_I915_GETPARAM, &getparam);
- intelScreen->has_resource_streamer = val > 0;
- }
+ dri_screen->extensions = !screen->has_context_reset_notification
+ ? screenExtensions : intelRobustScreenExtensions;
- return (const __DRIconfig**) intel_screen_make_configs(psp);
+ screen->compiler = brw_compiler_create(screen, devinfo);
+ screen->compiler->shader_debug_log = shader_debug_log_mesa;
+ screen->compiler->shader_perf_log = shader_perf_log_mesa;
+ screen->program_id = 1;
+
+ screen->has_exec_fence =
+ intel_get_boolean(screen, I915_PARAM_HAS_EXEC_FENCE);
+
+ return (const __DRIconfig**) intel_screen_make_configs(dri_screen);
}
struct intel_buffer {
};
static __DRIbuffer *
-intelAllocateBuffer(__DRIscreen *screen,
+intelAllocateBuffer(__DRIscreen *dri_screen,
unsigned attachment, unsigned format,
int width, int height)
{
struct intel_buffer *intelBuffer;
- struct intel_screen *intelScreen = screen->driverPrivate;
+ struct intel_screen *screen = dri_screen->driverPrivate;
assert(attachment == __DRI_BUFFER_FRONT_LEFT ||
attachment == __DRI_BUFFER_BACK_LEFT);
if (intelBuffer == NULL)
return NULL;
- /* The front and back buffers are color buffers, which are X tiled. */
+ /* The front and back buffers are color buffers, which are X tiled. GEN9+
+ * supports Y tiled and compressed buffers, but there is no way to plumb that
+ * through to here. */
uint32_t tiling = I915_TILING_X;
unsigned long pitch;
int cpp = format / 8;
- intelBuffer->bo = drm_intel_bo_alloc_tiled(intelScreen->bufmgr,
+ intelBuffer->bo = drm_intel_bo_alloc_tiled(screen->bufmgr,
"intelAllocateBuffer",
width,
height,
}
static void
-intelReleaseBuffer(__DRIscreen *screen, __DRIbuffer *buffer)
+intelReleaseBuffer(__DRIscreen *dri_screen, __DRIbuffer *buffer)
{
struct intel_buffer *intelBuffer = (struct intel_buffer *) buffer;