[Fix, Test] (MG_Backend/DirectGLES, MG_Util, MG_Impl): widen three-channel render targets wherever the driver refuses them

Complementary Reimagined would not load through Espryt on Mali: Iris got
GL_FRAMEBUFFER_UNSUPPORTED building its composite framebuffer, because
colortex1 is RGB8_SNORM and colortex2 is RGB16F - three-channel formats
that no real ES driver can render to (EXT_render_snorm covers R/RG/RGBA
only, and the float extensions exclude the RGB forms). The frontend's
probe cache diagnosed this correctly and then had nothing to offer: the
NoThreeChannelRenderTarget widening machinery existed but was gated to
multisample targets alone. llvmpipe turns out to refuse most of the same
attachments - CI retrace stayed green only because a replay never
branches on glCheckFramebufferStatus - so this was never a desktop-vs-
device split, just an unlit path.

The widening now applies to every color-attachable image, renderbuffers
included, riding the driver-probe branch so the native format is still
tried first and substituted only on refusal. One ThreeChannelWidening
table owns the widened (internalformat, format, type) triple per source
format - the previous per-case branches disagreed with each other and
could emit an unuploadable (RGBA16F, GL_RGB, GL_BYTE) combination or
widen into another three-channel format the driver refuses just the
same. Uploads repack three-component client data to four with the
format's own one in the alpha channel (127 is not 1 for RGB8I - the
integer arms carry integer ones); readback drops the synthetic alpha,
derived from the actual image being read, not the bound framebuffer,
so glGetTexImage through a scratch FBO cannot be confused by an
unrelated widened attachment.

Stored alpha on a widened attachment is now an invariant 1.0 rather
than an accident: the color-mask sync clears the alpha bit per draw
buffer (glColorMaski for MRT mixes), and clears route through
glClearBufferfv with alpha substituted on widened slots only -
scissored clears inherit the discipline for free, integer color
buffers keep their explicit integer-clear path, and glGet still
answers the application's own mask. GL_DST_ALPHA blending, blits and
readback therefore all see 1.0 without further interception.

DriverPost grows the rows this bug earned: EXT_color_buffer_float
detection (previously unreferenced anywhere) with a FAIL row when
absent, the missing EXT_render_snorm row, and a three-channel-
attachment row that reports one representative per widening class -
graded so a half-float-only driver warns about the 32-bit float gap
instead of being declared unsupported.

Gates: 606/606 unit at default and with the async kill switch; full
retrace, both backends - the complementary fixtures now run with the
widening ACTIVE on llvmpipe and pass with a slightly better SSIM than
before; ext caselist DirectGLES holds 3914/4867 with zero set drift
while 54 cases move from NotSupported to genuinely passing; on the
Mali-G77 device, Complementary Reimagined builds its pipeline and
renders in-world through Espryt (md5-verified build), BSL still green.
A new ThreeChannelAttachmentScenario pins the frontend answer -
COMPLETE where it used to say UNSUPPORTED - on the real driver.
This commit is contained in:
BZLZHH
2026-08-09 15:28:34 -04:00
parent 0f394fa46f
commit d8d7530011
17 changed files with 2066 additions and 143 deletions
+291
View File
@@ -15,6 +15,7 @@
#include <Config.h>
#include <MG_Backend/BackendObjects.h>
#include <MG_Backend/DirectGLES/Managers.h>
#include <MG_Backend/DirectGLES/Utils.h>
#include <MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.h>
#include <MG_Impl/GLImpl/Getter/GL_Getter.h>
#include <MG_Impl/GLImpl/RenderState/GL_RenderState.h>
@@ -27,6 +28,7 @@
#include <MG_Util/Converters/GLToMG/TextureEnumConverter.h>
#include <MG_Util/Converters/MGToGL/TextureEnumConverter.h>
#include <MG_Util/Converters/MGToMG/TextureEnumConverter.h>
#include <MG_Util/Converters/MGToStr/TextureEnumConverter.h>
#include <MG_Util/Math/SmallFloat.h>
#include <MG_Util/Texture/PixelStoreProcessor.h>
#include <MG_Util/Texture/TextureFormatProcessor.h>
@@ -2885,3 +2887,292 @@ TEST_F(TextureTest, NamedTextureCallKeepsUnitBindingAccountingCoherent) {
MG_Impl::GLImpl::DeleteTextures(2, names);
DrainPendingGlErrors();
}
// ---- Three-channel colour-renderable widening (Complementary Reimagined / Iris) ----------------
//
// No real OpenGL ES driver renders to a three-channel image, so a colour attachment the
// application asked for as GL_RGB8_SNORM or GL_RGB16F has to be stored in the four-channel
// sibling. The bit that says so used to be reachable for multisample storage only, which is why
// an ordinary GL_TEXTURE_2D attachment in one of those formats had no fallback at all and the
// frontend could only answer GL_FRAMEBUFFER_UNSUPPORTED.
TEST_F(TextureTest, ColorAttachableTargetsRequestTheThreeChannelWidening) {
using MobileGL::MG_Backend::DirectGLES::TextureImpl::GetRenderTargetNormalizeOptions;
using MobileGL::MG_Backend::DirectGLES::TextureImpl::TargetRequiresRenderableFormat;
MG_External::GLESCapabilities capabilities{};
capabilities.SupportsRenderSnorm = true;
capabilities.SupportsNorm16Texture = true;
// Every image that can be a colour attachment, not just the multisample pair: an ordinary 2D
// texture is what Iris attaches, and it used to be excluded.
for (const TextureTarget target : {TextureTarget::Texture2D, TextureTarget::Texture3D,
TextureTarget::TextureCubeMap, TextureTarget::Texture2DArray,
TextureTarget::TextureCubeMapArray, TextureTarget::Texture2DMultisample,
TextureTarget::Texture2DMultisampleArray, TextureTarget::Texture1D,
TextureTarget::Texture1DArray, TextureTarget::TextureRectangle}) {
const SizeT targetIndex = MobileGL::MG_Backend::GetFormatCapabilityTargetIndex(target);
EXPECT_TRUE(TargetRequiresRenderableFormat(targetIndex))
<< "target " << MG_Util::ConvertTextureTargetToString(target);
EXPECT_TRUE(GetRenderTargetNormalizeOptions(capabilities, targetIndex) &
PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget)
<< "target " << MG_Util::ConvertTextureTargetToString(target);
}
// A renderbuffer exists only to be attached.
EXPECT_TRUE(TargetRequiresRenderableFormat(MobileGL::MG_Backend::GetRenderbufferFormatCapabilityTargetIndex()));
// A buffer texture is the one image that can never be an attachment; its storage belongs to
// the buffer object, so widening it would misdescribe the application's data.
const SizeT bufferIndex = MobileGL::MG_Backend::GetFormatCapabilityTargetIndex(TextureTarget::TextureBuffer);
EXPECT_FALSE(TargetRequiresRenderableFormat(bufferIndex));
EXPECT_FALSE(GetRenderTargetNormalizeOptions(capabilities, bufferIndex));
// Without EXT_render_snorm a 16-bit SNORM render target cannot keep its encoding either.
MG_External::GLESCapabilities noSnormCapabilities{};
const SizeT texture2DIndex = MobileGL::MG_Backend::GetFormatCapabilityTargetIndex(TextureTarget::Texture2D);
EXPECT_TRUE(GetRenderTargetNormalizeOptions(noSnormCapabilities, texture2DIndex) &
PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget);
EXPECT_FALSE(GetRenderTargetNormalizeOptions(capabilities, texture2DIndex) &
PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget);
}
TEST_F(TextureTest, ThreeChannelRenderTargetOptionAppliesToEveryDeniedThreeChannelFormat) {
using MG_Util::TextureFormatProcessor::GetApplicablePixelFormatNormalizeOptions;
const Flags<PixelFormatNormalizeOptionBit> requested =
PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
// GL_RGB16F in particular matched no case at all, so no option could ever apply to it and it
// fell through NormalizePixelFormat's default passthrough unchanged.
for (const GLenum internalFormat : {GL_RGB8_SNORM, GL_RGB16_SNORM, GL_RGB16, GL_RGB10, GL_RGB12, GL_RGB16F,
GL_RGB32F, GL_SRGB8, GL_RGB8I, GL_RGB8UI, GL_RGB16I, GL_RGB16UI, GL_RGB32I,
GL_RGB32UI}) {
EXPECT_TRUE(GetApplicablePixelFormatNormalizeOptions(internalFormat, requested) &
PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget)
<< "internalformat 0x" << std::hex << internalFormat;
}
// Four-channel and shared-exponent formats are not widened: RGBA8_SNORM has its own always-on
// fallback, and GL_RGB9_E5 has no four-channel sibling that would not need the shared exponent
// unpacked on every transfer (nothing renders to it on desktop GL either).
for (const GLenum internalFormat : {GL_RGBA8_SNORM, GL_RGBA16F, GL_RGBA8, GL_RGB8, GL_RGB9_E5}) {
EXPECT_FALSE(GetApplicablePixelFormatNormalizeOptions(internalFormat, requested) &
PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget)
<< "internalformat 0x" << std::hex << internalFormat;
}
}
TEST_F(TextureTest, ThreeChannelWideningRetargetsInternalFormatAndTransferPairTogether) {
using MG_Util::TextureFormatProcessor::NormalizePixelFormat;
struct Case {
GLenum requested;
Flags<PixelFormatNormalizeOptionBit> options;
GLenum internalFormat;
GLenum format;
GLenum type;
};
const Flags<PixelFormatNormalizeOptionBit> widen = PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget;
const Flags<PixelFormatNormalizeOptionBit> widenNoSnorm16 =
PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget |
PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget;
const Case cases[] = {
// Complementary's colortex1 and colortex2. The transfer pair used to stay three-channel
// and keep the *source* component type, emitting (GL_RGBA16F, GL_RGB, GL_BYTE) - which ES
// rejects for glTexImage2D outright, and which only went unnoticed because the bit was
// reachable for multisample storage alone (glTexStorage*Multisample takes no pair).
{GL_RGB8_SNORM, widen, GL_RGBA16F, GL_RGBA, GL_FLOAT},
{GL_RGB16F, widen, GL_RGBA16F, GL_RGBA, GL_HALF_FLOAT},
{GL_RGB32F, widen, GL_RGBA32F, GL_RGBA, GL_FLOAT},
// 16-bit SNORM keeps its encoding where EXT_render_snorm can render to it; a half float's
// 11-bit mantissa cannot represent a 16-bit SNORM channel exactly.
{GL_RGB16_SNORM, widen, GL_RGBA16_SNORM, GL_RGBA, GL_SHORT},
{GL_RGB16_SNORM, widenNoSnorm16, GL_RGBA16F, GL_RGBA, GL_FLOAT},
// 16-bit UNORM and the legacy 10/12-bit formats stored as RGB16.
{GL_RGB16, widen, GL_RGBA32F, GL_RGBA, GL_FLOAT},
{GL_RGB10, widen, GL_RGBA32F, GL_RGBA, GL_FLOAT},
{GL_RGB12, widen, GL_RGBA32F, GL_RGBA, GL_FLOAT},
// sRGB and the integer formats: the base format has to move to the four-channel one of the
// right class, GL_RGBA_INTEGER included.
{GL_SRGB8, widen, GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE},
{GL_RGB8I, widen, GL_RGBA8I, GL_RGBA_INTEGER, GL_BYTE},
{GL_RGB8UI, widen, GL_RGBA8UI, GL_RGBA_INTEGER, GL_UNSIGNED_BYTE},
{GL_RGB16I, widen, GL_RGBA16I, GL_RGBA_INTEGER, GL_SHORT},
{GL_RGB16UI, widen, GL_RGBA16UI, GL_RGBA_INTEGER, GL_UNSIGNED_SHORT},
{GL_RGB32I, widen, GL_RGBA32I, GL_RGBA_INTEGER, GL_INT},
{GL_RGB32UI, widen, GL_RGBA32UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT},
// The widening outranks the other fallbacks, which all pick a three-channel storage the
// driver still refuses to render to (GL_RGB8_SNORM -> GL_RGB16F, GL_RGB16 -> GL_RGB32F).
{GL_RGB8_SNORM, widen | PixelFormatNormalizeOptionBit::NoSnorm8, GL_RGBA16F, GL_RGBA, GL_FLOAT},
{GL_RGB16, widen | PixelFormatNormalizeOptionBit::NoNorm16, GL_RGBA32F, GL_RGBA, GL_FLOAT},
// Control: without the bit nothing moves. The bit is only ever set for a target whose
// native probe failed, so this is the shape every driver that does render to the
// three-channel form keeps - per format, not per platform (llvmpipe renders to GL_RGB16F
// but not to GL_RGB8_SNORM, GL_SRGB8, GL_RGB32F or the RGB integer formats).
{GL_RGB8_SNORM, PixelFormatNormalizeOptionBit::None, GL_RGB8_SNORM, GL_RGB, GL_BYTE},
{GL_RGB16F, PixelFormatNormalizeOptionBit::None, GL_RGB16F, GL_RGB, GL_HALF_FLOAT},
{GL_RGB32F, PixelFormatNormalizeOptionBit::None, GL_RGB32F, GL_RGB, GL_FLOAT},
{GL_SRGB8, PixelFormatNormalizeOptionBit::None, GL_SRGB8, GL_RGB, GL_UNSIGNED_BYTE},
// Not widened even under the bit: no four-channel shared-exponent sibling exists.
{GL_RGB9_E5, widen, GL_RGB9_E5, GL_RGB, GL_UNSIGNED_INT_5_9_9_9_REV},
// Four-channel formats are unaffected by the bit; RGBA8_SNORM keeps its own fallback.
{GL_RGBA8_SNORM, widen, GL_RGBA8_SNORM, GL_RGBA, GL_BYTE},
{GL_RGBA8_SNORM, widen | PixelFormatNormalizeOptionBit::NoRGBA8Snorm, GL_RGBA16F, GL_RGBA, GL_FLOAT},
};
for (const auto& testCase : cases) {
GLenum internalFormat = 0;
GLenum format = 0;
GLenum type = 0;
NormalizePixelFormat(testCase.requested, testCase.options, &internalFormat, &format, &type);
EXPECT_EQ(internalFormat, testCase.internalFormat) << "requested 0x" << std::hex << testCase.requested;
EXPECT_EQ(format, testCase.format) << "requested 0x" << std::hex << testCase.requested;
EXPECT_EQ(type, testCase.type) << "requested 0x" << std::hex << testCase.requested;
}
}
TEST_F(TextureTest, WidenedRenderTargetUploadExpandsThreeChannelDataWithOpaqueAlpha) {
using MobileGL::MG_Backend::DirectGLES::TextureImpl::GetWidenableClientComponentCount;
using MobileGL::MG_Backend::DirectGLES::TextureImpl::PrepareChannelWidenedUpload;
// Only the three-channel formats that can be widened report a source component count; the
// repack is what keeps the driver from walking three texels' worth of data per four-texel row.
for (const TextureInternalFormat format :
{TextureInternalFormat::RGB8Snorm, TextureInternalFormat::RGB16F, TextureInternalFormat::RGB32F,
TextureInternalFormat::RGB16Snorm, TextureInternalFormat::RGB16, TextureInternalFormat::SRGB8,
TextureInternalFormat::RGB8UI, TextureInternalFormat::RGB32I}) {
EXPECT_EQ(GetWidenableClientComponentCount(format), 3u)
<< MG_Util::ConvertTextureInternalFormatToString(format);
}
EXPECT_EQ(GetWidenableClientComponentCount(TextureInternalFormat::RGBA8), 0u);
EXPECT_EQ(GetWidenableClientComponentCount(TextureInternalFormat::RGBA8Snorm), 0u);
EXPECT_EQ(GetWidenableClientComponentCount(TextureInternalFormat::RGB9E5), 0u);
const IntVec3 texelSize(2, 1, 1);
// GL_RGB8_SNORM -> GL_RGBA16F: PrepareNormFloatFallbackUpload has already turned the Int8
// shadow into floats, so what arrives here is three floats per texel.
{
const Float source[] = {0.25f, -0.5f, 0.75f, -1.0f, 0.0f, 1.0f};
Vector<Uint8> widened;
const auto* result = static_cast<const Float*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_FLOAT, widened));
ASSERT_NE(result, static_cast<const void*>(source));
ASSERT_EQ(widened.size(), 8 * sizeof(Float));
const Float expected[] = {0.25f, -0.5f, 0.75f, 1.0f, -1.0f, 0.0f, 1.0f, 1.0f};
for (SizeT i = 0; i < 8; ++i) {
EXPECT_FLOAT_EQ(result[i], expected[i]) << "component " << i;
}
}
// GL_RGB16F -> GL_RGBA16F uploads halves untouched, so the synthetic alpha is the half
// encoding of 1.0 rather than a saturated field.
{
const Uint16 source[] = {0x0001, 0x0002, 0x0003, 0x0004, 0x0005, 0x0006};
Vector<Uint8> widened;
const auto* result = static_cast<const Uint16*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_HALF_FLOAT, widened));
ASSERT_NE(result, static_cast<const void*>(source));
const Uint16 expected[] = {0x0001, 0x0002, 0x0003, 0x3C00, 0x0004, 0x0005, 0x0006, 0x3C00};
for (SizeT i = 0; i < 8; ++i) {
EXPECT_EQ(result[i], expected[i]) << "component " << i;
}
}
// GL_SRGB8 -> GL_SRGB8_ALPHA8: fixed-point one is the saturated field.
{
const Uint8 source[] = {1, 2, 3, 4, 5, 6};
Vector<Uint8> widened;
const auto* result = static_cast<const Uint8*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_UNSIGNED_BYTE, widened));
const Uint8 expected[] = {1, 2, 3, 0xFF, 4, 5, 6, 0xFF};
ASSERT_NE(result, static_cast<const void*>(source));
EXPECT_EQ(std::memcmp(result, expected, sizeof(expected)), 0);
}
// GL_RGB16_SNORM -> GL_RGBA16_SNORM keeps GL_SHORT, whose 1.0 is the positive maximum.
{
const Int16 source[] = {-1, 2, -3, 4, -5, 6};
Vector<Uint8> widened;
const auto* result = static_cast<const Int16*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_SHORT, widened));
const Int16 expected[] = {-1, 2, -3, 0x7FFF, 4, -5, 6, 0x7FFF};
ASSERT_NE(result, static_cast<const void*>(source));
EXPECT_EQ(std::memcmp(result, expected, sizeof(expected)), 0);
}
// An integer format's added channel carries the integer one, not a saturated field.
{
const Uint32 source[] = {10, 20, 30, 40, 50, 60};
Vector<Uint8> widened;
const auto* result = static_cast<const Uint32*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_UNSIGNED_INT, widened, /*integerData=*/true));
const Uint32 expected[] = {10, 20, 30, 1, 40, 50, 60, 1};
ASSERT_NE(result, static_cast<const void*>(source));
EXPECT_EQ(std::memcmp(result, expected, sizeof(expected)), 0);
}
// GL_RGB8I -> GL_RGBA8I uploads as GL_BYTE, the very type GL_RGB8_SNORM uses, so the type
// alone cannot decide the added channel's value: the integer format's one is 1, the
// signed-normalized format's is 0x7F. Getting this wrong is invisible through sampling and
// glGetTexImage (both answer the alpha with the format's implied one) but escapes through a
// blit or glCopyTexSubImage out of the widened attachment.
{
const Int8 source[] = {-1, 2, -3, 4, -5, 6};
Vector<Uint8> widened;
const auto* asInteger = static_cast<const Int8*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_BYTE, widened, /*integerData=*/true));
const Int8 expectedInteger[] = {-1, 2, -3, 1, 4, -5, 6, 1};
ASSERT_NE(asInteger, static_cast<const void*>(source));
EXPECT_EQ(std::memcmp(asInteger, expectedInteger, sizeof(expectedInteger)), 0);
Vector<Uint8> widenedNorm;
const auto* asNormalized = static_cast<const Int8*>(PrepareChannelWidenedUpload(
3, texelSize, source, sizeof(source), GL_BYTE, widenedNorm, /*integerData=*/false));
const Int8 expectedNormalized[] = {-1, 2, -3, 0x7F, 4, -5, 6, 0x7F};
EXPECT_EQ(std::memcmp(asNormalized, expectedNormalized, sizeof(expectedNormalized)), 0);
}
// Which class a widenable format belongs to.
for (const TextureInternalFormat format :
{TextureInternalFormat::RGB8I, TextureInternalFormat::RGB8UI, TextureInternalFormat::RGB16I,
TextureInternalFormat::RGB16UI, TextureInternalFormat::RGB32I, TextureInternalFormat::RGB32UI}) {
EXPECT_TRUE(MobileGL::MG_Backend::DirectGLES::TextureImpl::IsIntegerWidenableFormat(format))
<< MG_Util::ConvertTextureInternalFormatToString(format);
}
for (const TextureInternalFormat format :
{TextureInternalFormat::RGB8Snorm, TextureInternalFormat::RGB16Snorm, TextureInternalFormat::RGB16,
TextureInternalFormat::RGB16F, TextureInternalFormat::RGB32F, TextureInternalFormat::SRGB8}) {
EXPECT_FALSE(MobileGL::MG_Backend::DirectGLES::TextureImpl::IsIntegerWidenableFormat(format))
<< MG_Util::ConvertTextureInternalFormatToString(format);
}
// The destination is sized from the level, never from the source. The driver reads a full
// width*height*4 components for the transfer it was handed, so a short source must still
// leave a full buffer behind - sizing it from the source would hand the driver a buffer it
// runs off the end of.
{
const Float shortSource[] = {0.5f, 0.25f, 0.125f};
Vector<Uint8> widened;
const auto* result = static_cast<const Float*>(PrepareChannelWidenedUpload(
3, IntVec3(2, 2, 1), shortSource, sizeof(shortSource), GL_FLOAT, widened));
ASSERT_NE(result, static_cast<const void*>(shortSource));
ASSERT_EQ(widened.size(), 4 * 4 * sizeof(Float));
const Float expected[] = {0.5f, 0.25f, 0.125f, 1.0f, 0.0f, 0.0f, 0.0f, 1.0f,
0.0f, 0.0f, 0.0f, 1.0f, 0.0f, 0.0f, 0.0f, 1.0f};
for (SizeT i = 0; i < 16; ++i) {
EXPECT_FLOAT_EQ(result[i], expected[i]) << "component " << i;
}
}
// No widening in effect (or nothing to convert): the caller's pointer comes straight back, so
// the sub-rect upload fast path still recognises an unconverted level.
{
const Float source[] = {1.0f, 2.0f, 3.0f, 4.0f};
Vector<Uint8> widened;
EXPECT_EQ(PrepareChannelWidenedUpload(4, texelSize, source, sizeof(source), GL_FLOAT, widened),
static_cast<const void*>(source));
EXPECT_EQ(PrepareChannelWidenedUpload(0, texelSize, source, sizeof(source), GL_FLOAT, widened),
static_cast<const void*>(source));
EXPECT_EQ(PrepareChannelWidenedUpload(3, texelSize, nullptr, 0, GL_FLOAT, widened), nullptr);
}
}