/****************************************************************************** * The MIT License (MIT) * * Copyright (c) 2019-2022 Baldur Karlsson * * Permission is hereby granted, free of charge, to any person obtaining a copy * of this software and associated documentation files (the "Software"), to deal * in the Software without restriction, including without limitation the rights * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell * copies of the Software, and to permit persons to whom the Software is * furnished to do so, subject to the following conditions: * * The above copyright notice and this permission notice shall be included in * all copies or substantial portions of the Software. * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN * THE SOFTWARE. ******************************************************************************/ #include #include #include #include "QRDUtils.h" struct StructFormatData { ShaderConstant structDef; uint32_t pointerTypeId = 0; uint32_t offset = 0; uint32_t alignment = 0; uint32_t paddedStride = 0; bool singleDef = false; bool singleMember = false; }; GraphicsAPI BufferFormatter::m_API; static bool MatchBaseTypeDeclaration(QString basetype, const bool isUnsigned, ShaderConstant &el) { if(basetype == lit("bool")) { el.type.descriptor.type = VarType::Bool; } else if(basetype == lit("byte") || basetype == lit("char")) { el.type.descriptor.type = VarType::SByte; if(isUnsigned) el.type.descriptor.type = VarType::UByte; } else if(basetype == lit("ubyte") || basetype == lit("xbyte")) { el.type.descriptor.type = VarType::UByte; } else if(basetype == lit("short")) { el.type.descriptor.type = VarType::SShort; if(isUnsigned) el.type.descriptor.type = VarType::UShort; } else if(basetype == lit("ushort") || basetype == lit("xshort")) { el.type.descriptor.type = VarType::UShort; } else if(basetype == lit("long")) { el.type.descriptor.type = VarType::SLong; if(isUnsigned) el.type.descriptor.type = VarType::ULong; } else if(basetype == lit("ulong") || basetype == lit("xlong")) { el.type.descriptor.type = VarType::ULong; } else if(basetype == lit("int") || basetype == lit("ivec") || basetype == lit("imat")) { el.type.descriptor.type = VarType::SInt; if(isUnsigned) el.type.descriptor.type = VarType::UInt; } else if(basetype == lit("uint") || basetype == lit("xint") || basetype == lit("uvec") || basetype == lit("umat")) { el.type.descriptor.type = VarType::UInt; } else if(basetype == lit("half")) { el.type.descriptor.type = VarType::Half; } else if(basetype == lit("float") || basetype == lit("vec") || basetype == lit("mat")) { el.type.descriptor.type = VarType::Float; } else if(basetype == lit("double") || basetype == lit("dvec") || basetype == lit("dmat")) { el.type.descriptor.type = VarType::Double; } else if(basetype == lit("unormh")) { el.type.descriptor.type = VarType::UShort; el.type.descriptor.flags |= ShaderVariableFlags::UNorm; } else if(basetype == lit("unormb")) { el.type.descriptor.type = VarType::UByte; el.type.descriptor.flags |= ShaderVariableFlags::UNorm; } else if(basetype == lit("snormh")) { el.type.descriptor.type = VarType::SShort; el.type.descriptor.flags |= ShaderVariableFlags::SNorm; } else if(basetype == lit("snormb")) { el.type.descriptor.type = VarType::SByte; el.type.descriptor.flags |= ShaderVariableFlags::SNorm; } else if(basetype == lit("uintten")) { el.type.descriptor.type = VarType::UInt; el.type.descriptor.flags |= ShaderVariableFlags::R10G10B10A2; el.type.descriptor.columns = 4; } else if(basetype == lit("unormten")) { el.type.descriptor.type = VarType::UInt; el.type.descriptor.flags |= ShaderVariableFlags::R10G10B10A2; el.type.descriptor.flags |= ShaderVariableFlags::UNorm; el.type.descriptor.columns = 4; } else if(basetype == lit("floateleven")) { el.type.descriptor.type = VarType::Float; el.type.descriptor.flags |= ShaderVariableFlags::R11G11B10; el.type.descriptor.columns = 3; } else { return false; } return true; } void BufferFormatter::EstimatePackingRules(Packing::Rules &pack, const ShaderConstant &constant) { // see if this constant violates any of the packing rules we are currently checking for. // We can't *prove* a rule is followed just from one example, we can only see if it is never // *disproved*. This does mean we won't necessarily determine the exact packing scheme, e.g if // scalar packing was used but it was only three float4 vectors then it will look like the most // conservative std140/scalar. if(!pack.vector_align_component || !pack.vector_straddle_16b) { // column major matrices have vectors that are 'rows' long. Everything else is vectors of // 'columns' long uint8_t vecSize = constant.type.descriptor.columns; if(constant.type.descriptor.rows > 1 && constant.type.descriptor.ColMajor()) vecSize = constant.type.descriptor.rows; if(vecSize > 1) { // is this a vector that's only component aligned and NOT vector aligned. If so, // vector_align_component is true const uint32_t vec4Size = VarTypeByteSize(constant.type.descriptor.type) * 4; const uint32_t offsModVec = (constant.byteOffset % vec4Size); // if it's a vec3 or vec4 and its offset is not purely aligned, it's only component aligned if(vecSize >= 3 && offsModVec != 0) pack.vector_align_component = true; // if it's a vec2 and its offset is not either 0 or half the total size, it's also only // component aligned. vec2s without this allowance must be aligned to the vec2 size if(vecSize == 2 && offsModVec != 0 && offsModVec != vec4Size / 2) pack.vector_align_component = true; // while we're here, check if the vector straddles a 16-byte boundary const uint32_t low16b = (constant.byteOffset / 16); const uint32_t high16b = ((constant.byteOffset + VarTypeByteSize(constant.type.descriptor.type) * vecSize - 1) / 16); // if the vector crosses a 16-byte boundary, vectors can straddle them if(low16b != high16b) pack.vector_straddle_16b = true; } } if(!pack.tight_arrays && constant.type.descriptor.elements > 1) { // if the array has a byte stride less than 16, it must be non-tight packed if(constant.type.descriptor.arrayByteStride < 16) pack.tight_arrays = true; } } QString BufferFormatter::DeclarePacking(Packing::Rules pack) { if(pack == Packing::D3DCB) return lit("#pack(cbuffer)"); else if(pack == Packing::std140) return lit("#pack(std140)"); else if(pack == Packing::std430) return lit("#pack(std430)"); else if(pack == Packing::D3DUAV) // this is also C but we call it 'structured' for D3D return lit("#pack(structured)"); else if(pack == Packing::Scalar) return lit("#pack(scalar)"); // packing doesn't match a premade ruleset. Emit individual specifiers QString ret; if(pack.vector_align_component) ret += lit("#pack(vector_align_component) // vectors are aligned to their component\n"); else ret += lit("#pack(no_vector_align_component) // vectors are aligned evenly (float3 as float4)\n"); if(pack.tight_arrays) ret += lit("#pack(tight_arrays) // arrays are packed tightly\n"); else ret += lit("#pack(no_tight_arrays) // arrays are padded to 16-byte boundaries\n"); if(pack.vector_straddle_16b) ret += lit("#pack(vector_straddle_16b) // vectors can straddle 16-byte boundaries\n"); else ret += lit("#pack(no_vector_straddle_16b) // vectors cannot straddle 16-byte boundaries\n"); if(pack.trailing_overlap) ret += lit("#pack(trailing_overlap) // variables can overlap trailing padding after " "arrays/structs\n"); else ret += lit("#pack(no_trailing_overlap) // variables cannot overlap trailing padding after " "arrays/structs\n"); return ret.trimmed(); } Packing::Rules BufferFormatter::EstimatePackingRules(const rdcarray &members) { Packing::Rules pack; // start from the most conservative ruleset. We will iteratively turn off any rules which are // violated to end up with the most conservative ruleset which is still valid for the described // variable // D3D shouldn't really need to be estimating, because it's implicit from how this is bound // (cbuffer or structured resource) if(IsD3D(m_API)) pack = Packing::D3DCB; else pack = Packing::std140; for(size_t i = 0; i < members.size(); i++) { // check this constant EstimatePackingRules(pack, members[i]); // check for trailing array/struct use if(i > 0) { const uint32_t prevOffset = members[i - 1].byteOffset; const uint32_t prevArrayCount = members[i - 1].type.descriptor.elements; const uint32_t prevArrayStride = members[i - 1].type.descriptor.arrayByteStride; // if we overlap into the previous element, trailing padding is not reserved // this works for structs too, as the array stride *includes* padding if(prevArrayCount > 1 && members[i].byteOffset < (prevOffset + prevArrayCount * prevArrayStride)) { pack.trailing_overlap = true; } } // if we've degenerated to scalar we can't get any more lenient, stop checking rules if(pack == Packing::Scalar) break; } // only return a 'real' ruleset. Don't revert to individually setting rules if we can help it // since that's a mess. The worst case is if someone is really using a custom packing format then // we add some extra offset decorations // only look for layouts typical of the API in use if(IsD3D(m_API)) { // scalar is technically more lenient than anything D3D allows, as D3DUAV requires padding after // structs (it's closer to C packing) if(pack == Packing::D3DCB || pack == Packing::D3DUAV || pack == Packing::Scalar) return pack; // shouldn't end up with these as we started at D3DCB, but just for safety if(pack == Packing::std140) return Packing::D3DCB; if(pack == Packing::std430) return Packing::D3DUAV; } else { if(pack == Packing::std140 || pack == Packing::std430 || pack == Packing::Scalar) return pack; if(m_API == GraphicsAPI::Vulkan) { if(pack == Packing::D3DCB || pack == Packing::D3DUAV) return pack; // on vulkan HLSL shaders may use relaxed block layout, which is not wholly represented here. // it doesn't actually allow trailing overlap but this lets us check if we're 'almost' cbuffer // rules, at which point any instances where trailing overlap would be used will look just // like manual padding/offsetting Packing::Rules mod = pack; mod.trailing_overlap = true; if(mod == Packing::D3DCB) return Packing::D3DCB; } } // don't explicitly use C layout, revert to scalar which is more typical in graphics // the worst case is that some elements that would be in trailing padding in structs get explicit // offset annotations to move them out, since in C that would be implicit. // // note, D3DUAV is treated the same as C but we checked for it above so we'd only get here on // non-D3D if(pack == Packing::C) return Packing::Scalar; // our ruleset doesn't match exactly to a premade one. Check the rules to see which properties we // have. // Currently this always means devolving to scalar, but we lay it out explicitly like this in case // other rulesets are added in future. // only scalar layouts allow straddling 16 byte alignment, it would be very strange to allow // straddling 16 bytes but e.g. not have tight arrays or component-aligned vectors. Possibly no // arrays were seen so tight arrays couldn't be explicitly determined. So regardless of what else // we found return scalar if(pack.vector_straddle_16b) return Packing::Scalar; // trailing overlap is allowed in any D3D layout, but for non-D3D only in scalar layout. // Since we know from above that either we're not using D3D or we aren't an exact match for D3DCB, // assume we're in scalar one way or another. // This could be e.g. D3DUAV with tight arrays but vector straddling wasn't seen explicitly if(pack.trailing_overlap) return Packing::Scalar; // the exact same logic as above applies to component-aligned vectors. Allowed in any D3D layout, // but for non-D3D only in scalar layout. if(pack.vector_align_component) return Packing::Scalar; // For non-D3D: if we have tight arrays, this is possible in std430 - however since we didn't // match std430 above there must be some other allowance. That means we must devolve to scalar // For D3D this is possible only in D3DUAV (which is equivalent to scalar) if(pack.tight_arrays) return Packing::Scalar; // shouldn't get here, but just for safety return the ruleset we derived return pack; } bool BufferFormatter::ContainsUnbounded(const rdcarray &members) { for(size_t i = 0; i < members.size(); i++) { if(members[i].type.descriptor.elements == ~0U) return true; if(ContainsUnbounded(members[i].type.members)) return true; } return false; } bool BufferFormatter::CheckInvalidUnbounded(const ShaderConstant &structDef, QString &errors) { for(size_t i = 0; i < structDef.type.members.size(); i++) { const bool isLast = i == structDef.type.members.size() - 1; // if it's not the last member and it's unbounded, that's a problem! if(!isLast && structDef.type.members[i].type.descriptor.elements == ~0U) { errors = tr("%1 in %2 can't be unbounded when not the last member.\n") .arg(structDef.type.members[i].name) .arg(structDef.type.descriptor.name); return false; } // if it's not the last member, no child can have an unbounded array if(!isLast && ContainsUnbounded(structDef.type.members[i].type.members)) { errors = tr("%1 in %2 can't have unbounded array in a child.\n") .arg(structDef.type.members[i].name) .arg(structDef.type.descriptor.name); return false; } if(!CheckInvalidUnbounded(structDef.type.members[i], errors)) return false; } return true; } rdcpair BufferFormatter::ParseFormatString( const QString &formatString, uint64_t maxLen, bool cbuffer, QString &errors) { StructFormatData root; StructFormatData *cur = &root; QMap structelems; QString lastStruct; // regex doesn't account for trailing or preceeding whitespace, or comments QRegularExpression regExpr( lit("^" // start of the line "(?row_major\\s+|column_major\\s+)?" // matrix majorness "(?unsigned\\s+|signed\\s+)?" // allow 'signed int' or 'unsigned char' "(?rgb\\s+)?" // rgb element colourising "(?" // group the options for the type "uintten|unormten" // R10G10B10A2 types "|floateleven" // R11G11B10 special type "|unormh|unormb" // UNORM 16-bit and 8-bit types "|snormh|snormb" // SNORM 16-bit and 8-bit types "|bool" // bool is stored as 4-byte int "|byte|short|int|long|char" // signed ints "|ubyte|ushort|uint|ulong" // unsigned ints "|xbyte|xshort|xint|xlong" // hex ints "|half|float|double" // float types "|vec|uvec|ivec|dvec" // OpenGL vector types "|mat|umat|imat|dmat" // OpenGL matrix types ")" // end of the type group "(?[1-9])?" // might be a vector "(?x[1-9])?" // or a matrix "(?\\s+[A-Za-z@_][A-Za-z0-9@_]*)?" // get identifier name "(?\\s*\\[[0-9]*\\])?" // optional array dimension "(\\s*:\\s*" // optional specifier after : "(" // bitfield or semantic "(?[1-9][0-9]*)|" // bitfield packing "(?[A-Za-z_][A-Za-z0-9_]*)" // semantic to ignore ")" // end bitfield or semantic ")?" "$")); bool success = true; errors = QString(); QString text = formatString; QRegularExpression c_comments(lit("/\\*[^*]*\\*+(?:[^*/][^*]*\\*+)*/")); QRegularExpression cpp_comments(lit("//.*")); // remove all comments text = text.replace(c_comments, QString()).replace(cpp_comments, QString()); // ensure braces are forced onto separate lines so we can parse them text = text.replace(QLatin1Char('{'), lit("\n{\n")).replace(QLatin1Char('}'), lit("\n}\n")); // treat commas as semi-colons for simplicity of parsing enum declarations and struct declarations text = text.replace(QLatin1Char(','), QLatin1Char(';')); QRegularExpression annotationRegex( lit("^" // start of the line "\\[\\[" // opening [[ "(?[a-zA-Z0-9_-]+)" // annotation name "(\\((?[^)]+)\\))?" // optional parameter in ()s "\\]\\]" // closing ]] "\\s*")); QRegularExpression structDeclRegex( lit("^(struct|enum)\\s+([A-Za-z_][A-Za-z0-9_]*)(\\s*:\\s*([a-z]+))?$")); QRegularExpression structUseRegex( lit("^" // start of the line "([A-Za-z_][A-Za-z0-9_]*)" // struct type name "\\s*(\\*)?" // maybe a pointer "\\s+([A-Za-z@_][A-Za-z0-9@_]*)?" // variable name "(\\s*\\[[0-9]*\\])?" // optional array dimension "(\\s*:\\s*([1-9][0-9]*))?" // optional bitfield packing "$")); QRegularExpression enumValueRegex( lit("^" // start of the line "([A-Za-z_][A-Za-z0-9_]*)" // value name "\\s*=\\s*" // maybe a pointer "(0x[0-9a-fA-F]+|[0-9]+)" // numerical value "$")); QRegularExpression bitfieldSkipRegex( lit("^" // start of the line "(unsigned\\s+|signed\\s+)?" // allow 'signed int' or 'unsigned char' "(" // type group "|bool" // bool is stored as 4-byte int "|byte|short|int|long|char" // signed ints "|ubyte|ushort|uint|ulong" // unsigned ints "|xbyte|xshort|xint|xlong" // hex ints ")" // end of the type group // no variable name "\\s*:\\s*([1-9][0-9]*)" // bitfield packing "$")); QRegularExpression packingRegex( lit("^" // start of the line "#\\s*pack\\s*\\(" // #pack( "(?[a-zA-Z0-9_]+)" // packing ruleset or individual rule "\\)" // ) "$")); uint32_t bitfieldCurPos = ~0U; struct Annotation { QString name; QString param; }; // default to scalar (tight packing) if nothing else is specified at all. The expectation is // anything that needs a better default will insert that into the format string for the user Packing::Rules pack = Packing::Scalar; QList annotations; // get each line and parse it to determine the format the user wanted for(QString &l : text.split(QRegularExpression(lit("[;\n\r]")))) { QString line = l.trimmed(); if(line.isEmpty()) continue; do { QRegularExpressionMatch match = annotationRegex.match(line); if(!match.hasMatch()) break; annotations.push_back({match.captured(lit("name")), match.captured(lit("param"))}); line.remove(match.capturedStart(0), match.capturedLength(0)); } while(true); if(line.isEmpty()) continue; { QRegularExpressionMatch match = packingRegex.match(line); if(match.hasMatch()) { if(cur != &root) { errors = tr("Packing rules can only be changed at global scope: %1\n").arg(line); success = false; break; } QString packrule = match.captured(lit("rule")).toLower(); // try to pick up common aliases that people might use if(packrule == lit("d3dcbuffer") || packrule == lit("cbuffer") || packrule == lit("cb")) pack = Packing::D3DCB; else if(packrule == lit("d3duav") || packrule == lit("uav") || packrule == lit("structured")) pack = Packing::D3DUAV; else if(packrule == lit("std140") || packrule == lit("ubo") || packrule == lit("gl") || packrule == lit("gles") || packrule == lit("opengl") || packrule == lit("glsl")) pack = Packing::std140; else if(packrule == lit("std430") || packrule == lit("ssbo")) pack = Packing::std430; else if(packrule == lit("scalar")) pack = Packing::Scalar; else if(packrule == lit("c")) pack = Packing::C; // we also allow toggling the individual rules else if(packrule == lit("vector_align_component")) pack.vector_align_component = true; else if(packrule == lit("no_vector_align_component")) pack.vector_align_component = false; else if(packrule == lit("tight_arrays")) pack.tight_arrays = true; else if(packrule == lit("no_tight_arrays")) pack.tight_arrays = false; else if(packrule == lit("vector_straddle_16b")) pack.vector_straddle_16b = true; else if(packrule == lit("no_vector_straddle_16b")) pack.vector_straddle_16b = false; else if(packrule == lit("trailing_overlap")) pack.trailing_overlap = true; else if(packrule == lit("no_trailing_overlap")) pack.trailing_overlap = false; else packrule = QString(); if(packrule.isEmpty()) { errors = tr("Unrecognised packing rule specifier: %1\n").arg(line); success = false; break; } continue; } } if(cur == &root) { // if we're not in a struct, ignore the braces if(line == lit("{") || line == lit("}")) continue; } else { // if we're in a struct, ignore the opening brace and revert back to root elements when we hit // the closing brace. No brace nesting is supported if(line == lit("{")) continue; if(line == lit("}")) { if(bitfieldCurPos != ~0U) { // update final offset to account for any bits consumed by a trailing bitfield, including // any bits in the last byte that weren't allocated cur->offset += (bitfieldCurPos + 7) / 8; // reset bitpacking state. bitfieldCurPos = ~0U; } if(cur->structDef.type.descriptor.type == VarType::Struct) { cur->structDef.type.descriptor.arrayByteStride = cur->offset; cur->alignment = GetAlignment(pack, cur->structDef); // if we don't have tight arrays, struct byte strides are always 16-byte aligned if(!pack.tight_arrays) { cur->alignment = 16; } cur->structDef.type.descriptor.arrayByteStride = AlignUp(cur->offset, cur->alignment); if(cur->paddedStride > 0) { // only pad up to the stride, not down if(cur->paddedStride >= cur->structDef.type.descriptor.arrayByteStride) { cur->structDef.type.descriptor.arrayByteStride = cur->paddedStride; } else { errors = tr("Declared struct %1 stride %2 is less than structure size %3\n") .arg(cur->structDef.type.descriptor.name) .arg(cur->paddedStride) .arg(cur->structDef.type.descriptor.arrayByteStride); success = false; break; } } cur->pointerTypeId = PointerTypeRegistry::GetTypeID(cur->structDef.type); } cur = &root; continue; } } if(line.startsWith(lit("struct")) || line.startsWith(lit("enum"))) { QRegularExpressionMatch match = structDeclRegex.match(line); if(match.hasMatch()) { QString name = match.captured(2); if(structelems.contains(name)) { errors = tr("Duplicate struct/enum definition: %1\n").arg(name); success = false; break; } cur = &structelems[name]; cur->structDef.type.descriptor.name = name; bitfieldCurPos = ~0U; if(match.captured(1) == lit("struct")) { lastStruct = name; cur->structDef.type.descriptor.type = VarType::Struct; for(const Annotation &annot : annotations) { if(annot.name == lit("size") || annot.name == lit("byte_size")) { cur->paddedStride = annot.param.toUInt(); } else if(annot.name == lit("single") || annot.name == lit("fixed")) { cur->singleDef = true; } else { errors = tr("Unrecognised annotation on struct definition: %1\n").arg(annot.name); success = false; break; } } annotations.clear(); if(!success) break; } else { cur->structDef.type.descriptor.type = VarType::Enum; for(const Annotation &annot : annotations) { if(false) { // no annotations supported currently on enums } else { errors = tr("Unrecognised annotation on enum definition: %1\n").arg(annot.name); success = false; break; } } annotations.clear(); if(!success) break; QString baseType = match.captured(4); if(baseType.isEmpty()) { errors = tr("Enum declarations require sized base type, see line: %1\n").arg(name); success = false; break; } ShaderConstant tmp; bool matched = MatchBaseTypeDeclaration(baseType, true, tmp); if(!matched) { errors = tr("Unknown enum base type on line: %1\n").arg(line); success = false; break; } cur->structDef.type.descriptor.arrayByteStride = VarTypeByteSize(tmp.type.descriptor.type); } continue; } } ShaderConstant el; if(cur->structDef.type.descriptor.type == VarType::Enum) { QRegularExpressionMatch enumMatch = enumValueRegex.match(line); if(!enumMatch.hasMatch()) { errors = tr("Couldn't parse enum value declaration on line: %1\n").arg(line); success = false; break; } bool ok = false; uint64_t val = enumMatch.captured(2).toULongLong(&ok, 0); if(!ok) { errors = tr("Couldn't parse enum numerical value on line: %1\n").arg(line); success = false; break; } el.name = enumMatch.captured(1); el.defaultValue = val; for(const Annotation &annot : annotations) { if(false) { // no annotations supported currently on enums } else { errors = tr("Unrecognised annotation on enum value: %1\n").arg(annot.name); success = false; break; } } annotations.clear(); if(!success) break; cur->structDef.type.members.push_back(el); continue; } QRegularExpressionMatch bitfieldSkipMatch = bitfieldSkipRegex.match(line); if(bitfieldSkipMatch.hasMatch()) { if(bitfieldCurPos == ~0U) bitfieldCurPos = 0; bitfieldCurPos += bitfieldSkipMatch.captured(3).toUInt(); for(const Annotation &annot : annotations) { if(false) { // no annotations supported currently on enums } else { errors = tr("Unrecognised annotation on bitfield skip: %1\n").arg(annot.name); success = false; break; } } annotations.clear(); if(!success) break; continue; } if(cur->singleMember) { errors = tr("[[single]] can only be used if there is only one variable in the root.\n" "Consider wrapping the variables in a struct and marking it as [[single]].\n"); success = false; break; } QRegularExpressionMatch structMatch = structUseRegex.match(line); bool isPadding = false; if(structMatch.hasMatch() && structelems.contains(structMatch.captured(1))) { StructFormatData &structContext = structelems[structMatch.captured(1)]; bool isPointer = !structMatch.captured(2).trimmed().isEmpty(); if(structContext.singleDef) { errors = tr("[[single]] annotated structs cannot be used, only defined.\n"); success = false; break; } QString varName = structMatch.captured(3).trimmed(); if(varName.isEmpty()) varName = lit("data"); uint32_t specifiedOffset = ~0U; for(const Annotation &annot : annotations) { if(annot.name == lit("offset") || annot.name == lit("byte_offset")) { specifiedOffset = annot.param.toUInt(); } else if(annot.name == lit("pad") || annot.name == lit("padding")) { isPadding = true; } else if(annot.name == lit("single") || annot.name == lit("fixed")) { if(cur != &root) { errors = tr("[[single]] can only be used on variables in the root\n"); success = false; break; } else if(!cur->structDef.type.members.empty()) { errors = tr("[[single]] can only be used if there is only one variable in the root.\n" "Consider wrapping the variables in a struct and marking it as [[single]].\n"); success = false; break; } else { cur->singleMember = true; } } else { errors = tr("Unrecognised annotation on variable: %1\n").arg(annot.name); success = false; break; } } if(!success) break; annotations.clear(); QString arrayDim = structMatch.captured(4).trimmed(); uint32_t arrayCount = 1; if(!arrayDim.isEmpty()) { arrayDim = arrayDim.mid(1, arrayDim.count() - 2); if(arrayDim.isEmpty()) arrayDim = lit("%1").arg(~0U); bool ok = false; arrayCount = arrayDim.toUInt(&ok); if(!ok) arrayCount = 1; } if(cur->singleMember && arrayCount == ~0U) { errors = tr("[[single]] can't be used on unbounded arrays.\n"); success = false; break; } QString bitfield = structMatch.captured(6).trimmed(); if(isPointer) { if(!bitfield.isEmpty()) { errors = tr("Bitfield packing is not allowed on pointers on line: %1\n").arg(line); success = false; break; } // align to scalar size cur->offset = AlignUp(cur->offset, 8U); if(specifiedOffset != ~0U) { if(specifiedOffset < cur->offset) { errors = tr("Offset %1 on variable %2 overlaps previous data\n").arg(specifiedOffset).arg(varName); success = false; break; } cur->offset = specifiedOffset; } el.name = varName; el.byteOffset = cur->offset; el.type.descriptor.pointerTypeID = structContext.pointerTypeId; el.type.descriptor.type = VarType::ULong; el.type.descriptor.flags |= ShaderVariableFlags::HexDisplay; el.type.descriptor.arrayByteStride = 8; el.type.descriptor.elements = arrayCount; cur->offset += 8; if(!isPadding) cur->structDef.type.members.push_back(el); continue; } else if(structContext.structDef.type.descriptor.type == VarType::Enum) { if(!bitfield.isEmpty() && !arrayDim.isEmpty()) { errors = tr("Bitfield packing is not allowed on arrays on line: %1\n").arg(line); success = false; break; } // align to scalar size (if not bit packing) if(bitfieldCurPos == ~0U) cur->offset = AlignUp(cur->offset, structContext.structDef.type.descriptor.arrayByteStride); if(specifiedOffset != ~0U) { uint32_t offs = cur->offset; if(bitfieldCurPos != ~0U) offs += (bitfieldCurPos + 7) / 8; if(specifiedOffset < offs) { errors = tr("Offset %1 on variable %2 overlaps previous data\n").arg(specifiedOffset).arg(varName); success = false; break; } cur->offset = specifiedOffset; // reset any bitfield packing to start at 0 at the new location if(bitfieldCurPos != ~0U) bitfieldCurPos = 0; } el = structContext.structDef; el.name = varName; el.byteOffset = cur->offset; el.type.descriptor.elements = arrayCount; bool ok = false; el.bitFieldSize = qMax(1U, bitfield.toUInt(&ok)); if(!ok) el.bitFieldSize = 0; // don't continue here - we will go through and handle bitfield packing like any other // scalar } else { if(!bitfield.isEmpty()) { errors = tr("Bitfield packing is not allowed on structs on line: %1\n").arg(line); success = false; break; } // all packing rules align structs in the same way as arrays. We already calculated this // when calculating the struct's alignment which will be padded to 16B for non-tight arrays cur->offset = AlignUp(cur->offset, structContext.alignment); if(specifiedOffset != ~0U) { if(specifiedOffset < cur->offset) { errors = tr("Offset %1 on variable %2 overlaps previous data\n").arg(specifiedOffset).arg(varName); success = false; break; } cur->offset = specifiedOffset; } el = structContext.structDef; el.name = varName; el.byteOffset = cur->offset; el.type.descriptor.elements = arrayCount; if(!isPadding) cur->structDef.type.members.push_back(el); // advance by the struct including any trailing padding cur->offset += el.type.descriptor.elements * el.type.descriptor.arrayByteStride; // if we allow trailing overlap, remove the padding if(pack.trailing_overlap) cur->offset -= el.type.descriptor.arrayByteStride - structContext.offset; continue; } } else { QRegularExpressionMatch match = regExpr.match(line); if(!match.hasMatch()) { errors = tr("Couldn't parse line: %1\n").arg(line); success = false; break; } el.name = !match.captured(lit("name")).isEmpty() ? match.captured(lit("name")).trimmed() : lit("data"); QString basetype = match.captured(lit("type")); if(match.captured(lit("major")).trimmed() == lit("row_major")) el.type.descriptor.flags |= ShaderVariableFlags::RowMajorMatrix; if(!match.captured(lit("rgb")).isEmpty()) el.type.descriptor.flags |= ShaderVariableFlags::RGBDisplay; QString firstDim = !match.captured(lit("vec")).isEmpty() ? match.captured(lit("vec")) : lit("1"); QString secondDim = !match.captured(lit("mat")).isEmpty() ? match.captured(lit("mat")).mid(1) : lit("1"); QString arrayDim = !match.captured(lit("array")).isEmpty() ? match.captured(lit("array")).trimmed() : lit("[1]"); { bool isArray = !arrayDim.isEmpty(); arrayDim = arrayDim.mid(1, arrayDim.count() - 2).trimmed(); if(isArray && arrayDim.isEmpty()) arrayDim = lit("%1").arg(~0U); } const bool isUnsigned = match.captured(lit("sign")).trimmed() == lit("unsigned"); QString bitfield = match.captured(lit("bitfield")); if(!bitfield.isEmpty() && !arrayDim.isEmpty()) { errors = tr("Bitfield packing is not allowed on arrays on line: %1\n").arg(line); success = false; break; } QString vecMatSizeSuffix; // if we have a matrix and it's not GL style, then typeAxB means A rows and B columns // for GL matAxB that means A columns and B rows. This is in contrast to typeA which means A // columns for HLSL and A columns for GLSL, hence only the swap for matrices if(!match.captured(lit("mat")).isEmpty() && basetype != lit("mat")) { vecMatSizeSuffix = match.captured(lit("vec")) + match.captured(lit("mat")); firstDim.swap(secondDim); } else { if(!match.captured(lit("mat")).isEmpty()) vecMatSizeSuffix = match.captured(lit("mat")).mid(1) + lit("x"); vecMatSizeSuffix += match.captured(lit("vec")); } // check for square matrix declarations like 'mat4' and 'mat3' if(basetype == lit("mat") && match.captured(lit("mat")).isEmpty()) { secondDim = firstDim; vecMatSizeSuffix = firstDim + lit("x") + firstDim; } // check for square matrix declarations like 'mat4' and 'mat3' if(basetype == lit("mat") && match.captured(lit("mat")).isEmpty()) secondDim = firstDim; // calculate format { bool ok = false; el.type.descriptor.columns = firstDim.toUInt(&ok); if(!ok) { errors = tr("Invalid vector dimension on line: %1\n").arg(line); success = false; break; } el.type.descriptor.elements = qMax(1U, arrayDim.toUInt(&ok)); if(!ok) el.type.descriptor.elements = 1; el.type.descriptor.rows = qMax(1U, secondDim.toUInt(&ok)); if(!ok) { errors = tr("Invalid matrix second dimension on line: %1\n").arg(line); success = false; break; } el.bitFieldSize = qMax(1U, bitfield.toUInt(&ok)); if(!ok) el.bitFieldSize = 0; // vectors are marked as row-major by convention if(el.type.descriptor.rows == 1) el.type.descriptor.flags |= ShaderVariableFlags::RowMajorMatrix; bool matched = MatchBaseTypeDeclaration(basetype, isUnsigned, el); if(!matched) { errors = tr("Unrecognised type on line: %1\n").arg(line); success = false; break; } } el.type.descriptor.name = ToStr(el.type.descriptor.type) + vecMatSizeSuffix; // process packing annotations first, so we have that information to validate e.g. [[unorm]] for(const Annotation &annot : annotations) { if(annot.name == lit("packed")) { if(annot.param.toLower() == lit("r11g11b10")) { if(el.type.descriptor.columns != 3 || el.type.descriptor.type != VarType::Float) { errors = tr("R11G11B10 packing must be specified on a 'float3' variable: %1\n").arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::R11G11B10; } else if(annot.param.toLower() == lit("r10g10b10a2") || annot.param.toLower() == lit("r10g10b10a2_uint")) { if(el.type.descriptor.columns != 4 || el.type.descriptor.type != VarType::UInt) { errors = tr("R10G10B10A2 packing must be specified on a 'uint4' variable " "(optionally with [[unorm]]): %1\n") .arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::R10G10B10A2; } else if(annot.param.toLower() == lit("r10g10b10a2_unorm")) { if(el.type.descriptor.columns != 4 || el.type.descriptor.type != VarType::UInt) { errors = tr("R10G10B10A2 packing must be specified on a 'uint4' variable " "(optionally with [[unorm]]): %1\n") .arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::UNorm; } else if(annot.param.toLower() == lit("r10g10b10a2_snorm")) { if(el.type.descriptor.columns != 4 || (el.type.descriptor.type != VarType::SInt && el.type.descriptor.type != VarType::UInt)) { errors = tr("R10G10B10A2 packing must be specified on a '[u]int4' variable " "when using [[snorm]]): %1\n") .arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::SNorm; } else { errors = tr("Unrecognised pack type: %1\n").arg(annot.param); success = false; break; } } } if(!success) break; for(const Annotation &annot : annotations) { if(annot.name == lit("rgb")) { el.type.descriptor.flags |= ShaderVariableFlags::RGBDisplay; } else if(annot.name == lit("hex") || annot.name == lit("hexadecimal")) { if(VarTypeCompType(el.type.descriptor.type) == CompType::Float) { errors = tr("Hex display is not supported on floating point formats on line: %1\n").arg(line); success = false; break; } if(el.type.descriptor.flags & (ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::R11G11B10)) { errors = tr("Hex display is not supported on packed formats on line: %1\n").arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::HexDisplay; if(el.type.descriptor.type == VarType::SLong) el.type.descriptor.type = VarType::ULong; else if(el.type.descriptor.type == VarType::SInt) el.type.descriptor.type = VarType::UInt; else if(el.type.descriptor.type == VarType::SShort) el.type.descriptor.type = VarType::UShort; else if(el.type.descriptor.type == VarType::SByte) el.type.descriptor.type = VarType::UByte; } else if(annot.name == lit("bin") || annot.name == lit("binary")) { if(VarTypeCompType(el.type.descriptor.type) == CompType::Float) { errors = tr("Binary display is not supported on floating point formats on line: %1\n").arg(line); success = false; break; } if(el.type.descriptor.flags & (ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::R11G11B10)) { errors = tr("Binary display is not supported on packed formats on line: %1\n").arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::BinaryDisplay; if(el.type.descriptor.type == VarType::SLong) el.type.descriptor.type = VarType::ULong; else if(el.type.descriptor.type == VarType::SInt) el.type.descriptor.type = VarType::UInt; else if(el.type.descriptor.type == VarType::SShort) el.type.descriptor.type = VarType::UShort; else if(el.type.descriptor.type == VarType::SByte) el.type.descriptor.type = VarType::UByte; } else if(annot.name == lit("unorm")) { if(!(el.type.descriptor.flags & ShaderVariableFlags::R10G10B10A2)) { // verify that we're integer typed and 1 or 2 bytes if(el.type.descriptor.type != VarType::UShort && el.type.descriptor.type != VarType::SShort && el.type.descriptor.type != VarType::UByte && el.type.descriptor.type != VarType::SByte) { errors = tr("UNORM packing is only supported on [u]byte and [u]short types: %1\n").arg(line); success = false; break; } } el.type.descriptor.flags |= ShaderVariableFlags::UNorm; } else if(annot.name == lit("snorm")) { if(!(el.type.descriptor.flags & ShaderVariableFlags::R10G10B10A2)) { // verify that we're integer typed and 1 or 2 bytes if(el.type.descriptor.type != VarType::UShort && el.type.descriptor.type != VarType::SShort && el.type.descriptor.type != VarType::UByte && el.type.descriptor.type != VarType::SByte) { errors = tr("SNORM packing is only supported on [u]byte and [u]short types: %1\n").arg(line); success = false; break; } } el.type.descriptor.flags |= ShaderVariableFlags::SNorm; } else if(annot.name == lit("row_major")) { if(el.type.descriptor.rows == 1) { errors = tr("Row major can only be specified on matrices: %1\n").arg(line); success = false; break; } el.type.descriptor.flags |= ShaderVariableFlags::RowMajorMatrix; } else if(annot.name == lit("packed")) { // already processed } else if(annot.name == lit("offset") || annot.name == lit("byte_offset")) { uint32_t specifiedOffset = annot.param.toUInt(); if(specifiedOffset < cur->offset) { errors = tr("Offset %1 on variable %2 overlaps previous data\n").arg(specifiedOffset).arg(el.name); success = false; break; } cur->offset = specifiedOffset; } else if(annot.name == lit("pad") || annot.name == lit("padding")) { isPadding = true; } else if(annot.name == lit("single") || annot.name == lit("fixed")) { if(cur != &root) { errors = tr("[[single]] can only be used on variables in the root\n"); success = false; break; } else if(!cur->structDef.type.members.empty()) { errors = tr("[[single]] can only be used if there is only one variable in the root.\n" "Consider wrapping the variables in a struct and marking it as [[single]].\n"); success = false; break; } else { cur->singleMember = true; } } else { errors = tr("Unrecognised annotation on variable: %1\n").arg(annot.name); success = false; break; } } annotations.clear(); if(!success) break; // validate that bitfields are only allowed for regular scalars if(el.bitFieldSize > 0) { if(el.type.descriptor.rows > 1 || el.type.descriptor.columns > 1) { errors = tr("Bitfield packing only allowed on scalar values on line: %1\n").arg(line); success = false; break; } if(el.type.descriptor.elements > 1) { errors = tr("Bitfield packing not allowed on arrays on line: %1\n").arg(line); success = false; break; } if(el.type.descriptor.flags & (ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::R11G11B10 | ShaderVariableFlags::UNorm | ShaderVariableFlags::SNorm)) { errors = tr("Bitfield packing not allowed on interpreted/packed formats on line: %1\n").arg(line); success = false; break; } if(VarTypeCompType(el.type.descriptor.type) == CompType::Float) { errors = tr("Bitfield packing not allowed on floating point formats on line: %1\n").arg(line); success = false; break; } } if(basetype == lit("xlong") || basetype == lit("xint") || basetype == lit("xshort") || basetype == lit("xbyte")) el.type.descriptor.flags |= ShaderVariableFlags::HexDisplay; } const bool packed32bit = bool(el.type.descriptor.flags & (ShaderVariableFlags::R10G10B10A2 | ShaderVariableFlags::R11G11B10)); // normally the array stride is the size of an element const uint32_t elAlignment = packed32bit ? sizeof(uint32_t) : GetAlignment(pack, el); const uint8_t vecSize = (el.type.descriptor.rows > 1 && el.type.descriptor.ColMajor()) ? el.type.descriptor.rows : el.type.descriptor.columns; const uint32_t elSize = packed32bit ? sizeof(uint32_t) : (pack.vector_align_component ? elAlignment * vecSize : elAlignment); // if we aren't using tight arrays the stride is at least 16 bytes el.type.descriptor.arrayByteStride = elAlignment; if(el.type.descriptor.rows > 1 || el.type.descriptor.columns > 1) el.type.descriptor.arrayByteStride = elSize; if(!pack.tight_arrays) el.type.descriptor.arrayByteStride = std::max(16U, el.type.descriptor.arrayByteStride); // matrices are always aligned like arrays of vectors if(el.type.descriptor.rows > 1) { // the alignment calculated above is the alignment of a vector, that's our matrix stride el.type.descriptor.matrixByteStride = el.type.descriptor.arrayByteStride; // the array stride is that alignment times the number of rows/columns if(el.type.descriptor.RowMajor()) el.type.descriptor.arrayByteStride *= el.type.descriptor.rows; else el.type.descriptor.arrayByteStride *= el.type.descriptor.columns; } if(el.bitFieldSize > 0) { // we can use the arrayByteStride since this is a scalar so no vector/arrays, this is just the // base size. It also works for enums as this is the byte size of the declared underlying type const uint32_t elemScalarBitSize = cur->structDef.type.descriptor.arrayByteStride * 8; // bitfields can't be larger than the base type if(el.bitFieldSize > elemScalarBitSize) { errors = tr("Bitfield cannot specify a larger size than the base type on line: %1\n").arg(line); success = false; break; } uint32_t start = bitfieldCurPos; if(start == ~0U) start = 0; // if we would end past the current base type size, first roll over and start at the next // byte // this could be: // unsigned int a : 24; // unsigned byte b : 4; // unsigned byte c : 4; // where we just 'rollover' the 3 bytes packed into the unsigned int and start the byte on // the next byte, there's no extra padding added // or it could be: // unsigned int a : 29; // unsigned byte b : 4; // unsigned byte c : 4; // where b would pass through the end of the fourth byte so there ends up being 3 bits of // padding between a and b when b is rolled onto the next byte // similarly this can happen if the types are the same: // unsigned int a : 29; // unsigned int b : 4; // since b would still pass through the end of the first dword. // similarly this allows 'more' padding when the types are bigger: // unsigned int a : 17; // unsigned int b : 17; // which would produce 15 bytes of padding // Note that if the types are the same and big enough we won't roll over, as in: // unsigned int a : 24; // unsigned int b : 4; // unsigned int c : 4; if(start + el.bitFieldSize > elemScalarBitSize) { // align the offset up to where this bitfield needs to start cur->offset += ((bitfieldCurPos + (elemScalarBitSize - 1)) / elemScalarBitSize) * (elemScalarBitSize / 8); // reset the current bitfield pos bitfieldCurPos = 0; } // if there's no previous bitpacking, nothing much to do if(bitfieldCurPos == ~0U) { // start the next bitfield at our size bitfieldCurPos = el.bitFieldSize; } else { // start the next bitfield at the end of the previous el.bitFieldOffset = bitfieldCurPos; // update by our size bitfieldCurPos += el.bitFieldSize; } } else { // this element is not bitpacked if(bitfieldCurPos != ~0U) { // update offset to account for any bits consumed by the previous bitfield, which won't have // happened yet, including any bits in the last byte that weren't allocated cur->offset += (bitfieldCurPos + 7) / 8; // reset bitpacking state. bitfieldCurPos = ~0U; } // align to our element's base alignment cur->offset = AlignUp(cur->offset, elAlignment); // if we have non-tight arrays, arrays (and matrices) always start on a 16-byte boundary if(!pack.tight_arrays && (el.type.descriptor.elements > 1 || el.type.descriptor.rows > 1)) cur->offset = AlignUp(cur->offset, 16U); // if vectors can't straddle 16-byte alignment, check to see if we're going to do that if(!pack.vector_straddle_16b) { if(cur->offset / 16 != (cur->offset + elSize - 1) / 16) { cur->offset = AlignUp(cur->offset, 16U); } } } el.byteOffset = cur->offset; if(!isPadding) cur->structDef.type.members.push_back(el); // if we're bitfield packing don't advance offset, otherwise advance to the end of this element if(bitfieldCurPos == ~0U) { // advance by the struct including any trailing padding cur->offset += GetVarSize(el); // if we allow trailing overlap in arrays/matrices, remove the padding. This is only possible // with non-tight arrays if(pack.trailing_overlap && !pack.tight_arrays && (el.type.descriptor.type == VarType::Struct || el.type.descriptor.elements > 1 || el.type.descriptor.rows > 1)) { // the padding is the stride (which is rounded up to 16 for non-tight arrays) minus the size // of the last vector (whether or not this is an array of scalars, vectors or matrices cur->offset -= 16 - elSize; } } } if(bitfieldCurPos != ~0U) { // update final offset to account for any bits consumed by a trailing bitfield, including any // bits in the last byte that weren't allocated cur->offset += (bitfieldCurPos + 7) / 8; // reset bitpacking state. bitfieldCurPos = ~0U; } // if we succeeded parsing but didn't get any root elements, use the last defined struct as the // definition if(success && root.structDef.type.members.isEmpty() && !lastStruct.isEmpty()) root = structelems[lastStruct]; root.structDef.type.descriptor.arrayByteStride = AlignUp(root.offset, GetAlignment(pack, root.structDef)); if(!root.structDef.type.members.isEmpty() && root.structDef.type.members.back().type.descriptor.elements == ~0U) { root.structDef.type.descriptor.arrayByteStride = AlignUp(root.structDef.type.members.back().type.descriptor.arrayByteStride, GetAlignment(pack, root.structDef)); } if(success) { // check that unbounded arrays are only the last member of each struct. Doing this separately // makes the below check easier since we only have to consider last members if(!CheckInvalidUnbounded(root.structDef, errors)) success = false; } // we only allow one 'infinite' array. You can't have an infinite member inside an already // infinite struct. E.g. this is invalid: // // struct foo { // uint a; // float b[]; // }; // // foo data[]; // // but it's valid to have either this: // // struct foo { // uint a; // float b[3]; // }; // // foo data[]; // // or this: // // struct foo { // uint a; // float b[]; // }; // // foo data; if(success) { ShaderConstant *iter = &root.structDef; bool foundInfinite = false; QString infiniteArrayName; while(iter) { if(iter->type.descriptor.elements == ~0U) { if(foundInfinite) { success = false; errors = tr("Can't have unbounded %1[] as child of an unbounded %2[].\n") .arg(iter->name) .arg(infiniteArrayName); break; } infiniteArrayName = iter->type.descriptor.name; if(infiniteArrayName.isEmpty()) infiniteArrayName = tr("implicit_root"); foundInfinite = true; } // if there are no more members, stop looking if(iter->type.members.empty()) break; iter = &iter->type.members.back(); } } // on D3D if we have an unbounded array it *must* be the root element, as D3D does not support // some fixed elements before it, structured buffers are strictly just an AoS. // we do allow specifying cbuffers which are all fixed and not unbounded, so we just check to see // that if there is an unbounded array that it's the root if( // on D3D IsD3D(m_API) && // if the parsing worked success && !root.structDef.type.members.empty() && // if we have an unbounded array somewhere (we know there's only one, from above) ContainsUnbounded(root.structDef.type.members) && // it must be in the root and it must be alone with no siblings !(root.structDef.type.members.size() == 1 && root.structDef.type.members[0].type.descriptor.elements == ~0U)) { errors += tr("On D3D an unbounded array must be only be used alone as the root element.\n"); success = false; } // when not viewing a cbuffer, if the root hasn't been explicitly marked as a single struct and we // don't have an unbounded array then consider it an AoS definition in all other cases as that is // very likely what the user expects if(success && !root.structDef.type.members.empty() && !ContainsUnbounded(root.structDef.type.members) && !root.singleMember && !root.singleDef && !cbuffer) { // if there's already only one root member just make it infinite if(root.structDef.type.members.size() == 1) { root.structDef.type.members[0].type.descriptor.elements = ~0U; } else { // otherwise wrap a struct around the members, to be the infinite AoS rdcarray inners; inners.swap(root.structDef.type.members); ShaderConstant el; el.byteOffset = 0; el.type.descriptor.type = VarType::Struct; el.type.descriptor.elements = ~0U; el.type.descriptor.arrayByteStride = root.structDef.type.descriptor.arrayByteStride; root.structDef.type.members.push_back(el); inners.swap(root.structDef.type.members[0].type.members); } } if(!success || root.structDef.type.members.isEmpty()) { root.structDef.type.members.clear(); root.structDef.type.descriptor.type = VarType::Struct; ShaderConstant el; el.byteOffset = 0; el.type.descriptor.flags |= ShaderVariableFlags::HexDisplay; el.name = "data"; el.type.descriptor.type = VarType::UInt; el.type.descriptor.columns = 4; el.type.descriptor.elements = ~0U; if(maxLen > 0 && maxLen < 16) el.type.descriptor.columns = 1; if(maxLen > 0 && maxLen < 4) el.type.descriptor.type = VarType::UByte; el.type.descriptor.arrayByteStride = el.type.descriptor.matrixByteStride = el.type.descriptor.columns * VarTypeByteSize(el.type.descriptor.type); root.structDef.type.members.push_back(el); root.structDef.type.descriptor.arrayByteStride = el.type.descriptor.arrayByteStride; } // split the struct definition we have now into fixed and repeating. We've enforced above that // there's only one struct which is unbounded (no other children at any level are unbounded) and // it's the last member, so we find it, remove it from the hierarchy, and present it separately. ShaderConstant repeat; { ShaderConstant *iter = &root.structDef; rdcstr addedprefix; while(iter) { // if there are no more members, stop looking, there's no repeated member if(iter->type.members.empty()) break; // add the prefix, so that the repeated element that's a child like buffer { foo { blah[] } } // shows up as buffer.foo.blah if(!iter->name.empty()) addedprefix += iter->name + "."; // we want to search the members so we can remove from the current iter if(iter->type.members.back().type.descriptor.elements == ~0U) { repeat = iter->type.members.back(); repeat.name = addedprefix + repeat.name; repeat.type.descriptor.elements = 1; iter->type.members.pop_back(); break; } iter = &iter->type.members.back(); } } return {root.structDef, repeat}; } QString BufferFormatter::GetTextureFormatString(const TextureDescription &tex) { QString baseType; QString varName = lit("pixels"); uint32_t w = tex.width; switch(tex.format.type) { case ResourceFormatType::BC1: case ResourceFormatType::BC2: case ResourceFormatType::BC3: case ResourceFormatType::BC4: case ResourceFormatType::BC5: case ResourceFormatType::BC6: case ResourceFormatType::BC7: case ResourceFormatType::ETC2: case ResourceFormatType::EAC: case ResourceFormatType::ASTC: case ResourceFormatType::PVRTC: varName = lit("block"); // display a 4x4 block at a time w /= 4; default: break; } switch(tex.format.type) { case ResourceFormatType::Regular: { if(tex.format.compByteWidth == 1) { if(tex.format.compType == CompType::UNorm || tex.format.compType == CompType::UNormSRGB) baseType = lit("[[unorm]] ubyte"); else if(tex.format.compType == CompType::SNorm) baseType = lit("[[snorm]] byte"); else if(tex.format.compType == CompType::SInt) baseType = lit("byte"); else baseType = lit("ubyte"); } else if(tex.format.compByteWidth == 2) { if(tex.format.compType == CompType::UNorm || tex.format.compType == CompType::UNormSRGB) baseType = lit("[[unorm]] ushort"); else if(tex.format.compType == CompType::SNorm) baseType = lit("[[snorm]] short"); else if(tex.format.compType == CompType::Float) baseType = lit("half"); else if(tex.format.compType == CompType::SInt) baseType = lit("short"); else baseType = lit("ushort"); } else if(tex.format.compByteWidth == 4) { if(tex.format.compType == CompType::Float) baseType = lit("float"); else if(tex.format.compType == CompType::SInt) baseType = lit("int"); else baseType = lit("uint"); } else { if(tex.format.compType == CompType::Float) baseType = lit("double"); else if(tex.format.compType == CompType::SInt) baseType = lit("long"); else baseType = lit("ulong"); } baseType = QFormatStr("[[rgb]] %1%2").arg(baseType).arg(tex.format.compCount); break; } // 2x4 byte block, for 64-bit block formats case ResourceFormatType::BC1: case ResourceFormatType::BC4: case ResourceFormatType::ETC2: case ResourceFormatType::EAC: case ResourceFormatType::PVRTC: baseType = lit("[[row_major]] [[hex]] int2"); break; // 4x4 byte block, for 128-bit block formats case ResourceFormatType::BC2: case ResourceFormatType::BC3: case ResourceFormatType::BC5: case ResourceFormatType::BC6: case ResourceFormatType::BC7: case ResourceFormatType::ASTC: baseType = lit("[[row_major]] [[hex]] int4"); break; case ResourceFormatType::R10G10B10A2: baseType = lit("[[packed(r10g10b10a2)]] "); if(tex.format.compType == CompType::UNorm) baseType += lit("[[unorm]] "); baseType += lit("uint4"); break; case ResourceFormatType::R11G11B10: baseType = lit("[[rgb]] [[packed(r11g11b10)]] float3"); break; case ResourceFormatType::R5G6B5: case ResourceFormatType::R5G5B5A1: baseType = lit("[[hex]] short"); break; case ResourceFormatType::R9G9B9E5: baseType = lit("[[hex]] int"); break; case ResourceFormatType::R4G4B4A4: baseType = lit("[[hex]] short"); break; case ResourceFormatType::R4G4: baseType = lit("[[hex]] byte"); break; case ResourceFormatType::D16S8: case ResourceFormatType::D24S8: case ResourceFormatType::D32S8: case ResourceFormatType::YUV8: baseType = lit("[[hex]] byte4"); break; case ResourceFormatType::YUV10: case ResourceFormatType::YUV12: case ResourceFormatType::YUV16: baseType = lit("[[hex]] short4"); break; case ResourceFormatType::A8: case ResourceFormatType::S8: case ResourceFormatType::Undefined: baseType = lit("[[hex]] byte"); break; } if(tex.type == TextureType::Buffer) return QFormatStr("%1 %2;").arg(baseType).arg(varName); return QFormatStr("%1 %2[%3];").arg(baseType).arg(varName).arg(w); } QString BufferFormatter::GetBufferFormatString(Packing::Rules pack, const ShaderResource &res, const ResourceFormat &viewFormat) { QString format; if(!res.variableType.members.empty()) { QString structName = res.variableType.descriptor.name; if(structName.isEmpty()) structName = lit("el"); QList declaredStructs; format = DeclareStruct(pack, declaredStructs, structName, res.variableType.members, 0, QString()); format = QFormatStr("%1\n\n%2\n\n%3 buffer[];").arg(DeclarePacking(pack)).arg(format).arg(structName); } else { const auto &desc = res.variableType.descriptor; if(viewFormat.type == ResourceFormatType::Undefined) { if(desc.type == VarType::Unknown) { format = desc.name; } else { if(desc.RowMajor() && desc.rows > 1 && desc.columns > 1) format += lit("[[row_major]] "); format += ToQStr(desc.type); if(desc.rows > 1 && desc.columns > 1) format += QFormatStr("%1x%2").arg(desc.rows).arg(desc.columns); else if(desc.columns > 1) format += QString::number(desc.columns); if(!desc.name.empty()) format += lit(" ") + desc.name; if(desc.elements > 1) format += QFormatStr("[%1]").arg(desc.elements); } } else { if(viewFormat.type == ResourceFormatType::R10G10B10A2) { if(viewFormat.compType == CompType::UInt) format = lit("[[packed(r10g10b10a2)]] uint4"); if(viewFormat.compType == CompType::UNorm) format = lit("[[packed(r10g10b10a2)]] [[unorm]] uint4"); } else if(viewFormat.type == ResourceFormatType::R11G11B10) { format = lit("[[packed(r11g11b10]] float3"); } else { switch(viewFormat.compByteWidth) { case 1: { if(viewFormat.compType == CompType::UNorm || viewFormat.compType == CompType::UNormSRGB) format = lit("[[unorm]] ubyte"); if(viewFormat.compType == CompType::SNorm) format = lit("[[snorm]] byte"); if(viewFormat.compType == CompType::UInt) format = lit("ubyte"); if(viewFormat.compType == CompType::SInt) format = lit("byte"); break; } case 2: { if(viewFormat.compType == CompType::UNorm || viewFormat.compType == CompType::UNormSRGB) format = lit("[[unorm]] ushort"); if(viewFormat.compType == CompType::SNorm) format = lit("[[snorm]] short"); if(viewFormat.compType == CompType::UInt) format = lit("ushort"); if(viewFormat.compType == CompType::SInt) format = lit("short"); if(viewFormat.compType == CompType::Float) format = lit("half"); break; } case 4: { if(viewFormat.compType == CompType::UNorm || viewFormat.compType == CompType::UNormSRGB) format = lit("unormf"); if(viewFormat.compType == CompType::SNorm) format = lit("snormf"); if(viewFormat.compType == CompType::UInt) format = lit("uint"); if(viewFormat.compType == CompType::SInt) format = lit("int"); if(viewFormat.compType == CompType::Float) format = lit("float"); break; } } format += QString::number(viewFormat.compCount); } } } return format; } uint32_t BufferFormatter::GetVarStraddleSize(const ShaderConstant &var) { if(var.type.descriptor.type == VarType::Enum) return var.type.descriptor.arrayByteStride; // structs don't themselves have a straddle size // this is fine because the struct members itself don't straddle, and the alignment of the max of // their members. A struct that contains a vector will have to satisfy that vector's alignment. on // std140/430 this means the struct will be aligned such that as long as its members don't // straddle then any aligned placement of the struct they also won't straddle. // for D3D cbuffers where vectors have float alignment, structs are aligned to 16 always. // all others are scalar so straddling is allowed - i.e there is no packing scheme that disallows // straddling but allows vector AND struct placement so freely that a vector member could avoid // straddling until the struct is placed at a particular offset. if(!var.type.members.empty()) return 0; if(var.type.descriptor.rows > 1) return var.type.descriptor.matrixByteStride; return VarTypeByteSize(var.type.descriptor.type) * var.type.descriptor.columns; } uint32_t BufferFormatter::GetVarSize(const ShaderConstant &var) { if(var.type.descriptor.elements > 1 && var.type.descriptor.elements != ~0U) return var.type.descriptor.arrayByteStride * var.type.descriptor.elements; if(var.type.descriptor.type == VarType::Enum) return var.type.descriptor.arrayByteStride; if(!var.type.members.empty()) return var.type.descriptor.arrayByteStride; if(var.type.descriptor.rows > 1) { if(var.type.descriptor.RowMajor()) return var.type.descriptor.matrixByteStride * var.type.descriptor.rows; else return var.type.descriptor.matrixByteStride * var.type.descriptor.columns; } return VarTypeByteSize(var.type.descriptor.type) * var.type.descriptor.columns; } uint32_t BufferFormatter::GetAlignment(Packing::Rules pack, const ShaderConstant &c) { uint32_t ret = 1; if(c.type.descriptor.type == VarType::Struct) { for(const ShaderConstant &m : c.type.members) ret = std::max(ret, GetAlignment(pack, m)); } else if(c.type.descriptor.type == VarType::Enum) { ret = c.type.descriptor.arrayByteStride; } else if(c.type.members.empty()) { uint32_t align = VarTypeByteSize(c.type.descriptor.type); // if vectors aren't component aligned we need to calculate the alignment based on the size of // the vectors if(!pack.vector_align_component) { // column major matrices have vectors that are 'rows' long. Everything else is vectors of // 'columns' long uint8_t vecSize = c.type.descriptor.columns; if(c.type.descriptor.rows > 1 && c.type.descriptor.ColMajor()) vecSize = c.type.descriptor.rows; // 3- and 4- vectors are 4-component aligned if(vecSize >= 3) align *= 4; // 2- vectors are 2-component aligned else if(vecSize == 2) align *= 2; } ret = std::max(ret, align); } return ret; } uint32_t BufferFormatter::GetUnpaddedStructSize(const rdcarray &members) { uint32_t lastMemberStart = 0; if(members.empty()) return 0; const ShaderConstant *lastChild = &members.back(); lastMemberStart += lastChild->byteOffset; while(lastChild->type.descriptor.type != VarType::Enum && !lastChild->type.members.isEmpty()) { if(lastChild->type.descriptor.elements != ~0U) lastMemberStart += (qMax(lastChild->type.descriptor.elements, 1U) - 1) * lastChild->type.descriptor.arrayByteStride; lastChild = &lastChild->type.members.back(); lastMemberStart += lastChild->byteOffset; } return lastMemberStart + GetVarSize(*lastChild); } QString BufferFormatter::DeclareStruct(Packing::Rules pack, QList &declaredStructs, const QString &name, const rdcarray &members, uint32_t requiredByteStride, QString innerSkippedPrefixString) { QString declarations; QString ret; ret = lit("struct %1\n{\n").arg(name); ret += innerSkippedPrefixString; uint32_t offset = 0; uint32_t structAlignment = 1; for(int i = 0; i < members.count(); i++) { const uint32_t alignment = GetAlignment(pack, members[i]); const uint32_t vecsize = GetVarStraddleSize(members[i]); const uint32_t size = GetVarSize(members[i]); structAlignment = std::max(structAlignment, alignment); offset = AlignUp(offset, alignment); // if things can't straddle 16-byte boundaries, check that and enforce if(!pack.vector_straddle_16b) { if(offset / 16 != (offset + vecsize - 1) / 16) offset = AlignUp(offset, 16U); } // if we don't have tight arrays, arrays and structs begin at 16-byte boundaries if(!pack.tight_arrays && (members[i].type.descriptor.type == VarType::Struct || members[i].type.descriptor.elements > 1 || members[i].type.descriptor.rows > 1)) { offset = AlignUp(offset, 16U); } // if this variable is placed later, add an offset annotation if(offset < members[i].byteOffset) ret += lit(" [[offset(%1)]]\n").arg(members[i].byteOffset); else if(offset > members[i].byteOffset) qCritical() << "Unexpected offset overlap at" << QString(members[i].name) << "in" << QString(name); offset = members[i].byteOffset; offset += size; // if we allow trailing overlap, remove the padding at the end of the struct/array if(pack.trailing_overlap) { if(members[i].type.descriptor.type == VarType::Struct) { offset -= (members[i].type.descriptor.arrayByteStride - GetUnpaddedStructSize(members[i].type.members)); } else if((members[i].type.descriptor.elements > 1 || members[i].type.descriptor.rows > 1) && !pack.tight_arrays) { uint8_t vecSize = members[i].type.descriptor.columns; if(members[i].type.descriptor.rows > 1 && members[i].type.descriptor.ColMajor()) vecSize = members[i].type.descriptor.rows; uint32_t elSize = GetAlignment(pack, members[i]); if(pack.vector_align_component) elSize *= vecSize; // the padding is the stride (which is rounded up to 16 for non-tight arrays) minus the size // of the last vector (whether or not this is an array of scalars, vectors or matrices offset -= 16 - elSize; } } QString arraySize; if(members[i].type.descriptor.elements > 1) { if(members[i].type.descriptor.elements != ~0U) arraySize = QFormatStr("[%1]").arg(members[i].type.descriptor.elements); else arraySize = lit("[]"); } QString varTypeName = members[i].type.descriptor.name; if(members[i].type.descriptor.pointerTypeID != ~0U) { const ShaderConstantType &pointeeType = PointerTypeRegistry::GetTypeDescriptor(members[i].type.descriptor.pointerTypeID); varTypeName = pointeeType.descriptor.name; varTypeName = varTypeName.replace(QLatin1Char('['), QLatin1Char('_')).replace(QLatin1Char(']'), QString()); if(!declaredStructs.contains(varTypeName)) { declaredStructs.push_back(varTypeName); declarations += DeclareStruct(pack, declaredStructs, varTypeName, pointeeType.members, pointeeType.descriptor.arrayByteStride, QString()) + lit("\n"); } varTypeName += lit("*"); } else if(members[i].type.descriptor.type == VarType::Struct) { // GL structs don't give us typenames (boo!) so give them unique names. This will mean some // structs get duplicated if they're used in multiple places, but not much we can do about // that. if(varTypeName.isEmpty() || varTypeName == lit("struct")) varTypeName = lit("anon%1").arg(declaredStructs.size()); varTypeName = varTypeName.replace(QLatin1Char('['), QLatin1Char('_')).replace(QLatin1Char(']'), QString()); if(!declaredStructs.contains(varTypeName)) { declaredStructs.push_back(varTypeName); declarations += DeclareStruct(pack, declaredStructs, varTypeName, members[i].type.members, members[i].type.descriptor.arrayByteStride, QString()) + lit("\n"); } } QString varName = members[i].name; if(varName.isEmpty()) varName = QFormatStr("_child%1").arg(i); if(varName[0] == QLatin1Char('[')) varName = varName.replace(QLatin1Char('['), QLatin1Char('_')).replace(QLatin1Char(']'), QString()); if(members[i].type.descriptor.rows > 1) { if(members[i].type.descriptor.RowMajor()) varTypeName = lit("[[row_major]] ") + varTypeName; uint32_t stride = GetAlignment(pack, members[i]); if(pack.vector_align_component) { if(members[i].type.descriptor.RowMajor()) stride *= members[i].type.descriptor.columns; else stride *= members[i].type.descriptor.rows; } if(!pack.tight_arrays) stride = 16; if(stride != members[i].type.descriptor.matrixByteStride) ret += lit("// unexpected matrix stride %1").arg(members[i].type.descriptor.matrixByteStride); } ret += QFormatStr(" %1 %2%3;\n").arg(varTypeName).arg(varName).arg(arraySize); } // if we don't have tight arrays, struct byte strides are always 16-byte aligned if(!pack.tight_arrays) { structAlignment = 16; } offset = AlignUp(offset, structAlignment); if(requiredByteStride > 0) { if(requiredByteStride > offset) ret = lit("[[size(%1)]]\n%2").arg(requiredByteStride).arg(ret); else if(requiredByteStride != offset) ret = lit("// Unexpected size of struct %1\n%2").arg(requiredByteStride).arg(ret); } ret += lit("}\n"); return declarations + ret; } QString BufferFormatter::DeclareStruct(Packing::Rules pack, const QString &name, const rdcarray &members, uint32_t requiredByteStride) { QList declaredStructs; QString structDef = DeclareStruct(pack, declaredStructs, name, members, requiredByteStride, QString()); return QFormatStr("%1\n\n%2").arg(DeclarePacking(pack)).arg(structDef); } ResourceFormat GetInterpretedResourceFormat(const ShaderConstant &elem) { ResourceFormat format; format.type = ResourceFormatType::Regular; if(elem.type.descriptor.flags & ShaderVariableFlags::R10G10B10A2) format.type = ResourceFormatType::R10G10B10A2; else if(elem.type.descriptor.flags & ShaderVariableFlags::R11G11B10) format.type = ResourceFormatType::R11G11B10; format.compType = VarTypeCompType(elem.type.descriptor.type); if(elem.type.descriptor.flags & ShaderVariableFlags::UNorm) format.compType = CompType::UNorm; else if(elem.type.descriptor.flags & ShaderVariableFlags::SNorm) format.compType = CompType::SNorm; format.compByteWidth = VarTypeByteSize(elem.type.descriptor.type); if(elem.type.descriptor.type == VarType::Enum) format.compByteWidth = elem.type.descriptor.arrayByteStride; if(elem.type.descriptor.RowMajor() || elem.type.descriptor.rows == 1) format.compCount = elem.type.descriptor.columns; else format.compCount = elem.type.descriptor.rows; return format; } static void FillShaderVarData(ShaderVariable &var, const ShaderConstant &elem, const byte *data, const byte *end) { int src = 0; uint32_t outerCount = elem.type.descriptor.rows; uint32_t innerCount = elem.type.descriptor.columns; bool colMajor = false; if(elem.type.descriptor.ColMajor() && outerCount > 1) { colMajor = true; std::swap(outerCount, innerCount); } QVariantList objs = GetVariants(GetInterpretedResourceFormat(elem), elem, data, end); if(objs.isEmpty()) { var.name = elem.name; var.value = ShaderValue(); var.flags |= ShaderVariableFlags::Truncated; return; } for(uint32_t outer = 0; outer < outerCount; outer++) { for(uint32_t inner = 0; inner < innerCount; inner++) { uint32_t dst = outer * innerCount + inner; QVariant o = objs[src]; src++; switch(var.type) { case VarType::Float: var.value.f32v[dst] = o.toFloat(); break; case VarType::Double: var.value.f64v[dst] = o.toDouble(); break; case VarType::Half: var.value.f16v[dst] = rdhalf::make(o.toFloat()); break; case VarType::Bool: var.value.u32v[dst] = o.toBool() ? 1 : 0; break; case VarType::ULong: var.value.u64v[dst] = o.toULongLong(); break; case VarType::UInt: var.value.u32v[dst] = o.toUInt(); break; case VarType::UShort: var.value.u16v[dst] = o.toUInt() & 0xffff; break; case VarType::UByte: var.value.u8v[dst] = o.toUInt() & 0xff; break; case VarType::SLong: var.value.s64v[dst] = o.toLongLong(); break; case VarType::SInt: var.value.s32v[dst] = o.toInt(); break; case VarType::SShort: var.value.u16v[dst] = (int16_t)qBound((int)INT16_MIN, o.toInt(), (int)INT16_MAX); break; case VarType::SByte: var.value.u8v[dst] = (int8_t)qBound((int)INT8_MIN, o.toInt(), (int)INT8_MAX); break; case VarType::Enum: case VarType::GPUPointer: // treat this as a 64-bit unsigned integer var.value.u64v[dst] = o.toULongLong(); break; case VarType::ConstantBlock: case VarType::ReadOnlyResource: case VarType::ReadWriteResource: case VarType::Sampler: case VarType::Unknown: case VarType::Struct: qCritical() << "Unexpected variable type" << ToQStr(var.type) << "in variable" << (QString)var.name; break; } } } } ShaderVariable InterpretShaderVar(const ShaderConstant &elem, const byte *data, const byte *end) { ShaderVariable ret; ret.name = elem.name; ret.type = elem.type.descriptor.type; ret.columns = qMin(elem.type.descriptor.columns, uint8_t(4)); ret.rows = qMin(elem.type.descriptor.rows, uint8_t(4)); ret.flags = elem.type.descriptor.flags; if(elem.type.descriptor.type != VarType::Enum && !elem.type.members.isEmpty()) { ret.rows = ret.columns = 0; if(elem.type.descriptor.elements > 1 && elem.type.descriptor.elements != ~0U) { rdcarray arrayElements; for(uint32_t a = 0; a < elem.type.descriptor.elements; a++) { rdcarray members; for(size_t m = 0; m < elem.type.members.size(); m++) { const ShaderConstant &member = elem.type.members[m]; members.push_back(InterpretShaderVar(member, data + member.byteOffset, end)); } arrayElements.push_back(ret); arrayElements.back().name = QFormatStr("%1[%2]").arg(ret.name).arg(a); arrayElements.back().members = members; data += elem.type.descriptor.arrayByteStride; } ret.members = arrayElements; } else { rdcarray members; for(size_t m = 0; m < elem.type.members.size(); m++) { const ShaderConstant &member = elem.type.members[m]; members.push_back(InterpretShaderVar(member, data + member.byteOffset, end)); } ret.members = members; } } else if(elem.type.descriptor.type == VarType::Struct && elem.type.members.isEmpty()) { } else if(elem.type.descriptor.elements > 1 && elem.type.descriptor.elements != ~0U) { rdcarray arrayElements; for(uint32_t a = 0; a < elem.type.descriptor.elements; a++) { arrayElements.push_back(ret); arrayElements.back().name = QFormatStr("%1[%2]").arg(ret.name).arg(a); FillShaderVarData(arrayElements.back(), elem, data, end); data += elem.type.descriptor.arrayByteStride; } ret.rows = ret.columns = 0; ret.members = arrayElements; } else { FillShaderVarData(ret, elem, data, end); } return ret; } static QVariant interpret(const ResourceFormat &f, uint16_t comp) { if(f.compByteWidth != 2 || f.compType == CompType::Float) return QVariant(); if(f.compType == CompType::SInt) { return (int16_t)comp; } else if(f.compType == CompType::UInt) { return comp; } else if(f.compType == CompType::SScaled) { return (float)((int16_t)comp); } else if(f.compType == CompType::UScaled) { return (float)comp; } else if(f.compType == CompType::UNorm || f.compType == CompType::UNormSRGB) { return (float)comp / (float)0xffff; } else if(f.compType == CompType::SNorm) { int16_t cast = (int16_t)comp; float ret = -1.0f; if(cast == -32768) ret = -1.0f; else ret = ((float)cast) / 32767.0f; return ret; } return QVariant(); } static QVariant interpret(const ResourceFormat &f, byte comp) { if(f.compByteWidth != 1 || f.compType == CompType::Float) return QVariant(); if(f.compType == CompType::SInt) { return (int8_t)comp; } else if(f.compType == CompType::UInt) { return comp; } else if(f.compType == CompType::SScaled) { return (float)((int8_t)comp); } else if(f.compType == CompType::UScaled) { return (float)comp; } else if(f.compType == CompType::UNorm || f.compType == CompType::UNormSRGB) { return ((float)comp) / 255.0f; } else if(f.compType == CompType::SNorm) { int8_t cast = (int8_t)comp; float ret = -1.0f; if(cast == -128) ret = -1.0f; else ret = ((float)cast) / 127.0f; return ret; } return QVariant(); } template inline T readObj(const byte *&data, const byte *end, bool &ok) { if(data + sizeof(T) > end) { ok = false; return T(); } T ret = *(T *)data; data += sizeof(T); return ret; } QVariantList GetVariants(ResourceFormat format, const ShaderConstant &var, const byte *&data, const byte *end) { const ShaderConstantDescriptor &varDesc = var.type.descriptor; QVariantList ret; bool ok = true; if(format.type == ResourceFormatType::R5G5B5A1) { uint16_t packed = readObj(data, end, ok); ret.push_back((float)((packed >> 0) & 0x1f) / 31.0f); ret.push_back((float)((packed >> 5) & 0x1f) / 31.0f); ret.push_back((float)((packed >> 10) & 0x1f) / 31.0f); ret.push_back(((packed & 0x8000) > 0) ? 1.0f : 0.0f); if(format.BGRAOrder()) { QVariant tmp = ret[2]; ret[2] = ret[0]; ret[0] = tmp; } } else if(format.type == ResourceFormatType::R5G6B5) { uint16_t packed = readObj(data, end, ok); ret.push_back((float)((packed >> 0) & 0x1f) / 31.0f); ret.push_back((float)((packed >> 5) & 0x3f) / 63.0f); ret.push_back((float)((packed >> 11) & 0x1f) / 31.0f); if(format.BGRAOrder()) { QVariant tmp = ret[2]; ret[2] = ret[0]; ret[0] = tmp; } } else if(format.type == ResourceFormatType::R4G4B4A4) { uint16_t packed = readObj(data, end, ok); ret.push_back((float)((packed >> 0) & 0xf) / 15.0f); ret.push_back((float)((packed >> 4) & 0xf) / 15.0f); ret.push_back((float)((packed >> 8) & 0xf) / 15.0f); ret.push_back((float)((packed >> 12) & 0xf) / 15.0f); if(format.BGRAOrder()) { QVariant tmp = ret[2]; ret[2] = ret[0]; ret[0] = tmp; } } else if(format.type == ResourceFormatType::R10G10B10A2) { // allow for vectors of this format - for raw buffer viewer for(int i = 0; i < int(format.compCount / 4); i++) { uint32_t packed = readObj(data, end, ok); uint32_t r = (packed >> 0) & 0x3ff; uint32_t g = (packed >> 10) & 0x3ff; uint32_t b = (packed >> 20) & 0x3ff; uint32_t a = (packed >> 30) & 0x003; if(format.BGRAOrder()) { uint32_t tmp = b; b = r; r = tmp; } if(format.compType == CompType::UInt) { ret.push_back(r); ret.push_back(g); ret.push_back(b); ret.push_back(a); } else if(format.compType == CompType::UScaled) { ret.push_back((float)r); ret.push_back((float)g); ret.push_back((float)b); ret.push_back((float)a); } else if(format.compType == CompType::SInt || format.compType == CompType::SScaled || format.compType == CompType::SNorm) { int ir, ig, ib, ia; // interpret RGB as 10-bit signed integers if(r <= 511) ir = (int)r; else ir = ((int)r) - 1024; if(g <= 511) ig = (int)g; else ig = ((int)g) - 1024; if(b <= 511) ib = (int)b; else ib = ((int)b) - 1024; // 2-bit signed integer if(a <= 1) ia = (int)a; else ia = ((int)a) - 4; if(format.compType == CompType::SInt) { ret.push_back(ir); ret.push_back(ig); ret.push_back(ib); ret.push_back(ia); } else if(format.compType == CompType::SScaled) { ret.push_back((float)ir); ret.push_back((float)ig); ret.push_back((float)ib); ret.push_back((float)ia); } else if(format.compType == CompType::SNorm) { if(ir == -512) ir = -511; if(ig == -512) ig = -511; if(ib == -512) ib = -511; if(ia == -2) ia = -1; ret.push_back((float)ir / 511.0f); ret.push_back((float)ig / 511.0f); ret.push_back((float)ib / 511.0f); ret.push_back((float)ia / 1.0f); } } else { ret.push_back((float)r / 1023.0f); ret.push_back((float)g / 1023.0f); ret.push_back((float)b / 1023.0f); ret.push_back((float)a / 3.0f); } } } else if(format.type == ResourceFormatType::R11G11B10) { uint32_t packed = readObj(data, end, ok); uint32_t mantissas[] = { (packed >> 0) & 0x3f, (packed >> 11) & 0x3f, (packed >> 22) & 0x1f, }; int32_t exponents[] = { int32_t(packed >> 6) & 0x1f, int32_t(packed >> 17) & 0x1f, int32_t(packed >> 27) & 0x1f, }; static const uint32_t leadbit[] = { 0x40, 0x40, 0x20, }; for(int i = 0; i < 3; i++) { if(mantissas[i] == 0 && exponents[i] == 0) { ret.push_back((float)0.0f); } else { if(exponents[i] == 0x1f) { // no sign bit, can't be negative infinity if(mantissas[i] == 0) ret.push_back((float)qInf()); else ret.push_back((float)qQNaN()); } else if(exponents[i] != 0) { // normal value, add leading bit uint32_t combined = leadbit[i] | mantissas[i]; // calculate value ret.push_back(((float)combined / (float)leadbit[i]) * qPow(2.0f, (float)exponents[i] - 15.0f)); } else if(exponents[i] == 0) { // we know xMantissa isn't 0 also, or it would have been caught above so // this is a subnormal value, pretend exponent is 1 and don't add leading bit ret.push_back(((float)mantissas[i] / (float)leadbit[i]) * qPow(2.0f, (float)1.0f - 15.0f)); } } } } else { const byte *base = data; uint32_t rowCount = varDesc.rows; uint32_t colCount = varDesc.columns; for(uint32_t row = 0; row < qMax(rowCount, 1U); row++) { for(uint32_t col = 0; col < qMax(colCount, 1U); col++) { if(varDesc.RowMajor() || rowCount == 1) data = base + row * varDesc.matrixByteStride + col * format.compByteWidth; else data = base + col * varDesc.matrixByteStride + row * format.compByteWidth; if(format.compType == CompType::Float) { if(format.compByteWidth == 8) ret.push_back(readObj(data, end, ok)); else if(format.compByteWidth == 4) ret.push_back(readObj(data, end, ok)); else if(format.compByteWidth == 2) ret.push_back((float)rdhalf::make(readObj(data, end, ok))); } else if(format.compType == CompType::SInt) { if(var.bitFieldSize == 0) { if(format.compByteWidth == 8) ret.push_back((qlonglong)readObj(data, end, ok)); else if(format.compByteWidth == 4) ret.push_back((int)readObj(data, end, ok)); else if(format.compByteWidth == 2) ret.push_back((int)readObj(data, end, ok)); else if(format.compByteWidth == 1) ret.push_back((int)readObj(data, end, ok)); } else { uint64_t uval = 0; if(format.compByteWidth == 8) uval = readObj(data, end, ok); else if(format.compByteWidth == 4) uval = readObj(data, end, ok); else if(format.compByteWidth == 2) uval = readObj(data, end, ok); else if(format.compByteWidth == 1) uval = readObj(data, end, ok); int64_t val = 0; if(ok) { // shift by the offset uval >>= var.bitFieldOffset; // mask by the size const uint64_t mask = ((1ULL << var.bitFieldSize) - 1ULL); uval &= mask; // sign extend by hand if(uval & (1ULL << (var.bitFieldSize - 1))) uval |= ~0ULL ^ mask; memcpy(&val, &uval, sizeof(uval)); } ret.push_back(val); } } else if(format.compType == CompType::UInt) { if(var.bitFieldSize == 0) { if(format.compByteWidth == 8) ret.push_back((qulonglong)readObj(data, end, ok)); else if(format.compByteWidth == 4) ret.push_back((uint32_t)readObj(data, end, ok)); else if(format.compByteWidth == 2) ret.push_back((uint32_t)readObj(data, end, ok)); else if(format.compByteWidth == 1) ret.push_back((uint32_t)readObj(data, end, ok)); } else { uint64_t val = 0; if(format.compByteWidth == 8) val = readObj(data, end, ok); else if(format.compByteWidth == 4) val = readObj(data, end, ok); else if(format.compByteWidth == 2) val = readObj(data, end, ok); else if(format.compByteWidth == 1) val = readObj(data, end, ok); if(ok) { // shift by the offset val >>= var.bitFieldOffset; // mask by the size val &= ((1ULL << var.bitFieldSize) - 1ULL); } else { val = 0; } ret.push_back(val); } if(var.type.descriptor.type == VarType::Enum) { uint64_t val = ret.back().toULongLong(); QString str = QApplication::translate("BufferFormatter", "Unknown %1 (%2)") .arg(QString(var.type.descriptor.name)) .arg(val); for(size_t i = 0; i < var.type.members.size(); i++) { if(val == var.type.members[i].defaultValue) { str = var.type.members[i].name; break; } } ret.back() = str; } } else if(format.compType == CompType::UScaled) { if(format.compByteWidth == 4) ret.push_back((float)readObj(data, end, ok)); else if(format.compByteWidth == 2) ret.push_back((float)readObj(data, end, ok)); else if(format.compByteWidth == 1) ret.push_back((float)readObj(data, end, ok)); } else if(format.compType == CompType::SScaled) { if(format.compByteWidth == 4) ret.push_back((float)readObj(data, end, ok)); else if(format.compByteWidth == 2) ret.push_back((float)readObj(data, end, ok)); else if(format.compByteWidth == 1) ret.push_back((float)readObj(data, end, ok)); } else if(format.compType == CompType::Depth) { if(format.compByteWidth == 4) { // 32-bit depth is native floats ret.push_back(readObj(data, end, ok)); } else if(format.compByteWidth == 3) { // 32-bit depth is normalised, masked against non-stencil bits uint32_t f = readObj(data, end, ok); f &= 0x00ffffff; ret.push_back((float)f / (float)0x00ffffff); } else if(format.compByteWidth == 2) { // 16-bit depth is normalised float f = (float)readObj(data, end, ok); ret.push_back(f / (float)0x0000ffff); } } else { // unorm/snorm if(format.compByteWidth == 4) { // should never hit this - no 32bit unorm/snorm type qCritical() << "Unexpected 4-byte unorm/snorm value"; ret.push_back((float)readObj(data, end, ok) / (float)0xffffffff); } else if(format.compByteWidth == 2) { ret.push_back(interpret(format, readObj(data, end, ok))); } else if(format.compByteWidth == 1) { ret.push_back(interpret(format, readObj(data, end, ok))); } } } } if(format.BGRAOrder()) { QVariant tmp = ret[2]; ret[2] = ret[0]; ret[0] = tmp; } } // we read off the end, return empty set if(!ok) ret.clear(); return ret; } QString TypeString(const ShaderVariable &v) { if(!v.members.isEmpty() || v.type == VarType::Struct) { if(v.type == VarType::Struct) { if(!v.members.empty() && v.members[0].name.contains('[')) return lit("struct[%2]").arg(v.members.count()); else return lit("struct"); } else { return QFormatStr("%1[%2]").arg(TypeString(v.members[0])).arg(v.members.count()); } } if(v.type == VarType::GPUPointer) return PointerTypeRegistry::GetTypeDescriptor(v.GetPointer()).descriptor.name + "*"; QString typeStr = ToQStr(v.type); if(v.type == VarType::ReadOnlyResource) typeStr = lit("Resource"); else if(v.type == VarType::ReadWriteResource) typeStr = lit("RW Resource"); else if(v.type == VarType::Sampler) typeStr = lit("Sampler"); else if(v.type == VarType::ConstantBlock) typeStr = lit("Constant Block"); if(v.type == VarType::Unknown) return lit("Typeless"); if(v.rows == 1 && v.columns == 1) return typeStr; if(v.rows == 1) return QFormatStr("%1%2").arg(typeStr).arg(v.columns); else return QFormatStr("%1%2x%3 (%4)") .arg(typeStr) .arg(v.rows) .arg(v.columns) .arg(v.RowMajor() ? lit("row_major") : lit("column_major")); } template static QString RowValuesToString(int cols, ShaderVariableFlags flags, el x, el y, el z, el w) { if(flags & ShaderVariableFlags::Truncated) { QString ret = lit("---"); for(int i = 1; i < cols; i++) ret += lit(", ---"); return ret; } const bool hex = bool(flags & ShaderVariableFlags::HexDisplay); if(bool(flags & ShaderVariableFlags::BinaryDisplay)) { if(cols == 1) return Formatter::BinFormat(x); else if(cols == 2) return QFormatStr("%1, %2").arg(Formatter::BinFormat(x)).arg(Formatter::BinFormat(y)); else if(cols == 3) return QFormatStr("%1, %2, %3") .arg(Formatter::BinFormat(x)) .arg(Formatter::BinFormat(y)) .arg(Formatter::BinFormat(z)); else return QFormatStr("%1, %2, %3, %4") .arg(Formatter::BinFormat(x)) .arg(Formatter::BinFormat(y)) .arg(Formatter::BinFormat(z)) .arg(Formatter::BinFormat(w)); } if(cols == 1) return Formatter::Format(x, hex); else if(cols == 2) return QFormatStr("%1, %2").arg(Formatter::Format(x, hex)).arg(Formatter::Format(y, hex)); else if(cols == 3) return QFormatStr("%1, %2, %3") .arg(Formatter::Format(x, hex)) .arg(Formatter::Format(y, hex)) .arg(Formatter::Format(z, hex)); else return QFormatStr("%1, %2, %3, %4") .arg(Formatter::Format(x, hex)) .arg(Formatter::Format(y, hex)) .arg(Formatter::Format(z, hex)) .arg(Formatter::Format(w, hex)); } QString RowString(const ShaderVariable &v, uint32_t row, VarType type) { if(type == VarType::Unknown) type = v.type; if(v.type == VarType::GPUPointer) return ToQStr(v.GetPointer()); if(v.type == VarType::Struct) return lit("{ ... }"); switch(type) { case VarType::Float: return RowValuesToString((int)v.columns, v.flags, v.value.f32v[row * v.columns + 0], v.value.f32v[row * v.columns + 1], v.value.f32v[row * v.columns + 2], v.value.f32v[row * v.columns + 3]); case VarType::Double: return RowValuesToString((int)v.columns, v.flags, v.value.f64v[row * v.columns + 0], v.value.f64v[row * v.columns + 1], v.value.f64v[row * v.columns + 2], v.value.f64v[row * v.columns + 3]); case VarType::Half: return RowValuesToString((int)v.columns, v.flags, v.value.f16v[row * v.columns + 0], v.value.f16v[row * v.columns + 1], v.value.f16v[row * v.columns + 2], v.value.f16v[row * v.columns + 3]); case VarType::Bool: return RowValuesToString((int)v.columns, v.flags, v.value.u32v[row * v.columns + 0] ? true : false, v.value.u32v[row * v.columns + 1] ? true : false, v.value.u32v[row * v.columns + 2] ? true : false, v.value.u32v[row * v.columns + 3] ? true : false); case VarType::ULong: return RowValuesToString((int)v.columns, v.flags, v.value.u64v[row * v.columns + 0], v.value.u64v[row * v.columns + 1], v.value.u64v[row * v.columns + 2], v.value.u64v[row * v.columns + 3]); case VarType::UInt: return RowValuesToString((int)v.columns, v.flags, v.value.u32v[row * v.columns + 0], v.value.u32v[row * v.columns + 1], v.value.u32v[row * v.columns + 2], v.value.u32v[row * v.columns + 3]); case VarType::UShort: return RowValuesToString((int)v.columns, v.flags, v.value.u16v[row * v.columns + 0], v.value.u16v[row * v.columns + 1], v.value.u16v[row * v.columns + 2], v.value.u16v[row * v.columns + 3]); case VarType::UByte: return RowValuesToString((int)v.columns, v.flags, v.value.u8v[row * v.columns + 0], v.value.u8v[row * v.columns + 1], v.value.u8v[row * v.columns + 2], v.value.u8v[row * v.columns + 3]); case VarType::SLong: return RowValuesToString((int)v.columns, v.flags, v.value.s64v[row * v.columns + 0], v.value.s64v[row * v.columns + 1], v.value.s64v[row * v.columns + 2], v.value.s64v[row * v.columns + 3]); case VarType::SInt: return RowValuesToString((int)v.columns, v.flags, v.value.s32v[row * v.columns + 0], v.value.s32v[row * v.columns + 1], v.value.s32v[row * v.columns + 2], v.value.s32v[row * v.columns + 3]); case VarType::SShort: return RowValuesToString((int)v.columns, v.flags, v.value.s16v[row * v.columns + 0], v.value.s16v[row * v.columns + 1], v.value.s16v[row * v.columns + 2], v.value.s16v[row * v.columns + 3]); case VarType::SByte: return RowValuesToString((int)v.columns, v.flags, v.value.s8v[row * v.columns + 0], v.value.s8v[row * v.columns + 1], v.value.s8v[row * v.columns + 2], v.value.s8v[row * v.columns + 3]); case VarType::GPUPointer: return ToQStr(v.GetPointer()); case VarType::Enum: case VarType::ConstantBlock: case VarType::ReadOnlyResource: case VarType::ReadWriteResource: case VarType::Sampler: case VarType::Unknown: case VarType::Struct: break; } return lit("???"); } QString VarString(const ShaderVariable &v) { if(!v.members.isEmpty()) return QString(); if(v.rows == 1) return RowString(v, 0); QString ret; for(int i = 0; i < (int)v.rows; i++) { if(i > 0) ret += lit("\n"); ret += lit("{") + RowString(v, i) + lit("}"); } return ret; } QString RowTypeString(const ShaderVariable &v) { if(!v.members.isEmpty() || v.type == VarType::Struct) { if(v.type == VarType::Struct) return lit("struct"); else return lit("flibbertygibbet"); } if(v.rows == 0 && v.columns == 0) return lit("-"); if(v.type == VarType::GPUPointer) return PointerTypeRegistry::GetTypeDescriptor(v.GetPointer()).descriptor.name + "*"; QString typeStr = ToQStr(v.type); if(v.columns == 1) return typeStr; return QFormatStr("%1%2").arg(typeStr).arg(v.columns); }