/****************************************************************************** * The MIT License (MIT) * * Copyright (c) 2016 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 "d3d12_command_queue.h" #include "d3d12_command_list.h" #include "d3d12_resources.h" void STDMETHODCALLTYPE WrappedID3D12CommandQueue::UpdateTileMappings( ID3D12Resource *pResource, UINT NumResourceRegions, const D3D12_TILED_RESOURCE_COORDINATE *pResourceRegionStartCoordinates, const D3D12_TILE_REGION_SIZE *pResourceRegionSizes, ID3D12Heap *pHeap, UINT NumRanges, const D3D12_TILE_RANGE_FLAGS *pRangeFlags, const UINT *pHeapRangeStartOffsets, const UINT *pRangeTileCounts, D3D12_TILE_MAPPING_FLAGS Flags) { D3D12NOTIMP("Tiled Resources"); m_pReal->UpdateTileMappings(Unwrap(pResource), NumResourceRegions, pResourceRegionStartCoordinates, pResourceRegionSizes, Unwrap(pHeap), NumRanges, pRangeFlags, pHeapRangeStartOffsets, pRangeTileCounts, Flags); } void STDMETHODCALLTYPE WrappedID3D12CommandQueue::CopyTileMappings( ID3D12Resource *pDstResource, const D3D12_TILED_RESOURCE_COORDINATE *pDstRegionStartCoordinate, ID3D12Resource *pSrcResource, const D3D12_TILED_RESOURCE_COORDINATE *pSrcRegionStartCoordinate, const D3D12_TILE_REGION_SIZE *pRegionSize, D3D12_TILE_MAPPING_FLAGS Flags) { D3D12NOTIMP("Tiled Resources"); m_pReal->CopyTileMappings(Unwrap(pDstResource), pDstRegionStartCoordinate, Unwrap(pSrcResource), pSrcRegionStartCoordinate, pRegionSize, Flags); } bool WrappedID3D12CommandQueue::Serialise_ExecuteCommandLists(UINT NumCommandLists, ID3D12CommandList *const *ppCommandLists) { SERIALISE_ELEMENT(ResourceId, queueId, GetResourceID()); SERIALISE_ELEMENT(UINT, numCmds, NumCommandLists); vector cmdIds; ID3D12GraphicsCommandList **cmds = m_State >= WRITING ? NULL : new ID3D12GraphicsCommandList *[numCmds]; if(m_State >= WRITING) { for(UINT i = 0; i < numCmds; i++) { D3D12ResourceRecord *record = GetRecord(ppCommandLists[i]); RDCASSERT(record->bakedCommands); if(record->bakedCommands) cmdIds.push_back(record->bakedCommands->GetResourceID()); } } m_pSerialiser->Serialise("ppCommandLists", cmdIds); if(m_State < WRITING) { for(UINT i = 0; i < numCmds; i++) { cmds[i] = cmdIds[i] != ResourceId() ? GetResourceManager()->GetLiveAs(cmdIds[i]) : NULL; } } const string desc = m_pSerialiser->GetDebugStr(); // debug messages { vector debugMessages; if(m_State == WRITING_CAPFRAME) debugMessages = m_pDevice->GetDebugMessages(); SERIALISE_ELEMENT(uint32_t, NumMessages, (uint32_t)debugMessages.size()); for(uint32_t i = 0; i < NumMessages; i++) { ScopedContext msgscope(m_pSerialiser, "DebugMessage", "DebugMessage", 0, false); string msgDesc; if(m_State >= WRITING) msgDesc = debugMessages[i].description.elems; SERIALISE_ELEMENT(uint32_t, Category, debugMessages[i].category); SERIALISE_ELEMENT(uint32_t, Severity, debugMessages[i].severity); SERIALISE_ELEMENT(uint32_t, ID, debugMessages[i].messageID); SERIALISE_ELEMENT(string, Description, msgDesc); if(m_State == READING) { DebugMessage msg; msg.source = eDbgSource_API; msg.category = (DebugMessageCategory)Category; msg.severity = (DebugMessageSeverity)Severity; msg.messageID = ID; msg.description = Description; m_Cmd.m_EventMessages.push_back(msg); } } } ID3D12CommandQueue *real = NULL; if(m_State <= EXECUTING) real = Unwrap(GetResourceManager()->GetLiveAs(queueId)); if(m_State == READING) { for(uint32_t i = 0; i < numCmds; i++) { if(m_Cmd.m_BakedCmdListInfo[cmdIds[i]].executeEvents.empty() || m_Cmd.m_BakedCmdListInfo[cmdIds[i]].executeEvents[0].patched) { ID3D12CommandList *list = Unwrap(cmds[i]); real->ExecuteCommandLists(1, &list); #if ENABLED(SINGLE_FLUSH_VALIDATE) m_pDevice->GPUSync(); #endif } else { BakedCmdListInfo &info = m_Cmd.m_BakedCmdListInfo[cmdIds[i]]; // execute the first half of the cracked list ID3D12CommandList *list = Unwrap(info.crackedLists[0]); real->ExecuteCommandLists(1, &list); for(size_t c = 1; c < info.crackedLists.size(); c++) { m_pDevice->GPUSync(); // readback the patch buffer and perform patching m_ReplayList->PatchExecuteIndirect(info, uint32_t(c - 1)); // execute next list with this indirect. list = Unwrap(info.crackedLists[c]); real->ExecuteCommandLists(1, &list); } #if ENABLED(SINGLE_FLUSH_VALIDATE) m_pDevice->GPUSync(); #endif } } for(uint32_t i = 0; i < numCmds; i++) { ResourceId cmd = GetResourceManager()->GetLiveID(cmdIds[i]); m_pDevice->ApplyBarriers(m_Cmd.m_BakedCmdListInfo[cmd].barriers); } m_Cmd.AddEvent(desc); // we're adding multiple events, need to increment ourselves m_Cmd.m_RootEventID++; string basename = "ExecuteCommandLists(" + ToStr::Get(numCmds) + ")"; for(uint32_t c = 0; c < numCmds; c++) { string name = StringFormat::Fmt("=> %s[%u]: ID3D12CommandList(%s)", basename.c_str(), c, ToStr::Get(cmdIds[c]).c_str()); // add a fake marker FetchDrawcall draw; draw.name = name; draw.flags = eDraw_PassBoundary | eDraw_BeginPass; m_Cmd.AddEvent(name); m_Cmd.AddDrawcall(draw, true); m_Cmd.m_RootEventID++; BakedCmdListInfo &cmdBufInfo = m_Cmd.m_BakedCmdListInfo[cmdIds[c]]; // insert the baked command list in-line into this list of notes, assigning new event and // drawIDs m_Cmd.InsertDrawsAndRefreshIDs(cmdIds[c], cmdBufInfo.draw->children); for(size_t e = 0; e < cmdBufInfo.draw->executedCmds.size(); e++) { vector &submits = m_Cmd.m_Partial[D3D12CommandData::Secondary].cmdListExecs[cmdBufInfo.draw->executedCmds[e]]; for(size_t s = 0; s < submits.size(); s++) submits[s] += m_Cmd.m_RootEventID; } for(size_t i = 0; i < cmdBufInfo.debugMessages.size(); i++) { DebugMessage msg = cmdBufInfo.debugMessages[i]; msg.eventID += m_Cmd.m_RootEventID; m_pDevice->AddDebugMessage(msg); } // only primary command lists can be submitted m_Cmd.m_Partial[D3D12CommandData::Primary].cmdListExecs[cmdIds[c]].push_back( m_Cmd.m_RootEventID); m_Cmd.m_RootEventID += cmdBufInfo.eventCount; m_Cmd.m_RootDrawcallID += cmdBufInfo.drawCount; name = StringFormat::Fmt("=> %s[%u]: Close(%s)", basename.c_str(), c, ToStr::Get(cmdIds[c]).c_str()); draw.name = name; draw.flags = eDraw_PassBoundary | eDraw_EndPass; m_Cmd.AddEvent(name); m_Cmd.AddDrawcall(draw, true); m_Cmd.m_RootEventID++; } // account for the outer loop thinking we've added one event and incrementing, // since we've done all the handling ourselves this will be off by one. m_Cmd.m_RootEventID--; } else if(m_State == EXECUTING) { // account for the queue submit event m_Cmd.m_RootEventID++; uint32_t startEID = m_Cmd.m_RootEventID; // advance m_CurEventID to match the events added when reading for(uint32_t c = 0; c < numCmds; c++) { // 2 extra for the virtual labels around the command list m_Cmd.m_RootEventID += 2 + m_Cmd.m_BakedCmdListInfo[cmdIds[c]].eventCount; m_Cmd.m_RootDrawcallID += 2 + m_Cmd.m_BakedCmdListInfo[cmdIds[c]].drawCount; } // same accounting for the outer loop as above m_Cmd.m_RootEventID--; if(numCmds == 0) { // do nothing, don't bother with the logic below } else if(m_Cmd.m_LastEventID <= startEID) { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit no replay %u == %u", m_Cmd.m_LastEventID, startEID); #endif } else if(m_Cmd.m_DrawcallCallback && m_Cmd.m_DrawcallCallback->RecordAllCmds()) { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit re-recording from %u", m_Cmd.m_RootEventID); #endif vector rerecordedCmds; for(uint32_t c = 0; c < numCmds; c++) { ID3D12CommandList *cmd = m_Cmd.RerecordCmdList(cmdIds[c]); ResourceId rerecord = GetResID(cmd); #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit fully re-recorded replay of %llu, using %llu", cmdIds[c], rerecord); #endif rerecordedCmds.push_back(Unwrap(cmd)); m_pDevice->ApplyBarriers(m_Cmd.m_BakedCmdListInfo[rerecord].barriers); } #if ENABLED(SINGLE_FLUSH_VALIDATE) for(size_t i = 0; i < rerecordedCmds.size(); i++) { real->ExecuteCommandLists(1, &rerecordedCmds[i]); m_pDevice->GPUSync(); } #else real->ExecuteCommandLists((UINT)rerecordedCmds.size(), &rerecordedCmds[0]); #endif } else if(m_Cmd.m_LastEventID > startEID && m_Cmd.m_LastEventID < m_Cmd.m_RootEventID) { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit partial replay %u < %u", m_Cmd.m_LastEventID, m_Cmd.m_RootEventID); #endif uint32_t eid = startEID; vector trimmedCmdIds; vector trimmedCmds; for(uint32_t c = 0; c < numCmds; c++) { // account for the virtual label at the start of the events here // so it matches up to baseEvent eid++; uint32_t end = eid + m_Cmd.m_BakedCmdListInfo[cmdIds[c]].eventCount; if(eid == m_Cmd.m_Partial[D3D12CommandData::Primary].baseEvent) { ID3D12GraphicsCommandList *list = m_Cmd.RerecordCmdList(cmdIds[c], D3D12CommandData::Primary); ResourceId partial = GetResID(list); #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit partial replay of %llu at %u, using %llu", cmdIds[c], eid, partial); #endif trimmedCmdIds.push_back(partial); trimmedCmds.push_back(Unwrap(list)); } else if(m_Cmd.m_LastEventID >= end) { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit full replay %llu", cmdIds[c]); #endif trimmedCmdIds.push_back(cmdIds[c]); trimmedCmds.push_back(Unwrap(GetResourceManager()->GetLiveAs(cmdIds[c]))); } else { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue not submitting %llu", cmdIds[c]); #endif } // 1 extra to account for the virtual end command list label (begin is accounted for // above) eid += 1 + m_Cmd.m_BakedCmdListInfo[cmdIds[c]].eventCount; } RDCASSERT(trimmedCmds.size() > 0); #if ENABLED(SINGLE_FLUSH_VALIDATE) for(size_t i = 0; i < trimmedCmds.size(); i++) { real->ExecuteCommandLists(1, &trimmedCmds[i]); m_pDevice->GPUSync(); } #else real->ExecuteCommandLists((UINT)trimmedCmds.size(), &trimmedCmds[0]); #endif for(uint32_t i = 0; i < trimmedCmdIds.size(); i++) { ResourceId cmd = trimmedCmdIds[i]; m_pDevice->ApplyBarriers(m_Cmd.m_BakedCmdListInfo[cmd].barriers); } } else { #if ENABLED(VERBOSE_PARTIAL_REPLAY) RDCDEBUG("Queue Submit full replay %u >= %u", m_Cmd.m_LastEventID, m_Cmd.m_RootEventID); #endif ID3D12CommandList **unwrapped = new ID3D12CommandList *[numCmds]; for(uint32_t i = 0; i < numCmds; i++) unwrapped[i] = Unwrap(cmds[i]); #if ENABLED(SINGLE_FLUSH_VALIDATE) for(UINT i = 0; i < numCmds; i++) { real->ExecuteCommandLists(1, &unwrapped[i]); m_pDevice->GPUSync(); } #else real->ExecuteCommandLists(numCmds, unwrapped); #endif SAFE_DELETE_ARRAY(unwrapped); for(uint32_t i = 0; i < numCmds; i++) { ResourceId cmd = GetResourceManager()->GetLiveID(cmdIds[i]); m_pDevice->ApplyBarriers(m_Cmd.m_BakedCmdListInfo[cmd].barriers); } } } SAFE_DELETE_ARRAY(cmds); return true; } void STDMETHODCALLTYPE WrappedID3D12CommandQueue::ExecuteCommandLists( UINT NumCommandLists, ID3D12CommandList *const *ppCommandLists) { ID3D12CommandList **unwrapped = m_pDevice->GetTempArray(NumCommandLists); for(UINT i = 0; i < NumCommandLists; i++) unwrapped[i] = Unwrap(ppCommandLists[i]); m_pReal->ExecuteCommandLists(NumCommandLists, unwrapped); if(m_State >= WRITING) { SCOPED_LOCK(m_Lock); SCOPED_LOCK(m_pDevice->GetCapTransitionLock()); bool capframe = (m_State == WRITING_CAPFRAME); set refdIDs; for(UINT i = 0; i < NumCommandLists; i++) { D3D12ResourceRecord *record = GetRecord(ppCommandLists[i]); if(record->ContainsExecuteIndirect) m_QueueRecord->ContainsExecuteIndirect = true; m_pDevice->ApplyBarriers(record->bakedCommands->cmdInfo->barriers); // need to lock the whole section of code, not just the check on // m_State, as we also need to make sure we don't check the state, // start marking dirty resources then while we're doing so the // state becomes capframe. // the next sections where we mark resources referenced and add // the submit chunk to the frame record don't have to be protected. // Only the decision of whether we're inframe or not, and marking // dirty. if(capframe) { for(auto it = record->bakedCommands->cmdInfo->dirtied.begin(); it != record->bakedCommands->cmdInfo->dirtied.end(); ++it) GetResourceManager()->MarkPendingDirty(*it); } else { for(auto it = record->bakedCommands->cmdInfo->dirtied.begin(); it != record->bakedCommands->cmdInfo->dirtied.end(); ++it) GetResourceManager()->MarkDirtyResource(*it); } if(capframe) { // any descriptor copies or writes could reference new resources not in the // bound descs list yet. So we take all of those referenced descriptors and // include them to see if we need to flush std::vector dynDescRefs; m_pDevice->GetDynamicDescriptorReferences(dynDescRefs); for(size_t d = 0; d < dynDescRefs.size(); d++) { ResourceId id, id2; FrameRefType ref = eFrameRef_Read; dynDescRefs[d].GetRefIDs(id, id2, ref); if(id != ResourceId()) { refdIDs.insert(id); GetResourceManager()->MarkResourceFrameReferenced(id, ref); } if(id2 != ResourceId()) { refdIDs.insert(id2); GetResourceManager()->MarkResourceFrameReferenced(id2, ref); } } // for each bound descriptor table, mark it referenced as well as all resources currently // bound to it for(auto it = record->bakedCommands->cmdInfo->boundDescs.begin(); it != record->bakedCommands->cmdInfo->boundDescs.end(); ++it) { D3D12Descriptor *desc = *it; ResourceId id, id2; FrameRefType ref = eFrameRef_Read; desc->GetRefIDs(id, id2, ref); if(id != ResourceId()) { refdIDs.insert(id); GetResourceManager()->MarkResourceFrameReferenced(id, ref); } if(id2 != ResourceId()) { refdIDs.insert(id2); GetResourceManager()->MarkResourceFrameReferenced(id2, ref); } } // pull in frame refs from this baked command list record->bakedCommands->AddResourceReferences(GetResourceManager()); record->bakedCommands->AddReferencedIDs(refdIDs); // reference all executed bundles as well for(size_t b = 0; b < record->bakedCommands->cmdInfo->bundles.size(); b++) { record->bakedCommands->cmdInfo->bundles[b]->bakedCommands->AddResourceReferences( GetResourceManager()); record->bakedCommands->cmdInfo->bundles[b]->bakedCommands->AddReferencedIDs(refdIDs); GetResourceManager()->MarkResourceFrameReferenced( record->bakedCommands->cmdInfo->bundles[b]->GetResourceID(), eFrameRef_Read); record->bakedCommands->cmdInfo->bundles[b]->bakedCommands->AddRef(); } { m_CmdListRecords.push_back(record->bakedCommands); for(size_t sub = 0; sub < record->bakedCommands->cmdInfo->bundles.size(); sub++) m_CmdListRecords.push_back(record->bakedCommands->cmdInfo->bundles[sub]->bakedCommands); } record->bakedCommands->AddRef(); } record->cmdInfo->dirtied.clear(); } if(capframe) { vector maps = m_pDevice->GetMaps(); for(auto it = maps.begin(); it != maps.end(); ++it) { WrappedID3D12Resource *res = it->res; UINT subres = it->subres; size_t size = (size_t)it->totalSize; // only need to flush memory that could affect this submitted batch of work if(refdIDs.find(res->GetResourceID()) == refdIDs.end()) { RDCDEBUG("Map of memory %llu not referenced in this queue - not flushing", res->GetResourceID()); continue; } size_t diffStart = 0, diffEnd = 0; bool found = true; byte *ref = res->GetShadow(subres); byte *data = res->GetMap(subres); if(ref) found = FindDiffRange(data, ref, size, diffStart, diffEnd); else diffEnd = size; if(found) { RDCLOG("Persistent map flush forced for %llu (%llu -> %llu)", res->GetResourceID(), (uint64_t)diffStart, (uint64_t)diffEnd); D3D12_RANGE range = {diffStart, diffEnd}; m_pDevice->MapDataWrite(res, subres, data, range); if(ref == NULL) { res->AllocShadow(subres, size); ref = res->GetShadow(subres); } // update comparison shadow for next time memcpy(ref, res->GetMap(subres), size); GetResourceManager()->MarkPendingDirty(res->GetResourceID()); } else { RDCDEBUG("Persistent map flush not needed for %llu", res->GetResourceID()); } } for(UINT i = 0; i < NumCommandLists; i++) { SCOPED_SERIALISE_CONTEXT(EXECUTE_CMD_LISTS); Serialise_ExecuteCommandLists(1, ppCommandLists + i); m_QueueRecord->AddChunk(scope.Get()); } } } } void STDMETHODCALLTYPE WrappedID3D12CommandQueue::SetMarker(UINT Metadata, const void *pData, UINT Size) { m_pReal->SetMarker(Metadata, pData, Size); } void STDMETHODCALLTYPE WrappedID3D12CommandQueue::BeginEvent(UINT Metadata, const void *pData, UINT Size) { m_pReal->BeginEvent(Metadata, pData, Size); } void STDMETHODCALLTYPE WrappedID3D12CommandQueue::EndEvent() { m_pReal->EndEvent(); } bool WrappedID3D12CommandQueue::Serialise_Signal(ID3D12Fence *pFence, UINT64 Value) { SERIALISE_ELEMENT(ResourceId, Fence, GetResID(pFence)); SERIALISE_ELEMENT(UINT64, val, Value); if(m_State <= EXECUTING && GetResourceManager()->HasLiveResource(Fence)) { pFence = GetResourceManager()->GetLiveAs(Fence); m_pReal->Signal(Unwrap(pFence), val); } return true; } HRESULT STDMETHODCALLTYPE WrappedID3D12CommandQueue::Signal(ID3D12Fence *pFence, UINT64 Value) { if(m_State == WRITING_CAPFRAME) { SCOPED_LOCK(m_Lock); SCOPED_SERIALISE_CONTEXT(SIGNAL); Serialise_Signal(pFence, Value); m_QueueRecord->AddChunk(scope.Get()); GetResourceManager()->MarkResourceFrameReferenced(GetResID(pFence), eFrameRef_Read); } return m_pReal->Signal(Unwrap(pFence), Value); } bool WrappedID3D12CommandQueue::Serialise_Wait(ID3D12Fence *pFence, UINT64 Value) { SERIALISE_ELEMENT(ResourceId, Fence, GetResID(pFence)); SERIALISE_ELEMENT(UINT64, val, Value); if(m_State <= EXECUTING && GetResourceManager()->HasLiveResource(Fence)) { // pFence = GetResourceManager()->GetLiveAs(Fence); m_pDevice->GPUSync(); } return true; } HRESULT STDMETHODCALLTYPE WrappedID3D12CommandQueue::Wait(ID3D12Fence *pFence, UINT64 Value) { if(m_State == WRITING_CAPFRAME) { SCOPED_LOCK(m_Lock); SCOPED_SERIALISE_CONTEXT(WAIT); Serialise_Wait(pFence, Value); m_QueueRecord->AddChunk(scope.Get()); GetResourceManager()->MarkResourceFrameReferenced(GetResID(pFence), eFrameRef_Read); } return m_pReal->Wait(Unwrap(pFence), Value); } HRESULT STDMETHODCALLTYPE WrappedID3D12CommandQueue::GetTimestampFrequency(UINT64 *pFrequency) { return m_pReal->GetTimestampFrequency(pFrequency); } HRESULT STDMETHODCALLTYPE WrappedID3D12CommandQueue::GetClockCalibration(UINT64 *pGpuTimestamp, UINT64 *pCpuTimestamp) { return m_pReal->GetClockCalibration(pGpuTimestamp, pCpuTimestamp); }