fix bug introduced by refactor of ICommandPoolCache

Author: devshgraphicsprogramming

include/nbl/video/utilities/IUtilities.h

@@ -105,6 +105,7 @@ class NBL_API2 IUtilities : public core::IReferenceCounted
                 m_device->mapMemory(memoryRange, access);
 
                 m_defaultDownloadBuffer = core::make_smart_refctd_ptr<StreamingTransientDataBufferMT<>>(asset::SBufferRange<video::IGPUBuffer>{0ull,downstreamSize,std::move(buffer)},maxStreamingBufferAllocationAlignment,minStreamingBufferAllocationSize);
+                m_defaultDownloadBuffer->getBuffer()->setObjectDebugName(("Default Download Buffer of Utilities "+std::to_string(ptrdiff_t(this))).c_str());
             }
             {
                 IGPUBuffer::SCreationParams streamingBufferCreationParams = {};
@@ -132,6 +133,7 @@ class NBL_API2 IUtilities : public core::IReferenceCounted
                 m_device->mapMemory(memoryRange, access);
 
                 m_defaultUploadBuffer = core::make_smart_refctd_ptr<StreamingTransientDataBufferMT<>>(asset::SBufferRange<video::IGPUBuffer>{0ull,upstreamSize,std::move(buffer)},maxStreamingBufferAllocationAlignment,minStreamingBufferAllocationSize);
+                m_defaultUploadBuffer->getBuffer()->setObjectDebugName(("Default Upload Buffer of Utilities "+std::to_string(ptrdiff_t(this))).c_str());
             }
             m_propertyPoolHandler = core::make_smart_refctd_ptr<CPropertyPoolHandler>(core::smart_refctd_ptr(m_device));
             // smaller workgroups fill occupancy gaps better, especially on new Nvidia GPUs, but we don't want too small workgroups on mobile

src/nbl/video/utilities/ICommandPoolCache.cpp

@@ -17,6 +17,7 @@ ICommandPoolCache::ICommandPoolCache(ILogicalDevice* device, const uint32_t queu
 void ICommandPoolCache::releaseSet(const uint32_t poolIx)
 {
 	m_cache[poolIx]->reset();
+	m_cmdPoolAllocator.free_addr(poolIx,1);
 }
 
 void ICommandPoolCache::DeferredCommandPoolResetter::operator()()