diff --git a/driver/ControlListUtil.c b/driver/ControlListUtil.c
index 1ec263f..6482c93 100644
--- a/driver/ControlListUtil.c
+++ b/driver/ControlListUtil.c
@@ -696,7 +696,8 @@ void clInsertGEMRelocations(ControlList* cl,
 void clInsertShaderRecord(ControlList* cls,
 						  ControlList* relocCl,
 						  ControlList* handlesCl,
-						  uint8_t* handlesBuf, uint32_t handlesSize,
+						  uint32_t handlesOffset,
+						  uint32_t handlesSize,
 						  uint32_t fragmentShaderIsSingleThreaded, //0/1
 						  uint32_t pointSizeIncludedInShadedVertexData, //0/1
 						  uint32_t enableClipping, //0/1
@@ -724,14 +725,14 @@ void clInsertShaderRecord(ControlList* cls,
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = 0; cls->nextFreeByteOffset++;
 	*(uint16_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = moveBits(fragmentNumberOfUsedUniforms, 16, 0); cls->nextFreeByteOffset++;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) |= fragmentNumberOfVaryings; cls->nextFreeByteOffset++;
-	clEmitShaderRelocation(relocCl, handlesCl, handlesSize, &fragmentCodeAddress);
+	clEmitShaderRelocation(relocCl, handlesCl, handlesOffset, handlesSize, &fragmentCodeAddress);
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = fragmentCodeAddress.offset; cls->nextFreeByteOffset += 4;
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = fragmentUniformsAddress; cls->nextFreeByteOffset += 4;
 
 	*(uint16_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = moveBits(vertexNumberOfUsedUniforms, 16, 0); cls->nextFreeByteOffset += 2;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = vertexAttributeArraySelectBits; cls->nextFreeByteOffset++;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = vertexTotalAttributesSize; cls->nextFreeByteOffset++;
-	clEmitShaderRelocation(relocCl, handlesCl, handlesSize, &vertexCodeAddress);
+	clEmitShaderRelocation(relocCl, handlesCl, handlesOffset, handlesSize, &vertexCodeAddress);
 	//wtf??? --> shader code will always have an offset of 0 so this is fine
 	uint32_t offset = moveBits(vertexCodeAddress.offset, 32, 0) | moveBits(vertexUniformsAddress, 32, 0);
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = offset; cls->nextFreeByteOffset += 4;
@@ -740,7 +741,7 @@ void clInsertShaderRecord(ControlList* cls,
 	*(uint16_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = moveBits(coordinateNumberOfUsedUniforms, 16, 0); cls->nextFreeByteOffset += 2;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = coordinateAttributeArraySelectBits; cls->nextFreeByteOffset++;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = coordinateTotalAttributesSize; cls->nextFreeByteOffset++;
-	clEmitShaderRelocation(relocCl, handlesCl, handlesSize, &coordinateCodeAddress);
+	clEmitShaderRelocation(relocCl, handlesCl, handlesOffset, handlesSize, &coordinateCodeAddress);
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = coordinateCodeAddress.offset; cls->nextFreeByteOffset += 4;
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = coordinateUniformsAddress; cls->nextFreeByteOffset += 4;
 }
@@ -749,7 +750,7 @@ void clInsertShaderRecord(ControlList* cls,
 void clInsertAttributeRecord(ControlList* cls,
 							 ControlList* relocCl,
 							 ControlList* handlesCl,
-							 uint8_t* handlesBuf, uint32_t handlesSize,
+							 uint32_t handlesOffset, uint32_t handlesSize,
 						  ControlListAddress address,
 						  uint32_t sizeBytes,
 						  uint32_t stride,
@@ -759,7 +760,7 @@ void clInsertAttributeRecord(ControlList* cls,
 	assert(cls);
 	assert(cls->CPA);
 	uint32_t sizeBytesMinusOne = sizeBytes - 1;
-	clEmitShaderRelocation(relocCl, handlesCl, handlesSize, &address);
+	clEmitShaderRelocation(relocCl, handlesCl, handlesOffset, handlesSize, &address);
 	*(uint32_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = address.offset; cls->nextFreeByteOffset += 4;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = sizeBytesMinusOne; cls->nextFreeByteOffset++;
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = stride; cls->nextFreeByteOffset++;
@@ -767,16 +768,16 @@ void clInsertAttributeRecord(ControlList* cls,
 	*(uint8_t*)getCPAptrFromOffset(cls->CPA, cls->nextFreeByteOffset) = coordinateVPMOffset; cls->nextFreeByteOffset++;
 }
 
-uint32_t clGetHandleIndex(ControlList* handlesCl, uint32_t handlesSize, uint32_t handle)
+uint32_t clGetHandleIndex(ControlList* handlesCl, uint32_t handlesOffset, uint32_t handlesSize, uint32_t handle)
 {
 	uint32_t c = 0;
 
 	//if curr marker is closed already we need to work with the stored size
-	uint32_t numHandles = (handlesSize ? handlesSize : (handlesCl->nextFreeByteOffset - handlesCl->offset)) / 4;
+	uint32_t numHandles = (handlesSize ? handlesSize : (handlesCl->nextFreeByteOffset - handlesOffset)) / 4;
 
 	for(; c < numHandles; ++c)
 	{
-		if(((uint32_t*)getCPAptrFromOffset(handlesCl->CPA, handlesCl->offset))[c] == handle)
+		if(((uint32_t*)getCPAptrFromOffset(handlesCl->CPA, handlesOffset))[c] == handle)
 		{
 			//found
 			return c;
@@ -791,7 +792,7 @@ uint32_t clGetHandleIndex(ControlList* handlesCl, uint32_t handlesSize, uint32_t
 }
 
 //input: 2 cls (cl + handles cl)
-inline void clEmitShaderRelocation(ControlList* relocCl, ControlList* handlesCl, uint32_t handlesSize, const ControlListAddress* address)
+inline void clEmitShaderRelocation(ControlList* relocCl, ControlList* handlesCl, uint32_t handlesOffset, uint32_t handlesSize, const ControlListAddress* address)
 {
 	assert(relocCl);
 	assert(relocCl->CPA);
@@ -801,7 +802,7 @@ inline void clEmitShaderRelocation(ControlList* relocCl, ControlList* handlesCl,
 	assert(address->handle);
 
 	//store offset within handles in cl
-	*(uint32_t*)getCPAptrFromOffset(relocCl->CPA, relocCl->nextFreeByteOffset) = clGetHandleIndex(handlesCl, handlesSize, address->handle);
+	*(uint32_t*)getCPAptrFromOffset(relocCl->CPA, relocCl->nextFreeByteOffset) = clGetHandleIndex(handlesCl, handlesOffset, handlesSize, address->handle);
 	relocCl->nextFreeByteOffset += 4;
 }
 
diff --git a/driver/ControlListUtil.h b/driver/ControlListUtil.h
index e9acf06..a9501a5 100644
--- a/driver/ControlListUtil.h
+++ b/driver/ControlListUtil.h
@@ -67,7 +67,7 @@ typedef struct ControlList
 	uint32_t currMarkerOffset;
 } ControlList;
 
-void clEmitShaderRelocation(ControlList* relocCl, ControlList* handlesCl, uint32_t handlesSize, const ControlListAddress* address);
+void clEmitShaderRelocation(ControlList* relocCl, ControlList* handlesCl, uint32_t handlesOffset, uint32_t handlesSize, const ControlListAddress* address);
 void clDummyRelocation(ControlList* relocCl, const ControlListAddress* address);
 
 #define __gen_user_data struct ControlList
@@ -187,7 +187,8 @@ void clInsertGEMRelocations(ControlList* cl,
 void clInsertShaderRecord(ControlList* cls,
 						  ControlList* relocCl,
 						  ControlList* handlesCl,
-						  uint8_t* handlesBuf, uint32_t handlesSize,
+						  uint32_t handlesOffset,
+						  uint32_t handlesSize,
 						  uint32_t fragmentShaderIsSingleThreaded, //0/1
 						  uint32_t pointSizeIncludedInShadedVertexData, //0/1
 						  uint32_t enableClipping, //0/1
@@ -208,13 +209,13 @@ void clInsertShaderRecord(ControlList* cls,
 void clInsertAttributeRecord(ControlList* cls,
 							 ControlList* relocCl,
 							 ControlList* handlesCl,
-							 uint8_t* handlesBuf, uint32_t handlesSize,
+							 uint32_t handlesOffset, uint32_t handlesSize,
 						  ControlListAddress address,
 						  uint32_t sizeBytes,
 						  uint32_t stride,
 						  uint32_t vertexVPMOffset,
 						  uint32_t coordinateVPMOffset);
-uint32_t clGetHandleIndex(ControlList* handlesCl, uint32_t handlesSize, uint32_t handle);
+uint32_t clGetHandleIndex(ControlList* handlesCl, uint32_t handlesOffset, uint32_t handlesSize, uint32_t handle);
 
 #if defined (__cplusplus)
 }
diff --git a/driver/command.c b/driver/command.c
index 4d59af2..c7b0b50 100644
--- a/driver/command.c
+++ b/driver/command.c
@@ -327,12 +327,12 @@ VKAPI_ATTR VkResult VKAPI_CALL rpi_vkQueueSubmit(
 			uint32_t readMSAAdepthStencilImage = marker->readMSAAdepthStencilImage;
 
 			//This should not result in an insertion!
-			uint32_t writeImageIdx = writeImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, writeImage->boundMem->bo) : 0;
-			uint32_t readImageIdx = readImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, readImage->boundMem->bo) : 0;
-			uint32_t writeDepthStencilImageIdx = writeDepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, writeDepthStencilImage->boundMem->bo) : 0;
-			uint32_t readDepthStencilImageIdx = readDepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, readDepthStencilImage->boundMem->bo) : 0;
-			uint32_t writeMSAAimageIdx = writeMSAAimage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, writeMSAAimage->boundMem->bo) : 0;
-			uint32_t writeMSAAdepthStencilImageIdx = writeMSAAdepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, writeMSAAdepthStencilImage->boundMem->bo) : 0;
+			uint32_t writeImageIdx = writeImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesBufOffset, marker->handlesSize, writeImage->boundMem->bo) : 0;
+			uint32_t readImageIdx = readImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesBufOffset, marker->handlesSize, readImage->boundMem->bo) : 0;
+			uint32_t writeDepthStencilImageIdx = writeDepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesBufOffset, marker->handlesSize, writeDepthStencilImage->boundMem->bo) : 0;
+			uint32_t readDepthStencilImageIdx = readDepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesBufOffset, marker->handlesSize, readDepthStencilImage->boundMem->bo) : 0;
+			uint32_t writeMSAAimageIdx = writeMSAAimage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesSize, marker->handlesBufOffset, writeMSAAimage->boundMem->bo) : 0;
+			uint32_t writeMSAAdepthStencilImageIdx = writeMSAAdepthStencilImage ? clGetHandleIndex(&cmdbuf->handlesCl, marker->handlesBufOffset, marker->handlesSize, writeMSAAdepthStencilImage->boundMem->bo) : 0;
 
 //			fprintf(stderr, "writeImage: %u\n", writeImage);
 //			fprintf(stderr, "readImage: %u\n", readImage);
diff --git a/driver/draw.c b/driver/draw.c
index 75ceac9..afcabcd 100644
--- a/driver/draw.c
+++ b/driver/draw.c
@@ -225,7 +225,7 @@ static uint32_t drawCommon(VkCommandBuffer commandBuffer, int32_t vertexOffset)
 	clInsertShaderRecord(&commandBuffer->shaderRecCl,
 						 &relocCl,
 						 &commandBuffer->handlesCl,
-						 getCPAptrFromOffset(cb->handlesCl.CPA, currMarker->handlesBufOffset),
+						 currMarker->handlesBufOffset,
 						 currMarker->handlesSize,
 						 !fragModule->hasThreadSwitch,
 						 0, //TODO point size included in shaded vertex data?
@@ -311,7 +311,7 @@ static uint32_t drawCommon(VkCommandBuffer commandBuffer, int32_t vertexOffset)
 			clInsertAttributeRecord(&commandBuffer->shaderRecCl,
 									&relocCl,
 									&commandBuffer->handlesCl,
-									getCPAptrFromOffset(cb->handlesCl.CPA, currMarker->handlesBufOffset),
+									currMarker->handlesBufOffset,
 									currMarker->handlesSize,
 									vertexBuffer, //reloc address
 									formatByteSize,
@@ -350,7 +350,7 @@ static uint32_t drawCommon(VkCommandBuffer commandBuffer, int32_t vertexOffset)
 
 				//emit reloc for texture BO
 				clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, di->imageView->image->boundMem->bo);
+				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, di->imageView->image->boundMem->bo);
 
 				//emit tex bo reloc index
 				clFit(commandBuffer, &commandBuffer->uniformsCl, 4);
@@ -369,7 +369,7 @@ static uint32_t drawCommon(VkCommandBuffer commandBuffer, int32_t vertexOffset)
 
 				//emit reloc for BO
 				clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, db->buffer->boundMem->bo);
+				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, db->buffer->boundMem->bo);
 
 				//emit bo reloc index
 				clFit(commandBuffer, &commandBuffer->uniformsCl, 4);
@@ -386,7 +386,7 @@ static uint32_t drawCommon(VkCommandBuffer commandBuffer, int32_t vertexOffset)
 
 				//emit reloc for BO
 				clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, dtb->bufferView->buffer->boundMem->bo);
+				uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, dtb->bufferView->buffer->boundMem->bo);
 
 				//emit bo reloc index
 				clFit(commandBuffer, &commandBuffer->uniformsCl, 4);
@@ -592,7 +592,7 @@ VKAPI_ATTR void VKAPI_CALL rpi_vkCmdDrawIndexed(
 	CLMarker* currMarker = getCPAptrFromOffset(cb->binCl.CPA, cb->binCl.currMarkerOffset);
 
 	clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-	uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, cb->indexBuffer->boundMem->bo);
+	uint32_t idx = clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, cb->indexBuffer->boundMem->bo);
 
 	clInsertGEMRelocations(&commandBuffer->binCl, idx, 0);
 
diff --git a/driver/renderpass.c b/driver/renderpass.c
index ef01159..133ab24 100644
--- a/driver/renderpass.c
+++ b/driver/renderpass.c
@@ -194,37 +194,37 @@ void rpi_vkCmdBeginRenderPass(VkCommandBuffer commandBuffer, const VkRenderPassB
 	if(writeImage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, writeImage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, writeImage->boundMem->bo);
 	}
 
 	if(readImage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, readImage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, readImage->boundMem->bo);
 	}
 
 	if(writeDepthStencilImage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, writeDepthStencilImage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, writeDepthStencilImage->boundMem->bo);
 	}
 
 	if(readDepthStencilImage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, readDepthStencilImage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, readDepthStencilImage->boundMem->bo);
 	}
 
 	if(writeMSAAimage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, writeMSAAimage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, writeMSAAimage->boundMem->bo);
 	}
 
 	if(writeMSAAdepthStencilImage)
 	{
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, writeMSAAdepthStencilImage->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, writeMSAAdepthStencilImage->boundMem->bo);
 	}
 
 	uint32_t bpp = 0;
diff --git a/driver/stateChange.c b/driver/stateChange.c
index 203acad..72aaeb2 100644
--- a/driver/stateChange.c
+++ b/driver/stateChange.c
@@ -518,7 +518,7 @@ VKAPI_ATTR void VKAPI_CALL rpi_vkCmdClearColorImage(
 
 		//insert reloc for render target
 		clFit(commandBuffer, &commandBuffer->handlesCl, 4);
-		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesSize, i->boundMem->bo);
+		clGetHandleIndex(&commandBuffer->handlesCl, currMarker->handlesBufOffset, currMarker->handlesSize, i->boundMem->bo);
 
 		clFit(commandBuffer, &commandBuffer->binCl, V3D21_TILE_BINNING_MODE_CONFIGURATION_length);
 		clInsertTileBinningModeConfiguration(&commandBuffer->binCl,