How to increase the speed for quad SPI receive with HAL_OSPI_Receive?

[fb]

I'm using this code to read from a flash chip with quad SPI:

	// send read command and read page data
	OSPI_RegularCmdTypeDef sCommand;
	memset(&sCommand, 0, sizeof(sCommand));
	sCommand.OperationType = HAL_OSPI_OPTYPE_COMMON_CFG;
	sCommand.FlashId = HAL_OSPI_FLASH_ID_1;
	sCommand.InstructionMode = HAL_OSPI_INSTRUCTION_1_LINE;
	sCommand.InstructionDtrMode = HAL_OSPI_INSTRUCTION_DTR_DISABLE;
	sCommand.InstructionSize = HAL_OSPI_INSTRUCTION_8_BITS;
	sCommand.Instruction = CMD_FAST_READ_QUAD_OUTPUT;
	sCommand.AddressMode = HAL_OSPI_ADDRESS_1_LINE;
	sCommand.AddressDtrMode = HAL_OSPI_ADDRESS_DTR_DISABLE;
	sCommand.AddressSize = HAL_OSPI_ADDRESS_16_BITS;
	sCommand.Address = flashSource >> 11;
	sCommand.AlternateBytesMode = HAL_OSPI_ALTERNATE_BYTES_NONE;
	sCommand.DataMode = HAL_OSPI_DATA_4_LINES;
	sCommand.DataDtrMode = HAL_OSPI_DATA_DTR_DISABLE;
	sCommand.DummyCycles = 8;
	sCommand.NbData = len;
	sCommand.DQSMode = HAL_OSPI_DQS_DISABLE;
	sCommand.SIOOMode = HAL_OSPI_SIOO_INST_EVERY_CMD;
	if (HAL_OSPI_Command(&hospi1, &sCommand, HAL_OSPI_TIMEOUT_DEFAULT_VALUE)
			!= HAL_OK) {
		error("HAL_OSPI_Command error");
	}
	if (HAL_OSPI_Receive(&hospi1, ramDestination,
	HAL_OSPI_TIMEOUT_DEFAULT_VALUE) != HAL_OK) {
		error("HAL_OSPI_Receive error");
	}

The clock is 25 MHz and looks like it works, the received data is what I wrote to it, but there are gaps when reading:

I already tried to enable D-cache and I-cache, and compiled it in release mode, but no changes. The destination RAM is RAM_D1. I also tried different "Fifo Threshold" settings in CubeMX, but no changes. How can I speed it up?

I saw in the HAL code that OSPI_WaitFlagStateUntilTimeout is called for each byte, and the function doesn't look nice. Maybe someone has already implemented faster code, or would there be always a gap and I need something different, like DMA transfer?

[fb]

I tried HAL_OSPI_Transmit_DMA, but the HAL_OSPI_TxCpltCallback is never called. How do I need to configure the peripherals with CubeMX for it? Do I need to call anything else for it?

Is there a website with good documentation for the HAL functions? In the comments in the implementation file it says I should call HAL_NVIC_SetPriority and HAL_NVIC_EnableIRQ. I've enabled the "OCTOSPI1 global interrupt" in the "NVIC Settings" in CubeMX. Looks like it is not sufficient, still no callback called.

[fb · Accepted solution]

I found the problem. I forgot to click the "MDMA global interrupt" checkbox in CubeMX.