On Fri, 30 May 2025 at 17:05, Jyothi Kumar Seerapu
<quic_jseer...@quicinc.com> wrote:
>
>
>
> On 5/9/2025 11:48 AM, Jyothi Kumar Seerapu wrote:
> >
> >
> > On 5/6/2025 5:02 PM, Dmitry Baryshkov wrote:
> >> On Tue, May 06, 2025 at 04:48:43PM +0530, Jyothi Kumar Seerapu wrote:
> >>> GSI hardware generates an interrupt for each transfer completion.
> >>> For multiple messages within a single transfer, this results in
> >>> N interrupts for N messages, leading to significant software
> >>> interrupt latency.
> >>>
> >>> To mitigate this latency, utilize Block Event Interrupt (BEI) mechanism.
> >>> Enabling BEI instructs the GSI hardware to prevent interrupt generation
> >>> and BEI is disabled when an interrupt is necessary.
> >>>
> >>> When using BEI, consider splitting a single multi-message transfer into
> >>> chunks of 8 messages internally and so interrupts are not expected for
> >>> the first 7 message completions, only the last message triggers
> >>> an interrupt, indicating the completion of 8 messages.
> >>>
> >>> This BEI mechanism enhances overall transfer efficiency.
> >>>
> >>> Signed-off-by: Jyothi Kumar Seerapu <quic_jseer...@quicinc.com>
> >>> ---
> >>> v5 ->v6:
> >>> - For updating the block event interrupt bit, instead of relying on
> >>> bei_flag, decision check is moved with DMA_PREP_INTERRUPT flag.
> >>> v4 -> v5:
> >>> - BEI flag naming changed from flags to bei_flag.
> >>> - QCOM_GPI_BLOCK_EVENT_IRQ macro is removed from qcom-gpi-dma.h
> >>> file, and Block event interrupt support is checked with bei_flag.
> >>>
> >>> v3 -> v4:
> >>> - API's added for Block event interrupt with multi descriptor
> >>> support for
> >>> I2C is moved from qcom-gpi-dma.h file to I2C geni qcom driver file.
> >>> - gpi_multi_xfer_timeout_handler function is moved from GPI driver to
> >>> I2C driver.
> >>>
> >>> v2-> v3:
> >>> - Renamed gpi_multi_desc_process to gpi_multi_xfer_timeout_handler
> >>> - MIN_NUM_OF_MSGS_MULTI_DESC changed from 4 to 2
> >>> - Added documentation for newly added changes in "qcom-gpi-dma.h"
> >>> file
> >>> - Updated commit description.
> >>>
> >>> v1 -> v2:
> >>> - Changed dma_addr type from array of pointers to array.
> >>> - To support BEI functionality with the TRE size of 64 defined in
> >>> GPI driver,
> >>> updated QCOM_GPI_MAX_NUM_MSGS to 16 and NUM_MSGS_PER_IRQ to 4.
> >>>
> >>> drivers/dma/qcom/gpi.c | 3 +++
> >>> include/linux/dma/qcom-gpi-dma.h | 2 ++
> >>> 2 files changed, 5 insertions(+)
> >>>
> >>> diff --git a/drivers/dma/qcom/gpi.c b/drivers/dma/qcom/gpi.c
> >>> index b1f0001cc99c..7e511f54166a 100644
> >>> --- a/drivers/dma/qcom/gpi.c
> >>> +++ b/drivers/dma/qcom/gpi.c
> >>> @@ -1695,6 +1695,9 @@ static int gpi_create_i2c_tre(struct gchan
> >>> *chan, struct gpi_desc *desc,
> >>> tre->dword[3] = u32_encode_bits(TRE_TYPE_DMA, TRE_FLAGS_TYPE);
> >>> tre->dword[3] |= u32_encode_bits(1, TRE_FLAGS_IEOT);
> >>> +
> >>> + if (!(i2c->dma_flags & DMA_PREP_INTERRUPT))
> >>> + tre->dword[3] |= u32_encode_bits(1, TRE_FLAGS_BEI);
> >>> }
> >>> for (i = 0; i < tre_idx; i++)
> >>> diff --git a/include/linux/dma/qcom-gpi-dma.h b/include/linux/dma/
> >>> qcom-gpi-dma.h
> >>> index 6680dd1a43c6..ebac0d3edff2 100644
> >>> --- a/include/linux/dma/qcom-gpi-dma.h
> >>> +++ b/include/linux/dma/qcom-gpi-dma.h
> >>> @@ -65,6 +65,7 @@ enum i2c_op {
> >>> * @rx_len: receive length for buffer
> >>> * @op: i2c cmd
> >>> * @muli-msg: is part of multi i2c r-w msgs
> >>> + * @dma_flags: Flags indicating DMA capabilities
> >>> */
> >>> struct gpi_i2c_config {
> >>> u8 set_config;
> >>> @@ -78,6 +79,7 @@ struct gpi_i2c_config {
> >>> u32 rx_len;
> >>> enum i2c_op op;
> >>> bool multi_msg;
> >>> + unsigned int dma_flags;
> >>
> >> Why do you need extra field instead of using
> >> dma_async_tx_descriptor.flags?
> >
> > In the original I2C QCOM GENI driver, using the local variable (unsigned
> > in flags) and updating the "DMA_PREP_INTERRUPT" flag.
> >
> > Sure, i will review if "dma_async_tx_descriptor.flags" can be retrieved
> > in GPI driver for DMA_PREP_INTERRUPT flag status.
>
> Hi Dmitry,
>
> In the I2C Geni driver, the dma flags are primarily used in the
> dmaengine_prep_slave_single() function, which expects the argument type
> to be unsigned int. Therefore, the flags should be defined either as
> enum dma_ctrl_flags, or unsigned int.
>
> In the GPI driver, specifically within the gpi_prep_slave_sg() function,
> the flags are correctly received from the I2C driver. However, these
> flags are not currently passed to the gpi_create_i2c_tre() function.
>
> If we pass the existing flags variable to the gpi_create_i2c_tre()
> function, we can retrieve the DMA flags information without introducing
> any additional or external variables.
>
> Please confirm if this approach—reusing the existing flags argument in
> the GPI driver—is acceptable and good to proceed with.
Could you please check how other drivers use the DMA_PREP_INTERRUPT
flag? That will answer your question.
>
> >>
> >>> };
> >>> #endif /* QCOM_GPI_DMA_H */
> >>> --
> >>> 2.17.1
> >>>
> >>
> >
> >
>
--
With best wishes
Dmitry
