| Message ID | 1483013971-26446-1-git-send-email-m.szyprowski@samsung.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 3d63a47a380a873408dad10ca62bd8299b2208f1 |
| Headers | show |
Hi Marek, On Thu, Dec 29, 2016 at 01:19:31PM +0100, Marek Szyprowski wrote: > Requesting a DMA channel might be a time consuming operation, so there is > no need to acquire and release DMA channel for each SPI transfer. > DMA channels can be requested during driver probe and kept all the time, > also because there are no shared nor dynamically allocated channels on > Samsung S3C/S5P/Exynos platforms. > > While moving dma_requrest_slave_channel calls, lets switch to > dma_request_slave_channel_reason(), which returns error codes on failure, > which can be properly propagated to the caller (this for example defers > SPI probe when DMA controller is not yet available). > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> Thanks! Reviewed-by: Andi Shyti <andi.shyti@samsung.com> Tested-by: Andi Shyti <andi.shyti@samsung.com> Andi -- To unsubscribe from this list: send the line "unsubscribe linux-spi" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Dec 29, 2016 at 01:19:31PM +0100, Marek Szyprowski wrote: > Requesting a DMA channel might be a time consuming operation, so there is > no need to acquire and release DMA channel for each SPI transfer. > DMA channels can be requested during driver probe and kept all the time, > also because there are no shared nor dynamically allocated channels on > Samsung S3C/S5P/Exynos platforms. > > While moving dma_requrest_slave_channel calls, lets switch to > dma_request_slave_channel_reason(), which returns error codes on failure, > which can be properly propagated to the caller (this for example defers > SPI probe when DMA controller is not yet available). > > Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> > --- > drivers/spi/spi-s3c64xx.c | 57 ++++++++++++++++++++++++----------------------- > 1 file changed, 29 insertions(+), 28 deletions(-) Reviewed-by: Krzysztof Kozlowski <krzk@kernel.org> Best regards, Krzysztof -- To unsubscribe from this list: send the line "unsubscribe linux-spi" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Dec 29, 2016 at 01:19:31PM +0100, Marek Szyprowski wrote: > Requesting a DMA channel might be a time consuming operation, so there is > no need to acquire and release DMA channel for each SPI transfer. This doesn't apply against current code, please check and resend.
Hi Mark, On 2016-12-31 19:26, Mark Brown wrote: > On Thu, Dec 29, 2016 at 01:19:31PM +0100, Marek Szyprowski wrote: >> Requesting a DMA channel might be a time consuming operation, so there is >> no need to acquire and release DMA channel for each SPI transfer. > This doesn't apply against current code, please check and resend. It fails to apply because of missing commit 6f8dc9d481759b428a8cb6a17b83821451415ba9 ("spi: s3c64xx: Do not use platform_data for DMA parameters") in your spi-next, which is already in v4.10-rc1 and came through dma-engine tree. My patch depends on it and removing this dependency will make a merge conflict later while merging to Linus if I rebase directly onto your spi-next. IMHO it would be better either to merge v4.10-rc1 to your spi-next branch or at least apply the mentioned patch to your branch before applying this one. Best regards
On Mon, Jan 09, 2017 at 08:43:16AM +0100, Marek Szyprowski wrote: > On 2016-12-31 19:26, Mark Brown wrote: > > On Thu, Dec 29, 2016 at 01:19:31PM +0100, Marek Szyprowski wrote: > > > Requesting a DMA channel might be a time consuming operation, so there is > > > no need to acquire and release DMA channel for each SPI transfer. > > This doesn't apply against current code, please check and resend. > It fails to apply because of missing commit > 6f8dc9d481759b428a8cb6a17b83821451415ba9 > ("spi: s3c64xx: Do not use platform_data for DMA parameters") in your > spi-next, When I said "please check and resend" the resend part was important...
diff --git a/drivers/spi/spi-s3c64xx.c b/drivers/spi/spi-s3c64xx.c index 28dfdce4beae..849ee82483e4 100644 --- a/drivers/spi/spi-s3c64xx.c +++ b/drivers/spi/spi-s3c64xx.c @@ -341,43 +341,16 @@ static void s3c64xx_spi_set_cs(struct spi_device *spi, bool enable) static int s3c64xx_spi_prepare_transfer(struct spi_master *spi) { struct s3c64xx_spi_driver_data *sdd = spi_master_get_devdata(spi); - struct device *dev = &sdd->pdev->dev; if (is_polling(sdd)) return 0; - /* Acquire DMA channels */ - sdd->rx_dma.ch = dma_request_slave_channel(dev, "rx"); - if (!sdd->rx_dma.ch) { - dev_err(dev, "Failed to get RX DMA channel\n"); - return -EBUSY; - } spi->dma_rx = sdd->rx_dma.ch; - - sdd->tx_dma.ch = dma_request_slave_channel(dev, "tx"); - if (!sdd->tx_dma.ch) { - dev_err(dev, "Failed to get TX DMA channel\n"); - dma_release_channel(sdd->rx_dma.ch); - return -EBUSY; - } spi->dma_tx = sdd->tx_dma.ch; return 0; } -static int s3c64xx_spi_unprepare_transfer(struct spi_master *spi) -{ - struct s3c64xx_spi_driver_data *sdd = spi_master_get_devdata(spi); - - /* Free DMA channels */ - if (!is_polling(sdd)) { - dma_release_channel(sdd->rx_dma.ch); - dma_release_channel(sdd->tx_dma.ch); - } - - return 0; -} - static bool s3c64xx_spi_can_dma(struct spi_master *master, struct spi_device *spi, struct spi_transfer *xfer) @@ -1094,7 +1067,6 @@ static int s3c64xx_spi_probe(struct platform_device *pdev) master->prepare_transfer_hardware = s3c64xx_spi_prepare_transfer; master->prepare_message = s3c64xx_spi_prepare_message; master->transfer_one = s3c64xx_spi_transfer_one; - master->unprepare_transfer_hardware = s3c64xx_spi_unprepare_transfer; master->num_chipselect = sci->num_cs; master->dma_alignment = 8; master->bits_per_word_mask = SPI_BPW_MASK(32) | SPI_BPW_MASK(16) | @@ -1161,6 +1133,24 @@ static int s3c64xx_spi_probe(struct platform_device *pdev) } } + if (!is_polling(sdd)) { + /* Acquire DMA channels */ + sdd->rx_dma.ch = dma_request_slave_channel_reason(&pdev->dev, + "rx"); + if (IS_ERR(sdd->rx_dma.ch)) { + dev_err(&pdev->dev, "Failed to get RX DMA channel\n"); + ret = PTR_ERR(sdd->rx_dma.ch); + goto err_disable_io_clk; + } + sdd->tx_dma.ch = dma_request_slave_channel_reason(&pdev->dev, + "tx"); + if (IS_ERR(sdd->tx_dma.ch)) { + dev_err(&pdev->dev, "Failed to get TX DMA channel\n"); + ret = PTR_ERR(sdd->tx_dma.ch); + goto err_release_tx_dma; + } + } + pm_runtime_set_autosuspend_delay(&pdev->dev, AUTOSUSPEND_TIMEOUT); pm_runtime_use_autosuspend(&pdev->dev); pm_runtime_set_active(&pdev->dev); @@ -1206,6 +1196,12 @@ static int s3c64xx_spi_probe(struct platform_device *pdev) pm_runtime_disable(&pdev->dev); pm_runtime_set_suspended(&pdev->dev); + if (!is_polling(sdd)) + dma_release_channel(sdd->rx_dma.ch); +err_release_tx_dma: + if (!is_polling(sdd)) + dma_release_channel(sdd->tx_dma.ch); +err_disable_io_clk: clk_disable_unprepare(sdd->ioclk); err_disable_src_clk: clk_disable_unprepare(sdd->src_clk); @@ -1226,6 +1222,11 @@ static int s3c64xx_spi_remove(struct platform_device *pdev) writel(0, sdd->regs + S3C64XX_SPI_INT_EN); + if (!is_polling(sdd)) { + dma_release_channel(sdd->rx_dma.ch); + dma_release_channel(sdd->tx_dma.ch); + } + clk_disable_unprepare(sdd->ioclk); clk_disable_unprepare(sdd->src_clk);
Requesting a DMA channel might be a time consuming operation, so there is no need to acquire and release DMA channel for each SPI transfer. DMA channels can be requested during driver probe and kept all the time, also because there are no shared nor dynamically allocated channels on Samsung S3C/S5P/Exynos platforms. While moving dma_requrest_slave_channel calls, lets switch to dma_request_slave_channel_reason(), which returns error codes on failure, which can be properly propagated to the caller (this for example defers SPI probe when DMA controller is not yet available). Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- drivers/spi/spi-s3c64xx.c | 57 ++++++++++++++++++++++++----------------------- 1 file changed, 29 insertions(+), 28 deletions(-)