kernel/linux.git/drivers/spi, branch v4.19.16

spi: bcm2835: Unbreak the build of esoteric configs

2019-01-09T16:38:49+00:00

commit 29bdedfd9cf40e59456110ca417a8cb672ac9b92 upstream. Commit e82b0b382845 ("spi: bcm2835: Fix race on DMA termination") broke the build with COMPILE_TEST=y on arches whose cmpxchg() requires 32-bit operands (xtensa, older arm ISAs). Fix by changing the dma_pending flag's type from bool to unsigned int. Fixes: e82b0b382845 ("spi: bcm2835: Fix race on DMA termination") Signed-off-by: Lukas Wunner Signed-off-by: Mark Brown Cc: Frank Pavlic Cc: Martin Sperl Cc: Noralf Trønnes Cc: Sudip Mukherjee Signed-off-by: Greg Kroah-Hartman

spi: bcm2835: Avoid finishing transfer prematurely in IRQ mode

2019-01-09T16:38:44+00:00

commit 56c1723426d3cfd4723bfbfce531d7b38bae6266 upstream. The IRQ handler bcm2835_spi_interrupt() first reads as much as possible from the RX FIFO, then writes as much as possible to the TX FIFO. Afterwards it decides whether the transfer is finished by checking if the TX FIFO is empty. If very few bytes were written to the TX FIFO, they may already have been transmitted by the time the FIFO's emptiness is checked. As a result, the transfer will be declared finished and the chip will be reset without reading the corresponding received bytes from the RX FIFO. The odds of this happening increase with a high clock frequency (such that the TX FIFO drains quickly) and either passing "threadirqs" on the command line or enabling CONFIG_PREEMPT_RT_BASE (such that the IRQ handler may be preempted between filling the TX FIFO and checking its emptiness). Fix by instead checking whether rx_len has reached zero, which means that the transfer has been received in full. This is also more efficient as it avoids one bus read access per interrupt. Note that bcm2835_spi_transfer_one_poll() likewise uses rx_len to determine whether the transfer has finished. Signed-off-by: Lukas Wunner Fixes: e34ff011c70e ("spi: bcm2835: move to the transfer_one driver model") Cc: stable@vger.kernel.org # v4.1+ Cc: Mathias Duckeck Cc: Frank Pavlic Cc: Martin Sperl Cc: Noralf Trønnes Signed-off-by: Mark Brown Signed-off-by: Greg Kroah-Hartman

spi: bcm2835: Fix book-keeping of DMA termination

2019-01-09T16:38:44+00:00

commit dbc944115eed48af110646992893dc43321368d8 upstream. If submission of a DMA TX transfer succeeds but submission of the corresponding RX transfer does not, the BCM2835 SPI driver terminates the TX transfer but neglects to reset the dma_pending flag to false. Thus, if the next transfer uses interrupt mode (because it is shorter than BCM2835_SPI_DMA_MIN_LENGTH) and runs into a timeout, dmaengine_terminate_all() will be called both for TX (once more) and for RX (which was never started in the first place). Fix it. Signed-off-by: Lukas Wunner Fixes: 3ecd37edaa2a ("spi: bcm2835: enable dma modes for transfers meeting certain conditions") Cc: stable@vger.kernel.org # v4.2+ Cc: Mathias Duckeck Cc: Frank Pavlic Cc: Martin Sperl Cc: Noralf Trønnes Signed-off-by: Mark Brown Signed-off-by: Greg Kroah-Hartman

spi: bcm2835: Fix race on DMA termination

2019-01-09T16:38:44+00:00

commit e82b0b3828451c1cd331d9f304c6078fcd43b62e upstream. If a DMA transfer finishes orderly right when spi_transfer_one_message() determines that it has timed out, the callbacks bcm2835_spi_dma_done() and bcm2835_spi_handle_err() race to call dmaengine_terminate_all(), potentially leading to double termination. Prevent by atomically changing the dma_pending flag before calling dmaengine_terminate_all(). Signed-off-by: Lukas Wunner Fixes: 3ecd37edaa2a ("spi: bcm2835: enable dma modes for transfers meeting certain conditions") Cc: stable@vger.kernel.org # v4.2+ Cc: Mathias Duckeck Cc: Frank Pavlic Cc: Martin Sperl Cc: Noralf Trønnes Signed-off-by: Mark Brown Signed-off-by: Greg Kroah-Hartman

spi: omap2-mcspi: Add missing suspend and resume calls

2018-12-17T08:24:32+00:00

[ Upstream commit 91b9deefedf4c35a01027ce38bed7299605026a3 ] I've been wondering still about omap2-mcspi related suspend and resume flakeyness and looks like we're missing calls to spi_master_suspend() and spi_master_resume(). Adding those and using pm_runtime_force_suspend() and pm_runtime_force_resume() makes things work for suspend and resume and allows us to stop using noirq suspend and resume. And while at it, let's use SET_SYSTEM_SLEEP_PM_OPS to simplify things further. Signed-off-by: Tony Lindgren Signed-off-by: Mark Brown Signed-off-by: Sasha Levin

spi: gpio: No MISO does not imply no RX

2018-11-13T19:08:28+00:00

[ Upstream commit abf5feef3ff0cefade0c76be53b59e55fdd46093 ] There is a logical problem in spi-gpio with host just assigning a MOSI line and no MISO: this is interpreted as the host cannot do RX and the host is flagged with SPI_MASTER_NO_RX. This is wrong: since GPIO lines can switch direction, in 3WIRE operation the host will simply reverse the direction of the GPIO line and start reading from it, there is even code for doing this in the driver, but it went unnoticed because it was tested by using a master with 4 wires but a device using just 3 wires. Remove the offending flag. Cc: Andrzej Hajda Cc: Lorenzo Bianconi Signed-off-by: Linus Walleij Signed-off-by: Mark Brown Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

spi: spi-ep93xx: Use dma_data_direction for ep93xx_spi_dma_{finish,prepare}

2018-11-13T19:08:22+00:00

[ Upstream commit a1108c7b2efb892350ba6a0e932dfd45622f4e2b ] Clang warns when one enumerated type is implicitly converted to another. drivers/spi/spi-ep93xx.c:342:62: warning: implicit conversion from enumeration type 'enum dma_transfer_direction' to different enumeration type 'enum dma_data_direction' [-Wenum-conversion] nents = dma_map_sg(chan->device->dev, sgt->sgl, sgt->nents, dir); ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~ ./include/linux/dma-mapping.h:428:58: note: expanded from macro 'dma_map_sg' #define dma_map_sg(d, s, n, r) dma_map_sg_attrs(d, s, n, r, 0) ~~~~~~~~~~~~~~~~ ^ drivers/spi/spi-ep93xx.c:348:57: warning: implicit conversion from enumeration type 'enum dma_transfer_direction' to different enumeration type 'enum dma_data_direction' [-Wenum-conversion] dma_unmap_sg(chan->device->dev, sgt->sgl, sgt->nents, dir); ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~ ./include/linux/dma-mapping.h:429:62: note: expanded from macro 'dma_unmap_sg' #define dma_unmap_sg(d, s, n, r) dma_unmap_sg_attrs(d, s, n, r, 0) ~~~~~~~~~~~~~~~~~~ ^ drivers/spi/spi-ep93xx.c:377:56: warning: implicit conversion from enumeration type 'enum dma_transfer_direction' to different enumeration type 'enum dma_data_direction' [-Wenum-conversion] dma_unmap_sg(chan->device->dev, sgt->sgl, sgt->nents, dir); ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~ ./include/linux/dma-mapping.h:429:62: note: expanded from macro 'dma_unmap_sg' #define dma_unmap_sg(d, s, n, r) dma_unmap_sg_attrs(d, s, n, r, 0) ~~~~~~~~~~~~~~~~~~ ^ 3 warnings generated. dma_{,un}map_sg expect an enum of type dma_data_direction but this driver uses dma_transfer_direction for everything. Convert the driver to use dma_data_direction for these two functions. There are two places that strictly require an enum of type dma_transfer_direction: the direction member in struct dma_slave_config and the direction parameter in dmaengine_prep_slave_sg. To avoid using an explicit cast, add a simple function, ep93xx_dma_data_to_trans_dir, to safely map between the two types because they are not 1 to 1 in meaning. Signed-off-by: Nathan Chancellor Reviewed-by: Nick Desaulniers Reviewed-by: Mika Westerberg Signed-off-by: Mark Brown Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

spi: bcm-qspi: fix calculation of address length

2018-11-13T19:08:15+00:00

commit 0976eda7915507fe94e07870c19d717c9994b57a upstream. During implementation of the new API bcm_qspi_bspi_set_flex_mode() has been modified breaking calculation of address length. An unnecessary multiplication was added breaking flash reads. Fixes: 5f195ee7d830 ("spi: bcm-qspi: Implement the spi_mem interface") Signed-off-by: Rafał Miłecki Reviewed-by: Boris Brezillon Signed-off-by: Mark Brown Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman

spi: bcm-qspi: switch back to reading flash using smaller chunks

2018-11-13T19:08:15+00:00

commit 940ec770c295682993d1cccce3081fd7c74fece8 upstream. Fixing/optimizing bcm_qspi_bspi_read() performance introduced two changes: 1) It added a loop to read all requested data using multiple BSPI ops. 2) It bumped max size of a single BSPI block request from 256 to 512 B. The later change resulted in occasional BSPI timeouts causing a regression. For some unknown reason hardware doesn't always handle reads as expected when using 512 B chunks. In such cases it may happen that BSPI returns amount of requested bytes without the last 1-3 ones. It provides the remaining bytes later but doesn't raise an interrupt until another LR start. Switching back to 256 B reads fixes that problem and regression. Fixes: 345309fa7c0c ("spi: bcm-qspi: Fix bcm_qspi_bspi_read() performance") Signed-off-by: Rafał Miłecki Signed-off-by: Mark Brown Cc: stable@vger.kernel.org Signed-off-by: Greg Kroah-Hartman

spi: spi-mem: Adjust op len based on message/transfer size limitations

2018-11-13T19:08:15+00:00

commit e757996cafbeb6b71234a17130674bcd8f44c59e upstream. We need that to adjust the len of the 2nd transfer (called data in spi-mem) if it's too long to fit in a SPI message or SPI transfer. Fixes: c36ff266dc82 ("spi: Extend the core to ease integration of SPI memory controllers") Cc: Signed-off-by: Chuanhua Han Reviewed-by: Boris Brezillon Signed-off-by: Mark Brown Signed-off-by: Greg Kroah-Hartman