kernel/drivers/dma/qcom/bam_dma.c, branch linux-rolling-stable

dmaengine: qcom: bam_dma: Fix DT error handling for num-channels/ees

2025-09-19T14:37:39Z

commit 5068b5254812433e841a40886e695633148d362d upstream. When we don't have a clock specified in the device tree, we have no way to ensure the BAM is on. This is often the case for remotely-controlled or remotely-powered BAM instances. In this case, we need to read num-channels from the DT to have all the necessary information to complete probing. However, at the moment invalid device trees without clock and without num-channels still continue probing, because the error handling is missing return statements. The driver will then later try to read the number of channels from the registers. This is unsafe, because it relies on boot firmware and lucky timing to succeed. Unfortunately, the lack of proper error handling here has been abused for several Qualcomm SoCs upstream, causing early boot crashes in several situations [1, 2]. Avoid these early crashes by erroring out when any of the required DT properties are missing. Note that this will break some of the existing DTs upstream (mainly BAM instances related to the crypto engine). However, clearly these DTs have never been tested properly, since the error in the kernel log was just ignored. It's safer to disable the crypto engine for these broken DTBs. [1]: https://lore.kernel.org/r/CY01EKQVWE36.B9X5TDXAREPF@fairphone.com/ [2]: https://lore.kernel.org/r/20230626145959.646747-1-krzysztof.kozlowski@linaro.org/ Cc: stable@vger.kernel.org Fixes: 48d163b1aa6e ("dmaengine: qcom: bam_dma: get num-channels and num-ees from dt") Signed-off-by: Stephan Gerhold Reviewed-by: Konrad Dybcio Link: https://lore.kernel.org/r/20250212-bam-dma-fixes-v1-8-f560889e65d8@linaro.org Signed-off-by: Vinod Koul Signed-off-by: Greg Kroah-Hartman

dmaengine: Revert "dmaengine: qcom: bam_dma: Avoid writing unavailable register"

2025-02-27T07:59:15Z

This commit causes a hard crash on sdm845 and likely other platforms. Revert it until a proper fix is found. This reverts commit 57a7138d0627: ("dmaengine: qcom: bam_dma: Avoid writing unavailable register") Signed-off-by: Caleb Connolly Fixes: 57a7138d0627 ("dmaengine: qcom: bam_dma: Avoid writing unavailable register") Tested-by: Neil Armstrong # on sdm845-DB845c Tested-by: David Heidelberg Link: https://lore.kernel.org/r/20250208223112.142567-1-caleb.connolly@linaro.org Signed-off-by: Vinod Koul

dmaengine: qcom: bam_dma: Avoid writing unavailable register

2024-12-24T09:36:40Z

Avoid writing unavailable register in BAM-Lite mode. BAM_DESC_CNT_TRSHLD register is unavailable in BAM-Lite mode. Its only available in BAM-NDP mode. So only write this register for clients who is using BAM-NDP. Signed-off-by: Md Sadre Alam Link: https://lore.kernel.org/r/20241220094203.3510335-1-quic_mdalam@quicinc.com Signed-off-by: Vinod Koul

dmaengine: Switch back to struct platform_driver::remove()

2024-10-14T18:20:34Z

After commit 0edb555a65d1 ("platform: Make platform_driver::remove() return void") .remove() is (again) the right callback to implement for platform drivers. Convert all platform drivers below drivers/dma after the previous conversion commits apart from the wireless drivers to use .remove(), with the eventual goal to drop struct platform_driver::remove_new(). As .remove() and .remove_new() have the same prototypes, conversion is done by just changing the structure member name in the driver initializer. Signed-off-by: Uwe Kleine-König Link: https://lore.kernel.org/r/20241004062227.187726-2-u.kleine-koenig@baylibre.com Signed-off-by: Vinod Koul

Merge tag 'dmaengine-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine

2024-09-23T21:08:08Z

Pull dmaengine updates from Vinod Koul: "Unusually, more new driver and device support than updates. Couple of new device support, AMD, Rcar, Intel and New drivers in Freescale, Loonsoon, AMD and LPC32XX with DT conversion and mode updates etc. New support: - Support for AMD Versal Gen 2 DMA IP - Rcar RZ/G3S SoC dma controller - Support for Intel Diamond Rapids and Granite Rapids-D dma controllers - Support for Freescale ls1021a-qdma controller - New driver for Loongson-1 APB DMA - New driver for AMD QDMA - Pl08x in LPC32XX router dma driver Updates: - Support for dpdma cyclic dma mode - XML conversion for marvell xor dma bindings - Dma clocks documentation for imx dma" * tag 'dmaengine-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (24 commits) dmaengine: loongson1-apb-dma: Fix the build warning caused by the size of pdev_irqname dmaengine: Fix spelling mistakes dmaengine: Add dma router for pl08x in LPC32XX SoC dmaengine: fsl-edma: add edma src ID check at request channel dmaengine: fsl-edma: change to guard(mutex) within fsl_edma3_xlate() dmaengine: avoid non-constant format string dmaengine: imx-dma: Remove i.MX21 support dt-bindings: dma: fsl,imx-dma: Document the DMA clocks dmaengine: Loongson1: Add Loongson-1 APB DMA driver dt-bindings: dma: Add Loongson-1 APB DMA dmaengine: zynqmp_dma: Add support for AMD Versal Gen 2 DMA IP dt-bindings: dmaengine: zynqmp_dma: Add a new compatible string dmaengine: idxd: Add new DSA and IAA device IDs for Diamond Rapids platform dmaengine: idxd: Add a new DSA device ID for Granite Rapids-D platform dmaengine: ti: k3-udma: Remove unused declarations dmaengine: amd: qdma: Add AMD QDMA driver dmaengine: xilinx: dpdma: Add support for cyclic dma mode dma: ipu: Remove include/linux/dma/ipu-dma.h dt-bindings: dma: fsl-mxs-dma: Add compatible string "fsl,imx8qxp-dma-apbh" dt-bindings: fsl-qdma: allow compatible string fallback to fsl,ls1021a-qdma ...

dmaengine: Fix spelling mistakes

2024-09-02T08:26:01Z

Correct spelling mistakes in the DMA engine to improve readability and clarity without altering functionality. Signed-off-by: Amit Vadhavana Reviewed-by: Kees Cook Reviewed-by: Dave Jiang Reviewed-by: Fenghua Yu Link: https://lore.kernel.org/r/20240831172949.13189-1-av2082000@gmail.com Signed-off-by: Vinod Koul

dma-mapping: don't return errors from dma_set_max_seg_size

2024-08-29T04:22:49Z

A NULL dev->dma_parms indicates either a bus that is not DMA capable or grave bug in the implementation of the bus code. There isn't much the driver can do in terms of error handling for either case, so just warn and continue as DMA operations will fail anyway. Signed-off-by: Christoph Hellwig Reviewed-by: Robin Murphy Reviewed-by: Martin K. Petersen Acked-by: Ulf Hansson # For MMC

dmaengine: qcom: bam_dma: Annotate struct bam_async_desc with __counted_by

2023-09-28T11:12:14Z

Prepare for the coming implementation by GCC and Clang of the __counted_by attribute. Flexible array members annotated with __counted_by can have their accesses bounds-checked at run-time checking via CONFIG_UBSAN_BOUNDS (for array indexing) and CONFIG_FORTIFY_SOURCE (for strcpy/memcpy-family functions). As found with Coccinelle[1], add __counted_by for struct bam_async_desc. [1] https://github.com/kees/kernel-tools/blob/trunk/coccinelle/examples/counted_by.cocci Cc: Vinod Koul Cc: Andy Gross Cc: Bjorn Andersson Cc: Konrad Dybcio Cc: linux-arm-msm@vger.kernel.org Cc: dmaengine@vger.kernel.org Signed-off-by: Kees Cook Reviewed-by: "Gustavo A. R. Silva" Link: https://lore.kernel.org/r/20230817235859.49846-7-keescook@chromium.org Signed-off-by: Vinod Koul

dmaengine: qcom: bam_dma: Convert to platform remove callback returning void

2023-09-28T07:40:53Z

The .remove() callback for a platform driver returns an int which makes many driver authors wrongly assume it's possible to do error handling by returning an error code. However the value returned is ignored (apart from emitting a warning) and this typically results in resource leaks. To improve here there is a quest to make the remove callback return void. In the first step of this quest all drivers are converted to .remove_new() which already returns void. Eventually after all drivers are converted, .remove_new() is renamed to .remove(). Trivially convert this driver from always returning zero in the remove callback to the void returning variant. Signed-off-by: Uwe Kleine-König Reviewed-by: Konrad Dybcio Link: https://lore.kernel.org/r/20230919133207.1400430-35-u.kleine-koenig@pengutronix.de Signed-off-by: Vinod Koul

dmaengine: qcom: bam_dma: allow omitting num-{channels,ees}

2023-05-24T06:52:43Z

The bam_dma driver needs to know the number of channels and execution environments (EEs) at probe time. If we are in full control of the BAM controller this information can be obtained from the BAM identification registers (BAM_REVISION/BAM_NUM_PIPES). When the BAM is "controlled remotely" it is more complicated. The BAM might not be on at probe time, so reading the registers could fail. This is why the information must be added to the device tree in this case, using "num-channels" and "qcom,num-ees". However, there are also some BAM instances that are initialized by something else but we still have a clock that allows to turn it on when needed. This can be set up in the DT with "qcom,controlled-remotely" and "clocks" and is already supported by the bam_dma driver. Examples for this are the typical BLSP BAM instances on older SoCs, QPIC BAM (for NAND) and the crypto BAM on some SoCs. In this case, there is no need to read "num-channels" and "qcom,num-ees" from the DT. The BAN can be turned on using the clock so we can just read it from the BAM registers like in the normal case. Check for the BAM clock earlier and skip reading "num-channels" and "qcom,num-ees" if it is present to allow simplifying the DT description a bit. Signed-off-by: Stephan Gerhold Reviewed-by: Bhupesh Sharma Link: https://lore.kernel.org/r/20230518-bamclk-dt-v2-1-a1a857b966ca@gerhold.net Signed-off-by: Vinod Koul