kernel/drivers/dma/dw-edma, branch 0x221E-v0.0.1-v6.19

dmaengine: dw-edma: Fix multiple times setting of the CYCLE_STATE and CYCLE_BIT bits for HDMA.

2026-04-02T11:25:54Z

[ Upstream commit 3f63297ff61a994b99d710dcb6dbde41c4003233 ] Others have submitted this issue (https://lore.kernel.org/dmaengine/ 20240722030405.3385-1-zhengdongxiong@gxmicro.cn/), but it has not been fixed yet. Therefore, more supplementary information is provided here. As mentioned in the "PCS-CCS-CB-TCB" Producer-Consumer Synchronization of "DesignWare Cores PCI Express Controller Databook, version 6.00a": 1. The Consumer CYCLE_STATE (CCS) bit in the register only needs to be initialized once; the value will update automatically to be ~CYCLE_BIT (CB) in the next chunk. 2. The Consumer CYCLE_BIT bit in the register is loaded from the LL element and tested against CCS. When CB = CCS, the data transfer is executed. Otherwise not. The current logic sets customer (HDMA) CS and CB bits to 1 in each chunk while setting the producer (software) CB of odd chunks to 0 and even chunks to 1 in the linked list. This is leading to a mismatch between the producer CB and consumer CS bits. This issue can be reproduced by setting the transmission data size to exceed one chunk. By the way, in the EDMA using the same "PCS-CCS-CB-TCB" mechanism, the CS bit is only initialized once and this issue was not found. Refer to drivers/dma/dw-edma/dw-edma-v0-core.c:dw_edma_v0_core_start. So fix this issue by initializing the CYCLE_STATE and CYCLE_BIT bits only once. Fixes: e74c39573d35 ("dmaengine: dw-edma: Add support for native HDMA") Signed-off-by: LUO Haowen Reviewed-by: Frank Li Link: https://patch.msgid.link/tencent_CB11AA9F3920C1911AF7477A9BD8EFE0AD05@qq.com Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin

dmaengine: dw-edma: Set status for callback_result

2025-09-02T07:14:45Z

DMA Engine has support for the callback_result which provides the status of the request and the residue. This helps in determining the correct status of the request and in efficient resource management of the request. The 'callback_result' method is preferred over the deprecated 'callback' method. Signed-off-by: Devendra K Verma Link: https://lore.kernel.org/r/20250821121505.318179-1-devverma@amd.com Signed-off-by: Vinod Koul

Merge tag 'dmaengine-6.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine

2025-08-01T19:35:12Z

Pull dmaengine updates from Vinod Koul: "Core: - Managed API for dma channel request New support: - Sophgo CV18XX/SG200X dmamux driver - Qualcomm Milos GPI, sc8280xp GPI support Updates: - Conversion of brcm,iproc-sba and marvell,orion-xor binding - Unused code cleanup across drivers" * tag 'dmaengine-6.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (23 commits) dt-bindings: dma: fsl-mxs-dma: allow interrupt-names for fsl,imx23-dma-apbx dmaengine: xdmac: make it selectable for ARCH_MICROCHIP dt-bindings: dma: Convert marvell,orion-xor to DT schema dt-bindings: dma: Convert brcm,iproc-sba to DT schema dmaengine: nbpfaxi: Add missing check after DMA map dmaengine: mv_xor: Fix missing check after DMA map and missing unmap dt-bindings: dma: qcom,gpi: document the Milos GPI DMA Engine dmaengine: idxd: Remove __packed from structures dmaengine: ti: Do not enable by default during compile testing dmaengine: sh: Do not enable SH_DMAE_BASE by default during compile testing dmaengine: idxd: Fix warning for deadcode.deadstore dmaengine: mmp: Fix again Wvoid-pointer-to-enum-cast warning dmaengine: fsl-qdma: Add missing fsl_qdma_format kerneldoc dmaengine: qcom: gpi: Drop unused gpi_write_reg_field() dmaengine: fsl-dpaa2-qdma: Drop unused mc_enc() dmaengine: dw-edma: Drop unused dchan2dev() and chan2dev() dmaengine: stm32: Don't use %pK through printk dmaengine: stm32-dma: configure next sg only if there are more than 2 sgs dmaengine: sun4i: Simplify error handling in probe() dt-bindings: dma: qcom,gpi: Document the sc8280xp GPI DMA engine ...

dma: dw-edma: Fix build warning in dw_edma_pcie_probe()

2025-07-15T15:11:49Z

The function dw_edma_pcie_probe() in dw-edma-pcie.c triggered a frame size warning: ld.lld:warning: drivers/dma/dw-edma/dw-edma-pcie.c:162:0: stack frame size (1040) exceeds limit (1024) in function 'dw_edma_pcie_probe' This patch reduces the stack usage by dynamically allocating the `vsec_data` structure using kmalloc(), rather than placing it on the stack. This eliminates the overflow warning and improves kernel robustness. Signed-off-by: Abinash Singh Acked-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/20250705160055.808165-1-abinashsinghlalotra@gmail.com Signed-off-by: Vinod Koul

dmaengine: dw-edma: Drop unused dchan2dev() and chan2dev()

2025-06-26T22:29:44Z

Static functions dchan2dev() and chan2dev() are not used, W=1 build: dw-edma-core.c:27:16: error: unused function 'dchan2dev' [-Werror,-Wunused-function] dw-edma-core.c:33:16: error: unused function 'chan2dev' [-Werror,-Wunused-function] Fixes: e63d79d1ffcd ("dmaengine: Add Synopsys eDMA IP core driver") Signed-off-by: Krzysztof Kozlowski Link: https://lore.kernel.org/r/20250525-dma-fixes-v1-1-89d06dac9bcb@linaro.org Signed-off-by: Vinod Koul

dmaengine: dw-edma: Add HDMA NATIVE map check

2025-05-14T13:45:36Z

The HDMA IP supports the HDMA_NATIVE map format as part of Vendor-Specific Extended Capability. Added the check for HDMA_NATIVE map format. The check for map format enables the IP specific function invocation during the DMA ops. Signed-off-by: Devendra K Verma Link: https://lore.kernel.org/r/20250513070314.577823-1-devverma@amd.com Signed-off-by: Vinod Koul

dmaengine: Use str_enable_disable-like helpers

2025-02-10T14:27:11Z

Replace ternary (condition ? "enable" : "disable") syntax with helpers from string_choices.h because: 1. Simple function call with one argument is easier to read. Ternary operator has three arguments and with wrapping might lead to quite long code. 2. Is slightly shorter thus also easier to read. 3. It brings uniformity in the text - same string. 4. Allows deduping by the linker, which results in a smaller binary file. Signed-off-by: Krzysztof Kozlowski Reviewed-by: Manivannan Sadhasivam #dw-edma Reviewed-by: Jernej Skrabec Link: https://lore.kernel.org/r/20250114191021.854080-1-krzysztof.kozlowski@linaro.org Signed-off-by: Vinod Koul

dmaengine: dw-edma: Do not enable watermark interrupts for HDMA

2024-08-28T13:10:17Z

DW_HDMA_V0_LIE and DW_HDMA_V0_RIE are initialized as BIT(3) and BIT(4) respectively in dw_hdma_control enum. But as per HDMA register these bits are corresponds to LWIE and RWIE bit i.e local watermark interrupt enable and remote watermarek interrupt enable. In linked list mode LWIE and RWIE bits only enable the local and remote watermark interrupt. Since the watermark interrupts are not used but enabled, this leads to spurious interrupts getting generated. So remove the code that enables them to avoid generating spurious watermark interrupts. And also rename DW_HDMA_V0_LIE to DW_HDMA_V0_LWIE and DW_HDMA_V0_RIE to DW_HDMA_V0_RWIE as there is no LIE and RIE bits in HDMA and those bits are corresponds to LWIE and RWIE bits. Fixes: e74c39573d35 ("dmaengine: dw-edma: Add support for native HDMA") cc: stable@vger.kernel.org Signed-off-by: Mrinmay Sarkar Reviewed-by: Manivannan Sadhasivam Reviewed-by: Serge Semin Link: https://lore.kernel.org/r/1724674261-3144-3-git-send-email-quic_msarkar@quicinc.com Signed-off-by: Vinod Koul

dmaengine: dw-edma: Fix unmasking STOP and ABORT interrupts for HDMA

2024-08-28T13:10:17Z

The current logic is enabling both STOP_INT_MASK and ABORT_INT_MASK bit. This is apparently masking those particular interrupts rather than unmasking the same. If the interrupts are masked, they would never get triggered. So fix the issue by unmasking the STOP and ABORT interrupts properly. Fixes: e74c39573d35 ("dmaengine: dw-edma: Add support for native HDMA") cc: stable@vger.kernel.org Signed-off-by: Mrinmay Sarkar Reviewed-by: Manivannan Sadhasivam Link: https://lore.kernel.org/r/1724674261-3144-2-git-send-email-quic_msarkar@quicinc.com Signed-off-by: Vinod Koul

dmaengine: dw-edma: eDMA: Add sync read before starting the DMA transfer in remote setup

2024-02-07T08:30:52Z

The Linked list element and pointer are not stored in the same memory as the eDMA controller register. If the doorbell register is toggled before the full write of the linked list a race condition error will occur. In remote setup we can only use a readl to the memory to assure the full write has occurred. Fixes: 7e4b8a4fbe2c ("dmaengine: Add Synopsys eDMA IP version 0 support") Reviewed-by: Serge Semin Reviewed-by: Manivannan Sadhasivam Signed-off-by: Kory Maincent Link: https://lore.kernel.org/r/20240129-b4-feature_hdma_mainline-v7-6-8e8c1acb7a46@bootlin.com Signed-off-by: Vinod Koul