kernel.git/drivers/dma/sf-pdma, branch linux-6.3.y

dmaengine: sf-pdma: pdma_desc memory leak fix

2023-02-16T13:15:48Z

Commit b2cc5c465c2c ("dmaengine: sf-pdma: Add multithread support for a DMA channel") changed sf_pdma_prep_dma_memcpy() to unconditionally allocate a new sf_pdma_desc each time it is called. The driver previously recycled descs, by checking the in_use flag, only allocating additional descs if the existing one was in use. This logic was removed in commit b2cc5c465c2c ("dmaengine: sf-pdma: Add multithread support for a DMA channel"), but sf_pdma_free_desc() was not changed to handle the new behaviour. As a result, each time sf_pdma_prep_dma_memcpy() is called, the previous descriptor is leaked, over time leading to memory starvation: unreferenced object 0xffffffe008447300 (size 192): comm "irq/39-mchp_dsc", pid 343, jiffies 4294906910 (age 981.200s) hex dump (first 32 bytes): 00 00 00 ff 00 00 00 00 b8 c1 00 00 00 00 00 00 ................ 00 00 70 08 10 00 00 00 00 00 00 c0 00 00 00 00 ..p............. backtrace: [<00000000064a04f4>] kmemleak_alloc+0x1e/0x28 [<00000000018927a7>] kmem_cache_alloc+0x11e/0x178 [<000000002aea8d16>] sf_pdma_prep_dma_memcpy+0x40/0x112 Add the missing kfree() to sf_pdma_free_desc(), and remove the redundant in_use flag. Fixes: b2cc5c465c2c ("dmaengine: sf-pdma: Add multithread support for a DMA channel") Signed-off-by: Shravan Chippa Reviewed-by: Conor Dooley Link: https://lore.kernel.org/r/20230120100623.3530634-1-shravan.chippa@microchip.com Signed-off-by: Vinod Koul

dmaengine: drivers: Use devm_platform_ioremap_resource()

2023-01-18T16:46:47Z

platform_get_resource() and devm_ioremap_resource() are wrapped up in the devm_platform_ioremap_resource() helper. Use the helper and get rid of the local variable for struct resource *. We now have a function call less. Signed-off-by: Tudor Ambarus Acked-by: Jernej Skrabec Acked-by: Peter Ujfalusi Link: https://lore.kernel.org/r/20221110152528.7821-1-tudor.ambarus@microchip.com Signed-off-by: Vinod Koul

dmaengine: sf-pdma:Remove the print function dev_err()

2022-09-05T06:20:38Z

>From the coccinelle check: ./drivers/dma/sf-pdma/sf-pdma.c Error:line 409 is redundant because platform_get_irq() already prints an error ./drivers/dma/sf-pdma/sf-pdma.c Error:line 424 is redundant because platform_get_irq() already prints an error So,remove the unnecessary print function dev_err() Reported-by: Zeal Robot Signed-off-by: ye xingchen Link: https://lore.kernel.org/r/20220810062532.13425-1-ye.xingchen@zte.com.cn Signed-off-by: Vinod Koul

dmaengine: sf-pdma: Add multithread support for a DMA channel

2022-07-06T12:40:34Z

When we get a DMA channel and try to use it in multiple threads it will cause oops and hanging the system. % echo 64 > /sys/module/dmatest/parameters/threads_per_chan % echo 10000 > /sys/module/dmatest/parameters/iterations % echo 1 > /sys/module/dmatest/parameters/run [ 89.480664] Unable to handle kernel NULL pointer dereference at virtual address 00000000000000a0 [ 89.488725] Oops [#1] [ 89.494708] CPU: 2 PID: 1008 Comm: dma0chan0-copy0 Not tainted 5.17.0-rc5 [ 89.509385] epc : vchan_find_desc+0x32/0x46 [ 89.513553] ra : sf_pdma_tx_status+0xca/0xd6 This happens because of data race. Each thread rewrite channels's descriptor as soon as device_prep_dma_memcpy() is called. It leads to the situation when the driver thinks that it uses right descriptor that actually is freed or substituted for other one. With current fixes a descriptor changes its value only when it has been used. A new descriptor is acquired from vc->desc_issued queue that is already filled with descriptors that are ready to be sent. Threads have no direct access to DMA channel descriptor. Now it is just possible to queue a descriptor for further processing. Fixes: 6973886ad58e ("dmaengine: sf-pdma: add platform DMA support for HiFive Unleashed A00") Signed-off-by: Viacheslav Mitrofanov Link: https://lore.kernel.org/r/20220701082942.12835-1-v.v.mitrofanov@yadro.com Signed-off-by: Vinod Koul

dmaengine: sf-pdma: Get number of channel by device tree

2022-04-08T13:13:29Z

It currently assumes that there are always four channels, it would cause the error if there is actually less than four channels. Change that by getting number of channel from device tree. For backwards-compatibility, it uses the default value (i.e. 4) when there is no 'dma-channels' information in dts. Signed-off-by: Zong Li Acked-by: Palmer Dabbelt Reviewed-by: Bin Meng Link: https://lore.kernel.org/r/f08a95b6582a51712c5b2c3cb859136d07bfa8b9.1648461096.git.zong.li@sifive.com Signed-off-by: Vinod Koul

Merge tag 'dmaengine-5.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine

2021-07-05T19:05:13Z

Pull dmaengine updates from Vinod Koul: "This time around we have a smaller pull request than usual and this includes code removal, so should be good! New drivers/devices - Support for QCOM SM8250 GPI DMA - removal of shdma-of driver and binding Updates: - arm-pl08x yaml binding move - altera-msgdma gained DT support - removal of imx-sdma platform data support - idxd and xilinx driver updates" * tag 'dmaengine-5.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (22 commits) dmaengine: imx-sdma: Remove platform data header dmaengine: xilinx: dpdma: Fix spacing around addr[i-1] dmaengine: xilinx: dpdma: Use kernel type u32 over uint32_t dmaengine: altera-msgdma: add OF support MAINTAINERS: add entry for Altera mSGDMA dt-bindings: dma: add schema for altera-msgdma dmaengine: xilinx: dpdma: fix kernel-doc dmaengine: sf-pdma: apply proper spinlock flags in sf_pdma_prep_dma_memcpy() dmaengine: sh: Remove unused shdma-of driver dt-bindings: dmaengine: Remove SHDMA Device Tree bindings dmaengine: qcom: gpi: Add SM8250 compatible dt-bindings: dmaengine: qcom: gpi: add compatible for sm8250 dmaengine: sun4i: Use list_move_tail instead of list_del/list_add_tail dmaengine: ti: omap-dma: Skip pointless cpu_pm context restore on errors dmaengine: hsu: Account transferred bytes dmaengine: Move kdoc description of struct dma_chan_percpu closer to it dmaengine: xilinx: dpdma: Print debug message when losing vsync race dmaengine: xilinx: dpdma: Print channel number in kernel log messages dt-bindings: dma: convert arm-pl08x to yaml dmaengine: idxd: remove devm allocation for idxd->int_handles ...

dmaengine: sf-pdma: apply proper spinlock flags in sf_pdma_prep_dma_memcpy()

2021-06-16T10:40:00Z

The second parameter of spinlock_irq[save/restore] function is flags, which is the last input parameter of sf_pdma_prep_dma_memcpy(). So declare local variable 'iflags' to be used as the second parameter of spinlock_irq[save/restore] function. Signed-off-by: Austin Kim Link: https://lore.kernel.org/r/20210611065336.GA1121@raspberrypi Signed-off-by: Vinod Koul

dmaengine: SF_PDMA depends on HAS_IOMEM

2021-05-31T04:14:14Z

When CONFIG_HAS_IOMEM is not set/enabled, certain iomap() family functions [including ioremap(), devm_ioremap(), etc.] are not available. Drivers that use these functions should depend on HAS_IOMEM so that they do not cause build errors. Mends this build error: s390-linux-ld: drivers/dma/sf-pdma/sf-pdma.o: in function `sf_pdma_probe': sf-pdma.c:(.text+0x1668): undefined reference to `devm_ioremap_resource' Fixes: 6973886ad58e ("dmaengine: sf-pdma: add platform DMA support for HiFive Unleashed A00") Signed-off-by: Randy Dunlap Reported-by: kernel test robot Cc: Green Wan Cc: Vinod Koul Cc: dmaengine@vger.kernel.org Link: https://lore.kernel.org/r/20210522021313.16405-4-rdunlap@infradead.org Signed-off-by: Vinod Koul

dmaengine: sf: drop of_match_ptr from of_device_id table

2020-11-24T17:32:20Z

The driver can match only via the DT table so the table should be always used and the of_match_ptr does not have any sense (this also allows ACPI matching via PRP0001, even though it is not relevant here). This fixes compile warning (!CONFIG_OF on x86_64): drivers/dma/sf-pdma/sf-pdma.c:576:34: warning: ‘sf_pdma_dt_ids’ defined but not used [-Wunused-const-variable=] Signed-off-by: Krzysztof Kozlowski Link: https://lore.kernel.org/r/20201120162303.482126-4-krzk@kernel.org Signed-off-by: Vinod Koul

dmaengine: sf-pdma: remove redundant irqsave and irqrestore in hardIRQ

2020-11-09T11:55:54Z

Running in hardIRQ, disabling IRQ is redundant since hardIRQ has disabled IRQ. This patch removes the irqsave and irqstore to save some instruction cycles. Signed-off-by: Barry Song Cc: Green Wan Link: https://lore.kernel.org/r/20201027215252.25820-4-song.bao.hua@hisilicon.com Signed-off-by: Vinod Koul