kernel/linux.git/drivers/dma/ioat, branch v4.14.217

dmaengine: ioat setting ioat timeout as module parameter

2020-07-29T05:42:55+00:00

[ Upstream commit 87730ccbddcb48478b1b88e88b14e73424130764 ] DMA transaction time to completion is a function of PCI bandwidth, transaction size and a queue depth. So hard coded value for timeouts might be wrong for some scenarios. Signed-off-by: Leonid Ravich Reviewed-by: Dave Jiang Link: https://lore.kernel.org/r/20200701184816.29138-1-leonid.ravich@dell.com Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin

ioat: ioat_alloc_ring() failure handling.

2020-01-17T18:45:54+00:00

[ Upstream commit b0b5ce1010ffc50015eaec72b0028aaae3f526bb ] If dma_alloc_coherent() returns NULL in ioat_alloc_ring(), ring allocation must not proceed. Until now, if the first call to dma_alloc_coherent() in ioat_alloc_ring() returned NULL, the processing could proceed, failing with NULL-pointer dereferencing further down the line. Signed-off-by: Alexander Barabash Acked-by: Dave Jiang Link: https://lore.kernel.org/r/75e9c0e84c3345d693c606c64f8b9ab5@x13pwhopdag1307.AMER.DELL.COM Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin

dmaengine: ioat: fix prototype of ioat_enumerate_channels

2019-11-24T07:23:13+00:00

[ Upstream commit f4d34aa8c887a8a2d23ef546da0efa10e3f77241 ] Signed-off-by: Rami Rosen Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin

driver/dma/ioat: Call del_timer_sync() without holding prep_lock

2018-11-13T19:14:59+00:00

[ Upstream commit cfb03be6c7e8a1591285849c361d67b09f5149f7 ] The following lockdep splat was observed: [ 1222.241750] ====================================================== [ 1222.271301] WARNING: possible circular locking dependency detected [ 1222.301060] 4.16.0-10.el8+5.x86_64+debug #1 Not tainted [ 1222.326659] ------------------------------------------------------ [ 1222.356565] systemd-shutdow/1 is trying to acquire lock: [ 1222.382660] ((&ioat_chan->timer)){+.-.}, at: [<00000000f71e1a28>] del_timer_sync+0x5/0xf0 [ 1222.422928] [ 1222.422928] but task is already holding lock: [ 1222.451743] (&(&ioat_chan->prep_lock)->rlock){+.-.}, at: [<000000008ea98b12>] ioat_shutdown+0x86/0x100 [ioatdma] : [ 1223.524987] Chain exists of: [ 1223.524987] (&ioat_chan->timer) --> &(&ioat_chan->cleanup_lock)->rlock --> &(&ioat_chan->prep_lock)->rlock [ 1223.524987] [ 1223.594082] Possible unsafe locking scenario: [ 1223.594082] [ 1223.622630] CPU0 CPU1 [ 1223.645080] ---- ---- [ 1223.667404] lock(&(&ioat_chan->prep_lock)->rlock); [ 1223.691535] lock(&(&ioat_chan->cleanup_lock)->rlock); [ 1223.728657] lock(&(&ioat_chan->prep_lock)->rlock); [ 1223.765122] lock((&ioat_chan->timer)); [ 1223.784095] [ 1223.784095] *** DEADLOCK *** [ 1223.784095] [ 1223.813492] 4 locks held by systemd-shutdow/1: [ 1223.834677] #0: (reboot_mutex){+.+.}, at: [<0000000056d33456>] SYSC_reboot+0x10f/0x300 [ 1223.873310] #1: (&dev->mutex){....}, at: [<00000000258dfdd7>] device_shutdown+0x1c8/0x660 [ 1223.913604] #2: (&dev->mutex){....}, at: [<0000000068331147>] device_shutdown+0x1d6/0x660 [ 1223.954000] #3: (&(&ioat_chan->prep_lock)->rlock){+.-.}, at: [<000000008ea98b12>] ioat_shutdown+0x86/0x100 [ioatdma] In the ioat_shutdown() function: spin_lock_bh(&ioat_chan->prep_lock); set_bit(IOAT_CHAN_DOWN, &ioat_chan->state); del_timer_sync(&ioat_chan->timer); spin_unlock_bh(&ioat_chan->prep_lock); According to the synchronization rule for the del_timer_sync() function, the caller must not hold locks which would prevent completion of the timer's handler. The timer structure has its own lock that manages its synchronization. Setting the IOAT_CHAN_DOWN bit should prevent other CPUs from trying to use that device anyway, there is probably no need to call del_timer_sync() while holding the prep_lock. So the del_timer_sync() call is now moved outside of the prep_lock critical section to prevent the circular lock dependency. Signed-off-by: Waiman Long Reviewed-by: Dave Jiang Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

dmaengine: ioat: Fix error handling path

2018-02-25T10:08:01+00:00

[ Upstream commit 5c9afbda911ce20b3f2181d1e440a0222e1027dd ] If the last test in 'ioat_dma_self_test()' fails, we must release all the allocated resources and not just part of them. Signed-off-by: Christophe JAILLET Acked-by: Dave Jiang Signed-off-by: Vinod Koul Signed-off-by: Sasha Levin Signed-off-by: Greg Kroah-Hartman

dmaengine: ioatdma: Add intr_coalesce sysfs entry

2017-08-23T16:02:21+00:00

We observed performance increase with DMA copy from memory to MMIO by changing the interrupt coalescing value to 0. The previous set value was projected on the C5xxx Xeon platform and no longer holds true. Removing hard coded value and providing a tune-able in sysfs in order to allow user to tune this on a per channel basis. By default this value will be set to 0. Example of sysfs variable importing for interrupt coalescing value from command line: echo 5> /sys/devices/pci0000:00/0000:00:04.0/dma/dma0chan0/ quickdata/intr_coalesce Reported-by: Nithin Sujir Signed-off-by: Ujjal Singh Acked-by: Dave Jiang Signed-off-by: Vinod Koul

dmaengine: ioat: constify pci_device_id.

2017-07-18T16:02:14+00:00

pci_device_id are not supposed to change at runtime. All functions working with pci_device_id provided by work with const pci_device_id. So mark the non-const structs as const. File size before: text data bss dec hex filename 12582 3056 16 15654 3d26 drivers/dma/ioat/init.o File size After adding 'const': text data bss dec hex filename 14773 865 16 15654 3d26 drivers/dma/ioat/init.o Signed-off-by: Arvind Yadav Signed-off-by: Vinod Koul

Merge tag 'dmaengine-4.13-rc1' of git://git.infradead.org/users/vkoul/slave-dma

2017-07-08T19:36:50+00:00

Pull dmaengine updates from Vinod Koul: - removal of AVR32 support in dw driver as AVR32 is gone - new driver for Broadcom stream buffer accelerator (SBA) RAID driver - add support for Faraday Technology FTDMAC020 in amba-pl08x driver - IOMMU support in pl330 driver - updates to bunch of drivers * tag 'dmaengine-4.13-rc1' of git://git.infradead.org/users/vkoul/slave-dma: (36 commits) dmaengine: qcom_hidma: correct API violation for submit dmaengine: zynqmp_dma: Remove max len check in zynqmp_dma_prep_memcpy dmaengine: tegra-apb: Really fix runtime-pm usage dmaengine: fsl_raid: make of_device_ids const. dmaengine: qcom_hidma: allow ACPI/DT parameters to be overridden dmaengine: fsldma: set BWC, DAHTS and SAHTS values correctly dmaengine: Kconfig: Simplify the help text for MXS_DMA dmaengine: pl330: Delete unused functions dmaengine: Replace WARN_TAINT_ONCE() with pr_warn_once() dmaengine: Kconfig: Extend the dependency for MXS_DMA dmaengine: mxs: Use %zu for printing a size_t variable dmaengine: ste_dma40: Cleanup scatterlist layering violations dmaengine: imx-dma: cleanup scatterlist layering violations dmaengine: use proper name for the R-Car SoC dmaengine: imx-sdma: Fix compilation warning. dmaengine: imx-sdma: Handle return value of clk_prepare_enable dmaengine: pl330: Add IOMMU support to slave tranfers dmaengine: DW DMAC: Handle return value of clk_prepare_enable dmaengine: pl08x: use GENMASK() to create bitmasks dmaengine: pl08x: Add support for Faraday Technology FTDMAC020 ...

dmaengine: ioat: don't use DMA_ERROR_CODE

2017-06-20T09:12:56+00:00

DMA_ERROR_CODE is not a public API and will go away. Instead properly unwind based on the loop counter. Signed-off-by: Christoph Hellwig Acked-by: Dave Jiang Acked-By: Vinod Koul

dmaengine: Replace WARN_TAINT_ONCE() with pr_warn_once()

2017-06-15T04:20:37+00:00

The WARN_TAINT_ONCE() prints out a loud stack trace on broken BIOSes. The systems that have this problem are several years out of support and no longer have BIOS updates available. The stack trace isn't necessary and a pr_warn_once() will do. Change WARN_TAINT_ONCE() to pr_warn_once() and taint. Signed-off-by: Prarit Bhargava Cc: Dan Williams Cc: Vinod Koul Cc: Duyck, Alexander H Signed-off-by: Vinod Koul