@@ -225,3 +225,12 @@ config DMA_API_DEBUG_SG
is technically out-of-spec.
If unsure, say N.
+
+config DMA_MAP_BENCHMARK
+ bool "Enable benchmarking of streaming DMA mapping"
+ depends on DEBUG_FS
+ help
+ Provides /sys/kernel/debug/dma_map_benchmark that helps with testing
+ performance of dma_(un)map_page.
+
+ See tools/testing/selftests/dma/dma_map_benchmark.c
@@ -10,3 +10,4 @@ obj-$(CONFIG_DMA_API_DEBUG) += debug.o
obj-$(CONFIG_SWIOTLB) += swiotlb.o
obj-$(CONFIG_DMA_COHERENT_POOL) += pool.o
obj-$(CONFIG_DMA_REMAP) += remap.o
+obj-$(CONFIG_DMA_MAP_BENCHMARK) += map_benchmark.o
new file mode 100644
@@ -0,0 +1,361 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright (C) 2020 Hisilicon Limited.
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/debugfs.h>
+#include <linux/delay.h>
+#include <linux/device.h>
+#include <linux/dma-mapping.h>
+#include <linux/kernel.h>
+#include <linux/kthread.h>
+#include <linux/math64.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/timekeeping.h>
+
+#define DMA_MAP_BENCHMARK _IOWR('d', 1, struct map_benchmark)
+#define DMA_MAP_MAX_THREADS 1024
+#define DMA_MAP_MAX_SECONDS 300
+
+#define DMA_MAP_BIDIRECTIONAL 0
+#define DMA_MAP_TO_DEVICE 1
+#define DMA_MAP_FROM_DEVICE 2
+
+struct map_benchmark {
+ __u64 avg_map_100ns; /* average map latency in 100ns */
+ __u64 map_stddev; /* standard deviation of map latency */
+ __u64 avg_unmap_100ns; /* as above */
+ __u64 unmap_stddev;
+ __u32 threads; /* how many threads will do map/unmap in parallel */
+ __u32 seconds; /* how long the test will last */
+ __s32 node; /* which numa node this benchmark will run on */
+ __u32 dma_bits; /* DMA addressing capability */
+ __u32 dma_dir; /* DMA data direction */
+ __u64 expansion[10]; /* For future use */
+};
+
+struct map_benchmark_data {
+ struct map_benchmark bparam;
+ struct device *dev;
+ struct dentry *debugfs;
+ enum dma_data_direction dir;
+ atomic64_t sum_map_100ns;
+ atomic64_t sum_unmap_100ns;
+ atomic64_t sum_sq_map;
+ atomic64_t sum_sq_unmap;
+ atomic64_t loops;
+};
+
+static int map_benchmark_thread(void *data)
+{
+ void *buf;
+ dma_addr_t dma_addr;
+ struct map_benchmark_data *map = data;
+ int ret = 0;
+
+ buf = (void *)__get_free_page(GFP_KERNEL);
+ if (!buf)
+ return -ENOMEM;
+
+ while (!kthread_should_stop()) {
+ u64 map_100ns, unmap_100ns, map_sq, unmap_sq;
+ ktime_t map_stime, map_etime, unmap_stime, unmap_etime;
+ ktime_t map_delta, unmap_delta;
+
+ /*
+ * for a non-coherent device, if we don't stain them in the
+ * cache, this will give an underestimate of the real-world
+ * overhead of BIDIRECTIONAL or TO_DEVICE mappings;
+ * 66 means evertything goes well! 66 is lucky.
+ */
+ if (map->dir != DMA_FROM_DEVICE)
+ memset(buf, 0x66, PAGE_SIZE);
+
+ map_stime = ktime_get();
+ dma_addr = dma_map_single(map->dev, buf, PAGE_SIZE, map->dir);
+ if (unlikely(dma_mapping_error(map->dev, dma_addr))) {
+ pr_err("dma_map_single failed on %s\n",
+ dev_name(map->dev));
+ ret = -ENOMEM;
+ goto out;
+ }
+ map_etime = ktime_get();
+ map_delta = ktime_sub(map_etime, map_stime);
+
+ unmap_stime = ktime_get();
+ dma_unmap_single(map->dev, dma_addr, PAGE_SIZE, map->dir);
+ unmap_etime = ktime_get();
+ unmap_delta = ktime_sub(unmap_etime, unmap_stime);
+
+ /* calculate sum and sum of squares */
+
+ map_100ns = div64_ul(map_delta, 100);
+ unmap_100ns = div64_ul(unmap_delta, 100);
+ map_sq = map_100ns * map_100ns;
+ unmap_sq = unmap_100ns * unmap_100ns;
+
+ atomic64_add(map_100ns, &map->sum_map_100ns);
+ atomic64_add(unmap_100ns, &map->sum_unmap_100ns);
+ atomic64_add(map_sq, &map->sum_sq_map);
+ atomic64_add(unmap_sq, &map->sum_sq_unmap);
+ atomic64_inc(&map->loops);
+ }
+
+out:
+ free_page((unsigned long)buf);
+ return ret;
+}
+
+static int do_map_benchmark(struct map_benchmark_data *map)
+{
+ struct task_struct **tsk;
+ int threads = map->bparam.threads;
+ int node = map->bparam.node;
+ const cpumask_t *cpu_mask = cpumask_of_node(node);
+ u64 loops;
+ int ret = 0;
+ int i;
+
+ tsk = kmalloc_array(threads, sizeof(tsk), GFP_KERNEL);
+ if (!tsk)
+ return -ENOMEM;
+
+ get_device(map->dev);
+
+ for (i = 0; i < threads; i++) {
+ tsk[i] = kthread_create_on_node(map_benchmark_thread, map,
+ map->bparam.node, "dma-map-benchmark/%d", i);
+ if (IS_ERR(tsk[i])) {
+ pr_err("create dma_map thread failed\n");
+ ret = PTR_ERR(tsk[i]);
+ goto out;
+ }
+
+ if (node != NUMA_NO_NODE)
+ kthread_bind_mask(tsk[i], cpu_mask);
+ }
+
+ /* clear the old value in the previous benchmark */
+ atomic64_set(&map->sum_map_100ns, 0);
+ atomic64_set(&map->sum_unmap_100ns, 0);
+ atomic64_set(&map->sum_sq_map, 0);
+ atomic64_set(&map->sum_sq_unmap, 0);
+ atomic64_set(&map->loops, 0);
+
+ for (i = 0; i < threads; i++)
+ wake_up_process(tsk[i]);
+
+ msleep_interruptible(map->bparam.seconds * 1000);
+
+ /* wait for the completion of benchmark threads */
+ for (i = 0; i < threads; i++) {
+ ret = kthread_stop(tsk[i]);
+ if (ret)
+ goto out;
+ }
+
+ loops = atomic64_read(&map->loops);
+ if (likely(loops > 0)) {
+ u64 map_variance, unmap_variance;
+ u64 sum_map = atomic64_read(&map->sum_map_100ns);
+ u64 sum_unmap = atomic64_read(&map->sum_unmap_100ns);
+ u64 sum_sq_map = atomic64_read(&map->sum_sq_map);
+ u64 sum_sq_unmap = atomic64_read(&map->sum_sq_unmap);
+
+ /* average latency */
+ map->bparam.avg_map_100ns = div64_u64(sum_map, loops);
+ map->bparam.avg_unmap_100ns = div64_u64(sum_unmap, loops);
+
+ /* standard deviation of latency */
+ map_variance = div64_u64(sum_sq_map, loops) -
+ map->bparam.avg_map_100ns *
+ map->bparam.avg_map_100ns;
+ unmap_variance = div64_u64(sum_sq_unmap, loops) -
+ map->bparam.avg_unmap_100ns *
+ map->bparam.avg_unmap_100ns;
+ map->bparam.map_stddev = int_sqrt64(map_variance);
+ map->bparam.unmap_stddev = int_sqrt64(unmap_variance);
+ }
+
+out:
+ put_device(map->dev);
+ kfree(tsk);
+ return ret;
+}
+
+static long map_benchmark_ioctl(struct file *filep, unsigned int cmd,
+ unsigned long arg)
+{
+ struct map_benchmark_data *map = filep->private_data;
+ u64 old_dma_mask;
+
+ int ret;
+
+ if (copy_from_user(&map->bparam, (void __user *)arg,
+ sizeof(map->bparam)))
+ return -EFAULT;
+
+ switch (cmd) {
+ case DMA_MAP_BENCHMARK:
+ if (map->bparam.threads == 0 ||
+ map->bparam.threads > DMA_MAP_MAX_THREADS) {
+ pr_err("invalid thread number\n");
+ return -EINVAL;
+ }
+
+ if (map->bparam.seconds == 0 ||
+ map->bparam.seconds > DMA_MAP_MAX_SECONDS) {
+ pr_err("invalid duration seconds\n");
+ return -EINVAL;
+ }
+
+ if (map->bparam.node != NUMA_NO_NODE &&
+ !node_possible(map->bparam.node)) {
+ pr_err("invalid numa node\n");
+ return -EINVAL;
+ }
+
+ switch (map->bparam.dma_dir) {
+ case DMA_MAP_BIDIRECTIONAL:
+ map->dir = DMA_BIDIRECTIONAL;
+ break;
+ case DMA_MAP_FROM_DEVICE:
+ map->dir = DMA_FROM_DEVICE;
+ break;
+ case DMA_MAP_TO_DEVICE:
+ map->dir = DMA_TO_DEVICE;
+ break;
+ default:
+ pr_err("invalid DMA direction\n");
+ return -EINVAL;
+ }
+
+ old_dma_mask = dma_get_mask(map->dev);
+
+ ret = dma_set_mask(map->dev,
+ DMA_BIT_MASK(map->bparam.dma_bits));
+ if (ret) {
+ pr_err("failed to set dma_mask on device %s\n",
+ dev_name(map->dev));
+ return -EINVAL;
+ }
+
+ ret = do_map_benchmark(map);
+
+ /*
+ * restore the original dma_mask as many devices' dma_mask are
+ * set by architectures, acpi, busses. When we bind them back
+ * to their original drivers, those drivers shouldn't see
+ * dma_mask changed by benchmark
+ */
+ dma_set_mask(map->dev, old_dma_mask);
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ if (copy_to_user((void __user *)arg, &map->bparam, sizeof(map->bparam)))
+ return -EFAULT;
+
+ return ret;
+}
+
+static const struct file_operations map_benchmark_fops = {
+ .open = simple_open,
+ .unlocked_ioctl = map_benchmark_ioctl,
+};
+
+static void map_benchmark_remove_debugfs(void *data)
+{
+ struct map_benchmark_data *map = (struct map_benchmark_data *)data;
+
+ debugfs_remove(map->debugfs);
+}
+
+static int __map_benchmark_probe(struct device *dev)
+{
+ struct dentry *entry;
+ struct map_benchmark_data *map;
+ int ret;
+
+ map = devm_kzalloc(dev, sizeof(*map), GFP_KERNEL);
+ if (!map)
+ return -ENOMEM;
+ map->dev = dev;
+
+ ret = devm_add_action(dev, map_benchmark_remove_debugfs, map);
+ if (ret) {
+ pr_err("Can't add debugfs remove action\n");
+ return ret;
+ }
+
+ /*
+ * we only permit a device bound with this driver, 2nd probe
+ * will fail
+ */
+ entry = debugfs_create_file("dma_map_benchmark", 0600, NULL, map,
+ &map_benchmark_fops);
+ if (IS_ERR(entry))
+ return PTR_ERR(entry);
+ map->debugfs = entry;
+
+ return 0;
+}
+
+static int map_benchmark_platform_probe(struct platform_device *pdev)
+{
+ return __map_benchmark_probe(&pdev->dev);
+}
+
+static struct platform_driver map_benchmark_platform_driver = {
+ .driver = {
+ .name = "dma_map_benchmark",
+ },
+ .probe = map_benchmark_platform_probe,
+};
+
+static int
+map_benchmark_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id)
+{
+ return __map_benchmark_probe(&pdev->dev);
+}
+
+static struct pci_driver map_benchmark_pci_driver = {
+ .name = "dma_map_benchmark",
+ .probe = map_benchmark_pci_probe,
+};
+
+static int __init map_benchmark_init(void)
+{
+ int ret;
+
+ ret = pci_register_driver(&map_benchmark_pci_driver);
+ if (ret)
+ return ret;
+
+ ret = platform_driver_register(&map_benchmark_platform_driver);
+ if (ret) {
+ pci_unregister_driver(&map_benchmark_pci_driver);
+ return ret;
+ }
+
+ return 0;
+}
+
+static void __exit map_benchmark_cleanup(void)
+{
+ platform_driver_unregister(&map_benchmark_platform_driver);
+ pci_unregister_driver(&map_benchmark_pci_driver);
+}
+
+module_init(map_benchmark_init);
+module_exit(map_benchmark_cleanup);
+
+MODULE_AUTHOR("Barry Song <song.bao.hua@hisilicon.com>");
+MODULE_DESCRIPTION("dma_map benchmark driver");
+MODULE_LICENSE("GPL");
Nowadays, there are increasing requirements to benchmark the performance of dma_map and dma_unmap particually while the device is attached to an IOMMU. This patch enables the support. Users can run specified number of threads to do dma_map_page and dma_unmap_page on a specific NUMA node with the specified duration. Then dma_map_benchmark will calculate the average latency for map and unmap. A difficulity for this benchmark is that dma_map/unmap APIs must run on a particular device. Each device might have different backend of IOMMU or non-IOMMU. So we use the driver_override to bind dma_map_benchmark to a particual device by: For platform devices: echo dma_map_benchmark > /sys/bus/platform/devices/xxx/driver_override echo xxx > /sys/bus/platform/drivers/xxx/unbind echo xxx > /sys/bus/platform/drivers/dma_map_benchmark/bind For PCI devices: echo dma_map_benchmark > /sys/bus/pci/devices/0000:00:01.0/driver_override echo 0000:00:01.0 > /sys/bus/pci/drivers/xxx/unbind echo 0000:00:01.0 > /sys/bus/pci/drivers/dma_map_benchmark/bind Cc: Will Deacon <will@kernel.org> Cc: Shuah Khan <shuah@kernel.org> Cc: Christoph Hellwig <hch@lst.de> Cc: Marek Szyprowski <m.szyprowski@samsung.com> Cc: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Barry Song <song.bao.hua@hisilicon.com> --- -v4: * add dma direction support according to Christoph Hellwig's comment; * add dma mask bit set according to Christoph Hellwig's comment; * make the benchmark depend on DEBUG_FS according to John Garry's comment; * strictly check parameters in ioctl; * fixed more than 80 char in one line; kernel/dma/Kconfig | 9 + kernel/dma/Makefile | 1 + kernel/dma/map_benchmark.c | 361 +++++++++++++++++++++++++++++++++++++ 3 files changed, 371 insertions(+) create mode 100644 kernel/dma/map_benchmark.c