From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 7CCF6C7619A for ; Thu, 6 Apr 2023 02:44:08 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S234955AbjDFCoH (ORCPT ); Wed, 5 Apr 2023 22:44:07 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:37434 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S234857AbjDFCoE (ORCPT ); Wed, 5 Apr 2023 22:44:04 -0400 Received: from dfw.source.kernel.org (dfw.source.kernel.org [139.178.84.217]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 9F74B8A46 for ; Wed, 5 Apr 2023 19:44:00 -0700 (PDT) Received: from smtp.kernel.org (relay.kernel.org [52.25.139.140]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dfw.source.kernel.org (Postfix) with ESMTPS id 386156420B for ; Thu, 6 Apr 2023 02:44:00 +0000 (UTC) Received: by smtp.kernel.org (Postfix) with ESMTPSA id 93AFEC4339C; Thu, 6 Apr 2023 02:43:59 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=linux-foundation.org; s=korg; t=1680749039; bh=+xiDmG7L1opwt3JGWgZ/TLz0SwyeHZGoADOejKerQxg=; h=Date:To:From:Subject:From; b=BeDz/cY5PxSd5ITfq2W7DT7eAHJl+/6R6nGMWg+IkVaS+wUM+V4zT8nzSZlQDlDqQ /6BXi2aLbMAtuBn0t5PKaYHSnyqPrYHjXPQNhRTI1VdCpTspR0CqhZmHfxQ2r/xwjy WaruEYph8PQK+sQ/QBfK6rYJxcHDmmFqN3Wp1cg0= Date: Wed, 05 Apr 2023 19:43:58 -0700 To: mm-commits@vger.kernel.org, willy@infradead.org, tonyb@cybernetics.com, hch@lst.de, kbusch@kernel.org, akpm@linux-foundation.org From: Andrew Morton Subject: [merged mm-stable] dmapool-add-alloc-free-performance-test.patch removed from -mm tree Message-Id: <20230406024359.93AFEC4339C@smtp.kernel.org> Precedence: bulk Reply-To: linux-kernel@vger.kernel.org List-ID: X-Mailing-List: mm-commits@vger.kernel.org The quilt patch titled Subject: dmapool: add alloc/free performance test has been removed from the -mm tree. Its filename was dmapool-add-alloc-free-performance-test.patch This patch was dropped because it was merged into the mm-stable branch of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm ------------------------------------------------------ From: Keith Busch Subject: dmapool: add alloc/free performance test Date: Thu, 26 Jan 2023 13:51:14 -0800 Patch series "dmapool enhancements", v4. Time spent in dma_pool alloc/free increases linearly with the number of pages backing the pool. We can reduce this to constant time with minor changes to how free pages are tracked. This patch (of 12): Provide a module that allocates and frees many blocks of various sizes and report how long it takes. This is intended to provide a consistent way to measure how changes to the dma_pool_alloc/free routines affect timing. Link: https://lkml.kernel.org/r/20230126215125.4069751-1-kbusch@meta.com Link: https://lkml.kernel.org/r/20230126215125.4069751-2-kbusch@meta.com Signed-off-by: Keith Busch Cc: Christoph Hellwig Cc: Matthew Wilcox Cc: Tony Battersby Signed-off-by: Andrew Morton --- mm/Kconfig | 9 ++ mm/Makefile | 1 mm/dmapool_test.c | 147 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 157 insertions(+) --- /dev/null +++ a/mm/dmapool_test.c @@ -0,0 +1,147 @@ +#include +#include +#include +#include +#include +#include +#include + +#define NR_TESTS (100) + +struct dma_pool_pair { + dma_addr_t dma; + void *v; +}; + +struct dmapool_parms { + size_t size; + size_t align; + size_t boundary; +}; + +static const struct dmapool_parms pool_parms[] = { + { .size = 16, .align = 16, .boundary = 0 }, + { .size = 64, .align = 64, .boundary = 0 }, + { .size = 256, .align = 256, .boundary = 0 }, + { .size = 1024, .align = 1024, .boundary = 0 }, + { .size = 4096, .align = 4096, .boundary = 0 }, + { .size = 68, .align = 32, .boundary = 4096 }, +}; + +static struct dma_pool *pool; +static struct device test_dev; +static u64 dma_mask; + +static inline int nr_blocks(int size) +{ + return clamp_t(int, (PAGE_SIZE / size) * 512, 1024, 8192); +} + +static int dmapool_test_alloc(struct dma_pool_pair *p, int blocks) +{ + int i; + + for (i = 0; i < blocks; i++) { + p[i].v = dma_pool_alloc(pool, GFP_KERNEL, + &p[i].dma); + if (!p[i].v) + goto pool_fail; + } + + for (i = 0; i < blocks; i++) + dma_pool_free(pool, p[i].v, p[i].dma); + + return 0; + +pool_fail: + for (--i; i >= 0; i--) + dma_pool_free(pool, p[i].v, p[i].dma); + return -ENOMEM; +} + +static int dmapool_test_block(const struct dmapool_parms *parms) +{ + int blocks = nr_blocks(parms->size); + ktime_t start_time, end_time; + struct dma_pool_pair *p; + int i, ret; + + p = kcalloc(blocks, sizeof(*p), GFP_KERNEL); + if (!p) + return -ENOMEM; + + pool = dma_pool_create("test pool", &test_dev, parms->size, + parms->align, parms->boundary); + if (!pool) { + ret = -ENOMEM; + goto free_pairs; + } + + start_time = ktime_get(); + for (i = 0; i < NR_TESTS; i++) { + ret = dmapool_test_alloc(p, blocks); + if (ret) + goto free_pool; + if (need_resched()) + cond_resched(); + } + end_time = ktime_get(); + + printk("dmapool test: size:%-4zu align:%-4zu blocks:%-4d time:%llu\n", + parms->size, parms->align, blocks, + ktime_us_delta(end_time, start_time)); + +free_pool: + dma_pool_destroy(pool); +free_pairs: + kfree(p); + return ret; +} + +static void dmapool_test_release(struct device *dev) +{ +} + +static int dmapool_checks(void) +{ + int i, ret; + + ret = dev_set_name(&test_dev, "dmapool-test"); + if (ret) + return ret; + + ret = device_register(&test_dev); + if (ret) { + printk("%s: register failed:%d\n", __func__, ret); + goto put_device; + } + + test_dev.release = dmapool_test_release; + set_dma_ops(&test_dev, NULL); + test_dev.dma_mask = &dma_mask; + ret = dma_set_mask_and_coherent(&test_dev, DMA_BIT_MASK(64)); + if (ret) { + printk("%s: mask failed:%d\n", __func__, ret); + goto del_device; + } + + for (i = 0; i < ARRAY_SIZE(pool_parms); i++) { + ret = dmapool_test_block(&pool_parms[i]); + if (ret) + break; + } + +del_device: + device_del(&test_dev); +put_device: + put_device(&test_dev); + return ret; +} + +static void dmapool_exit(void) +{ +} + +module_init(dmapool_checks); +module_exit(dmapool_exit); +MODULE_LICENSE("GPL"); --- a/mm/Kconfig~dmapool-add-alloc-free-performance-test +++ a/mm/Kconfig @@ -1100,6 +1100,15 @@ comment "GUP_TEST needs to have DEBUG_FS config GUP_GET_PXX_LOW_HIGH bool +config DMAPOOL_TEST + tristate "Enable a module to run time tests on dma_pool" + depends on HAS_DMA + help + Provides a test module that will allocate and free many blocks of + various sizes and report how long it takes. This is intended to + provide a consistent way to measure how changes to the + dma_pool_alloc/free routines affect performance. + config ARCH_HAS_PTE_SPECIAL bool --- a/mm/Makefile~dmapool-add-alloc-free-performance-test +++ a/mm/Makefile @@ -103,6 +103,7 @@ obj-$(CONFIG_MEMCG) += swap_cgroup.o endif obj-$(CONFIG_CGROUP_HUGETLB) += hugetlb_cgroup.o obj-$(CONFIG_GUP_TEST) += gup_test.o +obj-$(CONFIG_DMAPOOL_TEST) += dmapool_test.o obj-$(CONFIG_MEMORY_FAILURE) += memory-failure.o obj-$(CONFIG_HWPOISON_INJECT) += hwpoison-inject.o obj-$(CONFIG_DEBUG_KMEMLEAK) += kmemleak.o _ Patches currently in -mm which might be from kbusch@kernel.org are