From patchwork Wed Jan 15 01:26:44 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333041 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 1353D14B4 for ; Wed, 15 Jan 2020 01:26:57 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id D1A842467C for ; Wed, 15 Jan 2020 01:26:56 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="H32XDa3w" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728883AbgAOB04 (ORCPT ); Tue, 14 Jan 2020 20:26:56 -0500 Received: from mail-pf1-f201.google.com ([209.85.210.201]:44620 "EHLO mail-pf1-f201.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728844AbgAOB04 (ORCPT ); Tue, 14 Jan 2020 20:26:56 -0500 Received: by mail-pf1-f201.google.com with SMTP id r127so9896941pfc.11 for ; Tue, 14 Jan 2020 17:26:55 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:message-id:mime-version:subject:from:to:cc; bh=uSFoTkbV1/PTTT0lqNVksRGhMp4JZGlIBCnq9ick5SE=; b=H32XDa3wbAKwYRDGaboS6Vbs+Rab3/Z1J1+Q55Jnru+N47ZsgToNF9uIrvUa7wFEmm E2cFvS9SsVroWN6w8Szus5V4k58Bx6E942EBVJAL4l0RP7W1wF3zeoTuhKFNks/bcj7R LAsJFiq3wpWXArVaLN+S8UX5GUJavPEF1XVeVOQQv76zIDMQEsqVR4QmgwIraYEYfFfn 0LvUCB0pUG0QT/O8IwIz+vLXdXAJmWFvZZtGvpHhRU2SiWskBD6jEX5cvJ5Q7HaxbxsM DmsKy0mDjJZN2mNnZUMQF5ChZ9QiJptpK+SVk1WYRFDjikMe00f2pDvfhMcHw1iL7b16 KF4A== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:message-id:mime-version:subject:from:to:cc; bh=uSFoTkbV1/PTTT0lqNVksRGhMp4JZGlIBCnq9ick5SE=; b=s0DeUqzYFNh2GQN4PHXDXwR71ueti937+QAp1arKvMuVIscgVf/0v71/oWWUXWbGy4 Ki9kJcE0uQn5Fj22fLh8/GtQYf49hsGYnp2uXN5Wa5d87I4ALweHe1TDVFFIs7FhVRQg OkPoKC3Ahyb/sSEAueyZf2NuwsP523FZOaV/nQhHoOJm4yxKdXqKDm7LREvJ83SfrcPt pMZ7/w8kFbHRd/MLgOgVdKhXiNYoec+4qqeDurUdvM3pU9TSHA9kHL/nSdBTdYZnf0sp clng3bT3hzf1FqEAsPewue5puVaVAhadlLxAXwsQTywS4xif+0sZN5bRY2GuIOp22wdM ivAQ== X-Gm-Message-State: APjAAAX6E9M4HLBqt9FVsL66xd9jk9bntSZRTbTZfmFGF9Hyx9qH8zpj YK6pRlTp2tUxRVFEQZaJ/qX4U879Iv0pHyv0Ag== X-Google-Smtp-Source: APXvYqyeSbR9Pa9AJxqA89NEPoK8PBNVAt8KDeIlz15X2UTQQ0RUv1ScueDAz/ZP1mnMjtIc0Jj3kXOjYP1K9fW2cw== X-Received: by 2002:a63:3f47:: with SMTP id m68mr29983698pga.411.1579051615137; Tue, 14 Jan 2020 17:26:55 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:44 -0800 Message-Id: <20200115012651.228058-1-almasrymina@google.com> Mime-Version: 1.0 X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 1/8] hugetlb_cgroup: Add hugetlb_cgroup reservation counter From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org These counters will track hugetlb reservations rather than hugetlb memory faulted in. This patch only adds the counter, following patches add the charging and uncharging of the counter. This is patch 1 of an 8 patch series. Problem: Currently tasks attempting to reserve more hugetlb memory than is available get a failure at mmap/shmget time. This is thanks to Hugetlbfs Reservations [1]. However, if a task attempts to reserve hugetlb memory only more than its hugetlb_cgroup limit allows, the kernel will allow the mmap/shmget call, but will SIGBUS the task when it attempts to fault the memory in. We have users hitting their hugetlb_cgroup limits and thus we've been looking at this failure mode. We'd like to improve this behavior such that users violating the hugetlb_cgroup limits get an error on mmap/shmget time, rather than getting SIGBUS'd when they try to fault the excess memory in. This gives the user an opportunity to fallback more gracefully to non-hugetlbfs memory for example. The underlying problem is that today's hugetlb_cgroup accounting happens at hugetlb memory *fault* time, rather than at *reservation* time. Thus, enforcing the hugetlb_cgroup limit only happens at fault time, and the offending task gets SIGBUS'd. Proposed Solution: A new page counter named 'hugetlb.xMB.reservation_[limit|usage|max_usage]_in_bytes'. This counter has slightly different semantics than 'hugetlb.xMB.[limit|usage|max_usage]_in_bytes': - While usage_in_bytes tracks all *faulted* hugetlb memory, reservation_usage_in_bytes tracks all *reserved* hugetlb memory and hugetlb memory faulted in without a prior reservation. - If a task attempts to reserve more memory than limit_in_bytes allows, the kernel will allow it to do so. But if a task attempts to reserve more memory than reservation_limit_in_bytes, the kernel will fail this reservation. This proposal is implemented in this patch series, with tests to verify functionality and show the usage. Alternatives considered: 1. A new cgroup, instead of only a new page_counter attached to the existing hugetlb_cgroup. Adding a new cgroup seemed like a lot of code duplication with hugetlb_cgroup. Keeping hugetlb related page counters under hugetlb_cgroup seemed cleaner as well. 2. Instead of adding a new counter, we considered adding a sysctl that modifies the behavior of hugetlb.xMB.[limit|usage]_in_bytes, to do accounting at reservation time rather than fault time. Adding a new page_counter seems better as userspace could, if it wants, choose to enforce different cgroups differently: one via limit_in_bytes, and another via reservation_limit_in_bytes. This could be very useful if you're transitioning how hugetlb memory is partitioned on your system one cgroup at a time, for example. Also, someone may find usage for both limit_in_bytes and reservation_limit_in_bytes concurrently, and this approach gives them the option to do so. Testing: - Added tests passing. - Used libhugetlbfs for regression testing. [1]: https://www.kernel.org/doc/html/latest/vm/hugetlbfs_reserv.html Signed-off-by: Mina Almasry Acked-by: David Rientjes --- Changes in v10: - Renamed reservation_* to resv.* --- include/linux/hugetlb.h | 4 +- mm/hugetlb_cgroup.c | 115 +++++++++++++++++++++++++++++++++++----- 2 files changed, 104 insertions(+), 15 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index 1e897e4168ac1..dea6143aa0685 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -432,8 +432,8 @@ struct hstate { unsigned int surplus_huge_pages_node[MAX_NUMNODES]; #ifdef CONFIG_CGROUP_HUGETLB /* cgroup control files */ - struct cftype cgroup_files_dfl[5]; - struct cftype cgroup_files_legacy[5]; + struct cftype cgroup_files_dfl[7]; + struct cftype cgroup_files_legacy[9]; #endif char name[HSTATE_NAME_LEN]; }; diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c index e434b05416c68..209f9b9604d34 100644 --- a/mm/hugetlb_cgroup.c +++ b/mm/hugetlb_cgroup.c @@ -36,6 +36,11 @@ struct hugetlb_cgroup { */ struct page_counter hugepage[HUGE_MAX_HSTATE]; + /* + * the counter to account for hugepage reservations from hugetlb. + */ + struct page_counter reserved_hugepage[HUGE_MAX_HSTATE]; + atomic_long_t events[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; atomic_long_t events_local[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; @@ -55,6 +60,14 @@ struct hugetlb_cgroup { static struct hugetlb_cgroup *root_h_cgroup __read_mostly; +static inline struct page_counter * +hugetlb_cgroup_get_counter(struct hugetlb_cgroup *h_cg, int idx, bool reserved) +{ + if (reserved) + return &h_cg->reserved_hugepage[idx]; + return &h_cg->hugepage[idx]; +} + static inline struct hugetlb_cgroup *hugetlb_cgroup_from_css(struct cgroup_subsys_state *s) { @@ -295,28 +308,42 @@ void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages, enum { RES_USAGE, + RES_RESERVATION_USAGE, RES_LIMIT, + RES_RESERVATION_LIMIT, RES_MAX_USAGE, + RES_RESERVATION_MAX_USAGE, RES_FAILCNT, + RES_RESERVATION_FAILCNT, }; static u64 hugetlb_cgroup_read_u64(struct cgroup_subsys_state *css, struct cftype *cft) { struct page_counter *counter; + struct page_counter *reserved_counter; struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(css); counter = &h_cg->hugepage[MEMFILE_IDX(cft->private)]; + reserved_counter = &h_cg->reserved_hugepage[MEMFILE_IDX(cft->private)]; switch (MEMFILE_ATTR(cft->private)) { case RES_USAGE: return (u64)page_counter_read(counter) * PAGE_SIZE; + case RES_RESERVATION_USAGE: + return (u64)page_counter_read(reserved_counter) * PAGE_SIZE; case RES_LIMIT: return (u64)counter->max * PAGE_SIZE; + case RES_RESERVATION_LIMIT: + return (u64)reserved_counter->max * PAGE_SIZE; case RES_MAX_USAGE: return (u64)counter->watermark * PAGE_SIZE; + case RES_RESERVATION_MAX_USAGE: + return (u64)reserved_counter->watermark * PAGE_SIZE; case RES_FAILCNT: return counter->failcnt; + case RES_RESERVATION_FAILCNT: + return reserved_counter->failcnt; default: BUG(); } @@ -338,10 +365,16 @@ static int hugetlb_cgroup_read_u64_max(struct seq_file *seq, void *v) 1 << huge_page_order(&hstates[idx])); switch (MEMFILE_ATTR(cft->private)) { + case RES_RESERVATION_USAGE: + counter = &h_cg->reserved_hugepage[idx]; + /* Fall through. */ case RES_USAGE: val = (u64)page_counter_read(counter); seq_printf(seq, "%llu\n", val * PAGE_SIZE); break; + case RES_RESERVATION_LIMIT: + counter = &h_cg->reserved_hugepage[idx]; + /* Fall through. */ case RES_LIMIT: val = (u64)counter->max; if (val == limit) @@ -365,6 +398,7 @@ static ssize_t hugetlb_cgroup_write(struct kernfs_open_file *of, int ret, idx; unsigned long nr_pages; struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of)); + bool reserved = false; if (hugetlb_cgroup_is_root(h_cg)) /* Can't set limit on root */ return -EINVAL; @@ -378,9 +412,14 @@ static ssize_t hugetlb_cgroup_write(struct kernfs_open_file *of, nr_pages = round_down(nr_pages, 1 << huge_page_order(&hstates[idx])); switch (MEMFILE_ATTR(of_cft(of)->private)) { + case RES_RESERVATION_LIMIT: + reserved = true; + /* Fall through. */ case RES_LIMIT: mutex_lock(&hugetlb_limit_mutex); - ret = page_counter_set_max(&h_cg->hugepage[idx], nr_pages); + ret = page_counter_set_max(hugetlb_cgroup_get_counter(h_cg, idx, + reserved), + nr_pages); mutex_unlock(&hugetlb_limit_mutex); break; default: @@ -406,18 +445,26 @@ static ssize_t hugetlb_cgroup_reset(struct kernfs_open_file *of, char *buf, size_t nbytes, loff_t off) { int ret = 0; - struct page_counter *counter; + struct page_counter *counter, *reserved_counter; struct hugetlb_cgroup *h_cg = hugetlb_cgroup_from_css(of_css(of)); counter = &h_cg->hugepage[MEMFILE_IDX(of_cft(of)->private)]; + reserved_counter = + &h_cg->reserved_hugepage[MEMFILE_IDX(of_cft(of)->private)]; switch (MEMFILE_ATTR(of_cft(of)->private)) { case RES_MAX_USAGE: page_counter_reset_watermark(counter); break; + case RES_RESERVATION_MAX_USAGE: + page_counter_reset_watermark(reserved_counter); + break; case RES_FAILCNT: counter->failcnt = 0; break; + case RES_RESERVATION_FAILCNT: + reserved_counter->failcnt = 0; + break; default: ret = -EINVAL; break; @@ -472,7 +519,7 @@ static void __init __hugetlb_cgroup_file_dfl_init(int idx) struct hstate *h = &hstates[idx]; /* format the size */ - mem_fmt(buf, 32, huge_page_size(h)); + mem_fmt(buf, sizeof(buf), huge_page_size(h)); /* Add the limit file */ cft = &h->cgroup_files_dfl[0]; @@ -482,15 +529,30 @@ static void __init __hugetlb_cgroup_file_dfl_init(int idx) cft->write = hugetlb_cgroup_write_dfl; cft->flags = CFTYPE_NOT_ON_ROOT; - /* Add the current usage file */ + /* Add the reservation limit file */ cft = &h->cgroup_files_dfl[1]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.max", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_LIMIT); + cft->seq_show = hugetlb_cgroup_read_u64_max; + cft->write = hugetlb_cgroup_write_dfl; + cft->flags = CFTYPE_NOT_ON_ROOT; + + /* Add the current usage file */ + cft = &h->cgroup_files_dfl[2]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.current", buf); cft->private = MEMFILE_PRIVATE(idx, RES_USAGE); cft->seq_show = hugetlb_cgroup_read_u64_max; cft->flags = CFTYPE_NOT_ON_ROOT; + /* Add the current reservation usage file */ + cft = &h->cgroup_files_dfl[3]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.current", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_USAGE); + cft->seq_show = hugetlb_cgroup_read_u64_max; + cft->flags = CFTYPE_NOT_ON_ROOT; + /* Add the events file */ - cft = &h->cgroup_files_dfl[2]; + cft = &h->cgroup_files_dfl[4]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.events", buf); cft->private = MEMFILE_PRIVATE(idx, 0); cft->seq_show = hugetlb_events_show; @@ -498,7 +560,7 @@ static void __init __hugetlb_cgroup_file_dfl_init(int idx) cft->flags = CFTYPE_NOT_ON_ROOT; /* Add the events.local file */ - cft = &h->cgroup_files_dfl[3]; + cft = &h->cgroup_files_dfl[5]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.events.local", buf); cft->private = MEMFILE_PRIVATE(idx, 0); cft->seq_show = hugetlb_events_local_show; @@ -507,7 +569,7 @@ static void __init __hugetlb_cgroup_file_dfl_init(int idx) cft->flags = CFTYPE_NOT_ON_ROOT; /* NULL terminate the last cft */ - cft = &h->cgroup_files_dfl[4]; + cft = &h->cgroup_files_dfl[6]; memset(cft, 0, sizeof(*cft)); WARN_ON(cgroup_add_dfl_cftypes(&hugetlb_cgrp_subsys, @@ -521,7 +583,7 @@ static void __init __hugetlb_cgroup_file_legacy_init(int idx) struct hstate *h = &hstates[idx]; /* format the size */ - mem_fmt(buf, 32, huge_page_size(h)); + mem_fmt(buf, sizeof(buf), huge_page_size(h)); /* Add the limit file */ cft = &h->cgroup_files_legacy[0]; @@ -530,28 +592,55 @@ static void __init __hugetlb_cgroup_file_legacy_init(int idx) cft->read_u64 = hugetlb_cgroup_read_u64; cft->write = hugetlb_cgroup_write_legacy; - /* Add the usage file */ + /* Add the reservation limit file */ cft = &h->cgroup_files_legacy[1]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.limit_in_bytes", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_LIMIT); + cft->read_u64 = hugetlb_cgroup_read_u64; + cft->write = hugetlb_cgroup_write_legacy; + + /* Add the usage file */ + cft = &h->cgroup_files_legacy[2]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.usage_in_bytes", buf); cft->private = MEMFILE_PRIVATE(idx, RES_USAGE); cft->read_u64 = hugetlb_cgroup_read_u64; + /* Add the reservation usage file */ + cft = &h->cgroup_files_legacy[3]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.usage_in_bytes", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_USAGE); + cft->read_u64 = hugetlb_cgroup_read_u64; + /* Add the MAX usage file */ - cft = &h->cgroup_files_legacy[2]; + cft = &h->cgroup_files_legacy[4]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.max_usage_in_bytes", buf); cft->private = MEMFILE_PRIVATE(idx, RES_MAX_USAGE); cft->write = hugetlb_cgroup_reset; cft->read_u64 = hugetlb_cgroup_read_u64; + /* Add the MAX reservation usage file */ + cft = &h->cgroup_files_legacy[5]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.max_usage_in_bytes", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_MAX_USAGE); + cft->write = hugetlb_cgroup_reset; + cft->read_u64 = hugetlb_cgroup_read_u64; + /* Add the failcntfile */ - cft = &h->cgroup_files_legacy[3]; + cft = &h->cgroup_files_legacy[6]; snprintf(cft->name, MAX_CFTYPE_NAME, "%s.failcnt", buf); - cft->private = MEMFILE_PRIVATE(idx, RES_FAILCNT); + cft->private = MEMFILE_PRIVATE(idx, RES_FAILCNT); + cft->write = hugetlb_cgroup_reset; + cft->read_u64 = hugetlb_cgroup_read_u64; + + /* Add the reservation failcntfile */ + cft = &h->cgroup_files_legacy[7]; + snprintf(cft->name, MAX_CFTYPE_NAME, "%s.resv.failcnt", buf); + cft->private = MEMFILE_PRIVATE(idx, RES_RESERVATION_FAILCNT); cft->write = hugetlb_cgroup_reset; cft->read_u64 = hugetlb_cgroup_read_u64; /* NULL terminate the last cft */ - cft = &h->cgroup_files_legacy[4]; + cft = &h->cgroup_files_legacy[8]; memset(cft, 0, sizeof(*cft)); WARN_ON(cgroup_add_legacy_cftypes(&hugetlb_cgrp_subsys, From patchwork Wed Jan 15 01:26:45 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333071 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id ED023186E for ; Wed, 15 Jan 2020 01:27:27 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id CB0DC2468B for ; Wed, 15 Jan 2020 01:27:27 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="lGs4BLNH" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728882AbgAOB07 (ORCPT ); Tue, 14 Jan 2020 20:26:59 -0500 Received: from mail-pl1-f201.google.com ([209.85.214.201]:39386 "EHLO mail-pl1-f201.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728915AbgAOB06 (ORCPT ); Tue, 14 Jan 2020 20:26:58 -0500 Received: by mail-pl1-f201.google.com with SMTP id p15so937357plq.6 for ; Tue, 14 Jan 2020 17:26:57 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=VD9B9XQiTIZ/I3GgP2atJmksN68hbJNGgZeyi4+Kld8=; b=lGs4BLNH5UJp8PirX+0cIBqHW95lTXQIuPY018sNrBns1BSssPpXT3HuuO7QeQGuIe u9Q/LVnpf1MptbB+AzYedmLDV26mXSRP8MpWL9lTvLGWMGIbRwOGxSxnCBMk7sd+o4yo s2Zay07dkVOJQeRnbeIva9zpUHc1dkmAq8FzgR6QiQ46BAqfCR0kZEteEgb5gRb6U9oz WllB3SFiOGY/FGJCcJ5dmT553Kfl75iBs9yhf92PpZWZc/8Q5FtErBGT/5i4DdpAiMNI 3/B44ps4npXEDY0hN7Vl2HehrtsCkNx+S7DP4joBNKGGqr/75S+33NtwDRyTJNWWN4my NZtg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=VD9B9XQiTIZ/I3GgP2atJmksN68hbJNGgZeyi4+Kld8=; b=ozrrorm+xZKoKjx0dxkoyIcHVXLaVKRWPhDpSdkYuH9yOKXF0ytTjMq3qoqMBLE7Oh MeHwCXLc1BktbRpvW1L15Pgn2MNFCbs2v10oQVOMVYGGqZ+0U+I0cut8v3d3XJgOyhvL pkkviawdGDqlsSyWjNUx6N8/kxwllhzhaxinXjeuQ2HK5Ppde0M5jggLjjYSKuoCvzFM jLUCc26WBsnVos8jM7LX54KXGh6JTzVB5b6zOR7MKP8ztLfppNYgnXX7SWPCYEbXTCK3 Lw3fcA8qK3LSIbyxQxdTdd5Hd49uUVUSdFmNS45s9/dfOcjnRi9p8I8WUFGCUo4tWpdT pttQ== X-Gm-Message-State: APjAAAWPUFYwhtbBrUz/eP56Y4MsZJpxyNk6nPaPk05jLvPeoAX/oBOx aiAtpvAXBqASdjJPaizzpOJ0e5Sac/PzCXWduw== X-Google-Smtp-Source: APXvYqy8wu+rN6XIvdPZvljPhhlginlJ4IfOUPOeV7wb2cEpfAUMfQvkqZxabEz78XjhuB2+EONpJS7WCBe000q0UA== X-Received: by 2002:a63:e84d:: with SMTP id a13mr31663837pgk.274.1579051617276; Tue, 14 Jan 2020 17:26:57 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:45 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-2-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 2/8] hugetlb_cgroup: add interface for charge/uncharge hugetlb reservations From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org Augments hugetlb_cgroup_charge_cgroup to be able to charge hugetlb usage or hugetlb reservation counter. Adds a new interface to uncharge a hugetlb_cgroup counter via hugetlb_cgroup_uncharge_counter. Integrates the counter with hugetlb_cgroup, via hugetlb_cgroup_init, hugetlb_cgroup_have_usage, and hugetlb_cgroup_css_offline. Signed-off-by: Mina Almasry Acked-by: Mike Kravetz Acked-by: David Rientjes --- Changes in v10: - Added missing VM_BUG_ON Changes in V9: - Fixed HUGETLB_CGROUP_MIN_ORDER. - Minor variable name update. - Moved some init/cleanup code from later patches in the series to this patch. - Updated reparenting of reservation accounting. --- include/linux/hugetlb_cgroup.h | 68 ++++++++++++++--------- mm/hugetlb.c | 19 ++++--- mm/hugetlb_cgroup.c | 99 +++++++++++++++++++++++++--------- 3 files changed, 128 insertions(+), 58 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/include/linux/hugetlb_cgroup.h b/include/linux/hugetlb_cgroup.h index 063962f6dfc6a..eab8a70d5bcb5 100644 --- a/include/linux/hugetlb_cgroup.h +++ b/include/linux/hugetlb_cgroup.h @@ -20,29 +20,37 @@ struct hugetlb_cgroup; /* * Minimum page order trackable by hugetlb cgroup. - * At least 3 pages are necessary for all the tracking information. + * At least 4 pages are necessary for all the tracking information. */ #define HUGETLB_CGROUP_MIN_ORDER 2 #ifdef CONFIG_CGROUP_HUGETLB -static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page) +static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page, + bool reserved) { VM_BUG_ON_PAGE(!PageHuge(page), page); if (compound_order(page) < HUGETLB_CGROUP_MIN_ORDER) return NULL; - return (struct hugetlb_cgroup *)page[2].private; + if (reserved) + return (struct hugetlb_cgroup *)page[3].private; + else + return (struct hugetlb_cgroup *)page[2].private; } -static inline -int set_hugetlb_cgroup(struct page *page, struct hugetlb_cgroup *h_cg) +static inline int set_hugetlb_cgroup(struct page *page, + struct hugetlb_cgroup *h_cg, + bool reservation) { VM_BUG_ON_PAGE(!PageHuge(page), page); if (compound_order(page) < HUGETLB_CGROUP_MIN_ORDER) return -1; - page[2].private = (unsigned long)h_cg; + if (reservation) + page[3].private = (unsigned long)h_cg; + else + page[2].private = (unsigned long)h_cg; return 0; } @@ -52,26 +60,34 @@ static inline bool hugetlb_cgroup_disabled(void) } extern int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup **ptr); + struct hugetlb_cgroup **ptr, + bool reserved); extern void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages, struct hugetlb_cgroup *h_cg, - struct page *page); + struct page *page, bool reserved); extern void hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages, - struct page *page); + struct page *page, bool reserved); + extern void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup *h_cg); + struct hugetlb_cgroup *h_cg, + bool reserved); +extern void hugetlb_cgroup_uncharge_counter(struct page_counter *p, + unsigned long nr_pages, + struct cgroup_subsys_state *css); + extern void hugetlb_cgroup_file_init(void) __init; extern void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage); #else -static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page) +static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page, + bool reserved) { return NULL; } -static inline -int set_hugetlb_cgroup(struct page *page, struct hugetlb_cgroup *h_cg) +static inline int set_hugetlb_cgroup(struct page *page, + struct hugetlb_cgroup *h_cg, bool reserved) { return 0; } @@ -81,28 +97,30 @@ static inline bool hugetlb_cgroup_disabled(void) return true; } -static inline int -hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup **ptr) +static inline int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, + struct hugetlb_cgroup **ptr, + bool reserved) { return 0; } -static inline void -hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages, - struct hugetlb_cgroup *h_cg, - struct page *page) +static inline void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages, + struct hugetlb_cgroup *h_cg, + struct page *page, + bool reserved) { } -static inline void -hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages, struct page *page) +static inline void hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages, + struct page *page, + bool reserved) { } -static inline void -hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup *h_cg) +static inline void hugetlb_cgroup_uncharge_cgroup(int idx, + unsigned long nr_pages, + struct hugetlb_cgroup *h_cg, + bool reserved) { } diff --git a/mm/hugetlb.c b/mm/hugetlb.c index dd8737a94bec4..62a4cf3db4090 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -1068,7 +1068,8 @@ static void update_and_free_page(struct hstate *h, struct page *page) 1 << PG_active | 1 << PG_private | 1 << PG_writeback); } - VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page), page); + VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page, false), page); + VM_BUG_ON_PAGE(hugetlb_cgroup_from_page(page, true), page); set_compound_page_dtor(page, NULL_COMPOUND_DTOR); set_page_refcounted(page); if (hstate_is_gigantic(h)) { @@ -1178,8 +1179,8 @@ static void __free_huge_page(struct page *page) spin_lock(&hugetlb_lock); clear_page_huge_active(page); - hugetlb_cgroup_uncharge_page(hstate_index(h), - pages_per_huge_page(h), page); + hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), + page, false); if (restore_reserve) h->resv_huge_pages++; @@ -1253,7 +1254,8 @@ static void prep_new_huge_page(struct hstate *h, struct page *page, int nid) INIT_LIST_HEAD(&page->lru); set_compound_page_dtor(page, HUGETLB_PAGE_DTOR); spin_lock(&hugetlb_lock); - set_hugetlb_cgroup(page, NULL); + set_hugetlb_cgroup(page, NULL, false); + set_hugetlb_cgroup(page, NULL, true); h->nr_huge_pages++; h->nr_huge_pages_node[nid]++; spin_unlock(&hugetlb_lock); @@ -2039,7 +2041,8 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, gbl_chg = 1; } - ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg); + ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg, + false); if (ret) goto out_subpool_put; @@ -2063,7 +2066,8 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, list_move(&page->lru, &h->hugepage_activelist); /* Fall through */ } - hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page); + hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page, + false); spin_unlock(&hugetlb_lock); set_page_private(page, (unsigned long)spool); @@ -2087,7 +2091,8 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, return page; out_uncharge_cgroup: - hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg); + hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg, + false); out_subpool_put: if (map_chg || avoid_reserve) hugepage_subpool_put_pages(spool, 1); diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c index 209f9b9604d34..c434f69f38354 100644 --- a/mm/hugetlb_cgroup.c +++ b/mm/hugetlb_cgroup.c @@ -96,8 +96,12 @@ static inline bool hugetlb_cgroup_have_usage(struct hugetlb_cgroup *h_cg) int idx; for (idx = 0; idx < hugetlb_max_hstate; idx++) { - if (page_counter_read(&h_cg->hugepage[idx])) + if (page_counter_read( + hugetlb_cgroup_get_counter(h_cg, idx, true)) || + page_counter_read( + hugetlb_cgroup_get_counter(h_cg, idx, false))) { return true; + } } return false; } @@ -108,18 +112,33 @@ static void hugetlb_cgroup_init(struct hugetlb_cgroup *h_cgroup, int idx; for (idx = 0; idx < HUGE_MAX_HSTATE; idx++) { - struct page_counter *counter = &h_cgroup->hugepage[idx]; - struct page_counter *parent = NULL; + struct page_counter *fault_parent = NULL; + struct page_counter *reserved_parent = NULL; unsigned long limit; int ret; - if (parent_h_cgroup) - parent = &parent_h_cgroup->hugepage[idx]; - page_counter_init(counter, parent); + if (parent_h_cgroup) { + fault_parent = hugetlb_cgroup_get_counter( + parent_h_cgroup, idx, false); + reserved_parent = hugetlb_cgroup_get_counter( + parent_h_cgroup, idx, true); + } + page_counter_init(hugetlb_cgroup_get_counter(h_cgroup, idx, + false), + fault_parent); + page_counter_init(hugetlb_cgroup_get_counter(h_cgroup, idx, + true), + reserved_parent); limit = round_down(PAGE_COUNTER_MAX, 1 << huge_page_order(&hstates[idx])); - ret = page_counter_set_max(counter, limit); + + ret = page_counter_set_max( + hugetlb_cgroup_get_counter(h_cgroup, idx, false), + limit); + VM_BUG_ON(ret); + ret = page_counter_set_max( + hugetlb_cgroup_get_counter(h_cgroup, idx, true), limit); VM_BUG_ON(ret); } } @@ -149,7 +168,6 @@ static void hugetlb_cgroup_css_free(struct cgroup_subsys_state *css) kfree(h_cgroup); } - /* * Should be called with hugetlb_lock held. * Since we are holding hugetlb_lock, pages cannot get moved from @@ -165,7 +183,7 @@ static void hugetlb_cgroup_move_parent(int idx, struct hugetlb_cgroup *h_cg, struct hugetlb_cgroup *page_hcg; struct hugetlb_cgroup *parent = parent_hugetlb_cgroup(h_cg); - page_hcg = hugetlb_cgroup_from_page(page); + page_hcg = hugetlb_cgroup_from_page(page, false); /* * We can have pages in active list without any cgroup * ie, hugepage with less than 3 pages. We can safely @@ -184,7 +202,7 @@ static void hugetlb_cgroup_move_parent(int idx, struct hugetlb_cgroup *h_cg, /* Take the pages off the local counter */ page_counter_cancel(counter, nr_pages); - set_hugetlb_cgroup(page, parent); + set_hugetlb_cgroup(page, parent, false); out: return; } @@ -227,7 +245,7 @@ static inline void hugetlb_event(struct hugetlb_cgroup *hugetlb, int idx, } int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup **ptr) + struct hugetlb_cgroup **ptr, bool reserved) { int ret = 0; struct page_counter *counter; @@ -250,13 +268,20 @@ int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, } rcu_read_unlock(); - if (!page_counter_try_charge(&h_cg->hugepage[idx], nr_pages, - &counter)) { + if (!page_counter_try_charge(hugetlb_cgroup_get_counter(h_cg, idx, + reserved), + nr_pages, &counter)) { ret = -ENOMEM; hugetlb_event(hugetlb_cgroup_from_counter(counter, idx), idx, HUGETLB_MAX); + css_put(&h_cg->css); + goto done; } - css_put(&h_cg->css); + /* Reservations take a reference to the css because they do not get + * reparented. + */ + if (!reserved) + css_put(&h_cg->css); done: *ptr = h_cg; return ret; @@ -265,12 +290,12 @@ int hugetlb_cgroup_charge_cgroup(int idx, unsigned long nr_pages, /* Should be called with hugetlb_lock held */ void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages, struct hugetlb_cgroup *h_cg, - struct page *page) + struct page *page, bool reserved) { if (hugetlb_cgroup_disabled() || !h_cg) return; - set_hugetlb_cgroup(page, h_cg); + set_hugetlb_cgroup(page, h_cg, reserved); return; } @@ -278,23 +303,29 @@ void hugetlb_cgroup_commit_charge(int idx, unsigned long nr_pages, * Should be called with hugetlb_lock held */ void hugetlb_cgroup_uncharge_page(int idx, unsigned long nr_pages, - struct page *page) + struct page *page, bool reserved) { struct hugetlb_cgroup *h_cg; if (hugetlb_cgroup_disabled()) return; lockdep_assert_held(&hugetlb_lock); - h_cg = hugetlb_cgroup_from_page(page); + h_cg = hugetlb_cgroup_from_page(page, reserved); if (unlikely(!h_cg)) return; - set_hugetlb_cgroup(page, NULL); - page_counter_uncharge(&h_cg->hugepage[idx], nr_pages); + set_hugetlb_cgroup(page, NULL, reserved); + + page_counter_uncharge(hugetlb_cgroup_get_counter(h_cg, idx, reserved), + nr_pages); + + if (reserved) + css_put(&h_cg->css); + return; } void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages, - struct hugetlb_cgroup *h_cg) + struct hugetlb_cgroup *h_cg, bool reserved) { if (hugetlb_cgroup_disabled() || !h_cg) return; @@ -302,8 +333,22 @@ void hugetlb_cgroup_uncharge_cgroup(int idx, unsigned long nr_pages, if (huge_page_order(&hstates[idx]) < HUGETLB_CGROUP_MIN_ORDER) return; - page_counter_uncharge(&h_cg->hugepage[idx], nr_pages); - return; + page_counter_uncharge(hugetlb_cgroup_get_counter(h_cg, idx, reserved), + nr_pages); + + if (reserved) + css_put(&h_cg->css); +} + +void hugetlb_cgroup_uncharge_counter(struct page_counter *p, + unsigned long nr_pages, + struct cgroup_subsys_state *css) +{ + if (hugetlb_cgroup_disabled() || !p || !css) + return; + + page_counter_uncharge(p, nr_pages); + css_put(css); } enum { @@ -675,6 +720,7 @@ void __init hugetlb_cgroup_file_init(void) void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage) { struct hugetlb_cgroup *h_cg; + struct hugetlb_cgroup *h_cg_reservation; struct hstate *h = page_hstate(oldhpage); if (hugetlb_cgroup_disabled()) @@ -682,11 +728,12 @@ void hugetlb_cgroup_migrate(struct page *oldhpage, struct page *newhpage) VM_BUG_ON_PAGE(!PageHuge(oldhpage), oldhpage); spin_lock(&hugetlb_lock); - h_cg = hugetlb_cgroup_from_page(oldhpage); - set_hugetlb_cgroup(oldhpage, NULL); + h_cg = hugetlb_cgroup_from_page(oldhpage, false); + h_cg_reservation = hugetlb_cgroup_from_page(oldhpage, true); + set_hugetlb_cgroup(oldhpage, NULL, false); /* move the h_cg details to new cgroup */ - set_hugetlb_cgroup(newhpage, h_cg); + set_hugetlb_cgroup(newhpage, h_cg_reservation, true); list_move(&newhpage->lru, &h->hugepage_activelist); spin_unlock(&hugetlb_lock); return; From patchwork Wed Jan 15 01:26:46 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333069 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id BE95514B4 for ; Wed, 15 Jan 2020 01:27:27 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 938032467A for ; Wed, 15 Jan 2020 01:27:27 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="aQmn7cGy" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728890AbgAOB1C (ORCPT ); Tue, 14 Jan 2020 20:27:02 -0500 Received: from mail-ua1-f74.google.com ([209.85.222.74]:37911 "EHLO mail-ua1-f74.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728949AbgAOB1B (ORCPT ); Tue, 14 Jan 2020 20:27:01 -0500 Received: by mail-ua1-f74.google.com with SMTP id i16so1491621ual.5 for ; Tue, 14 Jan 2020 17:27:00 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=Kg+p/VmE2ROhKV/f4JfIzn9grk12cA50+GiyFmuccAU=; b=aQmn7cGyi4DTCWWDs/8MKLAnnwva8Ubb5n+C5BOd3SmYTQ9GvqFJGRc9/m0UDfBAUo egHMx514bdLFhyEiqHW3Z6seaUi1zIef7UrcDyE/3fiNXQla98SmA45FRe2iHeVwvAio PVSpj49XmlCQ4CrADmBE4g4SdQ5Kbzt73FH6yn5FvJdgXy3ETfbWjP4Rh2Pl/0326NRJ UpkwGmw5oLVNWMtH2KBOarSK1YxJheBtmQGEe0zctVuswXKNxBWM6DeX+IgYeM7oC/Gm /v9NF8x7Zx+XkVCLhCSh4E/+NhtJE3DTs5SlkNOb4Y50DYO+aXv2qjlcJ9qoREboroH2 1UoA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=Kg+p/VmE2ROhKV/f4JfIzn9grk12cA50+GiyFmuccAU=; b=KUNk9vvjjc1WlJsNcplp9gtoz1+DhL+XDoRlYf1443OFdSLz6Pw5M6sAySojkhi0JE EJJZSnv3ERoSgYBl81Kf7kvdcEpCFoOIxZBNfT278To6Eb6Mhl3hdNNbyDsbD7U/gHeo lKvBuJYzl1AmYY4zGK6ZHl0UOWRxpGeho4FmaCsvGUdHubKFcgevx7WdRc6pA0IovHtV fId3ouGnP/5pnq3UaxUiHFwtfQlzHWn8+jNO99UktdedGsVqa5EgnHoSuJWcdwI2StQm kXOpew14YKCcQw69m2euEhIlenw65fYQA5ekZmJP6XBMfOPX7DlNSLMzmk/ySfN1X03i uJjA== X-Gm-Message-State: APjAAAW1Jjc4QIyAI75MrOadBUPiU/6FQfMx1ERvvBXaojOCkEuluz8J qKZst5GunubQrW1nrWikUbbTOvUYCwOgsQ89ZA== X-Google-Smtp-Source: APXvYqzxDknpwcDc/7ZohHTw0rZ9c3Ju59PGISjRs0mxtbwIEst6J/yZEzTUCBawy1GQDoNg+Mk/LOZfZ23dttLoAw== X-Received: by 2002:ab0:7450:: with SMTP id p16mr16108545uaq.20.1579051620011; Tue, 14 Jan 2020 17:27:00 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:46 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-3-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 3/8] hugetlb_cgroup: add reservation accounting for private mappings From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org Normally the pointer to the cgroup to uncharge hangs off the struct page, and gets queried when it's time to free the page. With hugetlb_cgroup reservations, this is not possible. Because it's possible for a page to be reserved by one task and actually faulted in by another task. The best place to put the hugetlb_cgroup pointer to uncharge for reservations is in the resv_map. But, because the resv_map has different semantics for private and shared mappings, the code patch to charge/uncharge shared and private mappings is different. This patch implements charging and uncharging for private mappings. For private mappings, the counter to uncharge is in resv_map->reservation_counter. On initializing the resv_map this is set to NULL. On reservation of a region in private mapping, the tasks hugetlb_cgroup is charged and the hugetlb_cgroup is placed is resv_map->reservation_counter. On hugetlb_vm_op_close, we uncharge resv_map->reservation_counter. Signed-off-by: Mina Almasry Reviewed-by: Mike Kravetz Acked-by: David Rientjes --- Changes in v10: - Fixed cases where the mapping is private but cgroup accounting is disabled. Changes in V9: - Updated for reparenting of hugetlb reservation accounting. --- include/linux/hugetlb.h | 10 +++++++++ include/linux/hugetlb_cgroup.h | 27 ++++++++++++++++++++++++ mm/hugetlb.c | 38 +++++++++++++++++++++++++++++++++- mm/hugetlb_cgroup.c | 28 ------------------------- 4 files changed, 74 insertions(+), 29 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h index dea6143aa0685..5491932ea5758 100644 --- a/include/linux/hugetlb.h +++ b/include/linux/hugetlb.h @@ -46,6 +46,16 @@ struct resv_map { long adds_in_progress; struct list_head region_cache; long region_cache_count; +#ifdef CONFIG_CGROUP_HUGETLB + /* + * On private mappings, the counter to uncharge reservations is stored + * here. If these fields are 0, then either the mapping is shared, or + * cgroup accounting is disabled for this resv_map. + */ + struct page_counter *reservation_counter; + unsigned long pages_per_hpage; + struct cgroup_subsys_state *css; +#endif }; extern struct resv_map *resv_map_alloc(void); void resv_map_release(struct kref *ref); diff --git a/include/linux/hugetlb_cgroup.h b/include/linux/hugetlb_cgroup.h index eab8a70d5bcb5..8c320accefe87 100644 --- a/include/linux/hugetlb_cgroup.h +++ b/include/linux/hugetlb_cgroup.h @@ -25,6 +25,33 @@ struct hugetlb_cgroup; #define HUGETLB_CGROUP_MIN_ORDER 2 #ifdef CONFIG_CGROUP_HUGETLB +enum hugetlb_memory_event { + HUGETLB_MAX, + HUGETLB_NR_MEMORY_EVENTS, +}; + +struct hugetlb_cgroup { + struct cgroup_subsys_state css; + + /* + * the counter to account for hugepages from hugetlb. + */ + struct page_counter hugepage[HUGE_MAX_HSTATE]; + + /* + * the counter to account for hugepage reservations from hugetlb. + */ + struct page_counter reserved_hugepage[HUGE_MAX_HSTATE]; + + atomic_long_t events[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; + atomic_long_t events_local[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; + + /* Handle for "hugetlb.events" */ + struct cgroup_file events_file[HUGE_MAX_HSTATE]; + + /* Handle for "hugetlb.events.local" */ + struct cgroup_file events_local_file[HUGE_MAX_HSTATE]; +}; static inline struct hugetlb_cgroup *hugetlb_cgroup_from_page(struct page *page, bool reserved) diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 62a4cf3db4090..f1b63946ee95c 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -666,6 +666,17 @@ struct resv_map *resv_map_alloc(void) INIT_LIST_HEAD(&resv_map->regions); resv_map->adds_in_progress = 0; +#ifdef CONFIG_CGROUP_HUGETLB + /* + * Initialize these to 0. On shared mappings, 0's here indicate these + * fields don't do cgroup accounting. On private mappings, these will be + * re-initialized to the proper values, to indicate that hugetlb cgroup + * reservations are to be un-charged from here. + */ + resv_map->reservation_counter = NULL; + resv_map->pages_per_hpage = 0; + resv_map->css = NULL; +#endif INIT_LIST_HEAD(&resv_map->region_cache); list_add(&rg->link, &resv_map->region_cache); @@ -3194,7 +3205,11 @@ static void hugetlb_vm_op_close(struct vm_area_struct *vma) reserve = (end - start) - region_count(resv, start, end); - kref_put(&resv->refs, resv_map_release); +#ifdef CONFIG_CGROUP_HUGETLB + hugetlb_cgroup_uncharge_counter(resv->reservation_counter, + (end - start) * resv->pages_per_hpage, + resv->css); +#endif if (reserve) { /* @@ -3204,6 +3219,8 @@ static void hugetlb_vm_op_close(struct vm_area_struct *vma) gbl_reserve = hugepage_subpool_put_pages(spool, reserve); hugetlb_acct_memory(h, -gbl_reserve); } + + kref_put(&resv->refs, resv_map_release); } static int hugetlb_vm_op_split(struct vm_area_struct *vma, unsigned long addr) @@ -4550,6 +4567,7 @@ int hugetlb_reserve_pages(struct inode *inode, struct hstate *h = hstate_inode(inode); struct hugepage_subpool *spool = subpool_inode(inode); struct resv_map *resv_map; + struct hugetlb_cgroup *h_cg; long gbl_reserve; /* This should never happen */ @@ -4583,12 +4601,30 @@ int hugetlb_reserve_pages(struct inode *inode, chg = region_chg(resv_map, from, to); } else { + /* Private mapping. */ resv_map = resv_map_alloc(); if (!resv_map) return -ENOMEM; chg = to - from; + if (hugetlb_cgroup_charge_cgroup(hstate_index(h), + chg * pages_per_huge_page(h), + &h_cg, true)) { + kref_put(&resv_map->refs, resv_map_release); + return -ENOMEM; + } + +#ifdef CONFIG_CGROUP_HUGETLB + /* + * Since this branch handles private mappings, we attach the + * counter to uncharge for this reservation off resv_map. + */ + resv_map->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + resv_map->pages_per_hpage = pages_per_huge_page(h); +#endif + set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } diff --git a/mm/hugetlb_cgroup.c b/mm/hugetlb_cgroup.c index c434f69f38354..ddfdf19a9ad35 100644 --- a/mm/hugetlb_cgroup.c +++ b/mm/hugetlb_cgroup.c @@ -23,34 +23,6 @@ #include #include -enum hugetlb_memory_event { - HUGETLB_MAX, - HUGETLB_NR_MEMORY_EVENTS, -}; - -struct hugetlb_cgroup { - struct cgroup_subsys_state css; - - /* - * the counter to account for hugepages from hugetlb. - */ - struct page_counter hugepage[HUGE_MAX_HSTATE]; - - /* - * the counter to account for hugepage reservations from hugetlb. - */ - struct page_counter reserved_hugepage[HUGE_MAX_HSTATE]; - - atomic_long_t events[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; - atomic_long_t events_local[HUGE_MAX_HSTATE][HUGETLB_NR_MEMORY_EVENTS]; - - /* Handle for "hugetlb.events" */ - struct cgroup_file events_file[HUGE_MAX_HSTATE]; - - /* Handle for "hugetlb.events.local" */ - struct cgroup_file events_local_file[HUGE_MAX_HSTATE]; -}; - #define MEMFILE_PRIVATE(x, val) (((x) << 16) | (val)) #define MEMFILE_IDX(val) (((val) >> 16) & 0xffff) #define MEMFILE_ATTR(val) ((val) & 0xffff) From patchwork Wed Jan 15 01:26:47 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333067 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 593D714B4 for ; Wed, 15 Jan 2020 01:27:25 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 198B924685 for ; Wed, 15 Jan 2020 01:27:25 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="DLGiyQ3W" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728894AbgAOB1Y (ORCPT ); Tue, 14 Jan 2020 20:27:24 -0500 Received: from mail-pf1-f202.google.com ([209.85.210.202]:52866 "EHLO mail-pf1-f202.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1728988AbgAOB1D (ORCPT ); Tue, 14 Jan 2020 20:27:03 -0500 Received: by mail-pf1-f202.google.com with SMTP id 145so9933126pfx.19 for ; Tue, 14 Jan 2020 17:27:02 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=/xscfjBmff5JrVk2WzWVkCtQEIejbU+Dlihd/yDhdmo=; b=DLGiyQ3W5hFqlGsXB1DbkIIohGB0aKZBwD2zT3WXCGKx28+BfhxwQgXSzUCacMnE91 R8dhu73d/s9iRkjVWKQ0mx9tdV/MiKv/teCcNM+73S572HK4VZW0BivhI0xyUktXdUKk Lk+R9PkJCPhKUU/jf5znw5WQvGZ0VkZbLHWJM5SUF0y+ColKbxTToqwh6hMADafuFUua CG58dcaFd295jgx1s7WXaaceAlL8m7ZEbT416wvsSRHc/MMmnMLkY4mjrp7SojUeZaei 20BohH5J6v2dMOpW47yreqce78/e4o4zrkN6/MvPZVAZhVYv7obETJpbMbWoyUjzvjTq Pxqg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=/xscfjBmff5JrVk2WzWVkCtQEIejbU+Dlihd/yDhdmo=; b=fFjnT7PQvtYELbj1bevmbbQx37u3336sxgCv13AKP41EiCB+/L83Ns77lc85WR3gT2 6iEWqIwvZedtosaw1VBBsmwg0lG7kIvvVGE6JFo2zYn0c15agZjqFKdbUum2G/4krDCM oF6ELX+2XJuJQ6P6LvMRz8J+Tt3w1Co4Lv7LTy5nC+exzbtzClrsB4WwzIp+xL91ghpG 6B7CONnmsHHUm5Q/yUZwGs4vwjlMhQ9IovzGJSmz2xfjbiEy9IE80KD2Q9T08whLvLU2 oTOKWnWdcPP3czTTaz7jyLdy/aDgYdJsIBdJz2yqcOCiEEolSLNPZq5OVlc/7odPhs67 e8ug== X-Gm-Message-State: APjAAAUWG9IWk8GREF2E1yagaGsa3FcDkgsc+LtrMUBlnohvN/kb5Y+H j541GfVd0MXtjE0wGJsdP0OwlGPPRy7hrtRw2w== X-Google-Smtp-Source: APXvYqyEo4pbySguc1WgtPlpiTL5DPJW0pPTOESC6K3DXPCJ8u5VwXBLbXSNlucqH+lC8lOt7ZuSN01aeJqk6TQ7Og== X-Received: by 2002:a65:5786:: with SMTP id b6mr30489168pgr.316.1579051622257; Tue, 14 Jan 2020 17:27:02 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:47 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-4-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 4/8] hugetlb: disable region_add file_region coalescing From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org A follow up patch in this series adds hugetlb cgroup uncharge info the file_region entries in resv->regions. The cgroup uncharge info may differ for different regions, so they can no longer be coalesced at region_add time. So, disable region coalescing in region_add in this patch. Behavior change: Say a resv_map exists like this [0->1], [2->3], and [5->6]. Then a region_chg/add call comes in region_chg/add(f=0, t=5). Old code would generate resv->regions: [0->5], [5->6]. New code would generate resv->regions: [0->1], [1->2], [2->3], [3->5], [5->6]. Special care needs to be taken to handle the resv->adds_in_progress variable correctly. In the past, only 1 region would be added for every region_chg and region_add call. But now, each call may add multiple regions, so we can no longer increment adds_in_progress by 1 in region_chg, or decrement adds_in_progress by 1 after region_add or region_abort. Instead, region_chg calls add_reservation_in_range() to count the number of regions needed and allocates those, and that info is passed to region_add and region_abort to decrement adds_in_progress correctly. We've also modified the assumption that region_add after region_chg never fails. region_chg now pre-allocates at least 1 region for region_add. If region_add needs more regions than region_chg has allocated for it, then it may fail. Signed-off-by: Mina Almasry Reviewed-by: Mike Kravetz --- Changes in v9: - Added clarifications in the comments and addressed minor issues from code review. Changes in v7: - region_chg no longer allocates (t-f) / 2 file_region entries. Changes in v6: - Fix bug in number of region_caches allocated by region_chg --- mm/hugetlb.c | 327 +++++++++++++++++++++++++++++++++++---------------- 1 file changed, 223 insertions(+), 104 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/mm/hugetlb.c b/mm/hugetlb.c index f1b63946ee95c..de0028e9a8630 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -245,110 +245,179 @@ struct file_region { long to; }; +/* Helper that removes a struct file_region from the resv_map cache and returns + * it for use. + */ +static struct file_region * +get_file_region_entry_from_cache(struct resv_map *resv, long from, long to) +{ + struct file_region *nrg = NULL; + + VM_BUG_ON(resv->region_cache_count <= 0); + + resv->region_cache_count--; + nrg = list_first_entry(&resv->region_cache, struct file_region, link); + VM_BUG_ON(!nrg); + list_del(&nrg->link); + + nrg->from = from; + nrg->to = to; + + return nrg; +} + /* Must be called with resv->lock held. Calling this with count_only == true * will count the number of pages to be added but will not modify the linked - * list. + * list. If regions_needed != NULL and count_only == true, then regions_needed + * will indicate the number of file_regions needed in the cache to carry out to + * add the regions for this range. */ static long add_reservation_in_range(struct resv_map *resv, long f, long t, - bool count_only) + long *regions_needed, bool count_only) { - long chg = 0; + long add = 0; struct list_head *head = &resv->regions; + long last_accounted_offset = f; struct file_region *rg = NULL, *trg = NULL, *nrg = NULL; - /* Locate the region we are before or in. */ - list_for_each_entry(rg, head, link) - if (f <= rg->to) - break; - - /* Round our left edge to the current segment if it encloses us. */ - if (f > rg->from) - f = rg->from; + if (regions_needed) + *regions_needed = 0; - chg = t - f; + /* In this loop, we essentially handle an entry for the range + * [last_accounted_offset, rg->from), at every iteration, with some + * bounds checking. + */ + list_for_each_entry_safe(rg, trg, head, link) { + /* Skip irrelevant regions that start before our range. */ + if (rg->from < f) { + /* If this region ends after the last accounted offset, + * then we need to update last_accounted_offset. + */ + if (rg->to > last_accounted_offset) + last_accounted_offset = rg->to; + continue; + } - /* Check for and consume any regions we now overlap with. */ - nrg = rg; - list_for_each_entry_safe(rg, trg, rg->link.prev, link) { - if (&rg->link == head) - break; + /* When we find a region that starts beyond our range, we've + * finished. + */ if (rg->from > t) break; - /* We overlap with this area, if it extends further than - * us then we must extend ourselves. Account for its - * existing reservation. + /* Add an entry for last_accounted_offset -> rg->from, and + * update last_accounted_offset. */ - if (rg->to > t) { - chg += rg->to - t; - t = rg->to; + if (rg->from > last_accounted_offset) { + add += rg->from - last_accounted_offset; + if (!count_only) { + nrg = get_file_region_entry_from_cache( + resv, last_accounted_offset, rg->from); + list_add(&nrg->link, rg->link.prev); + } else if (regions_needed) + *regions_needed += 1; } - chg -= rg->to - rg->from; - if (!count_only && rg != nrg) { - list_del(&rg->link); - kfree(rg); - } + last_accounted_offset = rg->to; } - if (!count_only) { - nrg->from = f; - nrg->to = t; + /* Handle the case where our range extends beyond + * last_accounted_offset. + */ + if (last_accounted_offset < t) { + add += t - last_accounted_offset; + if (!count_only) { + nrg = get_file_region_entry_from_cache( + resv, last_accounted_offset, t); + list_add(&nrg->link, rg->link.prev); + } else if (regions_needed) + *regions_needed += 1; } - return chg; + return add; } /* * Add the huge page range represented by [f, t) to the reserve - * map. Existing regions will be expanded to accommodate the specified - * range, or a region will be taken from the cache. Sufficient regions - * must exist in the cache due to the previous call to region_chg with - * the same range. + * map. Regions will be taken from the cache to fill in this range. + * Sufficient regions should exist in the cache due to the previous + * call to region_chg with the same range, but in some cases the cache will not + * have sufficient entries due to races with other code doing region_add or + * region_del. The extra needed entries will be allocated. * - * Return the number of new huge pages added to the map. This - * number is greater than or equal to zero. + * regions_needed is the out value provided by a previous call to region_chg. + * + * Return the number of new huge pages added to the map. This number is greater + * than or equal to zero. If file_region entries needed to be allocated for + * this operation and we were not able to allocate, it ruturns -ENOMEM. + * region_add of regions of length 1 never allocate file_regions and cannot + * fail; region_chg will always allocate at least 1 entry and a region_add for + * 1 page will only require at most 1 entry. */ -static long region_add(struct resv_map *resv, long f, long t) +static long region_add(struct resv_map *resv, long f, long t, + long in_regions_needed) { - struct list_head *head = &resv->regions; - struct file_region *rg, *nrg; - long add = 0; + long add = 0, actual_regions_needed = 0, i = 0; + struct file_region *trg = NULL, *rg = NULL; + struct list_head allocated_regions; + + INIT_LIST_HEAD(&allocated_regions); spin_lock(&resv->lock); - /* Locate the region we are either in or before. */ - list_for_each_entry(rg, head, link) - if (f <= rg->to) - break; +retry: + + /* Count how many regions are actually needed to execute this add. */ + add_reservation_in_range(resv, f, t, &actual_regions_needed, true); /* - * If no region exists which can be expanded to include the - * specified range, pull a region descriptor from the cache - * and use it for this range. + * Check for sufficient descriptors in the cache to accommodate + * this add operation. Note that actual_regions_needed may be greater + * than in_regions_needed. In this case, we need to make sure that we + * allocate extra entries, such that we have enough for all the + * existing adds_in_progress, plus the excess needed for this + * operation. */ - if (&rg->link == head || t < rg->from) { - VM_BUG_ON(resv->region_cache_count <= 0); + if (resv->region_cache_count < + resv->adds_in_progress + + (actual_regions_needed - in_regions_needed)) { + /* region_add operation of range 1 should never need to + * allocate file_region entries. + */ + VM_BUG_ON(t - f <= 1); - resv->region_cache_count--; - nrg = list_first_entry(&resv->region_cache, struct file_region, - link); - list_del(&nrg->link); + /* Must drop lock to allocate a new descriptor. */ + spin_unlock(&resv->lock); + for (i = 0; i < (actual_regions_needed - in_regions_needed); + i++) { + trg = kmalloc(sizeof(*trg), GFP_KERNEL); + if (!trg) + goto out_of_memory; + list_add(&trg->link, &allocated_regions); + } + spin_lock(&resv->lock); - nrg->from = f; - nrg->to = t; - list_add(&nrg->link, rg->link.prev); + list_for_each_entry_safe(rg, trg, &allocated_regions, link) { + list_del(&rg->link); + list_add(&rg->link, &resv->region_cache); + resv->region_cache_count++; + } - add += t - f; - goto out_locked; + goto retry; } - add = add_reservation_in_range(resv, f, t, false); + add = add_reservation_in_range(resv, f, t, NULL, false); + + resv->adds_in_progress -= in_regions_needed; -out_locked: - resv->adds_in_progress--; spin_unlock(&resv->lock); VM_BUG_ON(add < 0); return add; + +out_of_memory: + list_for_each_entry_safe(rg, trg, &allocated_regions, link) { + list_del(&rg->link); + kfree(rg); + } + return -ENOMEM; } /* @@ -358,49 +427,79 @@ static long region_add(struct resv_map *resv, long f, long t) * call to region_add that will actually modify the reserve * map to add the specified range [f, t). region_chg does * not change the number of huge pages represented by the - * map. A new file_region structure is added to the cache - * as a placeholder, so that the subsequent region_add - * call will have all the regions it needs and will not fail. + * map. A number of new file_region structures is added to the cache as a + * placeholder, for the subsequent region_add call to use. At least 1 + * file_region structure is added. + * + * out_regions_needed is the number of regions added to the + * resv->adds_in_progress. This value needs to be provided to a follow up call + * to region_add or region_abort for proper accounting. * * Returns the number of huge pages that need to be added to the existing * reservation map for the range [f, t). This number is greater or equal to * zero. -ENOMEM is returned if a new file_region structure or cache entry * is needed and can not be allocated. */ -static long region_chg(struct resv_map *resv, long f, long t) +static long region_chg(struct resv_map *resv, long f, long t, + long *out_regions_needed) { - long chg = 0; + struct file_region *trg = NULL, *rg = NULL; + long chg = 0, i = 0, to_allocate = 0; + struct list_head allocated_regions; + + INIT_LIST_HEAD(&allocated_regions); spin_lock(&resv->lock); -retry_locked: - resv->adds_in_progress++; + + /* Count how many hugepages in this range are NOT respresented. */ + chg = add_reservation_in_range(resv, f, t, out_regions_needed, true); + + if (*out_regions_needed == 0) + *out_regions_needed = 1; + + resv->adds_in_progress += *out_regions_needed; /* * Check for sufficient descriptors in the cache to accommodate * the number of in progress add operations. */ - if (resv->adds_in_progress > resv->region_cache_count) { - struct file_region *trg; - - VM_BUG_ON(resv->adds_in_progress - resv->region_cache_count > 1); - /* Must drop lock to allocate a new descriptor. */ - resv->adds_in_progress--; + while (resv->region_cache_count < resv->adds_in_progress) { + to_allocate = resv->adds_in_progress - resv->region_cache_count; + + /* Must drop lock to allocate a new descriptor. Note that even + * though we drop the lock here, we do not make another call to + * add_reservation_in_range after re-acquiring the lock. + * Essentially this branch makes sure that we have enough + * descriptors in the cache as suggested by the first call to + * add_reservation_in_range. If more regions turn out to be + * required, region_add will deal with it. + */ spin_unlock(&resv->lock); - - trg = kmalloc(sizeof(*trg), GFP_KERNEL); - if (!trg) - return -ENOMEM; + for (i = 0; i < to_allocate; i++) { + trg = kmalloc(sizeof(*trg), GFP_KERNEL); + if (!trg) + goto out_of_memory; + list_add(&trg->link, &allocated_regions); + } spin_lock(&resv->lock); - list_add(&trg->link, &resv->region_cache); - resv->region_cache_count++; - goto retry_locked; - } - chg = add_reservation_in_range(resv, f, t, true); + list_for_each_entry_safe(rg, trg, &allocated_regions, link) { + list_del(&rg->link); + list_add(&rg->link, &resv->region_cache); + resv->region_cache_count++; + } + } spin_unlock(&resv->lock); return chg; + +out_of_memory: + list_for_each_entry_safe(rg, trg, &allocated_regions, link) { + list_del(&rg->link); + kfree(rg); + } + return -ENOMEM; } /* @@ -408,17 +507,20 @@ static long region_chg(struct resv_map *resv, long f, long t) * of the resv_map keeps track of the operations in progress between * calls to region_chg and region_add. Operations are sometimes * aborted after the call to region_chg. In such cases, region_abort - * is called to decrement the adds_in_progress counter. + * is called to decrement the adds_in_progress counter. regions_needed + * is the value returned by the region_chg call, it is used to decrement + * the adds_in_progress counter. * * NOTE: The range arguments [f, t) are not needed or used in this * routine. They are kept to make reading the calling code easier as * arguments will match the associated region_chg call. */ -static void region_abort(struct resv_map *resv, long f, long t) +static void region_abort(struct resv_map *resv, long f, long t, + long regions_needed) { spin_lock(&resv->lock); VM_BUG_ON(!resv->region_cache_count); - resv->adds_in_progress--; + resv->adds_in_progress -= regions_needed; spin_unlock(&resv->lock); } @@ -1883,6 +1985,7 @@ static long __vma_reservation_common(struct hstate *h, struct resv_map *resv; pgoff_t idx; long ret; + long dummy_out_regions_needed; resv = vma_resv_map(vma); if (!resv) @@ -1891,20 +1994,29 @@ static long __vma_reservation_common(struct hstate *h, idx = vma_hugecache_offset(h, vma, addr); switch (mode) { case VMA_NEEDS_RESV: - ret = region_chg(resv, idx, idx + 1); + ret = region_chg(resv, idx, idx + 1, &dummy_out_regions_needed); + /* We assume that vma_reservation_* routines always operate on + * 1 page, and that adding to resv map a 1 page entry can only + * ever require 1 region. + */ + VM_BUG_ON(dummy_out_regions_needed != 1); break; case VMA_COMMIT_RESV: - ret = region_add(resv, idx, idx + 1); + ret = region_add(resv, idx, idx + 1, 1); + /* region_add calls of range 1 should never fail. */ + VM_BUG_ON(ret < 0); break; case VMA_END_RESV: - region_abort(resv, idx, idx + 1); + region_abort(resv, idx, idx + 1, 1); ret = 0; break; case VMA_ADD_RESV: - if (vma->vm_flags & VM_MAYSHARE) - ret = region_add(resv, idx, idx + 1); - else { - region_abort(resv, idx, idx + 1); + if (vma->vm_flags & VM_MAYSHARE) { + ret = region_add(resv, idx, idx + 1, 1); + /* region_add calls of range 1 should never fail. */ + VM_BUG_ON(ret < 0); + } else { + region_abort(resv, idx, idx + 1, 1); ret = region_del(resv, idx, idx + 1); } break; @@ -4563,12 +4675,12 @@ int hugetlb_reserve_pages(struct inode *inode, struct vm_area_struct *vma, vm_flags_t vm_flags) { - long ret, chg; + long ret, chg, add = -1; struct hstate *h = hstate_inode(inode); struct hugepage_subpool *spool = subpool_inode(inode); struct resv_map *resv_map; struct hugetlb_cgroup *h_cg; - long gbl_reserve; + long gbl_reserve, regions_needed = 0; /* This should never happen */ if (from > to) { @@ -4598,7 +4710,7 @@ int hugetlb_reserve_pages(struct inode *inode, */ resv_map = inode_resv_map(inode); - chg = region_chg(resv_map, from, to); + chg = region_chg(resv_map, from, to, ®ions_needed); } else { /* Private mapping. */ @@ -4668,9 +4780,14 @@ int hugetlb_reserve_pages(struct inode *inode, * else has to be done for private mappings here */ if (!vma || vma->vm_flags & VM_MAYSHARE) { - long add = region_add(resv_map, from, to); - - if (unlikely(chg > add)) { + add = region_add(resv_map, from, to, regions_needed); + + if (unlikely(add < 0)) { + hugetlb_acct_memory(h, -gbl_reserve); + /* put back original number of pages, chg */ + (void)hugepage_subpool_put_pages(spool, chg); + goto out_err; + } else if (unlikely(chg > add)) { /* * pages in this range were added to the reserve * map between region_chg and region_add. This @@ -4688,9 +4805,11 @@ int hugetlb_reserve_pages(struct inode *inode, return 0; out_err: if (!vma || vma->vm_flags & VM_MAYSHARE) - /* Don't call region_abort if region_chg failed */ - if (chg >= 0) - region_abort(resv_map, from, to); + /* Only call region_abort if the region_chg succeeded but the + * region_add failed or didn't run. + */ + if (chg >= 0 && add < 0) + region_abort(resv_map, from, to, regions_needed); if (vma && is_vma_resv_set(vma, HPAGE_RESV_OWNER)) kref_put(&resv_map->refs, resv_map_release); return ret; From patchwork Wed Jan 15 01:26:48 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333051 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 276BD14B4 for ; Wed, 15 Jan 2020 01:27:08 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id E666E2467E for ; Wed, 15 Jan 2020 01:27:07 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="g4vrCCuF" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729017AbgAOB1H (ORCPT ); Tue, 14 Jan 2020 20:27:07 -0500 Received: from mail-pf1-f201.google.com ([209.85.210.201]:37483 "EHLO mail-pf1-f201.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729016AbgAOB1F (ORCPT ); Tue, 14 Jan 2020 20:27:05 -0500 Received: by mail-pf1-f201.google.com with SMTP id d85so9934126pfd.4 for ; Tue, 14 Jan 2020 17:27:05 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=4/iY+6M9fGnOfYT38lRw2eCjml2wSZh+p85JWa17Yuo=; b=g4vrCCuF7G6cMYYudAo5h/E+v30Ehekj3WxyXuXzVQwxDVcWyVoAAN3OFGHHNOsiAg oAkKe7Ko+zrccunYtqksk8HBagVGIFSKE8PCZJdKXVvwp+PHcES2ZZnq7AebZfeeckKi jgYiRZkTukb34BC9YGdLOqqdR2cP2F1lsj4ZvttUyUT5J2k23i0ut8y1fVXCw/CxoBBG lFs1rvXjdPKkzNClMppobOlE59ccjODz4up1ITe2phRIQAELd6kv9Xv6kTa4ghgsB5Yr t6KzckbvynWqurVbOA8YhGSuyV/B7OfODYL5MB0+J1WVdTUIuuy8zZpwg5He71suUDrC jgIw== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=4/iY+6M9fGnOfYT38lRw2eCjml2wSZh+p85JWa17Yuo=; b=gNXg2Ly9xVFTFqOmJr8JDhc45DaL8TufRylDay4Ers0gSrDmkZnQERlpqgcl02CGHs /uceOWlj729fpwthFrqRTuzOuaOgV4xfnfkg/Bf7FKFL8DovliOThhUZ8Fyh3xN4xQrQ 7449JPK9FN5ui/MU0D6h98Akf6uaiXC9ehlWmaBhT4wHWUE+U2vFu/C4pnBqSukXswDd 6rtDe4CuNmzLUxtWFu9WWzsdlwgg2CfTYsS5hA0tLl43FaxmxvX7HB5I7FlQ1Yh4/zOh 5W1yiqIifopP0V5FpsTDg7qZim557nme9PitjQU436SO7mTBak//+RxTsOkIlKglV9tU 4HuQ== X-Gm-Message-State: APjAAAUq1ld13H3OW04TFUbPWHJmyFFE+s4NYo4KkmyUHXDy7jIX4nTj BYI5zGwS5fkckQLb2XmMFxFx35cRQb1FXH7wew== X-Google-Smtp-Source: APXvYqwddvd+8Tr4pUJA1VUpFUOfuOZzunJnUAn2p1fyIp9b1Iu3ONU8F5HWpFwkrI3ylCktac5hVZz59lyTHGk5vg== X-Received: by 2002:a63:2949:: with SMTP id p70mr30626707pgp.191.1579051624531; Tue, 14 Jan 2020 17:27:04 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:48 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-5-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 5/8] hugetlb_cgroup: add accounting for shared mappings From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org For shared mappings, the pointer to the hugetlb_cgroup to uncharge lives in the resv_map entries, in file_region->reservation_counter. After a call to region_chg, we charge the approprate hugetlb_cgroup, and if successful, we pass on the hugetlb_cgroup info to a follow up region_add call. When a file_region entry is added to the resv_map via region_add, we put the pointer to that cgroup in file_region->reservation_counter. If charging doesn't succeed, we report the error to the caller, so that the kernel fails the reservation. On region_del, which is when the hugetlb memory is unreserved, we also uncharge the file_region->reservation_counter. Signed-off-by: Mina Almasry --- Changes in v10: - Deleted duplicated code snippet. Changes in V9: - Updated for hugetlb reservation repareting. --- mm/hugetlb.c | 156 ++++++++++++++++++++++++++++++++++++++++----------- 1 file changed, 124 insertions(+), 32 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/mm/hugetlb.c b/mm/hugetlb.c index de0028e9a8630..9bcfc12c5d214 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -243,6 +243,16 @@ struct file_region { struct list_head link; long from; long to; +#ifdef CONFIG_CGROUP_HUGETLB + /* + * On shared mappings, each reserved region appears as a struct + * file_region in resv_map. These fields hold the info needed to + * uncharge each reservation. + */ + struct page_counter *reservation_counter; + unsigned long pages_per_hpage; + struct cgroup_subsys_state *css; +#endif }; /* Helper that removes a struct file_region from the resv_map cache and returns @@ -266,6 +276,25 @@ get_file_region_entry_from_cache(struct resv_map *resv, long from, long to) return nrg; } +/* Helper that records hugetlb_cgroup uncharge info. */ +static void record_hugetlb_cgroup_uncharge_info(struct hugetlb_cgroup *h_cg, + struct file_region *nrg, + struct hstate *h) +{ +#ifdef CONFIG_CGROUP_HUGETLB + if (h_cg) { + nrg->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + nrg->pages_per_hpage = pages_per_huge_page(h); + nrg->css = &h_cg->css; + } else { + nrg->reservation_counter = NULL; + nrg->pages_per_hpage = 0; + nrg->css = NULL; + } +#endif +} + /* Must be called with resv->lock held. Calling this with count_only == true * will count the number of pages to be added but will not modify the linked * list. If regions_needed != NULL and count_only == true, then regions_needed @@ -273,7 +302,9 @@ get_file_region_entry_from_cache(struct resv_map *resv, long from, long to) * add the regions for this range. */ static long add_reservation_in_range(struct resv_map *resv, long f, long t, - long *regions_needed, bool count_only) + struct hugetlb_cgroup *h_cg, + struct hstate *h, long *regions_needed, + bool count_only) { long add = 0; struct list_head *head = &resv->regions; @@ -312,6 +343,8 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, if (!count_only) { nrg = get_file_region_entry_from_cache( resv, last_accounted_offset, rg->from); + record_hugetlb_cgroup_uncharge_info(h_cg, nrg, + h); list_add(&nrg->link, rg->link.prev); } else if (regions_needed) *regions_needed += 1; @@ -328,11 +361,13 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, if (!count_only) { nrg = get_file_region_entry_from_cache( resv, last_accounted_offset, t); + record_hugetlb_cgroup_uncharge_info(h_cg, nrg, h); list_add(&nrg->link, rg->link.prev); } else if (regions_needed) *regions_needed += 1; } + VM_BUG_ON(add < 0); return add; } @@ -353,7 +388,8 @@ static long add_reservation_in_range(struct resv_map *resv, long f, long t, * fail; region_chg will always allocate at least 1 entry and a region_add for * 1 page will only require at most 1 entry. */ -static long region_add(struct resv_map *resv, long f, long t, +static long region_add(struct hstate *h, struct hugetlb_cgroup *h_cg, + struct resv_map *resv, long f, long t, long in_regions_needed) { long add = 0, actual_regions_needed = 0, i = 0; @@ -366,7 +402,8 @@ static long region_add(struct resv_map *resv, long f, long t, retry: /* Count how many regions are actually needed to execute this add. */ - add_reservation_in_range(resv, f, t, &actual_regions_needed, true); + add_reservation_in_range(resv, f, t, NULL, NULL, &actual_regions_needed, + true); /* * Check for sufficient descriptors in the cache to accommodate @@ -404,7 +441,7 @@ static long region_add(struct resv_map *resv, long f, long t, goto retry; } - add = add_reservation_in_range(resv, f, t, NULL, false); + add = add_reservation_in_range(resv, f, t, h_cg, h, NULL, false); resv->adds_in_progress -= in_regions_needed; @@ -452,7 +489,8 @@ static long region_chg(struct resv_map *resv, long f, long t, spin_lock(&resv->lock); /* Count how many hugepages in this range are NOT respresented. */ - chg = add_reservation_in_range(resv, f, t, out_regions_needed, true); + chg = add_reservation_in_range(resv, f, t, NULL, NULL, + out_regions_needed, true); if (*out_regions_needed == 0) *out_regions_needed = 1; @@ -524,6 +562,29 @@ static void region_abort(struct resv_map *resv, long f, long t, spin_unlock(&resv->lock); } +static void uncharge_cgroup_if_shared_mapping(struct resv_map *resv, + struct file_region *rg, + unsigned long nr_pages) +{ +#ifdef CONFIG_CGROUP_HUGETLB + /* + * If resv->reservation_counter is NULL, then this is either a shared + * reservation, or cgroup charging is disabled on this resv_map. + * + * If the cgroup charging is disabled, then rg->reservation_counter is + * NULL and the uncharge counter call is a no-op. If the mapping is + * shared then the reserved memory is tracked in the file_struct + * entries inside of resv_map. So we need to uncharge the memory here. + */ + if (rg->reservation_counter && rg->pages_per_hpage && nr_pages > 0 && + !resv->reservation_counter) { + hugetlb_cgroup_uncharge_counter(rg->reservation_counter, + nr_pages * rg->pages_per_hpage, + rg->css); + } +#endif +} + /* * Delete the specified range [f, t) from the reserve map. If the * t parameter is LONG_MAX, this indicates that ALL regions after f @@ -588,11 +649,22 @@ static long region_del(struct resv_map *resv, long f, long t) /* New entry for end of split region */ nrg->from = t; nrg->to = rg->to; + +#ifdef CONFIG_CGROUP_HUGETLB + nrg->reservation_counter = rg->reservation_counter; + nrg->pages_per_hpage = rg->pages_per_hpage; + nrg->css = rg->css; + css_get(rg->css); +#endif + INIT_LIST_HEAD(&nrg->link); /* Original entry is trimmed */ rg->to = f; + uncharge_cgroup_if_shared_mapping(resv, rg, + nrg->to - nrg->from); + list_add(&nrg->link, &rg->link); nrg = NULL; break; @@ -600,6 +672,8 @@ static long region_del(struct resv_map *resv, long f, long t) if (f <= rg->from && t >= rg->to) { /* Remove entire region */ del += rg->to - rg->from; + uncharge_cgroup_if_shared_mapping(resv, rg, + rg->to - rg->from); list_del(&rg->link); kfree(rg); continue; @@ -608,14 +682,20 @@ static long region_del(struct resv_map *resv, long f, long t) if (f <= rg->from) { /* Trim beginning of region */ del += t - rg->from; rg->from = t; + + uncharge_cgroup_if_shared_mapping(resv, rg, + t - rg->from); } else { /* Trim end of region */ del += rg->to - f; rg->to = f; + + uncharge_cgroup_if_shared_mapping(resv, rg, rg->to - f); } } spin_unlock(&resv->lock); kfree(nrg); + return del; } @@ -2002,7 +2082,7 @@ static long __vma_reservation_common(struct hstate *h, VM_BUG_ON(dummy_out_regions_needed != 1); break; case VMA_COMMIT_RESV: - ret = region_add(resv, idx, idx + 1, 1); + ret = region_add(NULL, NULL, resv, idx, idx + 1, 1); /* region_add calls of range 1 should never fail. */ VM_BUG_ON(ret < 0); break; @@ -2012,7 +2092,7 @@ static long __vma_reservation_common(struct hstate *h, break; case VMA_ADD_RESV: if (vma->vm_flags & VM_MAYSHARE) { - ret = region_add(resv, idx, idx + 1, 1); + ret = region_add(NULL, NULL, resv, idx, idx + 1, 1); /* region_add calls of range 1 should never fail. */ VM_BUG_ON(ret < 0); } else { @@ -4679,7 +4759,7 @@ int hugetlb_reserve_pages(struct inode *inode, struct hstate *h = hstate_inode(inode); struct hugepage_subpool *spool = subpool_inode(inode); struct resv_map *resv_map; - struct hugetlb_cgroup *h_cg; + struct hugetlb_cgroup *h_cg = NULL; long gbl_reserve, regions_needed = 0; /* This should never happen */ @@ -4720,23 +4800,6 @@ int hugetlb_reserve_pages(struct inode *inode, chg = to - from; - if (hugetlb_cgroup_charge_cgroup(hstate_index(h), - chg * pages_per_huge_page(h), - &h_cg, true)) { - kref_put(&resv_map->refs, resv_map_release); - return -ENOMEM; - } - -#ifdef CONFIG_CGROUP_HUGETLB - /* - * Since this branch handles private mappings, we attach the - * counter to uncharge for this reservation off resv_map. - */ - resv_map->reservation_counter = - &h_cg->reserved_hugepage[hstate_index(h)]; - resv_map->pages_per_hpage = pages_per_huge_page(h); -#endif - set_vma_resv_map(vma, resv_map); set_vma_resv_flags(vma, HPAGE_RESV_OWNER); } @@ -4746,6 +4809,26 @@ int hugetlb_reserve_pages(struct inode *inode, goto out_err; } + ret = hugetlb_cgroup_charge_cgroup( + hstate_index(h), chg * pages_per_huge_page(h), &h_cg, true); + + if (ret < 0) { + ret = -ENOMEM; + goto out_err; + } + +#ifdef CONFIG_CGROUP_HUGETLB + if (vma && !(vma->vm_flags & VM_MAYSHARE) && h_cg) { + /* For private mappings, the hugetlb_cgroup uncharge info hangs + * of the resv_map. + */ + resv_map->reservation_counter = + &h_cg->reserved_hugepage[hstate_index(h)]; + resv_map->pages_per_hpage = pages_per_huge_page(h); + resv_map->css = &h_cg->css; + } +#endif + /* * There must be enough pages in the subpool for the mapping. If * the subpool has a minimum size, there may be some global @@ -4754,7 +4837,7 @@ int hugetlb_reserve_pages(struct inode *inode, gbl_reserve = hugepage_subpool_get_pages(spool, chg); if (gbl_reserve < 0) { ret = -ENOSPC; - goto out_err; + goto out_uncharge_cgroup; } /* @@ -4763,9 +4846,7 @@ int hugetlb_reserve_pages(struct inode *inode, */ ret = hugetlb_acct_memory(h, gbl_reserve); if (ret < 0) { - /* put back original number of pages, chg */ - (void)hugepage_subpool_put_pages(spool, chg); - goto out_err; + goto out_put_pages; } /* @@ -4780,7 +4861,7 @@ int hugetlb_reserve_pages(struct inode *inode, * else has to be done for private mappings here */ if (!vma || vma->vm_flags & VM_MAYSHARE) { - add = region_add(resv_map, from, to, regions_needed); + add = region_add(h, h_cg, resv_map, from, to, regions_needed); if (unlikely(add < 0)) { hugetlb_acct_memory(h, -gbl_reserve); @@ -4797,12 +4878,23 @@ int hugetlb_reserve_pages(struct inode *inode, */ long rsv_adjust; - rsv_adjust = hugepage_subpool_put_pages(spool, - chg - add); + hugetlb_cgroup_uncharge_cgroup( + hstate_index(h), + (chg - add) * pages_per_huge_page(h), h_cg, + true); + + rsv_adjust = + hugepage_subpool_put_pages(spool, chg - add); hugetlb_acct_memory(h, -rsv_adjust); } } return 0; +out_put_pages: + /* put back original number of pages, chg */ + (void)hugepage_subpool_put_pages(spool, chg); +out_uncharge_cgroup: + hugetlb_cgroup_uncharge_cgroup( + hstate_index(h), chg * pages_per_huge_page(h), h_cg, true); out_err: if (!vma || vma->vm_flags & VM_MAYSHARE) /* Only call region_abort if the region_chg succeeded but the From patchwork Wed Jan 15 01:26:49 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333065 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 264A514B4 for ; Wed, 15 Jan 2020 01:27:24 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 0307A24658 for ; Wed, 15 Jan 2020 01:27:24 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="SbEo5zBp" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729028AbgAOB1I (ORCPT ); Tue, 14 Jan 2020 20:27:08 -0500 Received: from mail-pg1-f201.google.com ([209.85.215.201]:39203 "EHLO mail-pg1-f201.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729025AbgAOB1H (ORCPT ); Tue, 14 Jan 2020 20:27:07 -0500 Received: by mail-pg1-f201.google.com with SMTP id v2so9439485pgv.6 for ; Tue, 14 Jan 2020 17:27:07 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=+hDgtcJ/OAWe9zIJj2uUM/Qj+y1M7vqqOqi2MY740cw=; b=SbEo5zBpBq30EMJ5WGTJdUleiraqh7pRa2Nw0a4bsfHNDUmiBRMRcceqj5vgsp+yZD zA15qcUMkkqs0G0mlULqVgoPx8f6lINvEm/HUSYSvsASmziRYje1qMrSXwmErtDz2Bon ryd41KYz/6mj3AI2ESiO0S1m7FKOKJ0uP1mWiSBLxILSDpqupMzcecid/pl+I/2FMEBV BGJRAG0xlusuZ66v01XbYj+nn1YuYpw7QeNfBjSRZTV6hRnBYGFnFZzyXFp/XCy7SrX1 n+QZNB7SfXv9qStndFXqL7ojGlE9+1hkKxpS9zTRIp7b3i4VOp5qe8J+VekK5NT3ycMn 16RQ== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=+hDgtcJ/OAWe9zIJj2uUM/Qj+y1M7vqqOqi2MY740cw=; b=sPXQeRkJwU0sD1pzhOPWZsC7j7+Ijt/DA08JWkoDSg3AN1yc5dV82wteJlCkiuhX0p LM5FmZmDcF/IA6qnL/NybzL55enzEFkVdi3sO2rzBuZYtb1ZzPACr/NGeMNE9+JOygdp H4kfxxBmDFcKej5/cxK4SI+xYc8qAU6VfQ/4Icx8yWde/tpLl6Y0TbThLg0uLoxyGByx Y30c2poN+GrE2vUhIrigkh1Pyfn9iRbKYA6kNI4Gdh3VG8I4lWt0iXV3H7b3xEzIQZEF 9ZCul+N+dZK1iG2IMmfVY3zmOsy92yNDPpiV6eKLnJsInCIvouFCDrfNehM1VEaRvhe1 lf3A== X-Gm-Message-State: APjAAAUF3d44Vn3Z1NtV38wTbMHEqUiSMQU221lApjBlV4t+fNgR1rTy 8Jhj8V1RPgm7LDRMLVisyFcU23ZBHOCamVd/2Q== X-Google-Smtp-Source: APXvYqyyjl2zNq7/Fihix9iuEtq3Q06GE1o8i8KAInm5+9bDeWN7QRw6hAI7tkCO4X/1w9M2HxbiSoxA5H5cI3IpCQ== X-Received: by 2002:a65:4142:: with SMTP id x2mr29535809pgp.393.1579051626821; Tue, 14 Jan 2020 17:27:06 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:49 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-6-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 6/8] hugetlb_cgroup: support noreserve mappings From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org Support MAP_NORESERVE accounting as part of the new counter. For each hugepage allocation, at allocation time we check if there is a reservation for this allocation or not. If there is a reservation for this allocation, then this allocation was charged at reservation time, and we don't re-account it. If there is no reserevation for this allocation, we charge the appropriate hugetlb_cgroup. The hugetlb_cgroup to uncharge for this allocation is stored in page[3].private. We use new APIs added in an earlier patch to set this pointer. Signed-off-by: Mina Almasry --- Changes in v10: - Refactored deferred_reserve check. --- mm/hugetlb.c | 28 +++++++++++++++++++++++++++- 1 file changed, 27 insertions(+), 1 deletion(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 9bcfc12c5d214..d3f107626f927 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -1374,6 +1374,9 @@ static void __free_huge_page(struct page *page) clear_page_huge_active(page); hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), page, false); + hugetlb_cgroup_uncharge_page(hstate_index(h), pages_per_huge_page(h), + page, true); + if (restore_reserve) h->resv_huge_pages++; @@ -2207,6 +2210,7 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, long gbl_chg; int ret, idx; struct hugetlb_cgroup *h_cg; + bool deferred_reserve; idx = hstate_index(h); /* @@ -2244,10 +2248,20 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, gbl_chg = 1; } + /* If this allocation is not consuming a reservation, charge it now. + */ + deferred_reserve = map_chg || avoid_reserve || !vma_resv_map(vma); + if (deferred_reserve) { + ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), + &h_cg, true); + if (ret) + goto out_subpool_put; + } + ret = hugetlb_cgroup_charge_cgroup(idx, pages_per_huge_page(h), &h_cg, false); if (ret) - goto out_subpool_put; + goto out_uncharge_cgroup_reservation; spin_lock(&hugetlb_lock); /* @@ -2271,6 +2285,14 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, } hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, page, false); + /* If allocation is not consuming a reservation, also store the + * hugetlb_cgroup pointer on the page. + */ + if (deferred_reserve) { + hugetlb_cgroup_commit_charge(idx, pages_per_huge_page(h), h_cg, + page, true); + } + spin_unlock(&hugetlb_lock); set_page_private(page, (unsigned long)spool); @@ -2296,6 +2318,10 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, out_uncharge_cgroup: hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), h_cg, false); +out_uncharge_cgroup_reservation: + if (deferred_reserve) + hugetlb_cgroup_uncharge_cgroup(idx, pages_per_huge_page(h), + h_cg, true); out_subpool_put: if (map_chg || avoid_reserve) hugepage_subpool_put_pages(spool, 1); From patchwork Wed Jan 15 01:26:50 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333057 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id DA600184C for ; Wed, 15 Jan 2020 01:27:11 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 98AAC24684 for ; Wed, 15 Jan 2020 01:27:11 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="oMfPFcAv" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729073AbgAOB1K (ORCPT ); Tue, 14 Jan 2020 20:27:10 -0500 Received: from mail-pj1-f73.google.com ([209.85.216.73]:35777 "EHLO mail-pj1-f73.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729048AbgAOB1K (ORCPT ); Tue, 14 Jan 2020 20:27:10 -0500 Received: by mail-pj1-f73.google.com with SMTP id l8so9958294pje.0 for ; Tue, 14 Jan 2020 17:27:09 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=GBT2xfA+dnfRF2THo16AV8hNONxGZnDsFwLv2uCbQzA=; b=oMfPFcAv/vaC4oAvgrxT25sna9Zi56LUj3j/h+dBfBgNrVGxEaFjcgjn0cTWAIgJxK K0+oaDxpky26oVXAzoLbPD4Klizdq8RxuwSa7uWiHNXc5WfHv9ak/3B75txpSV3IsLFE m0sUf/HIH80Sei2hSsuttBmnmWvlQ5yNz3ErZsQoWsElQmxHTQvcn7q6c5Ozaqn9TKaw KPLNUsLF2A+Cwx80kCliyT7mAkN3HJBs3TFLwbdGpeLtFguKIxRNn3S1zAlz+5OtHFlr IInF6US9908wQjkMkZrDK5PKF7IG/gmzoUTZl2Ay8T2l7mmoUgDCqAGzHfYFq2roaZBm 4awA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=GBT2xfA+dnfRF2THo16AV8hNONxGZnDsFwLv2uCbQzA=; b=K5CZoJyM4Ck6HTaK1FZGQaRSs1r3drp7F7nn11F3pPVVINBpjpS5+/K0IPLSJZ/hgY gI67ADezqM56nPXMn0FkGrAqHg/omY6P2BQAAHuM3uRKvaigDtfSX4JcEfCwJuFyR/to 7ytlg1oiPzG5Uajc0lq3LC018CrtkroPkMtgKKPmH2gjSp5rHLij+PYZOxzAdqsj6uZk O21k4M75aWgOHwxbPU2WpG8DaL286IzwDyBBO7RKlVqid1XJ3jMXeZTcl00PDrE5xRe1 MlPJCBsYm0Xqyw/CZVC4TLMtaKcy2OI5D1MOTG0T7OlC3s8XXBB9fa6XGmcejsctTo3z gIEw== X-Gm-Message-State: APjAAAWM5msFfRbZxzcLpoqO4e+jN1ZLst1PGjecw/Kn7CHyO8UupBTl Fl111u1yOVzajNLbaClUcWedIzWf//lhRnvjIQ== X-Google-Smtp-Source: APXvYqxBDqBu5KZ8++eMJUOUrvpHWTDsruKcBC/a81Nka1SLN4EF/+X3233G9ftAgohp26Sbzy2Z7fM1I89oMtG06w== X-Received: by 2002:a63:364f:: with SMTP id d76mr30934995pga.215.1579051629106; Tue, 14 Jan 2020 17:27:09 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:50 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-7-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 7/8] hugetlb_cgroup: Add hugetlb_cgroup reservation tests From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org The tests use both shared and private mapped hugetlb memory, and monitors the hugetlb usage counter as well as the hugetlb reservation counter. They test different configurations such as hugetlb memory usage via hugetlbfs, or MAP_HUGETLB, or shmget/shmat, and with and without MAP_POPULATE. Also add test for hugetlb reservation reparenting, since this is a subtle issue. Signed-off-by: Mina Almasry --- Changes in v10: - Updated tests to resv.* name changes. Changes in v9: - Added tests for hugetlb reparenting. - Make tests explicitly support cgroup v1 and v2 via script argument. Changes in v6: - Updates tests for cgroups-v2 and NORESERVE allocations. --- tools/testing/selftests/vm/.gitignore | 1 + tools/testing/selftests/vm/Makefile | 1 + .../selftests/vm/charge_reserved_hugetlb.sh | 557 ++++++++++++++++++ .../selftests/vm/hugetlb_reparenting_test.sh | 235 ++++++++ .../selftests/vm/write_hugetlb_memory.sh | 23 + .../testing/selftests/vm/write_to_hugetlbfs.c | 261 ++++++++ 6 files changed, 1078 insertions(+) create mode 100755 tools/testing/selftests/vm/charge_reserved_hugetlb.sh create mode 100755 tools/testing/selftests/vm/hugetlb_reparenting_test.sh create mode 100644 tools/testing/selftests/vm/write_hugetlb_memory.sh create mode 100644 tools/testing/selftests/vm/write_to_hugetlbfs.c -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/tools/testing/selftests/vm/.gitignore b/tools/testing/selftests/vm/.gitignore index 31b3c98b6d34d..d3bed9407773c 100644 --- a/tools/testing/selftests/vm/.gitignore +++ b/tools/testing/selftests/vm/.gitignore @@ -14,3 +14,4 @@ virtual_address_range gup_benchmark va_128TBswitch map_fixed_noreplace +write_to_hugetlbfs diff --git a/tools/testing/selftests/vm/Makefile b/tools/testing/selftests/vm/Makefile index 7f9a8a8c31da9..662bb95e84c5d 100644 --- a/tools/testing/selftests/vm/Makefile +++ b/tools/testing/selftests/vm/Makefile @@ -22,6 +22,7 @@ TEST_GEN_FILES += userfaultfd ifneq (,$(filter $(ARCH),arm64 ia64 mips64 parisc64 ppc64 riscv64 s390x sh64 sparc64 x86_64)) TEST_GEN_FILES += va_128TBswitch TEST_GEN_FILES += virtual_address_range +TEST_GEN_FILES += write_to_hugetlbfs endif TEST_PROGS := run_vmtests diff --git a/tools/testing/selftests/vm/charge_reserved_hugetlb.sh b/tools/testing/selftests/vm/charge_reserved_hugetlb.sh new file mode 100755 index 0000000000000..8f2413352514b --- /dev/null +++ b/tools/testing/selftests/vm/charge_reserved_hugetlb.sh @@ -0,0 +1,557 @@ +#!/bin/sh +# SPDX-License-Identifier: GPL-2.0 + +set -e + +if [[ $(id -u) -ne 0 ]]; then + echo "This test must be run as root. Skipping..." + exit 0 +fi + +fault_limit_file=limit_in_bytes +reservation_limit_file=resv.limit_in_bytes +fault_usage_file=usage_in_bytes +reservation_usage_file=resv.usage_in_bytes + +if [[ "$1" == "-cgroup-v2" ]]; then + cgroup2=1 + fault_limit_file=max + reservation_limit_file=resv.max + fault_usage_file=current + reservation_usage_file=resv.current +fi + +cgroup_path=/dev/cgroup/memory +if [[ ! -e $cgroup_path ]]; then + mkdir -p $cgroup_path + if [[ $cgroup2 ]]; then + mount -t cgroup2 none $cgroup_path + else + mount -t cgroup memory,hugetlb $cgroup_path + fi +fi + +if [[ $cgroup2 ]]; then + echo "+hugetlb" >/dev/cgroup/memory/cgroup.subtree_control +fi + +cleanup() { + if [[ $cgroup2 ]]; then + echo $$ >$cgroup_path/cgroup.procs + else + echo $$ >$cgroup_path/tasks + fi + + if [[ -e /mnt/huge ]]; then + rm -rf /mnt/huge/* + umount /mnt/huge || echo error + rmdir /mnt/huge + fi + if [[ -e $cgroup_path/hugetlb_cgroup_test ]]; then + rmdir $cgroup_path/hugetlb_cgroup_test + fi + if [[ -e $cgroup_path/hugetlb_cgroup_test1 ]]; then + rmdir $cgroup_path/hugetlb_cgroup_test1 + fi + if [[ -e $cgroup_path/hugetlb_cgroup_test2 ]]; then + rmdir $cgroup_path/hugetlb_cgroup_test2 + fi + echo 0 >/proc/sys/vm/nr_hugepages + echo CLEANUP DONE +} + +function expect_equal() { + local expected="$1" + local actual="$2" + local error="$3" + + if [[ "$expected" != "$actual" ]]; then + echo "expected ($expected) != actual ($actual): $3" + cleanup + exit 1 + fi +} + +function setup_cgroup() { + local name="$1" + local cgroup_limit="$2" + local reservation_limit="$3" + + mkdir $cgroup_path/$name + + echo writing cgroup limit: "$cgroup_limit" + echo "$cgroup_limit" >$cgroup_path/$name/hugetlb.2MB.$fault_limit_file + + echo writing reseravation limit: "$reservation_limit" + echo "$reservation_limit" > \ + $cgroup_path/$name/hugetlb.2MB.$reservation_limit_file + + if [ -e "$cgroup_path/$name/cpuset.cpus" ]; then + echo 0 >$cgroup_path/$name/cpuset.cpus + fi + if [ -e "$cgroup_path/$name/cpuset.mems" ]; then + echo 0 >$cgroup_path/$name/cpuset.mems + fi +} + +function wait_for_hugetlb_memory_to_get_depleted() { + local cgroup="$1" + local path="/dev/cgroup/memory/$cgroup/hugetlb.2MB.$reservation_usage_file" + # Wait for hugetlbfs memory to get depleted. + while [ $(cat $path) != 0 ]; do + echo Waiting for hugetlb memory to get depleted. + cat $path + sleep 0.5 + done +} + +function wait_for_hugetlb_memory_to_get_reserved() { + local cgroup="$1" + local size="$2" + + local path="/dev/cgroup/memory/$cgroup/hugetlb.2MB.$reservation_usage_file" + # Wait for hugetlbfs memory to get written. + while [ $(cat $path) != $size ]; do + echo Waiting for hugetlb memory reservation to reach size $size. + cat $path + sleep 0.5 + done +} + +function wait_for_hugetlb_memory_to_get_written() { + local cgroup="$1" + local size="$2" + + local path="/dev/cgroup/memory/$cgroup/hugetlb.2MB.$fault_usage_file" + # Wait for hugetlbfs memory to get written. + while [ $(cat $path) != $size ]; do + echo Waiting for hugetlb memory to reach size $size. + cat $path + sleep 0.5 + done +} + +function write_hugetlbfs_and_get_usage() { + local cgroup="$1" + local size="$2" + local populate="$3" + local write="$4" + local path="$5" + local method="$6" + local private="$7" + local expect_failure="$8" + local reserve="$9" + + # Function return values. + reservation_failed=0 + oom_killed=0 + hugetlb_difference=0 + reserved_difference=0 + + local hugetlb_usage=$cgroup_path/$cgroup/hugetlb.2MB.$fault_usage_file + local reserved_usage=$cgroup_path/$cgroup/hugetlb.2MB.$reservation_usage_file + + local hugetlb_before=$(cat $hugetlb_usage) + local reserved_before=$(cat $reserved_usage) + + echo + echo Starting: + echo hugetlb_usage="$hugetlb_before" + echo reserved_usage="$reserved_before" + echo expect_failure is "$expect_failure" + + set +e + if [[ "$method" == "1" ]] || [[ "$method" == 2 ]] || + [[ "$private" == "-r" ]] && [[ "$expect_failure" != 1 ]]; then + + bash write_hugetlb_memory.sh "$size" "$populate" "$write" \ + "$cgroup" "$path" "$method" "$private" "-l" "$reserve" & + + local write_result=$? + + if [[ "$reserve" != "-n" ]]; then + wait_for_hugetlb_memory_to_get_reserved "$cgroup" "$size" + elif [[ "$populate" == "-o" ]] || [[ "$write" == "-w" ]]; then + wait_for_hugetlb_memory_to_get_written "$cgroup" "$size" + else + # This case doesn't produce visible effects, but we still have + # to wait for the async process to start and execute... + sleep 0.5 + fi + + echo write_result is $write_result + else + bash write_hugetlb_memory.sh "$size" "$populate" "$write" \ + "$cgroup" "$path" "$method" "$private" "$reserve" + local write_result=$? + + if [[ "$reserve" != "-n" ]]; then + wait_for_hugetlb_memory_to_get_reserved "$cgroup" "$size" + fi + fi + set -e + + if [[ "$write_result" == 1 ]]; then + reservation_failed=1 + fi + + # On linus/master, the above process gets SIGBUS'd on oomkill, with + # return code 135. On earlier kernels, it gets actual oomkill, with return + # code 137, so just check for both conditions in case we're testing + # against an earlier kernel. + if [[ "$write_result" == 135 ]] || [[ "$write_result" == 137 ]]; then + oom_killed=1 + fi + + local hugetlb_after=$(cat $hugetlb_usage) + local reserved_after=$(cat $reserved_usage) + + echo After write: + echo hugetlb_usage="$hugetlb_after" + echo reserved_usage="$reserved_after" + + hugetlb_difference=$(($hugetlb_after - $hugetlb_before)) + reserved_difference=$(($reserved_after - $reserved_before)) +} + +function cleanup_hugetlb_memory() { + set +e + local cgroup="$1" + if [[ "$(pgrep write_to_hugetlbfs)" != "" ]]; then + echo kiling write_to_hugetlbfs + killall -2 write_to_hugetlbfs + wait_for_hugetlb_memory_to_get_depleted $cgroup + fi + set -e + + if [[ -e /mnt/huge ]]; then + rm -rf /mnt/huge/* + umount /mnt/huge + rmdir /mnt/huge + fi +} + +function run_test() { + local size="$1" + local populate="$2" + local write="$3" + local cgroup_limit="$4" + local reservation_limit="$5" + local nr_hugepages="$6" + local method="$7" + local private="$8" + local expect_failure="$9" + local reserve="${10}" + + # Function return values. + hugetlb_difference=0 + reserved_difference=0 + reservation_failed=0 + oom_killed=0 + + echo nr hugepages = "$nr_hugepages" + echo "$nr_hugepages" >/proc/sys/vm/nr_hugepages + + setup_cgroup "hugetlb_cgroup_test" "$cgroup_limit" "$reservation_limit" + + mkdir -p /mnt/huge + mount -t hugetlbfs \ + -o pagesize=2M,size=256M none /mnt/huge + + write_hugetlbfs_and_get_usage "hugetlb_cgroup_test" "$size" "$populate" \ + "$write" "/mnt/huge/test" "$method" "$private" "$expect_failure" \ + "$reserve" + + cleanup_hugetlb_memory "hugetlb_cgroup_test" + + local final_hugetlb=$(cat $cgroup_path/hugetlb_cgroup_test/hugetlb.2MB.$fault_usage_file) + local final_reservation=$(cat $cgroup_path/hugetlb_cgroup_test/hugetlb.2MB.$reservation_usage_file) + + echo $hugetlb_difference + echo $reserved_difference + expect_equal "0" "$final_hugetlb" "final hugetlb is not zero" + expect_equal "0" "$final_reservation" "final reservation is not zero" +} + +function run_multiple_cgroup_test() { + local size1="$1" + local populate1="$2" + local write1="$3" + local cgroup_limit1="$4" + local reservation_limit1="$5" + + local size2="$6" + local populate2="$7" + local write2="$8" + local cgroup_limit2="$9" + local reservation_limit2="${10}" + + local nr_hugepages="${11}" + local method="${12}" + local private="${13}" + local expect_failure="${14}" + local reserve="${15}" + + # Function return values. + hugetlb_difference1=0 + reserved_difference1=0 + reservation_failed1=0 + oom_killed1=0 + + hugetlb_difference2=0 + reserved_difference2=0 + reservation_failed2=0 + oom_killed2=0 + + echo nr hugepages = "$nr_hugepages" + echo "$nr_hugepages" >/proc/sys/vm/nr_hugepages + + setup_cgroup "hugetlb_cgroup_test1" "$cgroup_limit1" "$reservation_limit1" + setup_cgroup "hugetlb_cgroup_test2" "$cgroup_limit2" "$reservation_limit2" + + mkdir -p /mnt/huge + mount -t hugetlbfs \ + -o pagesize=2M,size=256M none /mnt/huge + + write_hugetlbfs_and_get_usage "hugetlb_cgroup_test1" "$size1" \ + "$populate1" "$write1" "/mnt/huge/test1" "$method" "$private" \ + "$expect_failure" "$reserve" + + hugetlb_difference1=$hugetlb_difference + reserved_difference1=$reserved_difference + reservation_failed1=$reservation_failed + oom_killed1=$oom_killed + + local cgroup1_hugetlb_usage=$cgroup_path/hugetlb_cgroup_test1/hugetlb.2MB.$fault_usage_file + local cgroup1_reservation_usage=$cgroup_path/hugetlb_cgroup_test1/hugetlb.2MB.$reservation_usage_file + local cgroup2_hugetlb_usage=$cgroup_path/hugetlb_cgroup_test2/hugetlb.2MB.$fault_usage_file + local cgroup2_reservation_usage=$cgroup_path/hugetlb_cgroup_test2/hugetlb.2MB.$reservation_usage_file + + local usage_before_second_write=$(cat $cgroup1_hugetlb_usage) + local reservation_usage_before_second_write=$(cat \ + $cgroup1_reservation_usage) + + write_hugetlbfs_and_get_usage "hugetlb_cgroup_test2" "$size2" \ + "$populate2" "$write2" "/mnt/huge/test2" "$method" "$private" \ + "$expect_failure" "$reserve" + + hugetlb_difference2=$hugetlb_difference + reserved_difference2=$reserved_difference + reservation_failed2=$reservation_failed + oom_killed2=$oom_killed + + expect_equal "$usage_before_second_write" \ + "$(cat $cgroup1_hugetlb_usage)" "Usage changed." + expect_equal "$reservation_usage_before_second_write" \ + "$(cat $cgroup1_reservation_usage)" "Reservation usage changed." + + cleanup_hugetlb_memory + + local final_hugetlb=$(cat $cgroup1_hugetlb_usage) + local final_reservation=$(cat $cgroup1_reservation_usage) + + expect_equal "0" "$final_hugetlb" \ + "hugetlbt_cgroup_test1 final hugetlb is not zero" + expect_equal "0" "$final_reservation" \ + "hugetlbt_cgroup_test1 final reservation is not zero" + + local final_hugetlb=$(cat $cgroup2_hugetlb_usage) + local final_reservation=$(cat $cgroup2_reservation_usage) + + expect_equal "0" "$final_hugetlb" \ + "hugetlb_cgroup_test2 final hugetlb is not zero" + expect_equal "0" "$final_reservation" \ + "hugetlb_cgroup_test2 final reservation is not zero" +} + +cleanup + +for populate in "" "-o"; do + for method in 0 1 2; do + for private in "" "-r"; do + for reserve in "" "-n"; do + + # Skip mmap(MAP_HUGETLB | MAP_SHARED). Doesn't seem to be supported. + if [[ "$method" == 1 ]] && [[ "$private" == "" ]]; then + continue + fi + + # Skip populated shmem tests. Doesn't seem to be supported. + if [[ "$method" == 2"" ]] && [[ "$populate" == "-o" ]]; then + continue + fi + + if [[ "$method" == 2"" ]] && [[ "$reserve" == "-n" ]]; then + continue + fi + + cleanup + echo + echo + echo + echo Test normal case. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + run_test $((10 * 1024 * 1024)) "$populate" "" $((20 * 1024 * 1024)) \ + $((20 * 1024 * 1024)) 10 "$method" "$private" "0" "$reserve" + + echo Memory charged to hugtlb=$hugetlb_difference + echo Memory charged to reservation=$reserved_difference + + if [[ "$populate" == "-o" ]]; then + expect_equal "$((10 * 1024 * 1024))" "$hugetlb_difference" \ + "Reserved memory charged to hugetlb cgroup." + else + expect_equal "0" "$hugetlb_difference" \ + "Reserved memory charged to hugetlb cgroup." + fi + + if [[ "$reserve" != "-n" ]] || [[ "$populate" == "-o" ]]; then + expect_equal "$((10 * 1024 * 1024))" "$reserved_difference" \ + "Reserved memory not charged to reservation usage." + else + expect_equal "0" "$reserved_difference" \ + "Reserved memory not charged to reservation usage." + fi + + echo 'PASS' + + cleanup + echo + echo + echo + echo Test normal case with write. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + run_test $((10 * 1024 * 1024)) "$populate" '-w' $((20 * 1024 * 1024)) \ + $((20 * 1024 * 1024)) 10 "$method" "$private" "0" "$reserve" + + echo Memory charged to hugtlb=$hugetlb_difference + echo Memory charged to reservation=$reserved_difference + + expect_equal "$((10 * 1024 * 1024))" "$hugetlb_difference" \ + "Reserved memory charged to hugetlb cgroup." + + expect_equal "$((10 * 1024 * 1024))" "$reserved_difference" \ + "Reserved memory not charged to reservation usage." + + echo 'PASS' + + cleanup + continue + echo + echo + echo + echo Test more than reservation case. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + + if [ "$reserve" != "-n" ]; then + run_test "$((10 * 1024 * 1024))" "$populate" '' "$((20 * 1024 * 1024))" \ + "$((5 * 1024 * 1024))" "10" "$method" "$private" "1" "$reserve" + + expect_equal "1" "$reservation_failed" "Reservation succeeded." + fi + + echo 'PASS' + + cleanup + + echo + echo + echo + echo Test more than cgroup limit case. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + + # Not sure if shm memory can be cleaned up when the process gets sigbus'd. + if [[ "$method" != 2 ]]; then + run_test $((10 * 1024 * 1024)) "$populate" "-w" $((5 * 1024 * 1024)) \ + $((20 * 1024 * 1024)) 10 "$method" "$private" "1" "$reserve" + + expect_equal "1" "$oom_killed" "Not oom killed." + fi + echo 'PASS' + + cleanup + + echo + echo + echo + echo Test normal case, multiple cgroups. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + run_multiple_cgroup_test "$((6 * 1024 * 1024))" "$populate" "" \ + "$((20 * 1024 * 1024))" "$((20 * 1024 * 1024))" "$((10 * 1024 * 1024))" \ + "$populate" "" "$((20 * 1024 * 1024))" "$((20 * 1024 * 1024))" "10" \ + "$method" "$private" "0" "$reserve" + + echo Memory charged to hugtlb1=$hugetlb_difference1 + echo Memory charged to reservation1=$reserved_difference1 + echo Memory charged to hugtlb2=$hugetlb_difference2 + echo Memory charged to reservation2=$reserved_difference2 + + if [[ "$reserve" != "-n" ]] || [[ "$populate" == "-o" ]]; then + expect_equal "$((6 * 1024 * 1024))" "$reserved_difference1" \ + "Incorrect reservations charged to cgroup 1." + + expect_equal "$((10 * 1024 * 1024))" "$reserved_difference2" \ + "Incorrect reservation charged to cgroup 2." + + else + expect_equal "0" "$reserved_difference1" \ + "Incorrect reservations charged to cgroup 1." + + expect_equal "0" "$reserved_difference2" \ + "Incorrect reservation charged to cgroup 2." + fi + + if [[ "$populate" == "-o" ]]; then + expect_equal "$((6 * 1024 * 1024))" "$hugetlb_difference1" \ + "Incorrect hugetlb charged to cgroup 1." + + expect_equal "$((10 * 1024 * 1024))" "$hugetlb_difference2" \ + "Incorrect hugetlb charged to cgroup 2." + + else + expect_equal "0" "$hugetlb_difference1" \ + "Incorrect hugetlb charged to cgroup 1." + + expect_equal "0" "$hugetlb_difference2" \ + "Incorrect hugetlb charged to cgroup 2." + fi + echo 'PASS' + + cleanup + echo + echo + echo + echo Test normal case with write, multiple cgroups. + echo private=$private, populate=$populate, method=$method, reserve=$reserve + run_multiple_cgroup_test "$((6 * 1024 * 1024))" "$populate" "-w" \ + "$((20 * 1024 * 1024))" "$((20 * 1024 * 1024))" "$((10 * 1024 * 1024))" \ + "$populate" "-w" "$((20 * 1024 * 1024))" "$((20 * 1024 * 1024))" "10" \ + "$method" "$private" "0" "$reserve" + + echo Memory charged to hugtlb1=$hugetlb_difference1 + echo Memory charged to reservation1=$reserved_difference1 + echo Memory charged to hugtlb2=$hugetlb_difference2 + echo Memory charged to reservation2=$reserved_difference2 + + expect_equal "$((6 * 1024 * 1024))" "$hugetlb_difference1" \ + "Incorrect hugetlb charged to cgroup 1." + + expect_equal "$((6 * 1024 * 1024))" "$reserved_difference1" \ + "Incorrect reservation charged to cgroup 1." + + expect_equal "$((10 * 1024 * 1024))" "$hugetlb_difference2" \ + "Incorrect hugetlb charged to cgroup 2." + + expect_equal "$((10 * 1024 * 1024))" "$reserved_difference2" \ + "Incorrected reservation charged to cgroup 2." + echo 'PASS' + + cleanup + + done # reserve + done # private + done # populate +done # method + +umount $cgroup_path +rmdir $cgroup_path diff --git a/tools/testing/selftests/vm/hugetlb_reparenting_test.sh b/tools/testing/selftests/vm/hugetlb_reparenting_test.sh new file mode 100755 index 0000000000000..2be672c2b311a --- /dev/null +++ b/tools/testing/selftests/vm/hugetlb_reparenting_test.sh @@ -0,0 +1,235 @@ +#!/bin/bash +# SPDX-License-Identifier: GPL-2.0 + +set -e + +if [[ $(id -u) -ne 0 ]]; then + echo "This test must be run as root. Skipping..." + exit 0 +fi + +usage_file=usage_in_bytes + +if [[ "$1" == "-cgroup-v2" ]]; then + cgroup2=1 + usage_file=current +fi + +CGROUP_ROOT='/dev/cgroup/memory' +MNT='/mnt/huge/' + +if [[ ! -e $CGROUP_ROOT ]]; then + mkdir -p $CGROUP_ROOT + if [[ $cgroup2 ]]; then + mount -t cgroup2 none $CGROUP_ROOT + sleep 1 + echo "+hugetlb +memory" >$CGROUP_ROOT/cgroup.subtree_control + else + mount -t cgroup memory,hugetlb $CGROUP_ROOT + fi +fi + +function cleanup() { + echo cleanup + set +e + rm -rf "$MNT"/* 2>/dev/null + umount "$MNT" 2>/dev/null + rmdir "$MNT" 2>/dev/null + rmdir "$CGROUP_ROOT"/a/b 2>/dev/null + rmdir "$CGROUP_ROOT"/a 2>/dev/null + rmdir "$CGROUP_ROOT"/test1 2>/dev/null + echo 0 >/proc/sys/vm/nr_hugepages + set -e +} + +function assert_state() { + local expected_a="$1" + local expected_a_hugetlb="$2" + local expected_b="" + local expected_b_hugetlb="" + + if [ ! -z ${3:-} ] && [ ! -z ${4:-} ]; then + expected_b="$3" + expected_b_hugetlb="$4" + fi + local tolerance=$((5 * 1024 * 1024)) + + local actual_a + actual_a="$(cat "$CGROUP_ROOT"/a/memory.$usage_file)" + if [[ $actual_a -lt $(($expected_a - $tolerance)) ]] || + [[ $actual_a -gt $(($expected_a + $tolerance)) ]]; then + echo actual a = $((${actual_a%% *} / 1024 / 1024)) MB + echo expected a = $((${expected_a%% *} / 1024 / 1024)) MB + echo fail + + cleanup + exit 1 + fi + + local actual_a_hugetlb + actual_a_hugetlb="$(cat "$CGROUP_ROOT"/a/hugetlb.2MB.$usage_file)" + if [[ $actual_a_hugetlb -lt $(($expected_a_hugetlb - $tolerance)) ]] || + [[ $actual_a_hugetlb -gt $(($expected_a_hugetlb + $tolerance)) ]]; then + echo actual a hugetlb = $((${actual_a_hugetlb%% *} / 1024 / 1024)) MB + echo expected a hugetlb = $((${expected_a_hugetlb%% *} / 1024 / 1024)) MB + echo fail + + cleanup + exit 1 + fi + + if [[ -z "$expected_b" || -z "$expected_b_hugetlb" ]]; then + return + fi + + local actual_b + actual_b="$(cat "$CGROUP_ROOT"/a/b/memory.$usage_file)" + if [[ $actual_b -lt $(($expected_b - $tolerance)) ]] || + [[ $actual_b -gt $(($expected_b + $tolerance)) ]]; then + echo actual b = $((${actual_b%% *} / 1024 / 1024)) MB + echo expected b = $((${expected_b%% *} / 1024 / 1024)) MB + echo fail + + cleanup + exit 1 + fi + + local actual_b_hugetlb + actual_b_hugetlb="$(cat "$CGROUP_ROOT"/a/b/hugetlb.2MB.$usage_file)" + if [[ $actual_b_hugetlb -lt $(($expected_b_hugetlb - $tolerance)) ]] || + [[ $actual_b_hugetlb -gt $(($expected_b_hugetlb + $tolerance)) ]]; then + echo actual b hugetlb = $((${actual_b_hugetlb%% *} / 1024 / 1024)) MB + echo expected b hugetlb = $((${expected_b_hugetlb%% *} / 1024 / 1024)) MB + echo fail + + cleanup + exit 1 + fi +} + +function setup() { + echo 100 >/proc/sys/vm/nr_hugepages + mkdir "$CGROUP_ROOT"/a + sleep 1 + if [[ $cgroup2 ]]; then + echo "+hugetlb +memory" >$CGROUP_ROOT/a/cgroup.subtree_control + else + echo 0 >$CGROUP_ROOT/a/cpuset.mems + echo 0 >$CGROUP_ROOT/a/cpuset.cpus + fi + + mkdir "$CGROUP_ROOT"/a/b + + if [[ ! $cgroup2 ]]; then + echo 0 >$CGROUP_ROOT/a/b/cpuset.mems + echo 0 >$CGROUP_ROOT/a/b/cpuset.cpus + fi + + mkdir -p "$MNT" + mount -t hugetlbfs none "$MNT" +} + +write_hugetlbfs() { + local cgroup="$1" + local path="$2" + local size="$3" + + if [[ $cgroup2 ]]; then + echo $$ >$CGROUP_ROOT/$cgroup/cgroup.procs + else + echo 0 >$CGROUP_ROOT/$cgroup/cpuset.mems + echo 0 >$CGROUP_ROOT/$cgroup/cpuset.cpus + echo $$ >"$CGROUP_ROOT/$cgroup/tasks" + fi + ./write_to_hugetlbfs -p "$path" -s "$size" -m 0 -o + if [[ $cgroup2 ]]; then + echo $$ >$CGROUP_ROOT/cgroup.procs + else + echo $$ >"$CGROUP_ROOT/tasks" + fi + echo +} + +set -e + +size=$((2 * 1024 * 1024 * 25)) # 50MB = 25 * 2MB hugepages. + +cleanup + +echo +echo +echo Test charge, rmdir, uncharge +setup +echo mkdir +mkdir $CGROUP_ROOT/test1 + +echo write +write_hugetlbfs test1 "$MNT"/test $size + +echo rmdir +rmdir $CGROUP_ROOT/test1 +mkdir $CGROUP_ROOT/test1 + +echo uncharge +rm -rf /mnt/huge/* + +cleanup + +echo done +echo +echo +if [[ ! $cgroup2 ]]; then + echo "Test parent and child hugetlb usage" + setup + + echo write + write_hugetlbfs a "$MNT"/test $size + + echo Assert memory charged correctly for parent use. + assert_state 0 $size 0 0 + + write_hugetlbfs a/b "$MNT"/test2 $size + + echo Assert memory charged correctly for child use. + assert_state 0 $(($size * 2)) 0 $size + + rmdir "$CGROUP_ROOT"/a/b + sleep 5 + echo Assert memory reparent correctly. + assert_state 0 $(($size * 2)) + + rm -rf "$MNT"/* + umount "$MNT" + echo Assert memory uncharged correctly. + assert_state 0 0 + + cleanup +fi + +echo +echo +echo "Test child only hugetlb usage" +echo setup +setup + +echo write +write_hugetlbfs a/b "$MNT"/test2 $size + +echo Assert memory charged correctly for child only use. +assert_state 0 $(($size)) 0 $size + +rmdir "$CGROUP_ROOT"/a/b +echo Assert memory reparent correctly. +assert_state 0 $size + +rm -rf "$MNT"/* +umount "$MNT" +echo Assert memory uncharged correctly. +assert_state 0 0 + +cleanup + +echo ALL PASS + +umount $CGROUP_ROOT +rm -rf $CGROUP_ROOT diff --git a/tools/testing/selftests/vm/write_hugetlb_memory.sh b/tools/testing/selftests/vm/write_hugetlb_memory.sh new file mode 100644 index 0000000000000..d3d0d108924d4 --- /dev/null +++ b/tools/testing/selftests/vm/write_hugetlb_memory.sh @@ -0,0 +1,23 @@ +#!/bin/sh +# SPDX-License-Identifier: GPL-2.0 + +set -e + +size=$1 +populate=$2 +write=$3 +cgroup=$4 +path=$5 +method=$6 +private=$7 +want_sleep=$8 +reserve=$9 + +echo "Putting task in cgroup '$cgroup'" +echo $$ > /dev/cgroup/memory/"$cgroup"/cgroup.procs + +echo "Method is $method" + +set +e +./write_to_hugetlbfs -p "$path" -s "$size" "$write" "$populate" -m "$method" \ + "$private" "$want_sleep" "$reserve" diff --git a/tools/testing/selftests/vm/write_to_hugetlbfs.c b/tools/testing/selftests/vm/write_to_hugetlbfs.c new file mode 100644 index 0000000000000..85811c3384a10 --- /dev/null +++ b/tools/testing/selftests/vm/write_to_hugetlbfs.c @@ -0,0 +1,261 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * This program reserves and uses hugetlb memory, supporting a bunch of + * scenarios needed by the charged_reserved_hugetlb.sh test. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* Global definitions. */ +enum method { + HUGETLBFS, + MMAP_MAP_HUGETLB, + SHM, + MAX_METHOD +}; + + +/* Global variables. */ +static const char *self; +static char *shmaddr; +static int shmid; + +/* + * Show usage and exit. + */ +static void exit_usage(void) +{ + printf("Usage: %s -p -s " + "[-m <0=hugetlbfs | 1=mmap(MAP_HUGETLB)>] [-l] [-r] " + "[-o] [-w] [-n]\n", + self); + exit(EXIT_FAILURE); +} + +void sig_handler(int signo) +{ + printf("Received %d.\n", signo); + if (signo == SIGINT) { + printf("Deleting the memory\n"); + if (shmdt((const void *)shmaddr) != 0) { + perror("Detach failure"); + shmctl(shmid, IPC_RMID, NULL); + exit(4); + } + + shmctl(shmid, IPC_RMID, NULL); + printf("Done deleting the memory\n"); + } + exit(2); +} + +int main(int argc, char **argv) +{ + int fd = 0; + int key = 0; + int *ptr = NULL; + int c = 0; + int size = 0; + char path[256] = ""; + enum method method = MAX_METHOD; + int want_sleep = 0, private = 0; + int populate = 0; + int write = 0; + int reserve = 1; + + unsigned long i; + + if (signal(SIGINT, sig_handler) == SIG_ERR) + err(1, "\ncan't catch SIGINT\n"); + + /* Parse command-line arguments. */ + setvbuf(stdout, NULL, _IONBF, 0); + self = argv[0]; + + while ((c = getopt(argc, argv, "s:p:m:owlrn")) != -1) { + switch (c) { + case 's': + size = atoi(optarg); + break; + case 'p': + strncpy(path, optarg, sizeof(path)); + break; + case 'm': + if (atoi(optarg) >= MAX_METHOD) { + errno = EINVAL; + perror("Invalid -m."); + exit_usage(); + } + method = atoi(optarg); + break; + case 'o': + populate = 1; + break; + case 'w': + write = 1; + break; + case 'l': + want_sleep = 1; + break; + case 'r': + private + = 1; + break; + case 'n': + reserve = 0; + break; + default: + errno = EINVAL; + perror("Invalid arg"); + exit_usage(); + } + } + + if (strncmp(path, "", sizeof(path)) != 0) { + printf("Writing to this path: %s\n", path); + } else { + errno = EINVAL; + perror("path not found"); + exit_usage(); + } + + if (size != 0) { + printf("Writing this size: %d\n", size); + } else { + errno = EINVAL; + perror("size not found"); + exit_usage(); + } + + if (!populate) + printf("Not populating.\n"); + else + printf("Populating.\n"); + + if (!write) + printf("Not writing to memory.\n"); + + if (method == MAX_METHOD) { + errno = EINVAL; + perror("-m Invalid"); + exit_usage(); + } else + printf("Using method=%d\n", method); + + if (!private) + printf("Shared mapping.\n"); + else + printf("Private mapping.\n"); + + if (!reserve) + printf("NO_RESERVE mapping.\n"); + else + printf("RESERVE mapping.\n"); + + switch (method) { + case HUGETLBFS: + printf("Allocating using HUGETLBFS.\n"); + fd = open(path, O_CREAT | O_RDWR, 0777); + if (fd == -1) + err(1, "Failed to open file."); + + ptr = mmap(NULL, size, PROT_READ | PROT_WRITE, + (private ? MAP_PRIVATE : MAP_SHARED) | + (populate ? MAP_POPULATE : 0) | + (reserve ? 0 : MAP_NORESERVE), + fd, 0); + + if (ptr == MAP_FAILED) { + close(fd); + err(1, "Error mapping the file"); + } + break; + case MMAP_MAP_HUGETLB: + printf("Allocating using MAP_HUGETLB.\n"); + ptr = mmap(NULL, size, PROT_READ | PROT_WRITE, + (private ? (MAP_PRIVATE | MAP_ANONYMOUS) : + MAP_SHARED) | + MAP_HUGETLB | (populate ? MAP_POPULATE : 0) | + (reserve ? 0 : MAP_NORESERVE), + -1, 0); + + if (ptr == MAP_FAILED) + err(1, "mmap"); + + printf("Returned address is %p\n", ptr); + break; + case SHM: + printf("Allocating using SHM.\n"); + shmid = shmget(key, size, + SHM_HUGETLB | IPC_CREAT | SHM_R | SHM_W); + if (shmid < 0) { + shmid = shmget(++key, size, + SHM_HUGETLB | IPC_CREAT | SHM_R | SHM_W); + if (shmid < 0) + err(1, "shmget"); + } + printf("shmid: 0x%x, shmget key:%d\n", shmid, key); + + shmaddr = shmat(shmid, NULL, 0); + if (shmaddr == (char *)-1) { + perror("Shared memory attach failure"); + shmctl(shmid, IPC_RMID, NULL); + exit(2); + } + printf("shmaddr: %p\n", shmaddr); + + break; + default: + errno = EINVAL; + err(1, "Invalid method."); + } + + if (write) { + printf("Writing to memory.\n"); + if (method != SHM) { + memset(ptr, 1, size); + } else { + printf("Starting the writes:\n"); + for (i = 0; i < size; i++) { + shmaddr[i] = (char)(i); + if (!(i % (1024 * 1024))) + printf("."); + } + printf("\n"); + + printf("Starting the Check..."); + for (i = 0; i < size; i++) + if (shmaddr[i] != (char)i) { + printf("\nIndex %lu mismatched\n", i); + exit(3); + } + printf("Done.\n"); + } + } + + if (want_sleep) { + /* Signal to caller that we're done. */ + printf("DONE\n"); + + /* Hold memory until external kill signal is delivered. */ + while (1) + sleep(100); + } + + switch (method == HUGETLBFS) { + close(fd); + } + + return 0; +} From patchwork Wed Jan 15 01:26:51 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Mina Almasry X-Patchwork-Id: 11333063 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id D3DEB14B4 for ; Wed, 15 Jan 2020 01:27:19 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id A89AF24680 for ; Wed, 15 Jan 2020 01:27:19 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=google.com header.i=@google.com header.b="DdORwNw0" Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1729026AbgAOB1S (ORCPT ); Tue, 14 Jan 2020 20:27:18 -0500 Received: from mail-pf1-f202.google.com ([209.85.210.202]:37484 "EHLO mail-pf1-f202.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1729076AbgAOB1M (ORCPT ); Tue, 14 Jan 2020 20:27:12 -0500 Received: by mail-pf1-f202.google.com with SMTP id d85so9934289pfd.4 for ; Tue, 14 Jan 2020 17:27:12 -0800 (PST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20161025; h=date:in-reply-to:message-id:mime-version:references:subject:from:to :cc; bh=h8U2qNtEFbaq+hBB/JCIu5FRCXJjzgS5K1PNEW4ZKws=; b=DdORwNw0ndB7yPDnt8tT0WEqtdwD8V4r97JqJT/LXfvkDwe5s1Q21YiAh1Gty98/b0 UViZadY+jCLv9/GwljHX1EpDUv/iKm4sPdChU0zjoMGyXN7zg1cbfcpgPdzbX5qTrdF/ dbKOqYdZjkBCfPZQQ1YxpKX6DhhfqJq+713oEumbz1oluyU8GfaKlJqBtya86kphQBPp hQ6z3xF6QIrMNetm1VyXpVPsEnkMLh9SXbwVlRZzxr3OVTLRib9eqf1oTPYa4qshTYex xOfkILN202JFeCcVen+QZ9LO1j+FVNwWsgKIETR7mXnbAXLRN4ALA87go3rMV94BSNjZ 9NZA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:date:in-reply-to:message-id:mime-version :references:subject:from:to:cc; bh=h8U2qNtEFbaq+hBB/JCIu5FRCXJjzgS5K1PNEW4ZKws=; b=ZIYykBB2iAvrMCeRQzrVZP8XinlMlgQCiyIa5Jq+QrFT9sH8prncXbCeZ3TAqRwu59 3T7vJKhT4O4+1/iE/6mX79aWrWYzhOK5UxHLSAC5yMWPzao28ET7MrIjQHpggpMc+N5Z UIt6QJAQF9m/6egANVi7+96FNlDIYYCMyEKOCdhojj8NgzH/AANft/dAMHIISwTxdqvl 5tyBIz9QsOQCGYae3jpgX/TQFTJ0/gxUC0/rRhBLsjrt7sElK75WROp27xQVXs5XNifo kUB79w88h/abS9NnubcBB+2a8I2UBkvTkbTwmMHO95TbcZsQc6gQ8wGz6n7cqHU59Ck1 U1Yg== X-Gm-Message-State: APjAAAVIGK0PPVvtCKKd0J4R4kFnnmvQTwCRS/1nXVsIv4PbsG0V5G2q 5HfOl+2Ezh0aRMV1j3WMmpe37K0hRo5mn4/+Uw== X-Google-Smtp-Source: APXvYqxmMi2/ACArgSv7fIgXtolIquE4UCUC/iP8qxu7HL4rR76nCNRpBZfD1IoYHxOtfCJxpCr2SfVY1MIWGdoxtA== X-Received: by 2002:a63:4b49:: with SMTP id k9mr30074514pgl.269.1579051631919; Tue, 14 Jan 2020 17:27:11 -0800 (PST) Date: Tue, 14 Jan 2020 17:26:51 -0800 In-Reply-To: <20200115012651.228058-1-almasrymina@google.com> Message-Id: <20200115012651.228058-8-almasrymina@google.com> Mime-Version: 1.0 References: <20200115012651.228058-1-almasrymina@google.com> X-Mailer: git-send-email 2.25.0.rc1.283.g88dfdc4193-goog Subject: [PATCH v10 8/8] hugetlb_cgroup: Add hugetlb_cgroup reservation docs From: Mina Almasry To: mike.kravetz@oracle.com, rientjes@google.com, shakeelb@google.com Cc: shuah@kernel.org, almasrymina@google.com, gthelen@google.com, akpm@linux-foundation.org, linux-kernel@vger.kernel.org, linux-mm@kvack.org, linux-kselftest@vger.kernel.org, cgroups@vger.kernel.org, aneesh.kumar@linux.vnet.ibm.com Sender: linux-kselftest-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kselftest@vger.kernel.org Add docs for how to use hugetlb_cgroup reservations, and their behavior. Signed-off-by: Mina Almasry --- Changes in v10: - Clarify reparenting behavior. - Reword benefits of reservation limits. Changes in v6: - Updated docs to reflect the new design based on a new counter that tracks both reservations and faults. --- .../admin-guide/cgroup-v1/hugetlb.rst | 103 ++++++++++++++++-- 1 file changed, 92 insertions(+), 11 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog diff --git a/Documentation/admin-guide/cgroup-v1/hugetlb.rst b/Documentation/admin-guide/cgroup-v1/hugetlb.rst index a3902aa253a96..00dd4daf55f19 100644 --- a/Documentation/admin-guide/cgroup-v1/hugetlb.rst +++ b/Documentation/admin-guide/cgroup-v1/hugetlb.rst @@ -2,13 +2,6 @@ HugeTLB Controller ================== -The HugeTLB controller allows to limit the HugeTLB usage per control group and -enforces the controller limit during page fault. Since HugeTLB doesn't -support page reclaim, enforcing the limit at page fault time implies that, -the application will get SIGBUS signal if it tries to access HugeTLB pages -beyond its limit. This requires the application to know beforehand how much -HugeTLB pages it would require for its use. - HugeTLB controller can be created by first mounting the cgroup filesystem. # mount -t cgroup -o hugetlb none /sys/fs/cgroup @@ -28,10 +21,14 @@ process (bash) into it. Brief summary of control files:: - hugetlb..limit_in_bytes # set/show limit of "hugepagesize" hugetlb usage - hugetlb..max_usage_in_bytes # show max "hugepagesize" hugetlb usage recorded - hugetlb..usage_in_bytes # show current usage for "hugepagesize" hugetlb - hugetlb..failcnt # show the number of allocation failure due to HugeTLB limit + hugetlb..resv.limit_in_bytes # set/show limit of "hugepagesize" hugetlb reservations + hugetlb..resv.max_usage_in_bytes # show max "hugepagesize" hugetlb reservations and no-reserve faults + hugetlb..resv.usage_in_bytes # show current reservations and no-reserve faults for "hugepagesize" hugetlb + hugetlb..resv.failcnt # show the number of allocation failure due to HugeTLB reservation limit + hugetlb..limit_in_bytes # set/show limit of "hugepagesize" hugetlb faults + hugetlb..max_usage_in_bytes # show max "hugepagesize" hugetlb usage recorded + hugetlb..usage_in_bytes # show current usage for "hugepagesize" hugetlb + hugetlb..failcnt # show the number of allocation failure due to HugeTLB usage limit For a system supporting three hugepage sizes (64k, 32M and 1G), the control files include:: @@ -40,11 +37,95 @@ files include:: hugetlb.1GB.max_usage_in_bytes hugetlb.1GB.usage_in_bytes hugetlb.1GB.failcnt + hugetlb.1GB.resv.limit_in_bytes + hugetlb.1GB.resv.max_usage_in_bytes + hugetlb.1GB.resv.usage_in_bytes + hugetlb.1GB.resv.failcnt hugetlb.64KB.limit_in_bytes hugetlb.64KB.max_usage_in_bytes hugetlb.64KB.usage_in_bytes hugetlb.64KB.failcnt + hugetlb.64KB.resv.limit_in_bytes + hugetlb.64KB.resv.max_usage_in_bytes + hugetlb.64KB.resv.usage_in_bytes + hugetlb.64KB.resv.failcnt hugetlb.32MB.limit_in_bytes hugetlb.32MB.max_usage_in_bytes hugetlb.32MB.usage_in_bytes hugetlb.32MB.failcnt + hugetlb.32MB.resv.limit_in_bytes + hugetlb.32MB.resv.max_usage_in_bytes + hugetlb.32MB.resv.usage_in_bytes + hugetlb.32MB.resv.failcnt + + +1. Page fault accounting + +hugetlb..limit_in_bytes +hugetlb..max_usage_in_bytes +hugetlb..usage_in_bytes +hugetlb..failcnt + +The HugeTLB controller allows users to limit the HugeTLB usage (page fault) per +control group and enforces the limit during page fault. Since HugeTLB +doesn't support page reclaim, enforcing the limit at page fault time implies +that, the application will get SIGBUS signal if it tries to fault in HugeTLB +pages beyond its limit. Therefore the application needs to know exactly how many +HugeTLB pages it uses before hand, and the sysadmin needs to make sure that +there are enough available on the machine for all the users to avoid processes +getting SIGBUS. + + +2. Reservation accounting + +hugetlb..resv.limit_in_bytes +hugetlb..resv.max_usage_in_bytes +hugetlb..resv.usage_in_bytes +hugetlb..resv.failcnt + +The HugeTLB controller allows to limit the HugeTLB reservations per control +group and enforces the controller limit at reservation time and at the fault of +HugeTLB memory for which no reservation exists. Since reservation limits are +enforced at reservation time (on mmap or shget), reservation limits never causes +the application to get SIGBUS signal if the memory was reserved before hand. For +MAP_NORESERVE allocations, the reservation limit behaves the same as the fault +limit, enforcing memory usage at fault time and causing the application to +receive a SIGBUS if it's crossing its limit. + +Reservation limits are superior to page fault limits described above, since +reservation limits are enforced at reservation time (on mmap or shget), and +never causes the application to get SIGBUS signal if the memory was reserved +before hand. This allows for easier fallback to alternatives such as +non-HugeTLB memory for example. In the case of page fault accounting, it's very +hard to avoid processes getting SIGBUS since the sysadmin needs precisely know +the HugeTLB usage of all the tasks in the system and make sure there is enough +pages to satisfy all requests. Avoiding tasks getting SIGBUS on overcommited +systems is practically impossible with page fault accounting. + + +3. Caveats with shared memory + +For shared HugeTLB memory, both HugeTLB reservation and page faults are charged +to the first task that causes the memory to be reserved or faulted, and all +subsequent uses of this reserved or faulted memory is done without charging. + +Shared HugeTLB memory is only uncharged when it is unreserved or deallocated. +This is usually when the HugeTLB file is deleted, and not when the task that +caused the reservation or fault has exited. + + +4. Caveats with HugeTLB cgroup offline. + +When a HugeTLB cgroup goes offline with some reservations or faults still +charged to it, the behavior is as follows: + +- The fault charges are charged to the parent HugeTLB cgroup (reparented), +- the reservation charges remain on the offline HugeTLB cgroup. + +This means that if a HugeTLB cgroup gets offlined while there is still HugeTLB +reservations charged to it, that cgroup persists as a zombie until all HugeTLB +reservations are uncharged. HugeTLB reservations behave in this manner to match +the memory controller whose cgroups also persist as zombie until all charged +memory is uncharged. Also, the tracking of HugeTLB reservations is a bit more +complex compared to the tracking of HugeTLB faults, so it is significantly +harder to reparent reservations at offline time.