From: Muchun Song <songmuchun@bytedance.com> To: gregkh@linuxfoundation.org, rafael@kernel.org, adobriyan@gmail.com, akpm@linux-foundation.org, hannes@cmpxchg.org, mhocko@kernel.org, vdavydov.dev@gmail.com, hughd@google.com, shakeelb@google.com, guro@fb.com, samitolvanen@google.com, feng.tang@intel.com, neilb@suse.de, iamjoonsoo.kim@lge.com, rdunlap@infradead.org Cc: linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, linux-mm@kvack.org, cgroups@vger.kernel.org, Muchun Song <songmuchun@bytedance.com> Subject: [PATCH v5 6/7] mm: memcontrol: convert NR_FILE_PMDMAPPED account to pages Date: Thu, 17 Dec 2020 11:43:55 +0800 [thread overview] Message-ID: <20201217034356.4708-7-songmuchun@bytedance.com> (raw) In-Reply-To: <20201217034356.4708-1-songmuchun@bytedance.com> Currently we use struct per_cpu_nodestat to cache the vmstat counters, which leads to inaccurate statistics expecially THP vmstat counters. In the systems with hundreads of processors it can be GBs of memory. For example, for a 96 CPUs system, the threshold is the maximum number of 125. And the per cpu counters can cache 23.4375 GB in total. The THP page is already a form of batched addition (it will add 512 worth of memory in one go) so skipping the batching seems like sensible. Although every THP stats update overflows the per-cpu counter, resorting to atomic global updates. But it can make the statistics more accuracy for the THP vmstat counters. So we convert the NR_FILE_PMDMAPPED account to pages. This patch is consistent with 8f182270dfec ("mm/swap.c: flush lru pvecs on compound page arrival"). Doing this also can make the unit of vmstat counters more unified. Finally, the unit of the vmstat counters are pages, kB and bytes. The B/KB suffix can tell us that the unit is bytes or kB. The rest which is without suffix are pages. Signed-off-by: Muchun Song <songmuchun@bytedance.com> --- drivers/base/node.c | 3 +-- fs/proc/meminfo.c | 2 +- include/linux/mmzone.h | 3 ++- mm/rmap.c | 6 ++++-- 4 files changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/base/node.c b/drivers/base/node.c index 7a66aefe4e46..d02d86aec19f 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -465,8 +465,7 @@ static ssize_t node_read_meminfo(struct device *dev, nid, K(node_page_state(pgdat, NR_SHMEM_THPS)), nid, K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED)), nid, K(node_page_state(pgdat, NR_FILE_THPS)), - nid, K(node_page_state(pgdat, NR_FILE_PMDMAPPED) * - HPAGE_PMD_NR) + nid, K(node_page_state(pgdat, NR_FILE_PMDMAPPED)) #endif ); len += hugetlb_report_node_meminfo(buf, len, nid); diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c index c61f440570f9..6fa761c9cc78 100644 --- a/fs/proc/meminfo.c +++ b/fs/proc/meminfo.c @@ -137,7 +137,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v) show_val_kb(m, "FileHugePages: ", global_node_page_state(NR_FILE_THPS)); show_val_kb(m, "FilePmdMapped: ", - global_node_page_state(NR_FILE_PMDMAPPED) * HPAGE_PMD_NR); + global_node_page_state(NR_FILE_PMDMAPPED)); #endif #ifdef CONFIG_CMA diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 1ca6a34f6a4a..6139479ba417 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -219,7 +219,8 @@ static __always_inline bool vmstat_item_print_in_thp(enum node_stat_item item) return item == NR_ANON_THPS || item == NR_FILE_THPS || item == NR_SHMEM_THPS || - item == NR_SHMEM_PMDMAPPED; + item == NR_SHMEM_PMDMAPPED || + item == NR_FILE_PMDMAPPED; } /* diff --git a/mm/rmap.c b/mm/rmap.c index 1c1b576c0627..5ebf16fae4b9 100644 --- a/mm/rmap.c +++ b/mm/rmap.c @@ -1223,7 +1223,8 @@ void page_add_file_rmap(struct page *page, bool compound) __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED, nr_pages); else - __inc_node_page_state(page, NR_FILE_PMDMAPPED); + __mod_lruvec_page_state(page, NR_FILE_PMDMAPPED, + nr_pages); } else { if (PageTransCompound(page) && page_mapping(page)) { VM_WARN_ON_ONCE(!PageLocked(page)); @@ -1267,7 +1268,8 @@ static void page_remove_file_rmap(struct page *page, bool compound) __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED, -nr_pages); else - __dec_node_page_state(page, NR_FILE_PMDMAPPED); + __mod_lruvec_page_state(page, NR_FILE_PMDMAPPED, + -nr_pages); } else { if (!atomic_add_negative(-1, &page->_mapcount)) return; -- 2.11.0
WARNING: multiple messages have this Message-ID (diff)
From: Muchun Song <songmuchun-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org> To: gregkh-hQyY1W1yCW8ekmWlsbkhG0B+6BGkLq7r@public.gmane.org, rafael-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org, adobriyan-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org, akpm-de/tnXTf+JLsfHDXvbKv3WD2FQJk+8+b@public.gmane.org, hannes-druUgvl0LCNAfugRpC6u6w@public.gmane.org, mhocko-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org, vdavydov.dev-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org, hughd-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org, shakeelb-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org, guro-b10kYP2dOMg@public.gmane.org, samitolvanen-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org, feng.tang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org, neilb-l3A5Bk7waGM@public.gmane.org, iamjoonsoo.kim-Hm3cg6mZ9cc@public.gmane.org, rdunlap-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org Cc: linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-mm-Bw31MaZKKs3YtjvyW6yDsg@public.gmane.org, cgroups-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, Muchun Song <songmuchun-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org> Subject: [PATCH v5 6/7] mm: memcontrol: convert NR_FILE_PMDMAPPED account to pages Date: Thu, 17 Dec 2020 11:43:55 +0800 [thread overview] Message-ID: <20201217034356.4708-7-songmuchun@bytedance.com> (raw) In-Reply-To: <20201217034356.4708-1-songmuchun-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org> Currently we use struct per_cpu_nodestat to cache the vmstat counters, which leads to inaccurate statistics expecially THP vmstat counters. In the systems with hundreads of processors it can be GBs of memory. For example, for a 96 CPUs system, the threshold is the maximum number of 125. And the per cpu counters can cache 23.4375 GB in total. The THP page is already a form of batched addition (it will add 512 worth of memory in one go) so skipping the batching seems like sensible. Although every THP stats update overflows the per-cpu counter, resorting to atomic global updates. But it can make the statistics more accuracy for the THP vmstat counters. So we convert the NR_FILE_PMDMAPPED account to pages. This patch is consistent with 8f182270dfec ("mm/swap.c: flush lru pvecs on compound page arrival"). Doing this also can make the unit of vmstat counters more unified. Finally, the unit of the vmstat counters are pages, kB and bytes. The B/KB suffix can tell us that the unit is bytes or kB. The rest which is without suffix are pages. Signed-off-by: Muchun Song <songmuchun-EC8Uxl6Npydl57MIdRCFDg@public.gmane.org> --- drivers/base/node.c | 3 +-- fs/proc/meminfo.c | 2 +- include/linux/mmzone.h | 3 ++- mm/rmap.c | 6 ++++-- 4 files changed, 8 insertions(+), 6 deletions(-) diff --git a/drivers/base/node.c b/drivers/base/node.c index 7a66aefe4e46..d02d86aec19f 100644 --- a/drivers/base/node.c +++ b/drivers/base/node.c @@ -465,8 +465,7 @@ static ssize_t node_read_meminfo(struct device *dev, nid, K(node_page_state(pgdat, NR_SHMEM_THPS)), nid, K(node_page_state(pgdat, NR_SHMEM_PMDMAPPED)), nid, K(node_page_state(pgdat, NR_FILE_THPS)), - nid, K(node_page_state(pgdat, NR_FILE_PMDMAPPED) * - HPAGE_PMD_NR) + nid, K(node_page_state(pgdat, NR_FILE_PMDMAPPED)) #endif ); len += hugetlb_report_node_meminfo(buf, len, nid); diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c index c61f440570f9..6fa761c9cc78 100644 --- a/fs/proc/meminfo.c +++ b/fs/proc/meminfo.c @@ -137,7 +137,7 @@ static int meminfo_proc_show(struct seq_file *m, void *v) show_val_kb(m, "FileHugePages: ", global_node_page_state(NR_FILE_THPS)); show_val_kb(m, "FilePmdMapped: ", - global_node_page_state(NR_FILE_PMDMAPPED) * HPAGE_PMD_NR); + global_node_page_state(NR_FILE_PMDMAPPED)); #endif #ifdef CONFIG_CMA diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 1ca6a34f6a4a..6139479ba417 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -219,7 +219,8 @@ static __always_inline bool vmstat_item_print_in_thp(enum node_stat_item item) return item == NR_ANON_THPS || item == NR_FILE_THPS || item == NR_SHMEM_THPS || - item == NR_SHMEM_PMDMAPPED; + item == NR_SHMEM_PMDMAPPED || + item == NR_FILE_PMDMAPPED; } /* diff --git a/mm/rmap.c b/mm/rmap.c index 1c1b576c0627..5ebf16fae4b9 100644 --- a/mm/rmap.c +++ b/mm/rmap.c @@ -1223,7 +1223,8 @@ void page_add_file_rmap(struct page *page, bool compound) __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED, nr_pages); else - __inc_node_page_state(page, NR_FILE_PMDMAPPED); + __mod_lruvec_page_state(page, NR_FILE_PMDMAPPED, + nr_pages); } else { if (PageTransCompound(page) && page_mapping(page)) { VM_WARN_ON_ONCE(!PageLocked(page)); @@ -1267,7 +1268,8 @@ static void page_remove_file_rmap(struct page *page, bool compound) __mod_lruvec_page_state(page, NR_SHMEM_PMDMAPPED, -nr_pages); else - __dec_node_page_state(page, NR_FILE_PMDMAPPED); + __mod_lruvec_page_state(page, NR_FILE_PMDMAPPED, + -nr_pages); } else { if (!atomic_add_negative(-1, &page->_mapcount)) return; -- 2.11.0
next prev parent reply other threads:[~2020-12-17 3:47 UTC|newest] Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-12-17 3:43 [PATCH v5 0/7] Convert all THP vmstat counters to pages Muchun Song 2020-12-17 3:43 ` [PATCH v5 1/7] mm: memcontrol: fix NR_ANON_THPS accounting in charge moving Muchun Song 2020-12-23 21:01 ` Shakeel Butt 2020-12-23 21:01 ` Shakeel Butt 2020-12-17 3:43 ` [PATCH v5 2/7] mm: memcontrol: convert NR_ANON_THPS account to pages Muchun Song 2020-12-23 22:08 ` Shakeel Butt 2020-12-23 22:08 ` Shakeel Butt 2020-12-24 2:45 ` [External] " Muchun Song 2020-12-24 2:45 ` Muchun Song 2020-12-24 2:45 ` Muchun Song 2020-12-17 3:43 ` [PATCH v5 3/7] mm: memcontrol: convert NR_FILE_THPS " Muchun Song 2020-12-17 3:43 ` [PATCH v5 4/7] mm: memcontrol: convert NR_SHMEM_THPS " Muchun Song 2020-12-17 3:43 ` Muchun Song 2020-12-17 3:43 ` [PATCH v5 5/7] mm: memcontrol: convert NR_SHMEM_PMDMAPPED " Muchun Song 2020-12-17 3:43 ` Muchun Song [this message] 2020-12-17 3:43 ` [PATCH v5 6/7] mm: memcontrol: convert NR_FILE_PMDMAPPED " Muchun Song 2020-12-17 3:43 ` [PATCH v5 7/7] mm: memcontrol: make the slab calculation consistent Muchun Song 2020-12-23 21:21 ` Shakeel Butt 2020-12-23 21:21 ` Shakeel Butt 2020-12-23 21:21 ` Shakeel Butt 2020-12-24 2:43 ` [External] " Muchun Song 2020-12-24 2:43 ` Muchun Song 2020-12-24 2:43 ` Muchun Song
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20201217034356.4708-7-songmuchun@bytedance.com \ --to=songmuchun@bytedance.com \ --cc=adobriyan@gmail.com \ --cc=akpm@linux-foundation.org \ --cc=cgroups@vger.kernel.org \ --cc=feng.tang@intel.com \ --cc=gregkh@linuxfoundation.org \ --cc=guro@fb.com \ --cc=hannes@cmpxchg.org \ --cc=hughd@google.com \ --cc=iamjoonsoo.kim@lge.com \ --cc=linux-fsdevel@vger.kernel.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-mm@kvack.org \ --cc=mhocko@kernel.org \ --cc=neilb@suse.de \ --cc=rafael@kernel.org \ --cc=rdunlap@infradead.org \ --cc=samitolvanen@google.com \ --cc=shakeelb@google.com \ --cc=vdavydov.dev@gmail.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.