From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Morton Subject: [folded-merged] mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-v5.patch removed from -mm tree Date: Wed, 08 Apr 2020 17:57:37 -0700 Message-ID: <20200409005737.7PuVA0imE%akpm@linux-foundation.org> References: <20200406200254.a69ebd9e08c4074e41ddebaf@linux-foundation.org> Reply-To: linux-kernel@vger.kernel.org Return-path: Received: from mail.kernel.org ([198.145.29.99]:44262 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726527AbgDIA5i (ORCPT ); Wed, 8 Apr 2020 20:57:38 -0400 In-Reply-To: <20200406200254.a69ebd9e08c4074e41ddebaf@linux-foundation.org> Sender: mm-commits-owner@vger.kernel.org List-Id: mm-commits@vger.kernel.org To: andreas.schaufler@gmx.de, aslan@fb.com, guro@fb.com, js1304@gmail.com, mhocko@kernel.org, mike.kravetz@oracle.com, mm-commits@vger.kernel.org, rdunlap@infradead.org, riel@surriel.com The patch titled Subject: mm: hugetlb: optionally allocate gigantic hugepages using cma has been removed from the -mm tree. Its filename was mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-v5.patch This patch was dropped because it was folded into mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma.patch ------------------------------------------------------ From: Roman Gushchin Subject: mm: hugetlb: optionally allocate gigantic hugepages using cma Commit 944d9fec8d7a ("hugetlb: add support for gigantic page allocation at runtime") has added the run-time allocation of gigantic pages. However it actually works only at early stages of the system loading, when the majority of memory is free. After some time the memory gets fragmented by non-movable pages, so the chances to find a contiguous 1 GB block are getting close to zero. Even dropping caches manually doesn't help a lot. At large scale rebooting servers in order to allocate gigantic hugepages is quite expensive and complex. At the same time keeping some constant percentage of memory in reserved hugepages even if the workload isn't using it is a big waste: not all workloads can benefit from using 1 GB pages. The following solution can solve the problem: 1) On boot time a dedicated cma area* is reserved. The size is passed as a kernel argument. 2) Run-time allocations of gigantic hugepages are performed using the cma allocator and the dedicated cma area In this case gigantic hugepages can be allocated successfully with a high probability, however the memory isn't completely wasted if nobody is using 1GB hugepages: it can be used for pagecache, anon memory, THPs, etc. * On a multi-node machine a per-node cma area is allocated on each node. Following gigantic hugetlb allocation are using the first available numa node if the mask isn't specified by a user. Usage: 1) configure the kernel to allocate a cma area for hugetlb allocations: pass hugetlb_cma=10G as a kernel argument 2) allocate hugetlb pages as usual, e.g. echo 10 > /sys/kernel/mm/hugepages/hugepages-1048576kB/nr_hugepages If the option isn't enabled or the allocation of the cma area failed, the current behavior of the system is preserved. x86 and arm-64 are covered by this patch, other architectures can be trivially added later. The patch contains clean-ups and fixes proposed and implemented by Aslan Bakirov and Randy Dunlap. It also contains ideas and suggestions proposed by Rik van Riel, Michal Hocko and Mike Kravetz. Thanks! Link: http://lkml.kernel.org/r/20200407163840.92263-3-guro@fb.com Signed-off-by: Roman Gushchin Tested-by: Andreas Schaufler Acked-by: Mike Kravetz Acked-by: Michal Hocko Cc: Aslan Bakirov Cc: Randy Dunlap Cc: Rik van Riel Cc: Joonsoo Kim Signed-off-by: Andrew Morton --- Documentation/admin-guide/kernel-parameters.txt | 7 + include/linux/hugetlb.h | 4 + mm/hugetlb.c | 54 ++++++-------- 3 files changed, 34 insertions(+), 31 deletions(-) --- a/Documentation/admin-guide/kernel-parameters.txt~mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-v5 +++ a/Documentation/admin-guide/kernel-parameters.txt @@ -1475,12 +1475,13 @@ hpet_mmap= [X86, HPET_MMAP] Allow userspace to mmap HPET registers. Default set by CONFIG_HPET_MMAP_DEFAULT. - hugetlb_cma= [x86-64] The size of a cma area used for allocation + hugetlb_cma= [HW] The size of a cma area used for allocation of gigantic hugepages. Format: nn[KMGTPE] - If enabled, boot-time allocation of gigantic hugepages - is skipped. + Reserve a cma area of given size and allocate gigantic + hugepages using the cma allocator. If enabled, the + boot-time allocation of gigantic hugepages is skipped. hugepages= [HW,X86-32,IA-64] HugeTLB pages to allocate at boot. hugepagesz= [HW,IA-64,PPC,X86-64] The size of the HugeTLB pages. --- a/include/linux/hugetlb.h~mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-v5 +++ a/include/linux/hugetlb.h @@ -897,10 +897,14 @@ static inline spinlock_t *huge_pte_lock( #if defined(CONFIG_HUGETLB_PAGE) && defined(CONFIG_CMA) extern void __init hugetlb_cma_reserve(int order); +extern void __init hugetlb_cma_check(void); #else static inline __init void hugetlb_cma_reserve(int order) { } +static inline __init void hugetlb_cma_check(void) +{ +} #endif #endif /* _LINUX_HUGETLB_H */ --- a/mm/hugetlb.c~mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-v5 +++ a/mm/hugetlb.c @@ -1255,7 +1255,7 @@ static struct page *alloc_gigantic_page( for_each_node_mask(node, *nodemask) { if (!hugetlb_cma[node]) - break; + continue; page = cma_alloc(hugetlb_cma[node], nr_pages, huge_page_order(h), true); @@ -1308,8 +1308,14 @@ static void update_and_free_page(struct set_compound_page_dtor(page, NULL_COMPOUND_DTOR); set_page_refcounted(page); if (hstate_is_gigantic(h)) { + /* + * Temporarily drop the hugetlb_lock, because + * we might block in free_gigantic_page(). + */ + spin_unlock(&hugetlb_lock); destroy_compound_gigantic_page(page, huge_page_order(h)); free_gigantic_page(page, huge_page_order(h)); + spin_lock(&hugetlb_lock); } else { __free_pages(page, huge_page_order(h)); } @@ -3225,6 +3231,7 @@ static int __init hugetlb_init(void) default_hstate.max_huge_pages = default_hstate_max_huge_pages; } + hugetlb_cma_check(); hugetlb_init_hstates(); gather_bootmem_prealloc(); report_hugepages(); @@ -5540,6 +5547,7 @@ void move_hugetlb_state(struct page *old #ifdef CONFIG_CMA static unsigned long hugetlb_cma_size __initdata; +static bool cma_reserve_called __initdata; static int __init cmdline_parse_hugetlb_cma(char *p) { @@ -5554,6 +5562,8 @@ void __init hugetlb_cma_reserve(int orde unsigned long size, reserved, per_node; int nid; + cma_reserve_called = true; + if (!hugetlb_cma_size) return; @@ -5573,38 +5583,18 @@ void __init hugetlb_cma_reserve(int orde reserved = 0; for_each_node_state(nid, N_ONLINE) { - unsigned long start_pfn, end_pfn; - unsigned long min_pfn = 0, max_pfn = 0; - int res, i; - - for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) { - if (!min_pfn) - min_pfn = start_pfn; - max_pfn = end_pfn; - } + int res; size = min(per_node, hugetlb_cma_size - reserved); size = round_up(size, PAGE_SIZE << order); - if (size > ((max_pfn - min_pfn) << PAGE_SHIFT) / 2) { - pr_warn("hugetlb_cma: cma_area is too big, please try less than %lu MiB\n", - round_down(((max_pfn - min_pfn) << PAGE_SHIFT) * - nr_online_nodes / 2 / SZ_1M, - PAGE_SIZE << order)); - break; - } - - res = cma_declare_contiguous(PFN_PHYS(min_pfn), size, - PFN_PHYS(max_pfn), - PAGE_SIZE << order, - 0, false, - "hugetlb", &hugetlb_cma[nid]); + res = cma_declare_contiguous_nid(0, size, 0, PAGE_SIZE << order, + 0, false, "hugetlb", + &hugetlb_cma[nid], nid); if (res) { - phys_addr_t begpa = PFN_PHYS(min_pfn); - phys_addr_t endpa = PFN_PHYS(max_pfn); - pr_warn("%s: reservation failed: err %d, node %d, [%pap, %pap)\n", - __func__, res, nid, &begpa, &endpa); - break; + pr_warn("hugetlb_cma: reservation failed: err %d, node %d", + res, nid); + continue; } reserved += size; @@ -5616,4 +5606,12 @@ void __init hugetlb_cma_reserve(int orde } } +void __init hugetlb_cma_check(void) +{ + if (!hugetlb_cma_size || cma_reserve_called) + return; + + pr_warn("hugetlb_cma: the option isn't supported by current arch\n"); +} + #endif /* CONFIG_CMA */ _ Patches currently in -mm which might be from guro@fb.com are mmpage_alloccma-conditionally-prefer-cma-pageblocks-for-movable-allocations.patch mmpage_alloccma-conditionally-prefer-cma-pageblocks-for-movable-allocations-fix.patch mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma.patch mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-fix.patch mm-hugetlb-optionally-allocate-gigantic-hugepages-using-cma-fix-2.patch