From mboxrd@z Thu Jan 1 00:00:00 1970 From: Anatoly Burakov Subject: [PATCH 3/3] eal: attempt multiple hugepage allocations at init Date: Fri, 22 Feb 2019 16:14:03 +0000 Message-ID: <0a770fb3968c0a270eb57ff7764d6fba4cf6f76d.1550851998.git.anatoly.burakov@intel.com> References: Cc: maxime.leroy@6wind.com To: dev@dpdk.org Return-path: Received: from mga17.intel.com (mga17.intel.com [192.55.52.151]) by dpdk.org (Postfix) with ESMTP id BDF572C30 for ; Fri, 22 Feb 2019 17:14:06 +0100 (CET) In-Reply-To: In-Reply-To: References: List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" When requesting memory with ``-m`` or ``--socket-mem`` flags, currently the init will fail if the requested memory amount was bigger than any one memseg list, even if total amount of available memory was sufficient. Fix this by making EAL to attempt to allocate pages multiple times, until we either fulfill our memory requirements, or run out of hugepages to allocate. Bugzilla ID: 95 Signed-off-by: Anatoly Burakov --- lib/librte_eal/linuxapp/eal/eal_memory.c | 47 ++++++++++++++++-------- 1 file changed, 32 insertions(+), 15 deletions(-) diff --git a/lib/librte_eal/linuxapp/eal/eal_memory.c b/lib/librte_eal/linuxapp/eal/eal_memory.c index 1b96b576e..361109eb3 100644 --- a/lib/librte_eal/linuxapp/eal/eal_memory.c +++ b/lib/librte_eal/linuxapp/eal/eal_memory.c @@ -1784,30 +1784,47 @@ eal_hugepage_init(void) struct rte_memseg **pages; struct hugepage_info *hpi = &used_hp[hp_sz_idx]; unsigned int num_pages = hpi->num_pages[socket_id]; - int num_pages_alloc, i; + unsigned int num_pages_alloc; if (num_pages == 0) continue; - pages = malloc(sizeof(*pages) * num_pages); - RTE_LOG(DEBUG, EAL, "Allocating %u pages of size %" PRIu64 "M on socket %i\n", num_pages, hpi->hugepage_sz >> 20, socket_id); - num_pages_alloc = eal_memalloc_alloc_seg_bulk(pages, - num_pages, hpi->hugepage_sz, - socket_id, true); - if (num_pages_alloc < 0) { + /* we may not be able to allocate all pages in one go, + * because we break up our memory map into multiple + * memseg lists. therefore, try allocating multiple + * times and see if we can get the desired number of + * pages from multiple allocations. + */ + + num_pages_alloc = 0; + do { + int i, cur_pages, needed; + + needed = num_pages - num_pages_alloc; + + pages = malloc(sizeof(*pages) * needed); + + /* do not request exact number of pages */ + cur_pages = eal_memalloc_alloc_seg_bulk(pages, + needed, hpi->hugepage_sz, + socket_id, false); + if (cur_pages <= 0) { + free(pages); + return -1; + } + + /* mark preallocated pages as unfreeable */ + for (i = 0; i < cur_pages; i++) { + struct rte_memseg *ms = pages[i]; + ms->flags |= RTE_MEMSEG_FLAG_DO_NOT_FREE; + } free(pages); - return -1; - } - /* mark preallocated pages as unfreeable */ - for (i = 0; i < num_pages_alloc; i++) { - struct rte_memseg *ms = pages[i]; - ms->flags |= RTE_MEMSEG_FLAG_DO_NOT_FREE; - } - free(pages); + num_pages_alloc += cur_pages; + } while (num_pages_alloc != num_pages); } } /* if socket limits were specified, set them */ -- 2.17.1