Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751067AbdGMCOZ (ORCPT ); Wed, 12 Jul 2017 22:14:25 -0400 Received: from mail-pg0-f49.google.com ([74.125.83.49]:34269 "EHLO mail-pg0-f49.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750899AbdGMCOX (ORCPT ); Wed, 12 Jul 2017 22:14:23 -0400 From: Joel Fernandes To: linux-kernel@vger.kernel.org Cc: kernel-team@android.com, linux-mm@kvack.org, Joel Fernandes , Tim Murray , Ingo Molnar , Steven Rostedt , Andrew Morton Subject: [for-next][PATCH v2] tracing/ring_buffer: Try harder to allocate Date: Wed, 12 Jul 2017 19:14:16 -0700 Message-Id: <20170713021416.8897-1-joelaf@google.com> X-Mailer: git-send-email 2.13.2.932.g7449e964c-goog Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2584 Lines: 64 ftrace can fail to allocate per-CPU ring buffer on systems with a large number of CPUs coupled while large amounts of cache happening in the page cache. Currently the ring buffer allocation doesn't retry in the VM implementation even if direct-reclaim made some progress but still wasn't able to find a free page. On retrying I see that the allocations almost always succeed. The retry doesn't happen because __GFP_NORETRY is used in the tracer to prevent the case where we might OOM, however if we drop __GFP_NORETRY, we risk destabilizing the system if OOM killer is triggered. To prevent this situation, use the __GFP_RETRY_MAYFAIL flag introduced recently [1]. Tested the following still succeeds without destabilizing a system with 1GB memory. echo 300000 > /sys/kernel/debug/tracing/buffer_size_kb [1] https://marc.info/?l=linux-mm&m=149820805124906&w=2 Cc: Tim Murray Cc: Ingo Molnar Cc: Steven Rostedt Cc: Andrew Morton Acked-by: Vlastimil Babka Acked-by: Johannes Weiner Acked-by: Michal Hocko Signed-off-by: Joel Fernandes --- changes since v1: Added acks and dropped stable kernel/trace/ring_buffer.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index 4ae268e687fe..529cc50d7243 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -1136,12 +1136,12 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) for (i = 0; i < nr_pages; i++) { struct page *page; /* - * __GFP_NORETRY flag makes sure that the allocation fails - * gracefully without invoking oom-killer and the system is - * not destabilized. + * __GFP_RETRY_MAYFAIL flag makes sure that the allocation fails + * gracefully without invoking oom-killer and the system is not + * destabilized. */ bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()), - GFP_KERNEL | __GFP_NORETRY, + GFP_KERNEL | __GFP_RETRY_MAYFAIL, cpu_to_node(cpu)); if (!bpage) goto free_pages; @@ -1149,7 +1149,7 @@ static int __rb_allocate_pages(long nr_pages, struct list_head *pages, int cpu) list_add(&bpage->list, pages); page = alloc_pages_node(cpu_to_node(cpu), - GFP_KERNEL | __GFP_NORETRY, 0); + GFP_KERNEL | __GFP_RETRY_MAYFAIL, 0); if (!page) goto free_pages; bpage->page = page_address(page); -- 2.13.2.725.g09c95d1e9-goog