From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-9.5 required=3.0 tests=DKIMWL_WL_HIGH,DKIM_SIGNED, DKIM_VALID,DKIM_VALID_AU,INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY, SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id AACFEC43441 for ; Wed, 10 Oct 2018 23:14:08 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 324C92098A for ; Wed, 10 Oct 2018 23:14:08 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (1024-bit key) header.d=kernel.org header.i=@kernel.org header.b="OzQ4yEob" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 324C92098A Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=kernel.org Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727981AbeJKGi1 (ORCPT ); Thu, 11 Oct 2018 02:38:27 -0400 Received: from mail.kernel.org ([198.145.29.99]:36812 "EHLO mail.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726525AbeJKGi0 (ORCPT ); Thu, 11 Oct 2018 02:38:26 -0400 Received: from lerouge.suse.de (LFbn-NCY-1-241-207.w83-194.abo.wanadoo.fr [83.194.85.207]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-SHA256 (128/128 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPSA id E4D512087A; Wed, 10 Oct 2018 23:14:01 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=kernel.org; s=default; t=1539213244; bh=fGt5FUWysEUqGMoZVChWVVopR2CrsR/JYmgoOU7AkSA=; h=From:To:Cc:Subject:Date:In-Reply-To:References:From; b=OzQ4yEobeSfOqggMVvaNAzRxXoU7C7ReqlUQlpQ+QuFnBbcaicV573juvInYz4f70 kfQ/9uDPwgbian62DQ8/rCMKekAR2EsZ2rIFeDj2WinKcyMIKkSbYqQzhw4nNaE/ff dpFkKse2lRoKnkv1ntg1jjAoEN7vaYr0hIPjgnFU= From: Frederic Weisbecker To: LKML Cc: Frederic Weisbecker , Sebastian Andrzej Siewior , Peter Zijlstra , "David S . Miller" , Linus Torvalds , Thomas Gleixner , Frederic Weisbecker , "Paul E . McKenney" , Ingo Molnar , Mauro Carvalho Chehab Subject: [RFC PATCH 30/30] softirq: Tasklet/net-rx fixup Date: Thu, 11 Oct 2018 01:12:17 +0200 Message-Id: <1539213137-13953-31-git-send-email-frederic@kernel.org> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1539213137-13953-1-git-send-email-frederic@kernel.org> References: <1539213137-13953-1-git-send-email-frederic@kernel.org> Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Frederic Weisbecker Tasklets and net-rx vectors don't quite get along. If one is interrupted by another, we may run into a nasty spin_lock recursion: [ 135.427198] Call Trace: [ 135.429650] [ 135.431690] dump_stack+0x67/0x95 [ 135.435024] spin_bug+0x95/0xf0 [ 135.438187] do_raw_spin_lock+0x77/0xa0 [ 135.442079] _raw_spin_lock_nested+0x40/0x50 [ 135.446439] ? tcp_v4_rcv+0x9da/0xb10 [ 135.450131] tcp_v4_rcv+0x9da/0xb10 [ 135.453650] ? ip_local_deliver+0x78/0x260 [ 135.457758] ip_local_deliver+0xdf/0x260 [ 135.461728] ip_rcv+0x4e/0x80 [ 135.464716] __netif_receive_skb_one_core+0x55/0x80 [ 135.469623] __netif_receive_skb+0x1b/0x70 [ 135.473757] netif_receive_skb_internal+0x92/0x390 [ 135.478574] napi_gro_receive+0xdf/0x1a0 [ 135.482545] rtl8169_poll+0x2b8/0x670 [ 135.486211] net_rx_action+0x1f8/0x3e0 [ 135.489989] __do_softirq+0x1a0/0x63c [ 135.493691] irq_exit+0x10f/0x120 [ 135.497033] do_IRQ+0x71/0x130 [ 135.500137] common_interrupt+0xf/0xf [ 135.503839] RIP: 0010:_raw_spin_unlock_irqrestore+0x59/0x70 [ 135.509471] Code: 75 21 53 9d e8 e8 1d a5 ff bf 01 00 00 00 e8 8e f6 97 ff 65 8b 05 ef 06 8d 7e 85 c0 74 0e 5b 41 5c 5d c3 e8 c9 20 a5 ff 53 9d dd e8 90 d4 8b ff 5b 41 5c 5d c3 66 66 2e 0f 1f 84 00 00 00 00 [ 135.528347] RSP: 0018:ffff88021fb03d28 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffde [ 135.535989] RAX: ffff880217762480 RBX: 0000000000000246 RCX: 0000000000000002 [ 135.543201] RDX: 0000000000000000 RSI: ffff880217762c70 RDI: ffff880217762480 [ 135.550332] RBP: ffff88021fb03d38 R08: 0000000000000001 R09: 0000000000000000 [ 135.557519] R10: 0000000000000000 R11: 0000000000000000 R12: ffff88021fa59b40 [ 135.564719] R13: ffff88021fa59b40 R14: 00000000fffd78b5 R15: 000000000e000001 [ 135.571905] ? common_interrupt+0xa/0xf [ 135.575783] mod_timer+0x196/0x440 [ 135.579221] sk_reset_timer+0x18/0x30 [ 135.582940] tcp_schedule_loss_probe+0xe9/0x120 [ 135.587515] tcp_write_xmit+0x2c4/0x1240 [ 135.591468] tcp_tsq_write.part.46+0x5e/0xb0 [ 135.595756] tcp_tsq_handler+0xa3/0xb0 [ 135.599534] tcp_tasklet_func+0xdc/0x120 [ 135.603488] tasklet_action_common.isra.17+0xa3/0xb0 [ 135.608471] tasklet_action+0x2d/0x30 [ 135.612161] __do_softirq+0x1a0/0x63c [ 135.615847] irq_exit+0x10f/0x120 [ 135.619173] do_IRQ+0x71/0x130 [ 135.622251] common_interrupt+0xf/0xf [ 135.625949] This is an ugly workaround until we find a proper solution. Signed-off-by: Frederic Weisbecker Cc: Ingo Molnar Cc: Sebastian Andrzej Siewior Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Linus Torvalds Cc: David S. Miller Cc: Mauro Carvalho Chehab Cc: Paul E. McKenney --- kernel/softirq.c | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/kernel/softirq.c b/kernel/softirq.c index f4cb1ea..d95295f 100644 --- a/kernel/softirq.c +++ b/kernel/softirq.c @@ -320,6 +320,7 @@ asmlinkage __visible void __softirq_entry __do_softirq(void) unsigned long old_flags = current->flags; int max_restart = MAX_SOFTIRQ_RESTART; struct softirq_action *h; + bool tasklet_enabled = false, net_rx_enabled = false; bool in_hardirq; __u32 pending; int softirq_bit; @@ -338,6 +339,10 @@ asmlinkage __visible void __softirq_entry __do_softirq(void) in_hardirq = lockdep_softirq_start(); restart: + if (local_softirq_enabled() & TASKLET_SOFTIRQ) + tasklet_enabled = true; + if (local_softirq_enabled() & NET_RX_SOFTIRQ) + net_rx_enabled = true; /* Reset the pending bitmask before enabling irqs */ softirq_pending_nand(pending); @@ -358,8 +363,16 @@ asmlinkage __visible void __softirq_entry __do_softirq(void) trace_softirq_entry(vec_nr); softirq_enabled_nand(BIT(vec_nr)); + if (vec_nr == NET_RX_SOFTIRQ && tasklet_enabled) + softirq_enabled_nand(BIT(TASKLET_SOFTIRQ)); + if (vec_nr == TASKLET_SOFTIRQ && net_rx_enabled) + softirq_enabled_nand(BIT(NET_RX_SOFTIRQ)); barrier(); h->action(h); + if (vec_nr == TASKLET_SOFTIRQ && net_rx_enabled) + softirq_enabled_or(BIT(NET_RX_SOFTIRQ)); + if (vec_nr == NET_RX_SOFTIRQ && tasklet_enabled) + softirq_enabled_or(BIT(TASKLET_SOFTIRQ)); softirq_enabled_or(BIT(vec_nr)); trace_softirq_exit(vec_nr); if (unlikely(prev_count != preempt_count())) { -- 2.7.4