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 Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 98103EB8FB1 for ; Fri, 8 Sep 2023 00:26:30 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S232832AbjIHA0c (ORCPT ); Thu, 7 Sep 2023 20:26:32 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:33838 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S238560AbjIHA0c (ORCPT ); Thu, 7 Sep 2023 20:26:32 -0400 Received: from mail-io1-xd2f.google.com (mail-io1-xd2f.google.com [IPv6:2607:f8b0:4864:20::d2f]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id B81BC1BD3 for ; Thu, 7 Sep 2023 17:26:27 -0700 (PDT) Received: by mail-io1-xd2f.google.com with SMTP id ca18e2360f4ac-792623074edso67958839f.1 for ; Thu, 07 Sep 2023 17:26:27 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=joelfernandes.org; s=google; t=1694132787; x=1694737587; darn=vger.kernel.org; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:from:to:cc:subject:date:message-id:reply-to; bh=HVZN5vN8bRfWt9fXaRaMQ7INcbyUKWi7temjOgZW5Mk=; b=YLWL35y1Yh5HcyyrloFoTSDk57dkqiXMEmYi+VQY7J+NUFwoj+P7N/yEmM7Avamvtq O2ONTZjiTmLwiHbPPX2T0e3jU0D0Q2oluhHh8vW6Dtfb/MUxg2OVv7qiJS2JXLdGKJzu 9sIIul9sQthqwN5PXDVGLT/wazBQ+DqwbcEJs= X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1694132787; x=1694737587; h=in-reply-to:content-disposition:mime-version:references:message-id :subject:cc:to:from:date:x-gm-message-state:from:to:cc:subject:date :message-id:reply-to; bh=HVZN5vN8bRfWt9fXaRaMQ7INcbyUKWi7temjOgZW5Mk=; b=xMNkjFhrPPOnPUf7MQ2tH9tlV8mknMSDSZC5TNE33DEuR7b1jbsu9pySzSSaQckiVS BX1A1+zo++eYJ+Ytt8L5j296SGcEAKgxTR/MswiczL7rUfauvsnjvVRtcI3IatetIQEF Je++PO6TWNxm8EMC/ikldEVeJSoH3MwOyd+V3iVsZWuqvma7qcp33OJaDuB01HEs4slR 2TR7qODyOCeFVEBAoAaJGS/98IHhJ8WQF9+3W4+6DzuobMxN0uRKP65PLfkpm+8az0WO GW7B+MZlwfaofJH0ogGs2lO57z6v+XFTYgUqnoy55C5jO+FUbhn/JrOdWSRPyEkXLunb KUvg== X-Gm-Message-State: AOJu0YzMKW/E9gV3OgbsjbAy26qgK76NKrlU/9B0tdCGZO9DXJ0lJ9tL uZ61B43BTu4i4XZ76one0m7Wpg== X-Google-Smtp-Source: AGHT+IFRVFtwWrfcVc0Z7SXHe2h5aZvauo+UgAqJm05OoRDsjuegTeHmvw/xRiZ8zZNk0DTKgvpQMw== X-Received: by 2002:a5d:8048:0:b0:783:498c:9cf0 with SMTP id b8-20020a5d8048000000b00783498c9cf0mr1256406ior.4.1694132787021; Thu, 07 Sep 2023 17:26:27 -0700 (PDT) Received: from localhost (156.190.123.34.bc.googleusercontent.com. [34.123.190.156]) by smtp.gmail.com with ESMTPSA id x5-20020a6bda05000000b0076373f90e46sm168208iob.33.2023.09.07.17.26.26 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Thu, 07 Sep 2023 17:26:26 -0700 (PDT) Date: Fri, 8 Sep 2023 00:26:26 +0000 From: Joel Fernandes To: Vlastimil Babka Cc: Lorenzo Stoakes , linux-kernel@vger.kernel.org, Andrew Morton , "Paul E. McKenney" , Zqiang , Zhen Lei , rcu@vger.kernel.org, Matthew Wilcox , stable@vger.kernel.org, linux-mm@kvack.org Subject: Re: [PATCH v3 2/2] rcu: Dump vmalloc memory info safely Message-ID: <20230908002626.GB4088026@google.com> References: <20230904180806.1002832-1-joel@joelfernandes.org> <20230904180806.1002832-2-joel@joelfernandes.org> <9e329429-73a5-4926-af4f-edcf9e547101@lucifer.local> <20230905114841.GB3881391@google.com> <737d399d-c83a-3e66-ceb7-4ae7ba4acfb4@suse.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <737d399d-c83a-3e66-ceb7-4ae7ba4acfb4@suse.cz> Precedence: bulk List-ID: X-Mailing-List: rcu@vger.kernel.org On Thu, Sep 07, 2023 at 09:10:38AM +0200, Vlastimil Babka wrote: > On 9/6/23 21:18, Lorenzo Stoakes wrote: > > On Tue, 5 Sept 2023 at 12:48, Joel Fernandes wrote: > >> > >> On Tue, Sep 05, 2023 at 08:00:44AM +0100, Lorenzo Stoakes wrote: > >> > On Mon, Sep 04, 2023 at 06:08:05PM +0000, Joel Fernandes (Google) wrote: > >> > > From: Zqiang > >> > > > >> > > Currently, for double invoke call_rcu(), will dump rcu_head objects > >> > > memory info, if the objects is not allocated from the slab allocator, > >> > > the vmalloc_dump_obj() will be invoke and the vmap_area_lock spinlock > >> > > need to be held, since the call_rcu() can be invoked in interrupt context, > >> > > therefore, there is a possibility of spinlock deadlock scenarios. > >> > > > >> > > And in Preempt-RT kernel, the rcutorture test also trigger the following > >> > > lockdep warning: > >> > > > >> > > BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:48 > >> > > in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 1, name: swapper/0 > >> > > preempt_count: 1, expected: 0 > >> > > RCU nest depth: 1, expected: 1 > >> > > 3 locks held by swapper/0/1: > >> > > #0: ffffffffb534ee80 (fullstop_mutex){+.+.}-{4:4}, at: torture_init_begin+0x24/0xa0 > >> > > #1: ffffffffb5307940 (rcu_read_lock){....}-{1:3}, at: rcu_torture_init+0x1ec7/0x2370 > >> > > #2: ffffffffb536af40 (vmap_area_lock){+.+.}-{3:3}, at: find_vmap_area+0x1f/0x70 > >> > > irq event stamp: 565512 > >> > > hardirqs last enabled at (565511): [] __call_rcu_common+0x218/0x940 > >> > > hardirqs last disabled at (565512): [] rcu_torture_init+0x20b2/0x2370 > >> > > softirqs last enabled at (399112): [] __local_bh_enable_ip+0x126/0x170 > >> > > softirqs last disabled at (399106): [] inet_register_protosw+0x9/0x1d0 > >> > > Preemption disabled at: > >> > > [] rcu_torture_init+0x1f13/0x2370 > >> > > CPU: 0 PID: 1 Comm: swapper/0 Tainted: G W 6.5.0-rc4-rt2-yocto-preempt-rt+ #15 > >> > > Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.16.2-0-gea1b7a073390-prebuilt.qemu.org 04/01/2014 > >> > > Call Trace: > >> > > > >> > > dump_stack_lvl+0x68/0xb0 > >> > > dump_stack+0x14/0x20 > >> > > __might_resched+0x1aa/0x280 > >> > > ? __pfx_rcu_torture_err_cb+0x10/0x10 > >> > > rt_spin_lock+0x53/0x130 > >> > > ? find_vmap_area+0x1f/0x70 > >> > > find_vmap_area+0x1f/0x70 > >> > > vmalloc_dump_obj+0x20/0x60 > >> > > mem_dump_obj+0x22/0x90 > >> > > __call_rcu_common+0x5bf/0x940 > >> > > ? debug_smp_processor_id+0x1b/0x30 > >> > > call_rcu_hurry+0x14/0x20 > >> > > rcu_torture_init+0x1f82/0x2370 > >> > > ? __pfx_rcu_torture_leak_cb+0x10/0x10 > >> > > ? __pfx_rcu_torture_leak_cb+0x10/0x10 > >> > > ? __pfx_rcu_torture_init+0x10/0x10 > >> > > do_one_initcall+0x6c/0x300 > >> > > ? debug_smp_processor_id+0x1b/0x30 > >> > > kernel_init_freeable+0x2b9/0x540 > >> > > ? __pfx_kernel_init+0x10/0x10 > >> > > kernel_init+0x1f/0x150 > >> > > ret_from_fork+0x40/0x50 > >> > > ? __pfx_kernel_init+0x10/0x10 > >> > > ret_from_fork_asm+0x1b/0x30 > >> > > > >> > > > >> > > The previous patch fixes this by using the deadlock-safe best-effort > >> > > version of find_vm_area. However, in case of failure print the fact that > >> > > the pointer was a vmalloc pointer so that we print at least something. > >> > > > >> > > Reported-by: Zhen Lei > >> > > Cc: Paul E. McKenney > >> > > Cc: rcu@vger.kernel.org > >> > > Reviewed-by: Matthew Wilcox (Oracle) > >> > > Fixes: 98f180837a89 ("mm: Make mem_dump_obj() handle vmalloc() memory") > >> > > Cc: stable@vger.kernel.org > >> > > Signed-off-by: Zqiang > >> > > Signed-off-by: Joel Fernandes (Google) > >> > > --- > >> > > mm/util.c | 4 +++- > >> > > 1 file changed, 3 insertions(+), 1 deletion(-) > >> > > > >> > > diff --git a/mm/util.c b/mm/util.c > >> > > index dd12b9531ac4..406634f26918 100644 > >> > > --- a/mm/util.c > >> > > +++ b/mm/util.c > >> > > @@ -1071,7 +1071,9 @@ void mem_dump_obj(void *object) > >> > > if (vmalloc_dump_obj(object)) > >> > > return; > >> > > > >> > > - if (virt_addr_valid(object)) > >> > > + if (is_vmalloc_addr(object)) > >> > > + type = "vmalloc memory"; > >> > > + else if (virt_addr_valid(object)) > >> > > type = "non-slab/vmalloc memory"; > >> > > >> > I think you should update this to say non-slab/non-vmalloc memory (as much > >> > as that description sucks!) as this phrasing in the past meant to say > >> > 'non-slab or vmalloc memory' (already confusing phrasing) so better to be > >> > clear. > >> > >> True, though the issue you mentioned it is in existing code, a respin of this > >> patch could update it to say non-vmalloc. Good point, thanks for reviewing! > > > > No it's not, you're changing the meaning, because you changed the code > > that determines the output... > > I think it has always meant (but clearly it's not unambiguously worded) "not > slab && not vmalloc", that is before and after this patch. Only in case > patch 1 is applied and patch 2 not, can the output be wrong in that a > vmalloc pointer will (in case of trylock fail) be classified as "not slab && > not vmalloc", but seems fine to me after patch 2. > > I guess if we wanted, we could also rewrite it to be more like the kmem > check in the beginning of mem_dump_obj(), so there would be: > > if (is_vmalloc_addr(...)) { > vmalloc_dump_obj(...); > return; > } > > where vmalloc_dump_obj() itself would print at least "vmalloc memory" with > no further details in case of trylock failure. > > that assumes is_vmalloc_addr() is guaranteed to be true for all addresses > that __find_vmap_area resolves, otherwise it could miss something compared > to current code. Is it guaranteed? It is guaranteed based on my reading of the code. But maybe it may aid additional vmalloc-internals debugging if for some reason the address of the object stored in the vmalloc data structures is out of bound for some reason and the lookup actually succeded. That's just a hypothetical situation though and I don't think that that can actually happen. thanks, - Joel