From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S965534AbaGPP5F (ORCPT ); Wed, 16 Jul 2014 11:57:05 -0400 Received: from mail-yh0-f44.google.com ([209.85.213.44]:65078 "EHLO mail-yh0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965317AbaGPP5D (ORCPT ); Wed, 16 Jul 2014 11:57:03 -0400 Message-ID: <53C6A0E7.30006@gmail.com> Date: Wed, 16 Jul 2014 11:57:27 -0400 From: Pranith Kumar User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.6.0 MIME-Version: 1.0 To: paulmck@linux.vnet.ibm.com, Ingo Molnar CC: linux-kernel@vger.kernel.org, fabf@skynet.be, davidshan@tencent.com, joe@perches.com, keescook@chromium.org, Peter Zijlstra Subject: Re: [GIT PULL rcu/next] RCU commits for 3.17 References: <20140713184237.GA30855@linux.vnet.ibm.com> <20140716115046.GC28434@gmail.com> <20140716131322.GA16396@gmail.com> <20140716141707.GU8690@linux.vnet.ibm.com> <20140716152629.GA9340@linux.vnet.ibm.com> In-Reply-To: <20140716152629.GA9340@linux.vnet.ibm.com> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 07/16/2014 11:26 AM, Paul E. McKenney wrote: > On Wed, Jul 16, 2014 at 07:17:07AM -0700, Paul E. McKenney wrote: >> On Wed, Jul 16, 2014 at 03:13:22PM +0200, Ingo Molnar wrote: >>> >>> * Ingo Molnar wrote: >>> >>>> >>>> * Paul E. McKenney wrote: >>>> >>>>> Hello, Ingo, >>>>> >>>>> The changes in this series include: >>>>> >>>>> 1. Update RCU documentation. These were posted to LKML at >>>>> https://lkml.org/lkml/2014/7/7/650. >>>>> >>>>> 2. Miscellaneous fixes. These were posted to LKML at >>>>> https://lkml.org/lkml/2014/7/7/678. >>>>> >>>>> 3. Maintainership changes. These were posted to LKML at >>>>> https://lkml.org/lkml/2014/7/7/713, with a couple of >>>>> additional at https://lkml.org/lkml/2014/7/3/812 and >>>>> https://lkml.org/lkml/2014/6/2/585. >>>>> >>>>> 4. Torture-test updates. These were posted to LKML at >>>>> https://lkml.org/lkml/2014/7/7/816. >>>>> >>>>> 5. Callback-offloading changes. These were posted to LKML at >>>>> https://lkml.org/lkml/2014/7/7/1007. >>>>> >>>>> All of these have been exposed to -next testing. >>> >>> JFYI, the attached x86 (rand!-) config crashes on early bootup: >>> >>> [ 0.000000] RCU: Adjusting geometry for rcu_fanout_leaf=16, nr_cpu_ids=2 >>> [ 0.000000] ------------[ cut here ]------------ >>> [ 0.000000] WARNING: CPU: 0 PID: 0 at arch/x86/kernel/cpu/common.c:1439 warn_pre_alternatives+0x1e/0x20() >>> [ 0.000000] You're using static_cpu_has before alternatives have run! >>> [ 0.000000] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.16.0-rc5+ #236363 >>> [ 0.000000] Hardware name: System manufacturer System Product Name/A8N-E, BIOS ASUS A8N-E ACPI BIOS Revision 1008 08/22/2005 >>> [ 0.000000] 0000000000000000 ffffffff82787c60 ffffffff81ff4ec0 ffffffff82787ca8 >>> [ 0.000000] ffffffff82787c98 ffffffff810651ac ffffffff81011531 ffffffff82787e48 >>> [ 0.000000] 0000000000000000 0000000000000000 ffffffff827cf880 ffffffff82787cf8 >>> [ 0.000000] Call Trace: >>> [ 0.000000] [] dump_stack+0x4d/0x66 >>> [ 0.000000] [] warn_slowpath_common+0x7a/0x93 >>> [ 0.000000] [] ? warn_pre_alternatives+0x1e/0x20 >>> [ 0.000000] [] warn_slowpath_fmt+0x4c/0x4e >>> [ 0.000000] [] ? irq_return+0x7/0x7 >>> [ 0.000000] [] warn_pre_alternatives+0x1e/0x20 >>> [ 0.000000] [] __do_page_fault+0xc3/0x43f >>> [ 0.000000] [] ? print_context_stack+0x6a/0xb6 >>> [ 0.000000] [] ? dump_trace+0x27d/0x294 >>> [ 0.000000] [] ? number.isra.1+0x127/0x22c >>> [ 0.000000] [] ? print_time.part.5+0x58/0x5c >>> [ 0.000000] [] ? sched_clock_cpu+0x11/0xb9 >>> [ 0.000000] [] do_page_fault+0x1e/0x54 >>> [ 0.000000] [] ? irq_return+0x7/0x7 >>> [ 0.000000] [] page_fault+0x22/0x30 >>> [ 0.000000] [] ? __bitmap_or+0x15/0x28 >>> [ 0.000000] [] rcu_init_one+0x4c0/0x55d >>> [ 0.000000] [] rcu_init+0x270/0x2da >>> [ 0.000000] [] start_kernel+0x24f/0x4d2 >>> [ 0.000000] [] ? set_init_arg+0x53/0x53 >>> [ 0.000000] [] x86_64_start_reservations+0x2a/0x2c >>> [ 0.000000] [] x86_64_start_kernel+0xf1/0xf4 >>> [ 0.000000] ---[ end trace 4650963e41188009 ]--- >>> [ 0.000000] BUG: unable to handle kernel NULL pointer dereference at (null) >>> [ 0.000000] IP: [] __bitmap_or+0x15/0x28 >>> [ 0.000000] PGD 0 >>> [ 0.000000] Oops: 0000 [#1] SMP DEBUG_PAGEALLOC >>> [ 0.000000] CPU: 0 PID: 0 Comm: swapper/0 Tainted: G W 3.16.0-rc5+ #236363 >>> [ 0.000000] Hardware name: System manufacturer System Product Name/A8N-E, BIOS ASUS A8N-E ACPI BIOS Revision 1008 08/22/2005 >>> [ 0.000000] task: ffffffff827a3480 ti: ffffffff82784000 task.ti: ffffffff82784000 >>> [ 0.000000] RIP: 0010:[] [] __bitmap_or+0x15/0x28 >>> [ 0.000000] RSP: 0000:ffffffff82787ef8 EFLAGS: 00010002 >>> [ 0.000000] RAX: 0000000000000000 RBX: 00000000ffffffff RCX: 0000000000000001 >>> [ 0.000000] RDX: 0000000000000000 RSI: ffff880000019800 RDI: ffff880000019800 >>> [ 0.000000] RBP: ffffffff82787ef8 R08: 0000000000000000 R09: 0000000000000000 >>> [ 0.000000] R10: 0000000000000000 R11: 0000000000000000 R12: ffffffff827cf880 >>> [ 0.000000] R13: 0000000000000002 R14: ffffffff827cf880 R15: 00000000001cd000 >>> [ 0.000000] FS: 0000000000000000(0000) GS:ffff88003f800000(0000) knlGS:0000000000000000 >>> [ 0.000000] CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b >>> [ 0.000000] CR2: 0000000000000000 CR3: 000000000279e000 CR4: 00000000000006b0 >>> [ 0.000000] Stack: >>> [ 0.000000] ffffffff82787f50 ffffffff82c876f0 ffffffff83b7e7a0 0000000000000001 >>> [ 0.000000] 0000000000000082 0000000200000000 00000000ffffffff ffffffff82d37920 >>> [ 0.000000] ffff88003ffbba40 ffffffff82d3e890 0000000000000000 ffffffff82787f80 >>> [ 0.000000] Call Trace: >>> [ 0.000000] [] rcu_init_one+0x4c0/0x55d >>> [ 0.000000] [] rcu_init+0x270/0x2da >>> [ 0.000000] [] start_kernel+0x24f/0x4d2 >>> [ 0.000000] [] ? set_init_arg+0x53/0x53 >>> [ 0.000000] [] x86_64_start_reservations+0x2a/0x2c >>> [ 0.000000] [] x86_64_start_kernel+0xf1/0xf4 >>> [ 0.000000] Code: 4c 89 04 c7 4d 09 c1 48 ff c0 eb e8 31 c0 4d 85 c9 0f 95 c0 5d c3 55 48 63 c9 31 c0 48 83 c1 3f 48 89 e5 48 c1 e9 06 39 c1 7e 11 <4c> 8b 04 c2 4c 0b 04 c6 4c 89 04 c7 48 ff c0 eb eb 5d c3 55 48 >>> [ 0.000000] RIP [] __bitmap_or+0x15/0x28 >>> [ 0.000000] RSP >>> [ 0.000000] CR2: 0000000000000000 >>> [ 0.000000] ---[ end trace 4650963e4118800a ]--- >>> [ 0.000000] Kernel panic - not syncing: Attempted to kill the idle task! >>> [ 0.000000] Rebooting in 1 seconds..Press any key to enter the menu >>> >>> Excluding the new RCU bits from tip:master makes it boot. >>> >>> Any idea what's wrong? >> >> Looks like you have a setup that has NO_HZ_FULL=y, but that somehow >> avoids having a non-NULL tick_nohz_full_mask at rcu_init() time. But you >> probably knew that already. And of course when I test locally with the >> same RCU-related and NO_HZ_FULL-related configs, it all works just fine. >> Perhaps there is some interaction with some other code in -tip. >> >> So let's see... >> >> Now your .config has CONFIG_NO_HZ_FULL_ALL=y and therefore also has >> CONFIG_RCU_NOCB_CPU_ALL=y. In that case, there is no point in doing the >> cpumask_or() because all the bits are already set. So the only time >> that this cpumask_or() matters is when CONFIG_RCU_NOCB_CPU_ALL==n and >> an explicit nohz_full= mask was specified at boot time. So one reasonable >> change is to replace the #ifndef guarding the cpumask_or() with: >> >> #if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_RCU_NOCB_CPU_ALL) >> >> I am now looking to see how tick_nohz_full_mask might be NULL in this >> situation. Depending on what I find, I might insert a check for that. > > And of course if you don't actually specify a nohz_full= mask, then > tick_nohz_full_mask can be NULL at RCU initialization time, and if it > is also true that CONFIG_NO_HZ_FULL_ALL=n, this condition can persist > forever. > Hi Paul, The other location where tick_nohz_full_mask is being allocated is in tick_nohz_init_all(), called from tick_nohz_init(). rcu_init() is called before tick_nohz_init() in init/main.c. CONFIG_NO_HZ_FULL_ALL for allocation of the mask does not take effect when rcu_init() runs. So if nohz_full command line arg is not specified, tick_nohz_full_mask will always be NULL when rcu_init() runs. So just checking for tick_nohz_full_mask is NULL should be enough I guess. > Does the following patch on top of 1823172ab582 (Merge branches > 'doc.2014.07.08a', 'fixes.2014.07.09a', 'maintainers.2014.07.08b', > 'nocbs.2014.07.07a' and 'torture.2014.07.07a' into HEAD) fix the > problem? > > This may also be pulled from rcu/urgent in the -rcu git tree: > > git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git > > Thanx, Paul > > ------------------------------------------------------------------------ > > b/kernel/rcu/tree_plugin.h | 7 ++++--- > 1 file changed, 4 insertions(+), 3 deletions(-) > > rcu: Allow for NULL tick_nohz_full_mask when nohz_full= missing > > If there isn't a nohz_full= kernel parameter specified, then > tick_nohz_full_mask can legitimately be NULL. This can cause > problems when RCU's boot code tries to cpumask_or() this value into > rcu_nocb_mask. In addition, if NO_HZ_FULL_ALL=y, there is no point > in doing the cpumask_or() in the first place because this will cause > RCU_NOCB_CPU_ALL=y, which in turn will have all bits already set in > rcu_nocb_mask. > > This commit therefore avoids the cpumask_or() if NO_HZ_FULL_ALL=y > and checks for NULL tick_nohz_full_mask otherwise. > > Signed-off-by: Paul E. McKenney > > diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h > index f62b7f2f6abd..0f9ca12eabb9 100644 > --- a/kernel/rcu/tree_plugin.h > +++ b/kernel/rcu/tree_plugin.h > @@ -2479,9 +2479,10 @@ static void __init rcu_spawn_nocb_kthreads(struct rcu_state *rsp) > > if (rcu_nocb_mask == NULL) > return; > -#ifdef CONFIG_NO_HZ_FULL > - cpumask_or(rcu_nocb_mask, rcu_nocb_mask, tick_nohz_full_mask); > -#endif /* #ifdef CONFIG_NO_HZ_FULL */ > +#if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_NO_HZ_FULL_ALL) did you mean #if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_RCU_NOCB_CPU_ALL) > + if (tick_nohz_full_mask) > + cpumask_or(rcu_nocb_mask, rcu_nocb_mask, tick_nohz_full_mask); > +#endif /* #if defined(CONFIG_NO_HZ_FULL) && !defined(CONFIG_NO_HZ_FULL_ALL) */ > if (ls == -1) { > ls = int_sqrt(nr_cpu_ids); > rcu_nocb_leader_stride = ls; > -- Pranith