Re: [Qemu-devel] host stalls when qemu-system-aarch64 with kvm and pflash

[Marc Zyngier]

On 29/03/17 19:56, Christoffer Dall wrote:
> On Tue, Mar 28, 2017 at 01:24:15PM -0700, Radha Mohan wrote:
>> On Tue, Mar 28, 2017 at 1:16 PM, Christoffer Dall <address@hidden> wrote:
>>> Hi Radha,
>>>
>>> On Tue, Mar 28, 2017 at 12:58:24PM -0700, Radha Mohan wrote:
>>>> Hi,
>>>> I am seeing an issue with qemu-system-aarch64 when using pflash
>>>> (booting kernel via UEFI bios).
>>>>
>>>> Host kernel: 4.11.0-rc3-next-20170323
>>>> Qemu version: v2.9.0-rc1
>>>>
>>>> Command used:
>>>> ./aarch64-softmmu/qemu-system-aarch64 -cpu host -enable-kvm -M
>>>> virt,gic_version=3 -nographic -smp 1 -m 2048 -drive
>>>> if=none,id=hd0,file=/root/zesty-server-cloudimg-arm64.img,id=0 -device
>>>> virtio-blk-device,drive=hd0 -pflash /root/flash0.img -pflash
>>>> /root/flash1.img
>>>>
>>>>
>>>> As soon as the guest kernel boots the host starts to stall and prints
>>>> the below messages. And the system never recovers. I can neither
>>>> poweroff the guest nor the host. So I have resort to external power
>>>> reset of the host.
>>>>
>>>> ==================
>>>> [  116.199077] NMI watchdog: BUG: soft lockup - CPU#25 stuck for 23s!
>>>> [kworker/25:1:454]
>>>> [  116.206901] Modules linked in: binfmt_misc nls_iso8859_1 aes_ce_blk
>>>> shpchp crypto_simd gpio_keys cryptd aes_ce_cipher ghash_ce sha2_ce
>>>> sha1_ce uio_pdrv_genirq uio autofs4 btrfs raid10 rai
>>>> d456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor
>>>> raid6_pq libcrc32c raid1 raid0 multipath linear ast i2c_algo_bit ttm
>>>> drm_kms_helper syscopyarea sysfillrect sysimgblt fb_s
>>>> ys_fops drm nicvf ahci nicpf libahci thunder_bgx thunder_xcv
>>>> mdio_thunder mdio_cavium
>>>>
>>>> [  116.206995] CPU: 25 PID: 454 Comm: kworker/25:1 Not tainted
>>>> 4.11.0-rc3-next-20170323 #1
>>>> [  116.206997] Hardware name: www.cavium.com crb-1s/crb-1s, BIOS 0.3 Feb 
>>>> 23 2017
>>>> [  116.207010] Workqueue: events netstamp_clear
>>>> [  116.207015] task: ffff801f906b5400 task.stack: ffff801f901a4000
>>>> [  116.207020] PC is at smp_call_function_many+0x284/0x2e8
>>>> [  116.207023] LR is at smp_call_function_many+0x244/0x2e8
>>>> [  116.207026] pc : [<ffff000008156ecc>] lr : [<ffff000008156e8c>]
>>>> pstate: 80000145
>>>> [  116.207028] sp : ffff801f901a7be0
>>>> [  116.207030] x29: ffff801f901a7be0 x28: ffff000009139000
>>>> [  116.207036] x27: ffff000009139434 x26: 0000000000000080
>>>> [  116.207041] x25: 0000000000000000 x24: ffff0000081565d0
>>>> [  116.207047] x23: 0000000000000001 x22: ffff000008e11e00
>>>> [  116.207052] x21: ffff801f6d5cff00 x20: ffff801f6d5cff08
>>>> [  116.207057] x19: ffff000009138e38 x18: 0000000000000a03
>>>> [  116.207063] x17: 0000ffffb77c9028 x16: ffff0000082e81d8
>>>> [  116.207068] x15: 00003d0d6dd44d08 x14: 0036312196549b4a
>>>> [  116.207073] x13: 0000000058dabe4c x12: 0000000000000018
>>>> [  116.207079] x11: 00000000366e2f04 x10: 00000000000009f0
>>>> [  116.207084] x9 : ffff801f901a7d30 x8 : 0000000000000002
>>>> [  116.207089] x7 : 0000000000000000 x6 : 0000000000000000
>>>> [  116.207095] x5 : ffffffff00000000 x4 : 0000000000000020
>>>> [  116.207100] x3 : 0000000000000020 x2 : 0000000000000000
>>>> [  116.207105] x1 : ffff801f6d682578 x0 : 0000000000000003
>>>>
>>>> [  150.443116] INFO: rcu_sched self-detected stall on CPU
>>>> [  150.448261]  25-...: (14997 ticks this GP)
>>>> idle=47a/140000000000001/0 softirq=349/349 fqs=7495
>>>> [  150.451115] INFO: rcu_sched detected stalls on CPUs/tasks:
>>>> [  150.451123]  25-...: (14997 ticks this GP)
>>>> idle=47a/140000000000001/0 softirq=349/349 fqs=7495
>>>> [  150.451124]  (detected by 13, t=15002 jiffies, g=805, c=804, q=8384)
>>>> [  150.451136] Task dump for CPU 25:
>>>> [  150.451138] kworker/25:1    R  running task        0   454      2 
>>>> 0x00000002
>>>> [  150.451155] Workqueue: events netstamp_clear
>>>> [  150.451158] Call trace:
>>>> [  150.451164] [<ffff000008086188>] __switch_to+0x90/0xa8
>>>> [  150.451172] [<ffff0000081f6240>] static_key_slow_inc+0x128/0x138
>>>> [  150.451175] [<ffff0000081f6284>] static_key_enable+0x34/0x60
>>>> [  150.451178] [<ffff000008843268>] netstamp_clear+0x68/0x80
>>>> [  150.451181] [<ffff0000080e49a0>] process_one_work+0x158/0x478
>>>> [  150.451183] [<ffff0000080e4d10>] worker_thread+0x50/0x4a8
>>>> [  150.451187] [<ffff0000080ebd78>] kthread+0x108/0x138
>>>> [  150.451190] [<ffff0000080836c0>] ret_from_fork+0x10/0x50
>>>> [  150.477451]   (t=15008 jiffies g=805 c=804 q=8384)
>>>> [  150.482242] Task dump for CPU 25:
>>>> [  150.482245] kworker/25:1    R  running task        0   454      2 
>>>> 0x00000002
>>>> [  150.482259] Workqueue: events netstamp_clear
>>>> [  150.482264] Call trace:
>>>> [  150.482271] [<ffff00000808a530>] dump_backtrace+0x0/0x2b0
>>>> [  150.482277] [<ffff00000808a804>] show_stack+0x24/0x30
>>>> [  150.482281] [<ffff0000080fb750>] sched_show_task+0x128/0x178
>>>> [  150.482285] [<ffff0000080fd298>] dump_cpu_task+0x48/0x58
>>>> [  150.482288] [<ffff0000081f81e4>] rcu_dump_cpu_stacks+0xa0/0xe8
>>>> [  150.482297] [<ffff00000813983c>] rcu_check_callbacks+0x774/0x938
>>>> [  150.482305] [<ffff00000813fcb4>] update_process_times+0x34/0x60
>>>> [  150.482314] [<ffff000008151b80>] tick_sched_handle.isra.7+0x38/0x70
>>>> [  150.482319] [<ffff000008151c04>] tick_sched_timer+0x4c/0x98
>>>> [  150.482324] [<ffff000008140510>] __hrtimer_run_queues+0xd8/0x2b8
>>>> [  150.482328] [<ffff000008141180>] hrtimer_interrupt+0xa8/0x228
>>>> [  150.482334] [<ffff0000087f2a2c>] arch_timer_handler_phys+0x3c/0x50
>>>> [  150.482341] [<ffff00000812c194>] handle_percpu_devid_irq+0x8c/0x230
>>>> [  150.482344] [<ffff000008126174>] generic_handle_irq+0x34/0x50
>>>> [  150.482347] [<ffff000008126898>] __handle_domain_irq+0x68/0xc0
>>>> [  150.482351] [<ffff0000080817e4>] gic_handle_irq+0xc4/0x170
>>>> [  150.482356] Exception stack(0xffff801f901a7ab0 to 0xffff801f901a7be0)
>>>> [  150.482360] 7aa0:
>>>> 0000000000000003 ffff801f6d682578
>>>> [  150.482364] 7ac0: 0000000000000000 0000000000000020
>>>> 0000000000000020 ffffffff00000000
>>>> [  150.482367] 7ae0: 0000000000000000 0000000000000000
>>>> 0000000000000002 ffff801f901a7d30
>>>> [  150.482371] 7b00: 00000000000009f0 00000000366e2f04
>>>> 0000000000000018 0000000058dabe4c
>>>> [  150.482375] 7b20: 0036312196549b4a 00003d0d6dd44d08
>>>> ffff0000082e81d8 0000ffffb77c9028
>>>> [  150.482378] 7b40: 0000000000000a03 ffff000009138e38
>>>> ffff801f6d5cff08 ffff801f6d5cff00
>>>> [  150.482382] 7b60: ffff000008e11e00 0000000000000001
>>>> ffff0000081565d0 0000000000000000
>>>> [  150.482386] 7b80: 0000000000000080 ffff000009139434
>>>> ffff000009139000 ffff801f901a7be0
>>>> [  150.482390] 7ba0: ffff000008156e8c ffff801f901a7be0
>>>> ffff000008156ecc 0000000080000145
>>>> [  150.482394] 7bc0: ffff801f901a7be0 ffff000008156e68
>>>> ffffffffffffffff ffff000008156e8c
>>>> [  150.482397] [<ffff000008082ff4>] el1_irq+0xb4/0x140
>>>> [  150.482401] [<ffff000008156ecc>] smp_call_function_many+0x284/0x2e8
>>>> [  150.482405] [<ffff000008157020>] kick_all_cpus_sync+0x30/0x38
>>>> [  150.482409] [<ffff00000897c6cc>] aarch64_insn_patch_text+0xec/0xf8
>>>> [  150.482415] [<ffff000008095978>] arch_jump_label_transform+0x60/0x98
>>>> [  150.482420] [<ffff0000081f593c>] __jump_label_update+0x8c/0xa8
>>>> [  150.482423] [<ffff0000081f6088>] jump_label_update+0x58/0xe8
>>>> [  150.482429] [<ffff0000081f6240>] static_key_slow_inc+0x128/0x138
>>>> [  150.482434] [<ffff0000081f6284>] static_key_enable+0x34/0x60
>>>> [  150.482438] [<ffff000008843268>] netstamp_clear+0x68/0x80
>>>> [  150.482441] [<ffff0000080e49a0>] process_one_work+0x158/0x478
>>>> [  150.482444] [<ffff0000080e4d10>] worker_thread+0x50/0x4a8
>>>> [  150.482448] [<ffff0000080ebd78>] kthread+0x108/0x138
>>>> [  150.482451] [<ffff0000080836c0>] ret_from_fork+0x10/0x50
>>>>
>>>> ====================================
>>>>
>>>> I am observing that this usually happens when the guest tries to
>>>> bringup or use the default virtio-net interface.
>>>> And I am unable to reproduce this when directly booting the guest
>>>> kernel without UEFI BIOS.
>>>> So anyone observed similar issue ?
>>>>
>>>
>>> I haven't seen this myself.
>>>
>>> Have you tried with other versions of QEMU and the host kernel by any
>>> chance using the same setup?
>>
>> The earlier qemu version (atleast 2.5 which is default on ubuntu
>> 16.04) have different issue. The same thing happens when I use
>> vhost=on but without vhost and with virtio-net didn't see the issue.
>> The Armband folks have also reported the vhost issue here
>> https://bugs.launchpad.net/ubuntu/+source/edk2/+bug/1673564
> 
> So I just ran your setup using the flash emulation on the ThunderX in
> packet.net and my host is happily continuing to run hackbench.
> 
>>
>> So I am beginning to think the issue could be related to the flash
>> emulation but wanted to check on ml if anyone else is seeing it. But
>> if it is flash emulation how can it hose the host like this ?
> 
> Why do you think it's related to the flash emulation?
> 
> From what you're telling me it sounds much more like a vhost regression.
> 
> Can you try with v4.10 of the kernel with vhost=on ?

I'm unsure about this being directly related to vhost. vhost seems to 
be the trigger, but the issue seems that after a smp_call_function_many()
call, some of the target CPUs have either stopped responding, or missed
the IPI. The IPI initiator keeps on waiting forever.

I'm not seeing any of this on the HW I have access to, but this could 
be symptomatic of a SGI delivery problem.

Any chance you could try the following patchlet on the host kernel?

diff --git a/drivers/irqchip/irq-gic-v3.c b/drivers/irqchip/irq-gic-v3.c
index c132f29322cc..727c7dc7592a 100644
--- a/drivers/irqchip/irq-gic-v3.c
+++ b/drivers/irqchip/irq-gic-v3.c
@@ -617,6 +617,15 @@ static void gic_raise_softirq(const struct cpumask *mask, 
unsigned int irq)
         */
        smp_wmb();
 
+       /* Is it a broadcast? */
+       if (cpumask_weight(mask) == (num_online_cpus() - 1)) {
+               gic_write_sgi1r(1UL << ICC_SGI1R_IRQ_ROUTING_MODE_BIT |
+                               irq << ICC_SGI1R_SGI_ID_SHIFT);
+               isb();
+               return;
+       }
+
+
        for_each_cpu(cpu, mask) {
                unsigned long cluster_id = cpu_logical_map(cpu) & ~0xffUL;
                u16 tlist;

This shouldn't make any difference (other than being slightly faster in the
broadcast case, and slower in every other), but I'm prepared to try anything...

Thanks,

        M.
-- 
Jazz is not dead. It just smells funny...