mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Gavin Shan <gshan@redhat.com>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Will Deacon <will@kernel.org>,
	virtualization@lists.linux.dev, linux-kernel@vger.kernel.org,
	jasowang@redhat.com, xuanzhuo@linux.alibaba.com,
	yihyu@redhat.com, shan.gavin@gmail.com,
	linux-arm-kernel@lists.infradead.org,
	Catalin Marinas <catalin.marinas@arm.com>,
	mochs@nvidia.com
Subject: Re: [PATCH] virtio_ring: Fix the stale index in available ring
Date: Tue, 19 Mar 2024 16:49:50 +1000	[thread overview]
Message-ID: <d8387677-7025-4daa-b8b5-5a8f24a671d5@redhat.com> (raw)
In-Reply-To: <20240319024025-mutt-send-email-mst@kernel.org>


On 3/19/24 16:43, Michael S. Tsirkin wrote:
> On Tue, Mar 19, 2024 at 04:38:49PM +1000, Gavin Shan wrote:
>> On 3/19/24 16:09, Michael S. Tsirkin wrote:
>>
>>>>>> diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
>>>>>> index 49299b1f9ec7..7d852811c912 100644
>>>>>> --- a/drivers/virtio/virtio_ring.c
>>>>>> +++ b/drivers/virtio/virtio_ring.c
>>>>>> @@ -687,9 +687,15 @@ static inline int virtqueue_add_split(struct virtqueue *_vq,
>>>>>>     	avail = vq->split.avail_idx_shadow & (vq->split.vring.num - 1);
>>>>>>     	vq->split.vring.avail->ring[avail] = cpu_to_virtio16(_vq->vdev, head);
>>>>>> -	/* Descriptors and available array need to be set before we expose the
>>>>>> -	 * new available array entries. */
>>>>>> -	virtio_wmb(vq->weak_barriers);
>>>>>> +	/*
>>>>>> +	 * Descriptors and available array need to be set before we expose
>>>>>> +	 * the new available array entries. virtio_wmb() should be enough
>>>>>> +	 * to ensuere the order theoretically. However, a stronger barrier
>>>>>> +	 * is needed by ARM64. Otherwise, the stale data can be observed
>>>>>> +	 * by the host (vhost). A stronger barrier should work for other
>>>>>> +	 * architectures, but performance loss is expected.
>>>>>> +	 */
>>>>>> +	virtio_mb(false);
>>>>>>     	vq->split.avail_idx_shadow++;
>>>>>>     	vq->split.vring.avail->idx = cpu_to_virtio16(_vq->vdev,
>>>>>>     						vq->split.avail_idx_shadow);
>>>>>
>>>>> Replacing a DMB with a DSB is _very_ unlikely to be the correct solution
>>>>> here, especially when ordering accesses to coherent memory.
>>>>>
>>>>> In practice, either the larger timing different from the DSB or the fact
>>>>> that you're going from a Store->Store barrier to a full barrier is what
>>>>> makes things "work" for you. Have you tried, for example, a DMB SY
>>>>> (e.g. via __smb_mb()).
>>>>>
>>>>> We definitely shouldn't take changes like this without a proper
>>>>> explanation of what is going on.
>>>>>
>>>>
>>>> Thanks for your comments, Will.
>>>>
>>>> Yes, DMB should work for us. However, it seems this instruction has issues on
>>>> NVidia's grace-hopper. It's hard for me to understand how DMB and DSB works
>>>> from hardware level. I agree it's not the solution to replace DMB with DSB
>>>> before we fully understand the root cause.
>>>>
>>>> I tried the possible replacement like below. __smp_mb() can avoid the issue like
>>>> __mb() does. __ndelay(10) can avoid the issue, but __ndelay(9) doesn't.
>>>>
>>>> static inline int virtqueue_add_split(struct virtqueue *_vq, ...)
>>>> {
>>>>       :
>>>>           /* Put entry in available array (but don't update avail->idx until they
>>>>            * do sync). */
>>>>           avail = vq->split.avail_idx_shadow & (vq->split.vring.num - 1);
>>>>           vq->split.vring.avail->ring[avail] = cpu_to_virtio16(_vq->vdev, head);
>>>>
>>>>           /* Descriptors and available array need to be set before we expose the
>>>>            * new available array entries. */
>>>>           // Broken: virtio_wmb(vq->weak_barriers);
>>>>           // Broken: __dma_mb();
>>>>           // Work:   __mb();
>>>>           // Work:   __smp_mb();
>>>>           // Work:   __ndelay(100);
>>>>           // Work:   __ndelay(10);
>>>>           // Broken: __ndelay(9);
>>>>
>>>>          vq->split.avail_idx_shadow++;
>>>>           vq->split.vring.avail->idx = cpu_to_virtio16(_vq->vdev,
>>>>                                                   vq->split.avail_idx_shadow);
>>>
>>> What if you stick __ndelay here?
>>>
>>
>>         /* Put entry in available array (but don't update avail->idx until they
>>           * do sync). */
>>          avail = vq->split.avail_idx_shadow & (vq->split.vring.num - 1);
>>          vq->split.vring.avail->ring[avail] = cpu_to_virtio16(_vq->vdev, head);
>>
>>          /* Descriptors and available array need to be set before we expose the
>>           * new available array entries. */
>>          virtio_wmb(vq->weak_barriers);
>>          vq->split.avail_idx_shadow++;
>>          vq->split.vring.avail->idx = cpu_to_virtio16(_vq->vdev,
>>                                                  vq->split.avail_idx_shadow);
>>          /* Try __ndelay(x) here as Michael suggested
>>           *
>>           * Work:      __ndelay(200);    possiblly make it hard to reproduce
>>           * Broken:    __ndelay(100);
>>           * Broken:    __ndelay(20);
>>           * Broken:    __ndelay(10);
>>           */
>>          __ndelay(200);
> 
> So we see that just changing the timing masks the race.
> What are you using on the host side? vhost or qemu?
> 

__ndelay(200) may make the issue harder to be reproduce as I understand.
More delays here will give vhost relief, reducing the race.

The issue is only reproducible when vhost is turned on. Otherwise, we
aren't able to hit the issue.

    -netdev tap,id=vnet0,vhost=true,script=/etc/qemu-ifup,downscript=/etc/qemu-ifdown \
    -device virtio-net-pci,bus=pcie.8,netdev=vnet0,mac=52:54:00:f1:26:b0

>>
>>>
>>>>           vq->num_added++;
>>>>
>>>>           pr_debug("Added buffer head %i to %p\n", head, vq);
>>>>           END_USE(vq);
>>>>           :
>>>> }
>>>>
>>>> I also tried to measure the consumed time for various barrier-relative instructions using
>>>> ktime_get_ns() which should have consumed most of the time. __smb_mb() is slower than
>>>> __smp_wmb() but faster than __mb()
>>>>
>>>>       Instruction           Range of used time in ns
>>>>       ----------------------------------------------
>>>>       __smp_wmb()           [32  1128032]
>>>>       __smp_mb()            [32  1160096]
>>>>       __mb()                [32  1162496]
>>>>

Thanks,
Gavin


  reply	other threads:[~2024-03-19  6:50 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-03-14  7:49 Gavin Shan
2024-03-14  8:05 ` Michael S. Tsirkin
2024-03-14 10:15   ` Gavin Shan
2024-03-14 11:50     ` Michael S. Tsirkin
2024-03-14 12:50       ` Gavin Shan
2024-03-14 12:59         ` Michael S. Tsirkin
2024-03-15 10:45           ` Gavin Shan
2024-03-15 11:05             ` Michael S. Tsirkin
2024-03-15 11:24               ` Gavin Shan
2024-03-17 16:50                 ` Michael S. Tsirkin
2024-03-17 23:41                   ` Gavin Shan
2024-03-18  7:50                     ` Michael S. Tsirkin
2024-03-18 16:59 ` Will Deacon
2024-03-19  4:59   ` Gavin Shan
2024-03-19  6:09     ` Michael S. Tsirkin
2024-03-19  6:10       ` Michael S. Tsirkin
2024-03-19  6:54         ` Gavin Shan
2024-03-19  7:04           ` Michael S. Tsirkin
2024-03-19  7:41             ` Gavin Shan
2024-03-19  8:28           ` Michael S. Tsirkin
2024-03-19  6:38       ` Gavin Shan
2024-03-19  6:43         ` Michael S. Tsirkin
2024-03-19  6:49           ` Gavin Shan [this message]
2024-03-19  7:09             ` Michael S. Tsirkin
2024-03-19  8:08               ` Gavin Shan
2024-03-19  8:49                 ` Michael S. Tsirkin
2024-03-19 18:22     ` Will Deacon
2024-03-19 23:56       ` Gavin Shan
2024-03-20  0:49         ` Michael S. Tsirkin
2024-03-20  5:24           ` Gavin Shan
2024-03-20  7:14             ` Michael S. Tsirkin
2024-03-25  7:34               ` Gavin Shan
2024-03-26  7:49                 ` Michael S. Tsirkin
2024-03-26  9:38                   ` Keir Fraser
2024-03-26 11:43                     ` Will Deacon
2024-03-26 15:46                       ` Will Deacon
2024-03-26 23:14                         ` Gavin Shan
2024-03-27  0:01                           ` Gavin Shan
2024-03-27 11:56                         ` Michael S. Tsirkin
2024-03-20 17:15             ` Keir Fraser
2024-03-21 12:06               ` Gavin Shan
2024-03-19  7:36   ` Michael S. Tsirkin
2024-03-19 18:21     ` Will Deacon
2024-03-19  6:14 ` Michael S. Tsirkin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=d8387677-7025-4daa-b8b5-5a8f24a671d5@redhat.com \
    --to=gshan@redhat.com \
    --cc=catalin.marinas@arm.com \
    --cc=jasowang@redhat.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mochs@nvidia.com \
    --cc=mst@redhat.com \
    --cc=shan.gavin@gmail.com \
    --cc=virtualization@lists.linux.dev \
    --cc=will@kernel.org \
    --cc=xuanzhuo@linux.alibaba.com \
    --cc=yihyu@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®