diff mbox

net: fix qemu_announce_self not emitting packets

Message ID 1465285064-24833-1-git-send-email-pl@kamp.de (mailing list archive)
State New, archived
Headers show

Commit Message

Peter Lieven June 7, 2016, 7:37 a.m. UTC
commit fefe2a78 accidently dropped the code path for injecting
raw packets. This feature is needed for sending gratuitous ARPs
after an incoming migration has completed. The result is increased
network downtime for vservers where the network card is not virtio-net
with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.

Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
Cc: qemu-stable@nongnu.org
Cc: yanghy@cn.fujitsu.com
Signed-off-by: Peter Lieven <pl@kamp.de>
---
 net/net.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Comments

Paolo Bonzini June 7, 2016, 7:55 a.m. UTC | #1
On 07/06/2016 09:37, Peter Lieven wrote:
> commit fefe2a78 accidently dropped the code path for injecting
> raw packets. This feature is needed for sending gratuitous ARPs
> after an incoming migration has completed. The result is increased
> network downtime for vservers where the network card is not virtio-net
> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.
> 
> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
> Cc: qemu-stable@nongnu.org
> Cc: yanghy@cn.fujitsu.com
> Signed-off-by: Peter Lieven <pl@kamp.de>
> ---
>  net/net.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/net/net.c b/net/net.c
> index 5f3e5a9..d5834ea 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -722,7 +722,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>          return 0;
>      }
>  
> -    if (nc->info->receive_iov) {
> +    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
> +        nc->info->receive_raw) {
> +        /* this is required for qemu_announce_self() */
> +        ret = nc->info->receive_raw(nc, iov[0].iov_base, iov[0].iov_len);

Possibly move "iovcnt == 1" from the if condition to an assertion inside it?

Thanks,

Paolo

> +    } else  if (nc->info->receive_iov) {
>          ret = nc->info->receive_iov(nc, iov, iovcnt);
>      } else {
>          ret = nc_sendv_compat(nc, iov, iovcnt, flags);
>
Peter Lieven June 7, 2016, 8:02 a.m. UTC | #2
Am 07.06.2016 um 09:55 schrieb Paolo Bonzini:
>
> On 07/06/2016 09:37, Peter Lieven wrote:
>> commit fefe2a78 accidently dropped the code path for injecting
>> raw packets. This feature is needed for sending gratuitous ARPs
>> after an incoming migration has completed. The result is increased
>> network downtime for vservers where the network card is not virtio-net
>> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.
>>
>> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
>> Cc: qemu-stable@nongnu.org
>> Cc: yanghy@cn.fujitsu.com
>> Signed-off-by: Peter Lieven <pl@kamp.de>
>> ---
>>   net/net.c | 6 +++++-
>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>
>> diff --git a/net/net.c b/net/net.c
>> index 5f3e5a9..d5834ea 100644
>> --- a/net/net.c
>> +++ b/net/net.c
>> @@ -722,7 +722,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>>           return 0;
>>       }
>>   
>> -    if (nc->info->receive_iov) {
>> +    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
>> +        nc->info->receive_raw) {
>> +        /* this is required for qemu_announce_self() */
>> +        ret = nc->info->receive_raw(nc, iov[0].iov_base, iov[0].iov_len);
> Possibly move "iovcnt == 1" from the if condition to an assertion inside it?

I do not know if there is any other user that emits raw packets than qemu_announce_self.

The question is do you want to abort or just return -1 (which is the case without this patch).

Peter
Paolo Bonzini June 7, 2016, 8:23 a.m. UTC | #3
On 07/06/2016 10:02, Peter Lieven wrote:
> Am 07.06.2016 um 09:55 schrieb Paolo Bonzini:
>>
>> On 07/06/2016 09:37, Peter Lieven wrote:
>>> commit fefe2a78 accidently dropped the code path for injecting
>>> raw packets. This feature is needed for sending gratuitous ARPs
>>> after an incoming migration has completed. The result is increased
>>> network downtime for vservers where the network card is not virtio-net
>>> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.
>>>
>>> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
>>> Cc: qemu-stable@nongnu.org
>>> Cc: yanghy@cn.fujitsu.com
>>> Signed-off-by: Peter Lieven <pl@kamp.de>
>>> ---
>>>   net/net.c | 6 +++++-
>>>   1 file changed, 5 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/net/net.c b/net/net.c
>>> index 5f3e5a9..d5834ea 100644
>>> --- a/net/net.c
>>> +++ b/net/net.c
>>> @@ -722,7 +722,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState
>>> *sender,
>>>           return 0;
>>>       }
>>>   -    if (nc->info->receive_iov) {
>>> +    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
>>> +        nc->info->receive_raw) {
>>> +        /* this is required for qemu_announce_self() */
>>> +        ret = nc->info->receive_raw(nc, iov[0].iov_base,
>>> iov[0].iov_len);
>> Possibly move "iovcnt == 1" from the if condition to an assertion
>> inside it?
> 
> I do not know if there is any other user that emits raw packets than
> qemu_announce_self.

Currently all senders of raw packets go through qemu_send_packet_raw,
which has iovcnt == 1.

> The question is do you want to abort or just return -1 (which is the
> case without this patch).

I want to abort, since it's currently impossible to get into that
situation.  If someone wants to introduce
qemu_sendv_packet_async_with_flags, they must also introduce
nc->info->receive_raw_iov or something like that.

Thanks,

Paolo
Stefan Priebe - Profihost AG June 7, 2016, 9:33 a.m. UTC | #4
Am 07.06.2016 um 09:37 schrieb Peter Lieven:
> commit fefe2a78 accidently dropped the code path for injecting
> raw packets. This feature is needed for sending gratuitous ARPs
> after an incoming migration has completed. The result is increased
> network downtime for vservers where the network card is not virtio-net
> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.
> 
> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
> Cc: qemu-stable@nongnu.org
> Cc: yanghy@cn.fujitsu.com
> Signed-off-by: Peter Lieven <pl@kamp.de>
> ---
>  net/net.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/net/net.c b/net/net.c
> index 5f3e5a9..d5834ea 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -722,7 +722,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState *sender,
>          return 0;
>      }
>  
> -    if (nc->info->receive_iov) {
> +    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
> +        nc->info->receive_raw) {
> +        /* this is required for qemu_announce_self() */
> +        ret = nc->info->receive_raw(nc, iov[0].iov_base, iov[0].iov_len);
> +    } else  if (nc->info->receive_iov) {
>          ret = nc->info->receive_iov(nc, iov, iovcnt);
>      } else {
>          ret = nc_sendv_compat(nc, iov, iovcnt, flags);
> 

Thanks for the patch. Sadly it does not fix our problem. So we still
have another Problem.

Thanks!

Greets,
Stefan
Yang Hongyang June 7, 2016, 11:35 a.m. UTC | #5
Hi Peter,

Sorry for the inconvenience and thank you for the patch.

On Tue, Jun 7, 2016 at 3:37 PM, Peter Lieven <pl@kamp.de> wrote:

> commit fefe2a78 accidently dropped the code path for injecting
> raw packets. This feature is needed for sending gratuitous ARPs
> after an incoming migration has completed. The result is increased
> network downtime for vservers where the network card is not virtio-net
> with the VIRTIO_NET_F_GUEST_ANNOUNCE feature.
>
> Fixes: fefe2a78abde932e0f340b21bded2c86def1d242
> Cc: qemu-stable@nongnu.org
> Cc: yanghy@cn.fujitsu.com
> Signed-off-by: Peter Lieven <pl@kamp.de>
>

After addressing Paolo's comments
Reviewed-by: Yang Hongyang <hongyang.yang@easystack.cn>


> ---
>  net/net.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
>
> diff --git a/net/net.c b/net/net.c
> index 5f3e5a9..d5834ea 100644
> --- a/net/net.c
> +++ b/net/net.c
> @@ -722,7 +722,11 @@ ssize_t qemu_deliver_packet_iov(NetClientState
> *sender,
>          return 0;
>      }
>
> -    if (nc->info->receive_iov) {
> +    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
> +        nc->info->receive_raw) {
> +        /* this is required for qemu_announce_self() */
> +        ret = nc->info->receive_raw(nc, iov[0].iov_base, iov[0].iov_len);
> +    } else  if (nc->info->receive_iov) {
>          ret = nc->info->receive_iov(nc, iov, iovcnt);
>      } else {
>          ret = nc_sendv_compat(nc, iov, iovcnt, flags);
> --
> 1.9.1
>
>
>
diff mbox

Patch

diff --git a/net/net.c b/net/net.c
index 5f3e5a9..d5834ea 100644
--- a/net/net.c
+++ b/net/net.c
@@ -722,7 +722,11 @@  ssize_t qemu_deliver_packet_iov(NetClientState *sender,
         return 0;
     }
 
-    if (nc->info->receive_iov) {
+    if (flags & QEMU_NET_PACKET_FLAG_RAW && iovcnt == 1 &&
+        nc->info->receive_raw) {
+        /* this is required for qemu_announce_self() */
+        ret = nc->info->receive_raw(nc, iov[0].iov_base, iov[0].iov_len);
+    } else  if (nc->info->receive_iov) {
         ret = nc->info->receive_iov(nc, iov, iovcnt);
     } else {
         ret = nc_sendv_compat(nc, iov, iovcnt, flags);