Fwd: status in PACKET_RX_RING is actually a bit mask

Fwd: status in PACKET_RX_RING is actually a bit mask

[Stefan Puiu]

[re-sending since the original message was bounced back, something
about HTML subpart]

Hi,

I'm playing with the PACKET_RX_RING option to packet sockets and I
noticed that the status field from the tpacket_hdr structures inside
the mmaped buffer is actually a bit mask. I'm testing with 4Gbps
traffic rates and sometimes I get status=5. I'm using Ubuntu 12.04
with a 3.8.0-29 kernel, and at least from the code (I got it using
"sudo apt-get source linux-image-$(uname -r)") it seems the 5 means
TP_STATUS_USER | TP_STATUS_LOSING. The 3.8.0 kernel code has this line
in tpacket_rcv():

        /*
         * LOSING will be reported till you read the stats,
         * because it's COR - Clear On Read.
         * Anyways, moving it for V1/V2 only as V3 doesn't need this
         * at packet level.
         */
                if (po->stats.tp_drops)
                        status |= TP_STATUS_LOSING;

Also, in some source code example I found using google
(http://www.google.ro/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=0CEIQFjAE&url=http%3A%2F%2Fstuff.mit.edu%2Fafs%2Fsipb%2Fcontrib%2Flinux%2Ftools%2Ftesting%2Fselftests%2Fnet%2Fpsock_fanout.c&ei=lM8MU5yhB-iK7Aan7IGgAQ&usg=AFQjCNFtZBHSWgzHZkV3LNfLOSSTq0pV4Q&sig2=razCT0kUDv2-BIm_8gTE5g&bvm=bv.61725948,d.ZGU),
the status is checked like this:

while (header->tp_status & TP_STATUS_USER && count < RING_NUM_FRAMES) {
[...]

I haven't yet figured out how TP_STATUS_LOSING works, but I think it
would be worthwhile to at least document that the status is a bit mask
and not a value - that's what the patch at the end does.

Michael, maybe you can Cc somebody more knowledgeable on this to
confirm? IIRC there were some message exchanges on this on the list
recently...


diff --git a/man7/packet.7 b/man7/packet.7
index 1d3f222..6bac465 100644
--- a/man7/packet.7
+++ b/man7/packet.7
@@ -353,9 +353,9 @@ The packet socket owns all slots with status
 .BR TP_STATUS_KERNEL .
 After filling a slot, it changes the status of the slot to transfer
 ownership to the application.
-During normal operation, the new status is
-.BR TP_STATUS_USER ,
-to signal that a correctly received packet has been stored.
+During normal operation, the new status has the
+.BR TP_STATUS_USER
+bit set to signal that a correctly received packet has been stored.
 When the application has finished processing a packet, it transfers
 ownership of the slot back to the socket by setting the status to
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Michael Kerrisk (man-pages)]

Willem, Daniel, Carsten, Neil,

All of you have submitted patches in the recent past for packet.7, and
are presumably somewhat knowledgeable. Could one or more of your Ack
(or, if needed, improve) Stefan's proposed change?

Thanks,

Michael


On Tue, Feb 25, 2014 at 10:24 PM, Stefan Puiu <stefan.puiu-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> [re-sending since the original message was bounced back, something
> about HTML subpart]
>
> Hi,
>
> I'm playing with the PACKET_RX_RING option to packet sockets and I
> noticed that the status field from the tpacket_hdr structures inside
> the mmaped buffer is actually a bit mask. I'm testing with 4Gbps
> traffic rates and sometimes I get status=5. I'm using Ubuntu 12.04
> with a 3.8.0-29 kernel, and at least from the code (I got it using
> "sudo apt-get source linux-image-$(uname -r)") it seems the 5 means
> TP_STATUS_USER | TP_STATUS_LOSING. The 3.8.0 kernel code has this line
> in tpacket_rcv():
>
>         /*
>          * LOSING will be reported till you read the stats,
>          * because it's COR - Clear On Read.
>          * Anyways, moving it for V1/V2 only as V3 doesn't need this
>          * at packet level.
>          */
>                 if (po->stats.tp_drops)
>                         status |= TP_STATUS_LOSING;
>
> Also, in some source code example I found using google
> (http://www.google.ro/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=0CEIQFjAE&url=http%3A%2F%2Fstuff.mit.edu%2Fafs%2Fsipb%2Fcontrib%2Flinux%2Ftools%2Ftesting%2Fselftests%2Fnet%2Fpsock_fanout.c&ei=lM8MU5yhB-iK7Aan7IGgAQ&usg=AFQjCNFtZBHSWgzHZkV3LNfLOSSTq0pV4Q&sig2=razCT0kUDv2-BIm_8gTE5g&bvm=bv.61725948,d.ZGU),
> the status is checked like this:
>
> while (header->tp_status & TP_STATUS_USER && count < RING_NUM_FRAMES) {
> [...]
>
> I haven't yet figured out how TP_STATUS_LOSING works, but I think it
> would be worthwhile to at least document that the status is a bit mask
> and not a value - that's what the patch at the end does.
>
> Michael, maybe you can Cc somebody more knowledgeable on this to
> confirm? IIRC there were some message exchanges on this on the list
> recently...
>
>
> diff --git a/man7/packet.7 b/man7/packet.7
> index 1d3f222..6bac465 100644
> --- a/man7/packet.7
> +++ b/man7/packet.7
> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>  .BR TP_STATUS_KERNEL .
>  After filling a slot, it changes the status of the slot to transfer
>  ownership to the application.
> -During normal operation, the new status is
> -.BR TP_STATUS_USER ,
> -to signal that a correctly received packet has been stored.
> +During normal operation, the new status has the
> +.BR TP_STATUS_USER
> +bit set to signal that a correctly received packet has been stored.
>  When the application has finished processing a packet, it transfers
>  ownership of the slot back to the socket by setting the status to
> --
> To unsubscribe from this list: send the line "unsubscribe linux-man" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html



-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Daniel Borkmann]

Hi Michael et al,

On 04/24/2014 09:42 AM, Michael Kerrisk (man-pages) wrote:
> Willem, Daniel, Carsten, Neil,
>
> All of you have submitted patches in the recent past for packet.7, and
> are presumably somewhat knowledgeable. Could one or more of your Ack
> (or, if needed, improve) Stefan's proposed change?

Sorry for replying so late as I was just way to busy with other stuff.
Yes, I agree with you and Carsten that the man-page still need more
improvements, it's not perfect, and I'm glad that since a long time
Willem chipped in to at least bring the man page a bit more up to date.

I tried to update ...

  - http://lingrok.org/xref/linux-net-next/Documentation/networking/packet_mmap.txt
  - http://lingrok.org/xref/linux-net-next/Documentation/networking/filter.txt

... over time, both are somewhat tightly coupled as AF_PACKET users
heavily deal with BPF filters as well; so that at least it's a bit more
clear for other [kernel] developers as even the kernel documentation was
very terse for many years and without examples.

Now to your question. It can easily be seen from the if_packet.h header
file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
that TP_STATUS_* are individual bits that are set in tp_status field.

TP_STATUS_USER simply means a frame was received in the ring and is
ready for user space to be processed. If the field also indicates
TP_STATUS_LOSING then this means that there were packet drops in the
rx ring itself so a user knows it didn't get all traffic. This bit
is being reset on getsockopt() when querying PACKET_STATISTICS,
otherwise it stays. Drops occur when the ring buffer was not emptied
fast enough by user space (so no free slot with TP_STATUS_KERNEL), e.g.
due to high incoming traffic load. However, the current frame you're
reading that has TP_STATUS_USER|TP_STATUS_LOSING is fine by itself.

Hope that clarifies it?

Thanks,

Daniel

> Thanks,
>
> Michael
>
>
> On Tue, Feb 25, 2014 at 10:24 PM, Stefan Puiu <stefan.puiu-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>> [re-sending since the original message was bounced back, something
>> about HTML subpart]
>>
>> Hi,
>>
>> I'm playing with the PACKET_RX_RING option to packet sockets and I
>> noticed that the status field from the tpacket_hdr structures inside
>> the mmaped buffer is actually a bit mask. I'm testing with 4Gbps
>> traffic rates and sometimes I get status=5. I'm using Ubuntu 12.04
>> with a 3.8.0-29 kernel, and at least from the code (I got it using
>> "sudo apt-get source linux-image-$(uname -r)") it seems the 5 means
>> TP_STATUS_USER | TP_STATUS_LOSING. The 3.8.0 kernel code has this line
>> in tpacket_rcv():
>>
>>          /*
>>           * LOSING will be reported till you read the stats,
>>           * because it's COR - Clear On Read.
>>           * Anyways, moving it for V1/V2 only as V3 doesn't need this
>>           * at packet level.
>>           */
>>                  if (po->stats.tp_drops)
>>                          status |= TP_STATUS_LOSING;
>>
>> Also, in some source code example I found using google
>> (http://www.google.ro/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=0CEIQFjAE&url=http%3A%2F%2Fstuff.mit.edu%2Fafs%2Fsipb%2Fcontrib%2Flinux%2Ftools%2Ftesting%2Fselftests%2Fnet%2Fpsock_fanout.c&ei=lM8MU5yhB-iK7Aan7IGgAQ&usg=AFQjCNFtZBHSWgzHZkV3LNfLOSSTq0pV4Q&sig2=razCT0kUDv2-BIm_8gTE5g&bvm=bv.61725948,d.ZGU),
>> the status is checked like this:
>>
>> while (header->tp_status & TP_STATUS_USER && count < RING_NUM_FRAMES) {
>> [...]
>>
>> I haven't yet figured out how TP_STATUS_LOSING works, but I think it
>> would be worthwhile to at least document that the status is a bit mask
>> and not a value - that's what the patch at the end does.
>>
>> Michael, maybe you can Cc somebody more knowledgeable on this to
>> confirm? IIRC there were some message exchanges on this on the list
>> recently...
>>
>>
>> diff --git a/man7/packet.7 b/man7/packet.7
>> index 1d3f222..6bac465 100644
>> --- a/man7/packet.7
>> +++ b/man7/packet.7
>> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>>   .BR TP_STATUS_KERNEL .
>>   After filling a slot, it changes the status of the slot to transfer
>>   ownership to the application.
>> -During normal operation, the new status is
>> -.BR TP_STATUS_USER ,
>> -to signal that a correctly received packet has been stored.
>> +During normal operation, the new status has the
>> +.BR TP_STATUS_USER
>> +bit set to signal that a correctly received packet has been stored.
>>   When the application has finished processing a packet, it transfers
>>   ownership of the slot back to the socket by setting the status to
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-man" in
>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Michael Kerrisk (man-pages)]

> Now to your question. It can easily be seen from the if_packet.h header
> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
> that TP_STATUS_* are individual bits that are set in tp_status field.

So, I take it that that's an Ack-for the patch by Stefan?

Cheers,

Michael
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Daniel Borkmann]

On 04/24/2014 10:04 AM, Michael Kerrisk (man-pages) wrote:
>> Now to your question. It can easily be seen from the if_packet.h header
>> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
>> that TP_STATUS_* are individual bits that are set in tp_status field.
>
> So, I take it that that's an Ack-for the patch by Stefan?

Feel free to add my Ack. One nit below:

diff --git a/man7/packet.7 b/man7/packet.7
index 1d3f222..6bac465 100644
--- a/man7/packet.7
+++ b/man7/packet.7
@@ -353,9 +353,9 @@ The packet socket owns all slots with status
  .BR TP_STATUS_KERNEL .
  After filling a slot, it changes the status of the slot to transfer
  ownership to the application.
-During normal operation, the new status is
-.BR TP_STATUS_USER ,
-to signal that a correctly received packet has been stored.
+During normal operation, the new status has the
+.BR TP_STATUS_USER
+bit set to signal that a correctly received packet has been stored.

^--- I would drop the 'correctly' here as it rather raises questions
      what 'incorrectly' would mean in this context.

  When the application has finished processing a packet, it transfers
  ownership of the slot back to the socket by setting the status to


>
> Cheers,
>
> Michael
>
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Carsten Andrich]

I'd ack the patch, too.
However if we start clarifying the bitmask situation, we should do it
properly. packet.7 mentions TP_STATUS_xyz several times and never
actually differentiates between setting bits and setting the whole
tp_status variable. Actually both occurs inside the kernel:
http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#798
http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#2300

I could prepare another patch for review, that clarifies the tp_status
usage for all references to TP_STATUS_xyz values.
Unless Stefan would like to do this himself, since it's actually his
"discovery" ;)

Cheers,
Carsten

Daniel Borkmann schrieb:
> On 04/24/2014 10:04 AM, Michael Kerrisk (man-pages) wrote:
>>> Now to your question. It can easily be seen from the if_packet.h header
>>> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
>>> that TP_STATUS_* are individual bits that are set in tp_status field.
>>
>> So, I take it that that's an Ack-for the patch by Stefan?
> 
> Feel free to add my Ack. One nit below:
> 
> diff --git a/man7/packet.7 b/man7/packet.7
> index 1d3f222..6bac465 100644
> --- a/man7/packet.7
> +++ b/man7/packet.7
> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>  .BR TP_STATUS_KERNEL .
>  After filling a slot, it changes the status of the slot to transfer
>  ownership to the application.
> -During normal operation, the new status is
> -.BR TP_STATUS_USER ,
> -to signal that a correctly received packet has been stored.
> +During normal operation, the new status has the
> +.BR TP_STATUS_USER
> +bit set to signal that a correctly received packet has been stored.
> 
> ^--- I would drop the 'correctly' here as it rather raises questions
>      what 'incorrectly' would mean in this context.
> 
>  When the application has finished processing a packet, it transfers
>  ownership of the slot back to the socket by setting the status to
> 
> 
>>
>> Cheers,
>>
>> Michael
>>

--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Michael Kerrisk (man-pages)]

On 04/24/2014 11:17 AM, Carsten Andrich wrote:
> I'd ack the patch, too.
> However if we start clarifying the bitmask situation, we should do it
> properly. packet.7 mentions TP_STATUS_xyz several times and never
> actually differentiates between setting bits and setting the whole
> tp_status variable. Actually both occurs inside the kernel:
> http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#798
> http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#2300
> 
> I could prepare another patch for review, that clarifies the tp_status
> usage for all references to TP_STATUS_xyz values.
> Unless Stefan would like to do this himself, since it's actually his
> "discovery" ;)

I'd be happy to receive a patch from either of you.

Cheers,

Michael


> Daniel Borkmann schrieb:
>> On 04/24/2014 10:04 AM, Michael Kerrisk (man-pages) wrote:
>>>> Now to your question. It can easily be seen from the if_packet.h header
>>>> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
>>>> that TP_STATUS_* are individual bits that are set in tp_status field.
>>>
>>> So, I take it that that's an Ack-for the patch by Stefan?
>>
>> Feel free to add my Ack. One nit below:
>>
>> diff --git a/man7/packet.7 b/man7/packet.7
>> index 1d3f222..6bac465 100644
>> --- a/man7/packet.7
>> +++ b/man7/packet.7
>> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>>  .BR TP_STATUS_KERNEL .
>>  After filling a slot, it changes the status of the slot to transfer
>>  ownership to the application.
>> -During normal operation, the new status is
>> -.BR TP_STATUS_USER ,
>> -to signal that a correctly received packet has been stored.
>> +During normal operation, the new status has the
>> +.BR TP_STATUS_USER
>> +bit set to signal that a correctly received packet has been stored.
>>
>> ^--- I would drop the 'correctly' here as it rather raises questions
>>      what 'incorrectly' would mean in this context.
>>
>>  When the application has finished processing a packet, it transfers
>>  ownership of the slot back to the socket by setting the status to
>>
>>
>>>
>>> Cheers,
>>>
>>> Michael
>>>
> 
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Stefan Puiu]

Hi Carsten,

On Thu, Apr 24, 2014 at 12:17 PM, Carsten Andrich
<carsten.andrich-hs6bpBdVsEZfm0AUMx9V0g@public.gmane.org> wrote:
> I'd ack the patch, too.
> However if we start clarifying the bitmask situation, we should do it
> properly. packet.7 mentions TP_STATUS_xyz several times and never
> actually differentiates between setting bits and setting the whole
> tp_status variable. Actually both occurs inside the kernel:
> http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#798
> http://lingrok.org/xref/linux-net-next/net/packet/af_packet.c#2300

Note the section I touched covers PACKET_RX_RING. The second
occurrence you quote concerns PACKET_TX_RING - the send side. So far,
TP_STATUS_LOSING is the only status I've stumbled upon, but I know
there are other values possible.

>
> I could prepare another patch for review, that clarifies the tp_status
> usage for all references to TP_STATUS_xyz values.
> Unless Stefan would like to do this himself, since it's actually his
> "discovery" ;)

I don't have a patch cooking for the rest of the bits, so if you want
to have a stab at it, go ahead :).

Thanks for reviewing and submitting the patch, guys!

Stefan.
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Michael Kerrisk (man-pages)]

On 04/24/2014 10:14 AM, Daniel Borkmann wrote:
> On 04/24/2014 10:04 AM, Michael Kerrisk (man-pages) wrote:
>>> Now to your question. It can easily be seen from the if_packet.h header
>>> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
>>> that TP_STATUS_* are individual bits that are set in tp_status field.
>>
>> So, I take it that that's an Ack-for the patch by Stefan?
> 
> Feel free to add my Ack. One nit below:
> 
> diff --git a/man7/packet.7 b/man7/packet.7
> index 1d3f222..6bac465 100644
> --- a/man7/packet.7
> +++ b/man7/packet.7
> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>   .BR TP_STATUS_KERNEL .
>   After filling a slot, it changes the status of the slot to transfer
>   ownership to the application.
> -During normal operation, the new status is
> -.BR TP_STATUS_USER ,
> -to signal that a correctly received packet has been stored.
> +During normal operation, the new status has the
> +.BR TP_STATUS_USER
> +bit set to signal that a correctly received packet has been stored.
> 
> ^--- I would drop the 'correctly' here as it rather raises questions
>       what 'incorrectly' would mean in this context.
> 
>   When the application has finished processing a packet, it transfers
>   ownership of the slot back to the socket by setting the status to

Done!


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: status in PACKET_RX_RING is actually a bit mask

[Michael Kerrisk (man-pages)]

Stefan, Daniel,

Daniel: Thanks for the Ack.

Stefan: Thanks for the patch. Applied.

Cheers,

Michael


On 04/24/2014 10:02 AM, Daniel Borkmann wrote:
> Hi Michael et al,
> 
> On 04/24/2014 09:42 AM, Michael Kerrisk (man-pages) wrote:
>> Willem, Daniel, Carsten, Neil,
>>
>> All of you have submitted patches in the recent past for packet.7, and
>> are presumably somewhat knowledgeable. Could one or more of your Ack
>> (or, if needed, improve) Stefan's proposed change?
> 
> Sorry for replying so late as I was just way to busy with other stuff.
> Yes, I agree with you and Carsten that the man-page still need more
> improvements, it's not perfect, and I'm glad that since a long time
> Willem chipped in to at least bring the man page a bit more up to date.
> 
> I tried to update ...
> 
>   - http://lingrok.org/xref/linux-net-next/Documentation/networking/packet_mmap.txt
>   - http://lingrok.org/xref/linux-net-next/Documentation/networking/filter.txt
> 
> ... over time, both are somewhat tightly coupled as AF_PACKET users
> heavily deal with BPF filters as well; so that at least it's a bit more
> clear for other [kernel] developers as even the kernel documentation was
> very terse for many years and without examples.
> 
> Now to your question. It can easily be seen from the if_packet.h header
> file http://lingrok.org/xref/linux-net-next/include/uapi/linux/if_packet.h#93
> that TP_STATUS_* are individual bits that are set in tp_status field.
> 
> TP_STATUS_USER simply means a frame was received in the ring and is
> ready for user space to be processed. If the field also indicates
> TP_STATUS_LOSING then this means that there were packet drops in the
> rx ring itself so a user knows it didn't get all traffic. This bit
> is being reset on getsockopt() when querying PACKET_STATISTICS,
> otherwise it stays. Drops occur when the ring buffer was not emptied
> fast enough by user space (so no free slot with TP_STATUS_KERNEL), e.g.
> due to high incoming traffic load. However, the current frame you're
> reading that has TP_STATUS_USER|TP_STATUS_LOSING is fine by itself.
> 
> Hope that clarifies it?
> 
> Thanks,
> 
> Daniel
> 
>> Thanks,
>>
>> Michael
>>
>>
>> On Tue, Feb 25, 2014 at 10:24 PM, Stefan Puiu <stefan.puiu-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
>>> [re-sending since the original message was bounced back, something
>>> about HTML subpart]
>>>
>>> Hi,
>>>
>>> I'm playing with the PACKET_RX_RING option to packet sockets and I
>>> noticed that the status field from the tpacket_hdr structures inside
>>> the mmaped buffer is actually a bit mask. I'm testing with 4Gbps
>>> traffic rates and sometimes I get status=5. I'm using Ubuntu 12.04
>>> with a 3.8.0-29 kernel, and at least from the code (I got it using
>>> "sudo apt-get source linux-image-$(uname -r)") it seems the 5 means
>>> TP_STATUS_USER | TP_STATUS_LOSING. The 3.8.0 kernel code has this line
>>> in tpacket_rcv():
>>>
>>>          /*
>>>           * LOSING will be reported till you read the stats,
>>>           * because it's COR - Clear On Read.
>>>           * Anyways, moving it for V1/V2 only as V3 doesn't need this
>>>           * at packet level.
>>>           */
>>>                  if (po->stats.tp_drops)
>>>                          status |= TP_STATUS_LOSING;
>>>
>>> Also, in some source code example I found using google
>>> (http://www.google.ro/url?sa=t&rct=j&q=&esrc=s&source=web&cd=5&cad=rja&ved=0CEIQFjAE&url=http%3A%2F%2Fstuff.mit.edu%2Fafs%2Fsipb%2Fcontrib%2Flinux%2Ftools%2Ftesting%2Fselftests%2Fnet%2Fpsock_fanout.c&ei=lM8MU5yhB-iK7Aan7IGgAQ&usg=AFQjCNFtZBHSWgzHZkV3LNfLOSSTq0pV4Q&sig2=razCT0kUDv2-BIm_8gTE5g&bvm=bv.61725948,d.ZGU),
>>> the status is checked like this:
>>>
>>> while (header->tp_status & TP_STATUS_USER && count < RING_NUM_FRAMES) {
>>> [...]
>>>
>>> I haven't yet figured out how TP_STATUS_LOSING works, but I think it
>>> would be worthwhile to at least document that the status is a bit mask
>>> and not a value - that's what the patch at the end does.
>>>
>>> Michael, maybe you can Cc somebody more knowledgeable on this to
>>> confirm? IIRC there were some message exchanges on this on the list
>>> recently...
>>>
>>>
>>> diff --git a/man7/packet.7 b/man7/packet.7
>>> index 1d3f222..6bac465 100644
>>> --- a/man7/packet.7
>>> +++ b/man7/packet.7
>>> @@ -353,9 +353,9 @@ The packet socket owns all slots with status
>>>   .BR TP_STATUS_KERNEL .
>>>   After filling a slot, it changes the status of the slot to transfer
>>>   ownership to the application.
>>> -During normal operation, the new status is
>>> -.BR TP_STATUS_USER ,
>>> -to signal that a correctly received packet has been stored.
>>> +During normal operation, the new status has the
>>> +.BR TP_STATUS_USER
>>> +bit set to signal that a correctly received packet has been stored.
>>>   When the application has finished processing a packet, it transfers
>>>   ownership of the slot back to the socket by setting the status to
>>> --
>>> To unsubscribe from this list: send the line "unsubscribe linux-man" in
>>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>>
>>
> 


-- 
Michael Kerrisk
Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/
Linux/UNIX System Programming Training: http://man7.org/training/
--
To unsubscribe from this list: send the line "unsubscribe linux-man" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html