[P4-dev] What about a packet is preserved from ingress to egress with P4_16 and v1model.p4?

Mihai Budiu mbudiu at vmware.com
Mon Apr 24 12:56:25 EDT 2017


Your interpretation is correct. There is no documentation for v1model.p4 – except that it should model the P4-14 switch model accurately. So technically the P4-14 specification is the documentation for v1model.p4.

Mihai

From: P4-dev [mailto:p4-dev-bounces at lists.p4.org] On Behalf Of Andy Fingerhut
Sent: Sunday, April 23, 2017 11:25 PM
To: p4-dev <p4-dev at lists.p4.org>
Subject: [P4-dev] What about a packet is preserved from ingress to egress with P4_16 and v1model.p4?

Here is a link to an excerpt of v1model.p4 checked into the P4_16 compiler repository:

https://github.com/p4lang/p4c/blob/master/p4include/v1model.p4#L130-L152<https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_p4lang_p4c_blob_master_p4include_v1model.p4-23L130-2DL152&d=DwMFaQ&c=uilaK90D4TOVoH58JNXRgQ&r=tGW6TKXajnoXSyy1S1P4DHGPe8sj54GGvw-b21n7aWg&m=VLNxR1p9ZNvpRDsgSQ0VXjfc_xGBzyonlN-TAK_MmMw&s=glyJoKhmEYTFOu2KyCNnBL0n_rRZpGa9XcyfHO4fblE&e=>

From reading that, and from trying a few experiments, my guess is that whatever the ingress pipeline does to modify the parsed headers in type H, and however it modifies the metadata with type M, and the standard_metadata_t parameter, every bit of all of those types will become the values that the packet will begin with when the egress control block is run for that packet.

Does that sound correct?

Thanks,
Andy

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.p4.org/pipermail/p4-dev_lists.p4.org/attachments/20170424/9d456c57/attachment-0002.html>


More information about the P4-dev mailing list