iSER/iWARP Support in version 2.6.20

[vs18]

In version 2.6.20 kernel sources, struct iser_hdr defined in
iscsi_iser.h does not seem to match what the iSER spec has (http://
www.ietf.org/internet-drafts/draft-ietf-ips-iser-06.txt page 69). The
struct iser_header has two fields: write_va and read_va that I did not
expect per spec.

Also,

1. Do not see the iSER Hello Message/Reply processing is being done.

2. Unable to see where in code the payload described in section 4.1.3
(iWARP Message Format for SCSI Read Command PDU) of the iSER document
is setup. To be specific, where is the MPA header encapsulating the
iSER header is done?

Thanks

[vs18]

[Or Gerlitz]

vs18 wrote:
> In version 2.6.20 kernel sources, struct iser_hdr defined in
> iscsi_iser.h does not seem to match what the iSER spec has (http://
> www.ietf.org/internet-drafts/draft-ietf-ips-iser-06.txt page 69). The
> struct iser_header has two fields: write_va and read_va that I did not
> expect per spec.

The IBTA 1.2 spec annex A12: SUPPORT FOR ISCSI EXTENSIONS FOR RDMA
(ISER) table 4 "Expanded iSER Header for Supporting Virtual Address"
states that the iser header for Infiniband contains a read and write
virtual addresses, each 64bit. I realize now that there is no link from
draft-ietf-ips-iser-06.txt to the IBTA annex, something which need to be
fixed.

Or.

[Erez Zilber]

vs18 wrote:

> In version 2.6.20 kernel sources, struct iser_hdr defined in
> iscsi_iser.h does not seem to match what the iSER spec has (http://
> www.ietf.org/internet-drafts/draft-ietf-ips-iser-06.txt page 69). The
> struct iser_header has two fields: write_va and read_va that I did not
> expect per spec.
>
>

This version of iSER is meant to run over InfiniBand. Please refer to
the iSER for InfiniBand annex (table 4):
http://www.infinibandta.org/members/spec/Annex_iSER.PDF

> Also,
>
> 1. Do not see the iSER Hello Message/Reply processing is being done.
>

This is on our to-do list.

> 2. Unable to see where in code the payload described in section 4.1.3
> (iWARP Message Format for SCSI Read Command PDU) of the iSER document
> is setup. To be specific, where is the MPA header encapsulating the
> iSER header is done?
>

As mentioned above, this version of iSER is meant to run over
InfiniBand. Therefore, the MPA header is irrelevant.

--

____________________________________________________________

Erez Zilber | 972-9-971-7689

Software Engineer, Storage Team

Voltaire – _The Grid Backbone_

__

www.voltaire.com <http://www.voltaire.com/>

[vs18]

Thanks for both responses. I have couple of follow-up questions.

1. It looks like there is some iSER/iWARP related support in 2.6.20
and also Cheslso RNIC is supported. When (roughly) will open-iscsi
have complete support for iSER/iWARP?

2. Will there be two versions of iser_header based on the transport?

3. Can the iSER/iWARP todo-list be shared?

Thanks

vs

[Erez Zilber]

vs18 wrote:

> Thanks for both responses. I have couple of follow-up questions.
>
> 1. It looks like there is some iSER/iWARP related support in 2.6.20
> and also Cheslso RNIC is supported. When (roughly) will open-iscsi
> have complete support for iSER/iWARP?
>

We (Voltaire) may be involved in this in the future. I don't know if
there's anyone else working on iSER over iWARP.

> 2. Will there be two versions of iser_header based on the transport?
>

Mike - can you answer that? The complete thread (since you've just
joined) is here:
http://groups.google.com/group/open-iscsi/browse_thread/thread/23ee18054e8412e6

> 3. Can the iSER/iWARP todo-list be shared?
>

I don't know if this list exists. I was talking about the iSER over
InfiniBand list.

Erez