From: Steve Wise <swise-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW@public.gmane.org>
To: Sean Hefty <sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Cc: linux-rdma <linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>
Subject: Re: udaddy.c question
Date: Fri, 12 Mar 2010 11:06:44 -0600 [thread overview]
Message-ID: <4B9A74A4.8030404@opengridcomputing.com> (raw)
In-Reply-To: <87571B1E12964DEE8582290B804E03A8-Zpru7NauK7drdx17CPfAsdBPR1lH4CV8@public.gmane.org>
Sean Hefty wrote:
>> I'm trying to understand how IB UD service works with the rdmacm in
>> linux. According to the IBTA specification, I should be able to create
>> a single UD QP and exchange messages with any number of peers as long as
>> I know their qkeys. Yet when I look at udaddy.c (and even mckey.c), it
>> appears that the UD service in the rdmacm is really connection oriented
>> and a QP is required for each pair of processes exchanging data. I
>> think this is just the way udaddy.c was implemented, but I want to know
>> if it is possible with librdmacm to create a single cm_id with a single
>> QP, bind it to an IP address/port, and then send/recv to multiple peers
>> (via unicast). Is this possible?
>>
>
> It's intended to be possible. :) The rdma_cm uses a single qkey be default in
> order to handle mixed UD and multicast traffic. You could override this by
> exchanging whatever qkey you want via private data. But you must discover the
> remote qpn.
>
> To discover the remote qpn, you use the normal rdma calls (resolve_addr /
> resolve_route / connect or listen / accept). You need to do this for each
> remote QP that you want to exchange data with. You don't need a QP associated
> with the rdma_cm_id's when resolving the remote qpn, or even need to keep the
> rdma_cm_id's around afterwards. Actually, I don't think there's any reason why
> you can't create the UD QPs using ibv_create_qp, and only use the rdma cm to
> discover the QPNs.
>
> The udaddy code was derived from ucmatose, which is why it looks a lot like
> connection oriented code.
>
> - Sean
>
What if I don't want to use the qpn discovery mechanism from librdmacm?
IE I have have an out of band protocol for exchanging qkeys -and- qpns.
Can I avoid the connect on the active side, and the listen/accept on the
server side?
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2010-03-12 17:06 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-03-12 15:25 udaddy.c question Steve Wise
[not found] ` <4B9A5CDA.1030704-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW@public.gmane.org>
2010-03-12 16:37 ` Sean Hefty
[not found] ` <87571B1E12964DEE8582290B804E03A8-Zpru7NauK7drdx17CPfAsdBPR1lH4CV8@public.gmane.org>
2010-03-12 17:06 ` Steve Wise [this message]
[not found] ` <4B9A74A4.8030404-7bPotxP6k4+P2YhJcF5u+vpXobYPEAuW@public.gmane.org>
2010-03-12 17:36 ` Sean Hefty
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=4B9A74A4.8030404@opengridcomputing.com \
--to=swise-7bpotxp6k4+p2yhjcf5u+vpxobypeauw@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
/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