> From: Jason Gunthorpe
> > >fill mr->key by the lkey or rkey based on that and everything will
> > >work fine.
> >
> > But the ULP *can* register a memory buffer with local and remote
> > access permissions.
> Not in the new API.
>
> If a ULP ever comes along
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
> ow...@vger.kernel.org] On Behalf Of Doug Ledford
> These patches add the concept of duplicate GIDs that are differentiated by
> their RoCE version (also called network type). So, now, an incoming packet
> could match a couple
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
> ow...@vger.kernel.org] On Behalf Of Doug Ledford
> In particular, Liran piped up with this comment:
>
> "Also, I don't want to do any route resolution on the Rx path. A UD QP
> completion just reports the details of the packet it
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
> > Since you and Jason did not reach a consensus, I have to dig in and
> > see if these patches make it possible to break namespace confinement,
> > either accidentally or with intentionally tricky behavior. That's
> > going to take
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
>
> > You are pushing abstraction into provider code instead of handling it in a
> generic way.
>
> No, I am defining an API that *make sense* and doesn't leak useless details.
> Of course that doesn't force code duplication or
> From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com]
> Look, here is a concrete direction:
>
> Replace all the crap in
> ib_init_ah_from_wc/get_sgid_index_from_eth/rdma_addr_find_dmac_by_
> grh
>
> with a straightforward
>
>rdma_dgid_index_from_wc(
>
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
>
> The abstraction at the gid cache is making it too easy to make this mistake.
> It
> is enabling callers to do direct gid lookups without a route lookup, which is
> unconditionally wrong. Every call site into the gid cache I looked
From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com]
It does if you want a planned 'gental' removal to be possible..
There could be a lot of design options for a 'gentle' removal, such as first
sending a 'prepare' event, and only then doing the flow proposed here.
I do not want
From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
ow...@vger.kernel.org] On Behalf Of Jason Gunthorpe
I have no real problem with that, it would be nice to have an answer
to the uverbs vs ucma removal ordering question and the basic issue of
if we even want to do this so async
From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com]
After all, it is the payload that designates the entity that you
want to establish a connection to, rather than the packet headers,
which are just meant to relay the packet to the proper CM
No, that isn't right. The IBA uses
From: Jason Gunthorpe [mailto:jguntho...@obsidianresearch.com]
What is really missing here I guess is a mechanism that would
enforce containers to only use certain pkeys - perhaps with
something like an RDMA cgroup. It could force containers to only
use approved pkeys not only with
From: Weiny, Ira [mailto:ira.we...@intel.com]
ib_verbs define an *extensive* direct HW access API, which is constantly
evolving.
This is the problem with verbs...
Huh?
It is its strength, if you don't break backward compatibility...
You cannot describe the intricate object relations
From: Doug Ledford [mailto:dledf...@redhat.com]
But the node_type stands for more than just an abstract RDMA device:
In IB, it designates an instance of an industry-standard, well-defined,
device type: it's possible link types, transport, semantics, management,
everything.
It *should* be
From: Doug Ledford [mailto:dledf...@redhat.com]
OPA cannot impersonate IB; OPA node and link types have to be
designated as such. In terms of MAD processing flows, both
explicit (as in the handle_opa_smi() call below) and implicit code
paths (which share IB flows - there are
From: Ira Weiny ira.we...@intel.com
Hi Ira,
OPA cannot impersonate IB; OPA node and link types have to be designated as
such.
In terms of MAD processing flows, both explicit (as in the handle_opa_smi()
call below) and implicit code paths (which share IB flows - there are several
cases) must
Why do you have RDMA_NODE_IB_SWITCH related stuff inside the
handle_opa_smi() function?
Is there a node type of switch in OPA similar to IB?
Yes. OPA uses the same node types as IB.
Ira
No, OPA cannot impersonate IB.
It has to have distinct node and link types.
--Liran
--
To
From: Doug Ledford [mailto:dledf...@redhat.com]
I suppose that the main issue would be handling existing user memory
mappings,
which cannot be just invalidated -- the user-space driver may not be aware
of the
device removal and may access this memory concurrently, and we don't
want it
From: Doug Ledford [mailto:dledf...@redhat.com]
I suppose that the main issue would be handling existing user memory
mappings,
which cannot be just invalidated -- the user-space driver may not be aware
of the
device removal and may access this memory concurrently, and we don't
want it
From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
Add raw helpers:
rdma_tech_ib
rdma_tech_iboe
rdma_tech_iwarp
rdma_ib_or_iboe (transition, clean up later) To help us detect which
technology the port supported.
Replace rdma_tech_* with rdma_protocol_*.
From: Hefty, Sean [mailto:sean.he...@intel.com]
[snip]
So, I think that our old-transport below is just fine.
No need to change it (and you aren't, since it is currently
implemented
as a function).
I think there is a need to change this. Encoding the transport into
the node
From: Michael Wang [mailto:yun.w...@profitbricks.com]
[snip]
Depends on who is we.
For ULPs, you are probably right.
However, core services (e.g., mad management, CM, SA) do care about
various details.
In some cases, where it doesn't matter, this code will use management
helpers.
From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
[snip]
a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index
65994a1..d54f91e 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -75,10 +75,13 @@ enum rdma_node_type { };
enum rdma_transport_type {
+
From: Michael Wang [mailto:yun.w...@profitbricks.com]
Hi, Liran
Thanks for the comment :-)
On 04/22/2015 01:36 AM, Liran Liss wrote:
[snip]
(**) This has been extended to also encode the transport in the current
code.
At least for user-space visible APIs, we might chose to leave
Hi Michael,
The spirit of this patch-set is great, but I think that we need to clarify some
concepts.
Since this will affect the whole patch-set, I am laying out my concerns here
instead.
A suggestion for the resulting management helpers is given below.
I believe the result would be much more
RoCE v2 is really Infiniband over UDP over IP. Why don't we just
call
it IBoUDP like it is?
RoCEv2 is the name in the IBTA spec (Annex 17)
We call RoCE IBoE in the kernel, because that's what it is. RoCE is an IBTA
marketing name.
Looking through the Annex, I don't see where
The RoCE Verbs interface references the HCA GID table in QPs and AHs,
for all RoCE versions.
The IBTA specifically does not define software interfaces. The concern here
is
the architecture and definition of the linux rdma software stack, not verbs,
despite the fact that the layer is
Here are a few more clarifications:
1) ODP MRs can cover address ranges that do not have a mapping at registration
time.
This means that MPI can register in advance, say, the lower GB's of the address
space, covering malloc's primary arena.
Thus, there is no need to adjust to each increase in
-Original Message-
From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
ow...@vger.kernel.org] On Behalf Of Jeff Squyres (jsquyres)
Sent: Monday, June 10, 2013 5:50 PM
To: Jason Gunthorpe
Cc: Haggai Eran; Or Gerlitz; linux-rdma@vger.kernel.org; Shachar Raindel
Subject: Re:
On Fri, Dec 2, 2011 at 2:19 AM, Yevgeny Petrilin
yevge...@mellanox.co.il wrote:
When running in SRIOV mode, driver should not automatically
start/stop
the mlx4_core upon sensing an HCA internal error -- doing this
disables/enables
sriov, which will cause the hypervisor to hang if there
Small correction needed regarding the multicast forwarding.
Since we are talking about IPv6 multicast groups, which
translate to 33:33:xx:xx:xx:xx MAC address, the router
listener notification protocol is going to be MLD and not
IGMP. Still there are switches which support MLD
The text is saying that the specification does not use any of the
LID fields in the verbs interface, that is it. It isn't talking
about MAC addresses.
Exactly how and where the MAC address comes about was
never decided,
and at least some participants thought it should be a
But, we can't mandate an overload of the GID in a way that
it prevents its use as a true L3 address (eventually routable).
Actually I'm beginning to think that the only possible way we
can use the GID in IBoE is as a link-local IPv6 addresses
containing an Ethernet address. Trying
A quibble about multicast - AFAIK this is unsolved. I
think some spec needs to be agreed that documents what
sort of multicast snooping operations switches need to do,
ie if IGMP joins imply that IBoE traffic for the same DMAC
is included in the join, or if IBoE requires a
...A verbs consumer using a RoCE network relies strictly
on so-called
Layer 3 addressing (GIDs); layer 2 addresses (e.g. subnet local
identifiers) are not passed across the verbs interface...
Ah, hmm, well, I was on that list during this time and I don't
think
this
as possible, while preserving transparency to the
applications.
Comments are welcome.
Liran
-Original Message-
From: Or Gerlitz [mailto:ogerl...@voltaire.com]
Sent: Wednesday, July 07, 2010 9:00 AM
To: Liran Liss
Cc: Roland Dreier; Jason Gunthorpe; Hefty, Sean; Aleksey
Senin; linux-rdma
: Thursday, June 24, 2010 11:37 PM
To: Liran Liss
Cc: Hefty, Sean; Roland Dreier; Aleksey Senin; linux-rdma; mo...@voltaire.com;
aleks...@voltaire.com; yift...@voltaire.com; Tziporet Koren; al...@voltaire.com
Subject: Re: When IBoE will be merged to upstream?
The current behavior of ibv_create_ah
Regarding GID to Eth mappings, we discussed using the standard create_ah() Verb
for this.
In the kernel, create_ah() will call a generic address resolution function in
the cma.
The returned information will be copied back to user-space in a device-specific
structure (since address handles are
S.B.
--Liran
-Original Message-
From: Hefty, Sean [mailto:sean.he...@intel.com]
Sent: Thursday, June 24, 2010 9:06 PM
To: Liran Liss; Roland Dreier; Aleksey Senin
Cc: linux-rdma; mo...@voltaire.com; aleks...@voltaire.com;
yift...@voltaire.com; Tziporet Koren; al...@voltaire.com
Subject
If we have a dedicated ABI call for this mapping, then it seems reasonable to
have it device independent.
However, this mapping is really only used when creating address handles.
So, we can base the mapping on the (device specific) create_ah() flow, but
provide generic mapping functions for all
Existing apps rely on transport_type == IBV_TRANSPORT_IB to indicate
IB management is present. There are many examples of this.
The art of API compatability is to not break existing old apps, so you
don't get to change the meaning of transport_type == IBV_TRANSPORT_IB to
mean 'it is only IB
So? There are substantial semantic differences for *all* non-rdmacm
applications. Even common ones like OpenMPI. You propose to ignore them?
On the contrary! Any application that *does* care what the link layer is
can look up a new field in port_attr (rather than a new node transport
type).
Hi Paul, you are not missing anything - lookback communication will work
in RDMAoE just as in IB.
--Liran
-Original Message-
From: Paul Grun [mailto:pg...@systemfabricworks.com]
Sent: Wednesday, December 02, 2009 10:55 AM
To: 'Or Gerlitz'; Liran Liss
Cc: 'Sean Hefty'; 'Jason Gunthorpe
42 matches
Mail list logo