Title: Message
Cisco IB host drivers are available at http://www.cisco.com/cgi-bin/tablebuild.pl/sfs-linux and
http://www.cisco.com/cgi-bin/tablebuild.pl/sfs-win2K.
Scott
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Tziporet
KorenSent: Monday, July 24, 2006 2:
Quoting r. somenath <[EMAIL PROTECTED]>:
> thanks Tziporet.
> from ur link , I am looking here
> http://www.mellanox.com/support/firmware_table.php
>
> for the exact HCA I have from your list. I don't find the exact match of
> OPN and Card Rev (PSID) and I don't know how to find those.
> all I fo
Roland Dreier wrote:
> [CC'ing linux-scsi as well -- I think we'll get better insight from there]
>
> > The current SRP initiator code cannot work with several fail-over
> mechanisms.
> >
> > The current srp driver's behavior when a target off-line then online:
> > 1) The target is offline.
[CC'ing linux-scsi as well -- I think we'll get better insight from there]
> The current SRP initiator code cannot work with several fail-over
> mechanisms.
>
> The current srp driver's behavior when a target off-line then online:
> 1) The target is offline.
> 2) the initiator tries to re
>> Are you trying to have each port join the same multicast group?
>
>Possibly - could make some implementation easier, though for performance
> probably not.
If you only want to join a group once per system, then using the following calls
from the RDMA CM should work:
rdma_create_id();
rdma_res
Sean Hefty wrote:
>>Yes, that's what I've been doing, though you recommended using the RDMA
>>CM to set up multicast groups. What interface should I be using then?
>
>
> That's the problem. There is no raw userspace SA query interface. The only
> other alternative is to send MADs directly to t
Title: Message
Hi all,
This is the outcome of the meeting we had
today regarding OFED 1.1 schedule and features.
Tziporet
1. Schedule:
Target release date: 31-Aug
Intermediate milestones:
1.
Create 1.1 branch of user level code and rc1: 27-Jul
2.
Featur
OpenSM/osm_prtn.c: Set exactly selectors for MTU and rate as well as in
component mask prior to calling osm_mcmr_rcv_find_or_create_new_mgrp
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: opensm/osm_prtn.c
=
>Yes, that's what I've been doing, though you recommended using the RDMA
>CM to set up multicast groups. What interface should I be using then?
That's the problem. There is no raw userspace SA query interface. The only
other alternative is to send MADs directly to the SA through the SA MAD
inte
Sean Hefty wrote:
>>Not quite - that's backwards from how I'm looking to do it. I want to
>>use rdma_get_devices() (or something similar) to see which devices are
>>available, then get/create rdma_cm_id's for each of those devices,
>>without having something like an IP address beforehand.
>
>
>
>Not quite - that's backwards from how I'm looking to do it. I want to
>use rdma_get_devices() (or something similar) to see which devices are
>available, then get/create rdma_cm_id's for each of those devices,
>without having something like an IP address beforehand.
It sounds like you just want
Hi Hal,
> Are the network order routines ( ib_notice_set_prod_type and
> ib_notice_set_vend_id) still needed or should they be eliminated ?
[EZ] I propose to keep them.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman
Sean Hefty wrote:
>>What's the best way to map the ibv_context's return from
>>rdma_get_devices() to rdma_cm_id's for use with the rest of the RDMA
>>functions?
>
>
> Once an rdma_cm_id has been bound to a device, it will reference one of the
> devices (ibv_context) that's also return from rdma_g
>What's the best way to map the ibv_context's return from
>rdma_get_devices() to rdma_cm_id's for use with the rest of the RDMA
>functions?
Once an rdma_cm_id has been bound to a device, it will reference one of the
devices (ibv_context) that's also return from rdma_get_devices(). I.e. the
pointe
> One point to take into account, in my experience, is that to be fair with
> size
> savings checks, you should build with -Os.
Yes, I have CONFIG_CC_OPTIMIZE_FOR_SIZE=y
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mai
>with the following, which uninlines ib_response_mad() for the
>following size savings:
I've committed this change to svn as well. (I committed while writing the svn
log message, so it's truncated.)
- Sean
___
openib-general mailing list
openib-genera
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: 3 patches in mst-for-2.6.18
>
> > Roland, I have put the following patches in my mst-for-2.6.18 tree:
>
> OK, I have all three.
>
> I updated this:
>
> >[PATCH] IB/mad: validate MADs for spec compliance C13-18.1.1
>
> with the
Hi,
The current SRP initiator code cannot work with several fail-over mechanisms.
The current srp driver's behavior when a target off-line then online:
1) The target is offline.
2) the initiator tries to reconnect and fails
3) The initiator calls srp_remove_work that removes the scsi_host.
4) Th
What's the best way to map the ibv_context's return from
rdma_get_devices() to rdma_cm_id's for use with the rest of the RDMA
functions?
Should I allocate my own rdma_cm_id and fill it in (like
rdma_create_id()), or am I missing something?
Andrew
Sean Hefty wrote:
> The following patch adds c
Sean Hefty wrote:
> Yes - I inadvertently removed this fix when I removed libsysfs from the
> librdmacm. I will re-instate the fix.
7497 was a fix that added libsysfs to the librdmacm as a result of it being
removed from libibverbs. In any case, I applied a fix to assume the latest ABI
version
I reverted svn checkin 8629 and replaced it with check-in 8646, shown below.
This only causes a minor change to the CM event handling patch.
Return error code, rather than success (0), on failure allocating timewait_info
in ib_send_cm_req.
Signed-off-by: Sean Hefty <[EMAIL PROTECTED]>
---
Index:
Tziporet Koren wrote:
> somenath wrote:
>
>> I see this messages while booting my system:
>>
>> Jul 17 10:35:37 alekhine kernel: ib_mthca :03:00.0: HCA FW
>> version 4.6.0 is old (4.7.0 is current).
>> Jul 17 10:35:37 alekhine kernel: ib_mthca :03:00.0: If you have
>> problems, try updat
Sean Hefty wrote:
> The race can only happen on the passive side, but... for peer to peer
> connections, we won't know which side is passive until later, and this tracks
> the local QPN as part of timewait now. (Note that peer to peer support
> requires
> more work.)
For peer to peer, tracking t
Linus, please pull from
master.kernel.org:/pub/scm/linux/kernel/git/roland/infiniband.git for-linus
This tree is also available from kernel.org mirrors at:
git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband.git
for-linus
to get a few fixes:
Dotan Barak:
IB/mthca: Fi
Thanks, applied.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
> Roland, I have put the following patches in my mst-for-2.6.18 tree:
OK, I have all three.
I updated this:
>[PATCH] IB/mad: validate MADs for spec compliance C13-18.1.1
with the following, which uninlines ib_response_mad() for the
following size savings:
textdata bss dec
>Some things unclear to me:
>1. Is a user somehow *required* to call ib_cm_establish if he gets the event?
> E.g. what about cma users?
A user is not required to call ib_cm_establish. The patch fixes the issue where
CMA users receive a COMM_EST event, but do not have the ability to call
ib_cm_e
Just wanted to update people on this (still looking for some insight, but
not really expecting any).
I was successfully able to bring up the entire stack (all loadable modules)
without ifconfig'ing the ib0 interface. At this point, I'm able to
participate in the network (I can see the subnet mana
Title: Message
Dotan Barak <[EMAIL PROTECTED]> wrote: -Original Message-From: keshetti mahesh[mailto:[EMAIL PROTECTED] Sent: Monday, July 24, 2006 3:21PMTo: Dotan BarakSubject: RE: [openib-general]connection loss handling in mthcaDotanBarak <[EMAIL PROTECTED]>
Hi.On Monday 24 July 2006 13:50, keshetti mahesh wrote:> i have a query regarding the handling of asynchronous events in mthca driver> consider the situation, receiver has posted some 10 descriptors. and 5 out of them are completd successfully, after that connection is lost( in NIC level) due to s
Hi.
On Monday 24 July 2006 13:50, keshetti mahesh wrote:
> i have a query regarding the handling of asynchronous events in mthca driver
> consider the situation, receiver has posted some 10 descriptors. and 5 out of
> them are completd successfully, after that connection is lost( in NIC level)
>
OpenSM: Better handling for Path/MultiPath requests for raw traffic
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: opensm/osm_sa_path_record.c
===
--- opensm/osm_sa_path_r
i have a query regarding the handling of asynchronous events in mthca driverconsider the situation, receiver has posted some 10 descriptors. and 5 out of them are completd successfully, after that connection is lost( in NIC level) due to some reasonnow,1. how do the QP know about this(there is no
Hi Eitan,
On Sun, 2006-07-23 at 05:01, Eitan Zahavi wrote:
> Hi Hal
>
> When formating an SM generated report the ProducerType was using CL_NTOH32
> which can not be used to format a 24bit network order number.
>
> The propose fix also avoids the duplication of conversion to and from network
>
Or Gerlitz wrote:
>> Or Gerlitz wrote:
>>> Vladimir Sokolovsky wrote:
>
>>> Did you have any special reason to assign host1:ib1 an IP address
>>> ***before*** the failover? is the reason for that happen to be
>>> having it joins the IPv4 multicast group at "batch time", that is
>>> not during th
Quoting r. Or Gerlitz <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] IB/ipoib: fix crash on mcast join finish
>
> Michael S. Tsirkin wrote:
> > Quoting r. Or Gerlitz <[EMAIL PROTECTED]>:
>
> >> also since the crash does not allow
> >> to debug mcast, i guess the fix needs to go into OFED 1.1 as well
Roland, I have put the following patches in my mst-for-2.6.18 tree:
$git log --pretty=short origin..mst-for-2.6.18
commit cdfe3231414ee6dca31bb1fa3f9140408ae33a6a
Author: Or Gerlitz <[EMAIL PROTECTED]>
[PATCH] IB/ipoib: fix oops with ipoib_debug_mcast set
commit 7534914e26bd847d61e8de6db3ce4
Michael S. Tsirkin wrote:
> Quoting r. Or Gerlitz <[EMAIL PROTECTED]>:
>> also since the crash does not allow
>> to debug mcast, i guess the fix needs to go into OFED 1.1 as well.
> Then Cc me on this patch :)
OK, I see, so you are the maintainer of IPoIB in OFED? and of any
other modules as wel
Quoting r. Or Gerlitz <[EMAIL PROTECTED]>:
> This is 2.6.18 material, correct?
Looks like it.
> also since the crash does not allow
> to debug mcast, i guess the fix needs to go into OFED 1.1 as well.
>
> Or.
Then Cc me on this patch :)
--
MST
___
> Or Gerlitz wrote:
>> Vladimir Sokolovsky wrote:
>> Did you have any special reason to assign host1:ib1 an IP address
>> ***before*** the failover? is the reason for that happen to be having
>> it joins the IPv4 multicast group at "batch time", that is not during
>> the failover?
> ib1 interf
Am Freitag, den 21.07.2006, 09:42 -0700 schrieb Sean Hefty:
> Arne Redlich wrote:
> > I only took a rather superficial look at the code. But since the invalid
> > GID REJ is treated with such care, I guess it's intentional and not a
> > bug?
>
> I would lean more towards it being a bug. The data
Roland,
This crash happens 1:1 with setting ipoib_debug_mcast, the fix applied by the
patch below is to set mcast->ah before the debug code attempts to access it.
This is 2.6.18 material, correct? also since the crash does not allow
to debug mcast, i guess the fix needs to go into OFED 1.1 as wel
42 matches
Mail list logo