Sean Hefty wrote:
mckey is intended to be a fairly simple send/receive multicast test program.
What's the reasoning behind adding the event handling?
The librdmacm examples serve for multiple purposes, among them user education
on how to write rdmacm based apps and as a vehicle to
Eli Dorfman (Voltaire) wrote:
Yevgeny Kliteynik wrote:
Eli Dorfman (Voltaire) wrote:
Yevgeny Kliteynik wrote:
Eli Dorfman (Voltaire) wrote:
Yevgeny Kliteynik wrote:
Eli Dorfman (Voltaire) wrote:
Yevgeny Kliteynik wrote:
Yevgeny Kliteynik wrote:
Line Holen wrote:
On 11/ 4/09 04:54 PM,
On Thu, Nov 12, 2009 at 8:00 AM, Keshetti Mahesh
keshetti.mah...@gmail.com wrote:
I am observing difference in behavior of 'saquery' on real cluster and
ibsim. Whenever I query for path records on cluster using
#saquery -p
it returns path records of all possible source-destination
On 18:30 Thu 12 Nov , Keshetti Mahesh wrote:
Is there any limitation in ibsim which is not letting the saquery program
retrieve all path records available ?
Yes, ibsim doesn't support large ( 256 bytes) RMPP packets yet.
Sasha
--
To unsubscribe from this list: send the line unsubscribe
Thanks Hal for the quick reply.
ibsim does not support RMPP.
Is there any plan to add RMPP support to ibsim in the future ?
--
Keshetti Mahesh
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
On Sun, Nov 8, 2009 at 9:37 AM, Yevgeny Kliteynik
klit...@dev.mellanox.co.il wrote:
Hi Sasha,
I noticed that OpenSM doesn't send InformInfo on traps 129/130/131.
This is what osm_trap_rcv.c is doing:
322: static void trap_rcv_process_request(IN osm_sm_t * sm,
323:
On 08:46 Thu 12 Nov , Hal Rosenstock wrote:
Any particular reason why there's no reporting of these traps?
AFAIK it's been this way for at least the last 5 or 6 years.
Yes, this is what I found too tracking down git/svn history.
A practical consideration in changing this is that
The librdmacm examples serve for multiple purposes, among them user education
on how to write rdmacm based apps and as a vehicle to test/validate/reproduce
features/bugs/issues, for example a follow program claimed that she isn't sure
to get a multicast error event on her application when a port
On 14:04 Tue 10 Nov , Ira Weiny wrote:
From: Ira Weiny wei...@llnl.gov
Date: Tue, 3 Nov 2009 13:58:03 -0800
Subject: [PATCH] infiniband-diags/ibqueryerrors: Reformat the output of the
xmtdiscard details.
Old:
Errors for 0x2c9020040fec8 Infiniscale-IV Mellanox Technologies
GUID
Hefty, Sean wrote:
Did you send it before morning coffee or after ? :)
Maybe, because it seems obvious.
All are agree with the elimination of unnecessary #ifndef's.
As to me, go ahead, commit it.
It needs to be merged into the mgmt.git tree.
#include stdlib.h
+#include inttypes.h
On Thu, 12 Nov 2009 18:31:05 +0200
Sasha Khapyorsky sas...@voltaire.com wrote:
On 10:34 Fri 06 Nov , Al Chu wrote:
Why do you want to remove this? port-path_portid can be useful for
logging, specific querying, etc.. Even node-dnext can be helpful for
some advanced use too.
On 17:48 Mon 09 Nov , Stan C. Smith wrote:
In order to skip the #ifndef __WIN__ around #include stdint.h, inttypes.h
was used.
'inttypes.h' includes stdint.h and exists in the WinOF svn tree.
OFED opensm builds without problems on EL 5.3.
Signed-off-by: stan smith
| While this is true for SLtoVL, we create other files which are
| device specific under the port directory too.
| It seems like we might need to introduce a callback into the driver to
| create the port specific sysfs files.
|
| Umm, you could have said there were other
On 09:20 Mon 05 Oct , Sean Hefty wrote:
/**
**/
-static long compar_mgids(const void *m1, const void *m2)
+static intn_t compar_mgids(const void
int compar(const void *, const void *);
, or to not bother at all and leave it as 'long'.
Hmm, I would prefer to change to 'int' to prevent future confusing.
Thoughts?
int seems to make the most sense to me.
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the
On 09:09 Mon 05 Oct , Smith, Stan wrote:
--- a/opensm/opensm/osm_subnet.c2009-10-01
12:45:52.0 -0700 +++ b/opensm/opensm/osm_subnet.c
2009-10-01 14:24:18.0 -0700 @@ -397,7 +397,7 @@
/**
On 11:38 Thu 12 Nov , Hal Rosenstock wrote:
We've cared for several years now (babbling port policy was introduced
into the master on 7/6/09) but this issue still persists. Are you
saying no reports for these traps until the trap rate is obeyed ?
If existing trap rate filtering in OpenSM
On 18:59 Fri 06 Nov , Sasha Khapyorsky wrote:
On 09:42 Wed 04 Nov , Hal Rosenstock wrote:
When --details selected, if PortCounters has transmit discards, then
query PortXmitDiscardDetails. On reset, if --details is selected, then
PortXmitDiscardDetails will also be reset.
See section 4 in the paper called High Performance RDMA Based MPI Implementation over
InfiniBand on the MVAPICH web page for description of one implementation that polls on data
buffers. Specifically, look at text around the statement Although the approach uses the
in-order implementation of
On 14:01 Tue 10 Nov , Ira Weiny wrote:
Sasha,
This and the subsequent patch clean up the port info query code to be more
straight forward.
First off I got rid of the decode_port_info function as it does very little
and combined the smp_query_via call into a query_port_info
The
Some of the SPARC servers from Sun support an IO memory mapping
mechanism for marking pages for weak or strong ordering. It has no
dependency on the HCA. Marking memory pages for data buffers as weak
ordered delivers significantly better throughput on IB.
I assumed that there are many more
On 15:28 Sun 08 Nov , Yevgeny Kliteynik wrote:
Fixing bug in cl_list_insert_array_head/tail functions.
In case of failure, no cleanup is done.
Signed-off-by: Yevgeny Kliteynik klit...@dev.mellanox.co.il
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe
On 15:35 Fri 06 Nov , Hal Rosenstock wrote:
Per published MgtWG erratum RefID 4576 - Add PortInfo component
to allow SM to tell port to suppress Bad P_Key traps for multicast
limited members
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Applied. Thanks.
Sasha
--
To
This failure, while indicating fatal problems with the device, shouldn't
kill the system.
Signed-off-by: Steve Wise sw...@opengridcomputing.com
---
drivers/infiniband/hw/cxgb3/cxio_hal.c |1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git
Hi Al,
On 09:51 Thu 12 Nov , Al Chu wrote:
Really? I thought that it could be a useful data for advanced uses.
It was removed in commit 094f922a34d6378d6a3bd1d137f90d6530685f94. It
was a simpler version of a patch that Ira had proposed on the mailing
list.
This means that I'm
Making things private allows us to change the library without breaking the
interface.
I don't think
Furthermore, it simplifies the interface for users who want to
write code at a higher level.
I'm not asking to make high level life harder :). My point is to not
prevent from advanced
On 12:41 Fri 06 Nov , Michael Reed wrote:
opensm enters an uninterruptible loop when the user enters
-g 0 on the command line. The only way to kill opensm is to
background the process and send kill -9.
The reason is that the port chooser is running in code section where all
signals are
On Fri, Nov 13, 2009 at 04:55:27AM +0200, Sasha Khapyorsky wrote:
On 12:41 Fri 06 Nov , Michael Reed wrote:
opensm enters an uninterruptible loop when the user enters
-g 0 on the command line. The only way to kill opensm is to
background the process and send kill -9.
The reason is
On 20:22 Tue 03 Nov , Ira Weiny wrote:
Sasha,
2 clean up patches which apply in order.
Ira
From: Ira Weiny wei...@llnl.gov
Date: Tue, 3 Nov 2009 13:19:33 -0800
Subject: [PATCH] infiniband-diags/ibqueryerrors: Change realloc of suppressed
fields to a static array
Realloc
Move new MGID allocation code into separate function build_new_mgid().
Signed-off-by: Sasha Khapyorsky sas...@voltaire.com
---
opensm/opensm/osm_sa_mcmember_record.c | 68 +++-
1 files changed, 40 insertions(+), 28 deletions(-)
diff --git
If we are going to do many MGID to single MLID compression we cannot use
MLID value for unique MGID generation.
Instead use static counter and also add querying for potentially existing
multicast group with same MGID value (with 1000 attempts limit).
Signed-off-by: Sasha Khapyorsky
31 matches
Mail list logo
