Anatrim – The latest and most attracting product for weighty people is made
available now – As could be seen on Oprah.
Do you recall all the cases when you asked yourself to do anything for being
delivered from this fastly growing number of kilos? Luckily, now no great price
is to be paid.
On 4/19/07, Michael S. Tsirkin [EMAIL PROTECTED] wrote:
Quoting Roland Dreier [EMAIL PROTECTED]:
Subject: Re: pkey change handling patch
So since all this thread was started by Moni because of IPoIB,
the path is clear in that respect, and would already be a step in the
right
Quoting Moni Levy [EMAIL PROTECTED]:
Subject: Re: pkey change handling patch
On 4/19/07, Michael S. Tsirkin [EMAIL PROTECTED] wrote:
Quoting Roland Dreier [EMAIL PROTECTED]:
Subject: Re: pkey change handling patch
So since all this thread was started by Moni because of IPoIB,
Quoting Bryan Lawver [EMAIL PROTECTED]:
Subject: IPoIB forwarding
I have a small test bed with 2 nodes with IB/OFED1.2/connected mode and a
third node which has IP only and is connected to one of the IB nodes. In
between are DDR IB switch and 10GE IP switch. The node with both IP and IB
Hi Yevgeny,
On Wed, 2007-04-25 at 08:10, Yevgeny Kliteynik wrote:
Hi Hal,
[V2] - Fixing a problematic usage of sprintf() in osm_helper.c:
When using sprintf(), source and destination strings should
not overlap, otherwise the function behavior is undefined.
Please apply to ofed_1_2 and
Hal Rosenstock wrote:
Hi Yevgeny,
On Wed, 2007-04-25 at 08:10, Yevgeny Kliteynik wrote:
Hi Hal,
[V2] - Fixing a problematic usage of sprintf() in osm_helper.c:
When using sprintf(), source and destination strings should
not overlap, otherwise the function behavior is undefined.
Please
osm/include/iba/ib_types.h: ib_get_node_type_str is to be used for
decoding the node_type from the node info attribute only
Signed-off-by: Ira K. Weiny [EMAIL PROTECTED]
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
osm/include/iba/ib_types.h |6 +++---
1 files changed, 3 insertions(+), 3
OpenSM/osm_helper: osm_get_node_type_str_fixed_width is to be used for
decoding the node_type from the node info attribute only
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
diff --git a/osm/include/opensm/osm_helper.h b/osm/include/opensm/osm_helper.h
index 3eab913..7164cc3 100644
---
Modify SW reset to acquire the HW semaphore before hitting reset. This is
recommended to prevent SW reset of the device while the flash is being
programmed.
Avoid delay after reset - this is not required. Instead, wait until the
vendor ID is read.
Signed-off-by: Eli Cohen [EMAIL PROTECTED]
---
Sean,
The Lamprey guys uncovered a potential bug in the RDMA CM:
If an application binds to address 0, port 0 via a call to
rdma_bind_addr(), and then calls rdma_get_src_addr(), the port number is
still 0 even though the RDMA CM chose a port during the bind operation.
However, If the application
Hi Sean,
I have a question about hotplug event:
I want to handle RDMA_CM_EVENT_DEVICE_REMOVAL event in RDS. When I got
this event, I call to rdma_disconnect function in order to modify the QP
to the error state but I got -EINVAL because of the following code in
cma.c:
if
On Wed, 2007-04-25 at 08:10, Yevgeny Kliteynik wrote:
Hi Hal,
[V2] - Fixing a problematic usage of sprintf() in osm_helper.c:
When using sprintf(), source and destination strings should
not overlap, otherwise the function behavior is undefined.
Please apply to ofed_1_2 and to master.
Cheers Effie,Here you'll find the largest collection of top quality watches that look exactly like their highly expensive brand prototypes!
Go to :: http://h.parspr.com Model : Daytona Full 18k GoldPrice : 179.-
Perfect quality
Worldwide shipping
Secure payment
You won't see
Sean,
This patch enables rping between a BE and LE system. Tested on IBM
PPC64 - AMD64.
Transfer rkey/addr/len information in network byte order.
Signed-off-by: Steve Wise [EMAIL PROTECTED]
---
examples/rping.c |7 ---
1 files changed, 4 insertions(+), 3 deletions(-)
diff --git
The built-in atomics in gcc had some ia64 issues until gcc 4.1.1 so we
had no choice but to get down and do some bogus things. :^)
Actually I think you've underestimated the depth of the bogosity.
Many of the uses of atomics in dapl that a quick grep turns up look to
be cargo cult uses where
Vlad,
These changes are a set of bug fixes to the chelsio drivers as well as
support for their latest firmware. This is required for OFED-1.2.
Please pull from:
git://git.openfabrics.org/~swise/ofed_1_2 ofed_1_2
Thanks,
Steve.
-
Shortlog:
-
Divy Le Ray:
Reuse the
What really should happen is that the field Local Ack Timeout in REQ
should be (2 * PacketLifeTime + Local CA’s ACK delay) (see 12.7.34)
and then the responder should use this for it's QP.
Just to clarify, the value is _based_ on (2 * PacketLifeTime + local CA ack
delay). For example, if
Roland Dreier wrote:
./dapl/udapl/linux/dapl_osd.h:82:24: error: asm/atomic.h: No such file or
directory
In file included from ./dapl/include/dapl.h:50,
from dapl/udapl/dapl_init.c:39:
./dapl/udapl/linux/dapl_osd.h: In function 'dapl_os_atomic_inc':
Meanwhile, in all the stuff in docs/ at which I have glanced thusfar,
none of it seems to describe how to go about building a native SDP
application - one which creates an AF_INET_SDP socket explicitly. Ie
which include file(s) to include and any libraries against which it
should be
Should I move the QP to the error state in RDS or cma should handle this
state too.
Let me think about what to do here. I think the cma should perform this
transition if it makes sense.
- Sean
___
general mailing list
general@lists.openfabrics.org
There is no header, see https://bugs.openfabrics.org/show_bug.cgi?id=25.
Scott
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of
Roland Dreier (rdreier)
Sent: Wednesday, April 25, 2007 10:06 AM
To: Rick Jones
Cc: general@lists.openfabrics.org
+static struct miscdevice local_sa_misc = {
+ .minor = MISC_DYNAMIC_MINOR,
+ .name = ib_local_sa,
+};
I don't understand why you're registering a miscdevice etc. I don't
see any implementation of a character device or indeed any userspace
interface at
Scott Weitzenkamp (sweitzen) wrote:
There is no header, see https://bugs.openfabrics.org/show_bug.cgi?id=25.
Thanks - BTW my browser got a triffle cranky accessing that site
complaing that the cert presented for bugs.openfabrics.org belongs to
:staging.openfabrics.org
I guess that until it
That seems like an abuse of the miscdevice stuff, since you don't
actually have a device. Why not just use module parameters? The only
difference would be that the paths start /sys/module/ib_local_sa/parameters
instead. Or if you really wanted to, I guess a sysctl would be
appropriate. But I
I want changes to these values to force a cache update. Can you do that with
module parameters?
Sure... you'll have to implement your own set method but that's no
different from putting attributes under your miscdevice. Just look at
module_param_call() -- it's exactly what you want I think.
Tziporet,
Below is a patch to your docs git tree. It adds a cxgb3 release notes
file, and updates the OFED release and install docs.
---
Updates for chelsio.
- added cxgb3 release notes file
- updated ofed release notes and installation guide.
Signed-off-by: Steve Wise [EMAIL PROTECTED]
Sure... you'll have to implement your own set method but that's no
different from putting attributes under your miscdevice. Just look at
module_param_call() -- it's exactly what you want I think.
Thanks - I'll update this.
___
general mailing list
Thanks, I applied a version of this that checks the vendor field is !=
0x instead of checking it against the mellanox value, because
someone may build HCAs with a different vendor value.
Also is it worth taking the semaphore in the mthca reset function?
- R.
OFED 1.2 April 25 meeting summary
Main decisions:
1. RC was delayed to end of next week due to many critical open bugs
2. RC3 due date is May 3.
3. All code changes should be ready for May 2
4. Bug fixes after RC3 will have to be approved by the RM (Tziporet)
Cluster testing:
- Intel will
- Intel will test IPoIB, Intel MPI and MVAPICH on 256 nodes cluster -
was not started yet
This will begin starting tomorrow night. We had to delay a week to allow
them to upgrade to EL4-u4, since OFED 1.2 did not support EL4-U2, which
is what they were running.
woody
Hi Hal,
I am working on a new SGI product that will have two separate InfiniBand
fabrics.
Each of these fabrics may have a different topology and could be running one of
a number of routing engines (i.e. lash or up/dn) the Subnet Management for both
fabrics will run on one host (leader node).
Thanks for the input. I had found the debug switch and it pointed out some
issues. Drops at the routing node are many. It seems to work at debug rate
so I hope it is not a buffer pool issue.
At 05:46 AM 4/25/2007, Michael S. Tsirkin wrote:
Quoting Bryan Lawver [EMAIL PROTECTED]:
Subject:
On Mon, 2007-04-23 at 18:05, Sean Hefty wrote:
Has anyone thought about using replication rather than caching to
solve this problem? It seems to me it would be alot faster for some
single process in the network to fetch and keep a copy of the entire
SA route database, format it into a binary
Quoting Hal Rosenstock [EMAIL PROTECTED]:
Subject: RE: [RFC] [PATCH 0/3] 2.6.22 or 23 ib: add path record cache
On Mon, 2007-04-23 at 17:32, Sean Hefty wrote:
Isn't there a way to get notice for this?
The closest trap I'm aware of is GID in/out of service. See 14.2.5.1 and
14.4.9.
Quoting Roland Dreier [EMAIL PROTECTED]:
Subject: Re: [PATCH] net/mlx4: modify sw reset
Thanks, I applied a version of this that checks the vendor field is !=
0x instead of checking it against the mellanox value, because
someone may build HCAs with a different vendor value.
Also is
Also is it worth taking the semaphore in the mthca reset function?
Why not? PRM says we should - the point of this is protecting against someone
accidentally touching flash while device is reset,
which will result is hangs, system errors or even data corruption.
OK... that was just my
We can get notic on port state changes though, can't we?
Trap 128 (sent by switches) indicates that the link state at least one port of
switch at LIDADDR has changed. I think it would be difficult for all nodes
to determine which paths were affected.
- Sean
On Fri, 2007-04-20 at 15:32 -0700, Roland Dreier wrote:
Hook up mlx4_core and mlx4_ib drivers to Kconfig and Makefiles.
Signed-off-by: Roland Dreier [EMAIL PROTECTED]
---
infiniband/Kconfig |2 ++
infiniband/Makefile |1 +
infiniband/hw/mlx4/Kconfig |9
Hi John,
On Wed, 2007-04-25 at 16:05, John Partridge wrote:
Hi Hal,
I am working on a new SGI product that will have two separate InfiniBand
fabrics.
Each of these fabrics may have a different topology and could be running one
of
a number of routing engines (i.e. lash or up/dn) the
+config MLX4_CORE
+ tristate
+ depends on PCI
+ default n
+
No help menu for the core module?
It's an invisible option. You get it if you enable anything that uses
it (mlx4_ib or mlx4_eth), and you never even have to know about the
option.
- R.
I guess that until it is resolved I'll just kludge around it with my own
define.
Soo, I did a bunch of cut and paste in netperf, where I take what
getaddrinfo() returns and replace the -ai_family with AF_INET_SDP and
set the -ai_protocol to 0 (a guess since there isn't much in the way of
Rick, I still think this is unecessary copy-paste code duplication, just
so the report says SDP instead of TCP. Part of the advantage of
libsdp.so is you can use SDP w/o having to recode your application.
Do you really want to maintain duplicate code for all the
TCP_STREAM-vs-SDP_STREAM, etc.
Has anyone run into an issue with autotools not generating the .so extension to
built library files, or know how to fix such an issue?
- Sean
___
general mailing list
general@lists.openfabrics.org
On Mon, 2007-04-23 at 05:40, GREGOIRE Philippe wrote:
Hal Rosenstock a écrit :
The following changes are proposed for IB management (master branch of
my management git tree):
In order to better match package names, the following directory names to
be changed from-to:
osm-opensm
Can you give this a try?
The source address was being overwritten by whatever the user passed into
rdma_bind_addr.
Signed-off-by: Sean Hefty [EMAIL PROTECTED]
---
diff --git a/src/cma.c b/src/cma.c
index c5f8cd9..fdadb69 100644
--- a/src/cma.c
+++ b/src/cma.c
@@ -509,12 +509,7 @@ int
That works!
Acked-by: Steve Wise [EMAIL PROTECTED]
On Wed, 2007-04-25 at 15:47 -0700, Sean Hefty wrote:
Can you give this a try?
The source address was being overwritten by whatever the user passed into
rdma_bind_addr.
Signed-off-by: Sean Hefty [EMAIL PROTECTED]
---
diff --git
Roland Dreier wrote:
Is there a differentiation between multiple CQE's being in the CQ
vs. CQE's being arriving into the CQ when using completion
notifications?
For example, assume I have the following order of events:
2 CQEs arrive
select() returns readable for comp.
In this case it brings up an interesting question. If the adaptor will
only generate an event for a new CQE added, how do you tell that
you've successfully polled all the CQEs that triggered the first event?
You could continuously call ibv_poll_cq() until it returned zero, but
this has
Quoting Rick Jones [EMAIL PROTECTED]:
Subject: initial set of direct SDP tests in netperf
I guess that until it is resolved I'll just kludge around it with my own
define.
Soo, I did a bunch of cut and paste in netperf, where I take what
getaddrinfo() returns and replace the -ai_family
Quoting Hal Rosenstock [EMAIL PROTECTED]:
Subject: RE: [ofa-general] Re: [RFC] [PATCH 0/3] 2.6.22 or 23 ib: add
pathrecord cache
On Mon, 2007-04-23 at 18:51, Sean Hefty wrote:
We could solve this by implementing a process running on the same node as
the
SA.
And it's probably not
50 matches
Mail list logo
