Hello Sean, Thanks for replying.
Still I am not able to get it completely, I have seen through the SA
code please help me out in understanding SA design as a whole, my
query is as follows
If some client calls cma_resolve_ib_route(), and let's assume that its
local cache miss, and
Hi Hal,
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.
Thanks.
-- Yevgeny
Signed-off-by: Yevgeny Kliteynik [EMAIL
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
diags-openib-diags
Since opensm is a system daemon,
Quoting Yevgeny Kliteynik [EMAIL PROTECTED]:
Subject: [PATCH] osm: source and destination strings overlap when using
sprintf()
Hi Hal,
Fixing a problematic usage of sprintf() in osm_helper.c:
When using sprintf(), source and destination strings should
not overlap, otherwise the
Gentle reminder: there is no EWG teleconference today in observance
of the Israel holiday.
This week's teleconference was moved to Wednesday, 25 April:
- 11:30pm US Pacific
- 2:30pm US Eastern
- 9:30pm Israel
I just noticed that I was an hour off when I moved the meeting
invitation last
If this is a reasonable thing to do -can some one tell me how to change
this? I did not see any API available to do this?
Pradeep
[EMAIL PROTECTED]
Pradeep Satyanarayana/Beaverton/IBM
04/20/2007 01:35 PM
To
general@lists.openfabrics.org
cc
Subject
How to set local_ca_ack_delay?
This
Local CA ACK delay is not something that you set. It is read-only
information that the HCA is giving you about how long it will take in
the worst case to generate and send an ACK. If some HCA is returning
0 for this, then that is most likely a bug in the HCA or driver.
- R.
There are quite a few places in ipoib_cm.c where we know IRQs are
enabled because we do something that sleeps in the same function, so
we can convert several occurrences of spin_lock_irqsave() to a plain
spin_lock_irq(). This cleans up the source a little and makes the
code smaller too:
Add the missing device link from /sys/class/infiniband/* to the actual device.
Signed-off-by: Joachim Fenkes [EMAIL PROTECTED]
---
sysfs.c |1 +
1 file changed, 1 insertion(+)
--- linux-2.6.20/drivers/infiniband/core/sysfs.c.old2007-04-23
15:37:37.0 +0200
+++
Add Modify Port verb support to eHCA driver.
ib_cm needs this to initialize properly.
Signed-off-by: Joachim Fenkes [EMAIL PROTECTED]
---
ehca_hca.c | 48 ++--
hcp_if.c | 24
hcp_if.h |4
3 files changed, 74
Quoting Roland Dreier [EMAIL PROTECTED]:
Subject: [PATCH] IPoIB/cm: Convert spin_lock_irqsave() to spin_lock_irq()
There are quite a few places in ipoib_cm.c where we know IRQs are
enabled because we do something that sleeps in the same function, so
we can convert several occurrences of
If some client calls cma_resolve_ib_route(), and let's assume that its
local cache miss, and cma_query_ib_route() is called, this will send a
SA query to the SM node CMIIW, Now on SM node I am not able to figure
out that who will respond this GMP, and how requested attribute info
is collected?
Hi Joachim,
On Mon, 2007-04-23 at 12:23, Joachim Fenkes wrote:
Add Modify Port verb support to eHCA driver.
ib_cm needs this to initialize properly.
Signed-off-by: Joachim Fenkes [EMAIL PROTECTED]
---
ehca_hca.c | 48 ++--
hcp_if.c |
There are some signature differences between versions.
Since redirection exposes signatures it is not trivial
for a single redirection to support different signatures.
Is this really needed?
Thanks,
Arkady Kanevsky email: [EMAIL PROTECTED]
Network Appliance Inc.
1. What happens on e.g. a heterogenious network? It seems that
path to a specific GID might change e.g. MTU without GID
going in/out of service. How would this be handled?
If the path parameters change without a GID going in/out of service, then the
cached path records would be off. A
Hmm, I have links like this on my system already:
$ ls -l /sys/class/infiniband/mlx4_0/device
lrwxrwxrwx 1 root root 0 2007-04-23 12:14
/sys/class/infiniband/mlx4_0/device -
../../../devices/pci:00/:00:06.0/:0d:00.0
the patch actually looks sane but I don't understand why
+if (hipz_h_query_port(shca-ipz_hca_handle, port, rblock) != H_SUCCESS)
{
+ehca_err(shca-ib_device, Can't query port properties);
+ret = -EINVAL;
+goto modify_port1;
+}
+
+cap = (rblock-capability_mask | props-set_port_cap_mask)
+
A straight-forward approach would be to listen for port up/down events
rather than or in addition to GID in/out, and do network discovery by DR SMPs.
I'm not entirely following you. How would you listen for port up/down events?
And are you suggesting that all nodes do network discovery using DR
Michael S. Tsirkin [EMAIL PROTECTED] wrote on 04/19/2007 04:51:19
AM:
Quoting Pradeep Satyanarayana [EMAIL PROTECTED]:
Subject: IPOIB CM (NOSRQ)[PATCH V2] patch for review
Here is a second version of the IPOIB_CM_NOSRQ patch for review. This
patch will benefit adapters that do not
Quoting Sean Hefty [EMAIL PROTECTED]:
Subject: RE: [RFC] [PATCH 0/3] 2.6.22 or 23 ib: add path record cache
A straight-forward approach would be to listen for port up/down events
rather than or in addition to GID in/out, and do network discovery by DR
SMPs.
I'm not entirely following
This patch has been tested with linux-2.6.21-rc5 and rc7 with Topspin and
IBM HCAs on ppc64 machines. I have run
netperf between two IBM HCAs and two Topspin HCAs, as well as between IBM
and Topspin HCA.
Note 1: There was interesting discovery that I made when I ran netperf
Vlad,
I'm trying to build the src tree that is installed when you install the
kernel-ib-devel package and I'm hitting a problem. iw_cxgb3 fails to
load because the ib_core module doesn't have the genalloc code included
in it. I think the Makefile in drivers/infininband/core didn't get
patched
On Mon, Apr 23, 2007 at 11:20:59PM +0300, Michael S. Tsirkin wrote:
I haven't thought this through yet. Basically, I just note that
caching the path until GID goes out of service isn't right - since
path parameters such as MTU or rate might change without GID going
out of service.
So what
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. GID in/out of service is related to the existence of a path record
between the SGID and DGID. If the path record parameters change, I'm not sure
if the GID technically
4. Why do we need yet another API and yet another module to speed up just
RDMA/CM path record queries? We now get 2 ways to do this (with/without the
cache). Shouldn't there be just one?
I did consider this, but the cache operates synchronously, and ib_sa interface
is asynchronous. I
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 format and use RC RDMA to
transfer it to every node
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 format and use RC RDMA to
transfer it to every
We could solve this by implementing a process running on the same node as the
SA.
And it's probably not too hard to add a way for opensm to spit out
the table into an external file when it gets a signal or something.
I agree that there are ways to solve this, but those solutions won't work with
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
interfaces is simply a IP router in this test setup. The IB only node
On Tue, Apr 24, 2007 at 01:30:05AM +0300, Michael S. Tsirkin wrote:
For say, 1 nodes you could compact an any-to-any path table into
around 20 megabytes.
I wonder how do you propose to compact the path records that drastically?
Number of records seems to be 1 * 1 / 2 = 50
Maybe, but there might be several other reasons for this.
One might be that IPoIB is slower than link speeds,
so e.g. miscalculating the rate still does not cause network failures.
Another might be that people run TCP mostly, which
is very good at recovering from failures, so if you get the LID
Michael S. Tsirkin [EMAIL PROTECTED] wrote on 04/23/2007 01:50:32
PM:
This would only help if there are short bursts of high-speed
activity
on the receiving HCA: if the speed is different in the long run,
the right thing to do is to drop some packets and have TCP adjust
its
Y-yeah. orange yearly sea Dana was now label slightly trembling. IsJeff paused
annually spare for a shut fought moment to gather his thoughts drain bit Dana
gave corporeal her a wound stern look, Sometimes, a man'svespine I'll try pump
talking to wheel her tomorrow. sprang You know, my
33 matches
Mail list logo
