...@voltaire.com
Reviewd-by:Moni Shoua mo...@mellanox.com
--
Index: b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c
===
--- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c2008-11-26
18:28:04.0 +0200
+++ b/drivers/infiniband
+Signed-off-by: Moni Shoua mo...@voltaire.com
+--
+
+diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h
b/drivers/infiniband/ulp/ipoib/ipoib.h
+index ab97f92..fb1714f 100644
+--- a/drivers/infiniband/ulp/ipoib/ipoib.h
b/drivers/infiniband/ulp/ipoib/ipoib.h
+@@ -92,6 +92,7 @@ enum
maximum retry setting in
sys/module/ib_sa/parameters/mcast_leave_retries.
+Add a debug print when the maximum retry count is reached.
+
+Signed-off-by: Nir Muchtar n...@voltaire.com
+Reviewed-by: Moni Shoua mo...@voltaire.com
+--
+
+Index: ofa_kernel-1.5.2/drivers/infiniband/core/multicast.c
with EINVAL error then it tries to align
users memory range by huge page size and tries madvise() again.
Signed-off-by: Alex Vaynman al...@voltaire.com
Reviewed-by: Moni Shoua mo...@voltaire.com
---
src/memory.c | 69 +-
1 files
Signed-off-by: Aleksey Senin aleks...@voltaire.com
Reviewed-by: Moni Shoua mo...@voltaire.com
--
Index: libibverbs-1.1.3/include/infiniband/verbs.h
===
--- libibverbs-1.1.3.orig/include/infiniband/verbs.h2010-05-27
11:18
Add support for RAW_ETH QP. Also, consider the flag IBV_SEND_IP_CSUM in send
operation.
Signed-off-by: Aleksey Senin aleks...@voltaire.com
Reviewed-by: Moni Shoua mo...@voltaire.com
--
diff --git a/src/qp.c b/src/qp.c
index 4322513..d850b1c 100644
--- a/src/qp.c
+++ b/src/qp.c
@@ -229,6
Identify that VLAN is present in a completion and copy it along with the VLAN
ID to the user completion.
Signed-off-by: Aleksey Senin aleks...@voltaire.com
Reviewed-by: Moni Shoua mo...@voltaire.com
--
--- a/src/cq.c 2010-02-08 17:33:45.0 +0200
+++ b/src/cq.c 2010-02-10 15:12
Mike Heinz wrote:
Wouldn't the BUSY patch I proposed last year deal with this situation?
Can you please send a link to this patch?
-Original Message-
From: ewg-boun...@lists.openfabrics.org
[mailto:ewg-boun...@lists.openfabrics.org] On Behalf Of Moni Shoua
Sent: Wednesday
Add a new flag that can come with a completion to state a presence of a VLAN
Signed-off-by: Aleksey Senin aleks...@voltaire.com
Reviewed-by: Moni Shoua mo...@voltaire.com
--
--- a/include/infiniband/verbs.h2010-01-19 16:12:06.0 +0200
+++ b/include/infiniband/verbs.h
Mike Heinz wrote:
It was discussed in the Linux-RDMA list for many months. You can find a list
of the archived messages here:
http://www.mail-archive.com/search?q=SA+Busyl=linux-r...@vger.kernel.org
The most recent version of the patch is here:
Overriding the distro open-iscsi RPM is no longer required.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
install.pl | 13 +
1 file changed, 1 insertion(+), 12 deletions(-)
--- install.pl.orig 2010-07-08 12:44:27.0 +0300
+++ install.pl 2010-07-08 15:07:23.0
Moni,
May open-iscsi be removed from OFED now?
This is the general idea.
However, Or Gerlitz needs some more time to make a final decision.
Can we postpone the answer to this until next OFED?
Moni
___
ewg mailing list
ewg@lists.openfabrics.org
feature they will benefit from the common part.
---
Moni Shoua| +972-54-5567934
-Original Message-
From: Roland Dreier [mailto:rdre...@cisco.com]
Sent: Wednesday, June 23, 2010 8:32 PM
To: Moni Shoua
Cc: Aleksey Senin; Eli Cohen; e
Hi,
In yesterday's EWG meeting we agreed that a better name is required instead of
IBV_QPT_RAW_ETH.
The qp type IBV_QPT_RAW_ETH was introduced lately and it represents a QP that
doesn't add headers to the data in post_send operation. Today, this QP is used
in applications that want to send
It doesn't even look like this patch and the mlx4 patch were ever posted
to linux-rdma. Only to the EWG list.
Not 100% correct. See thread from April 30.
Patches to core, libibverbs and NES driver were presented there.
Granted our dev process may not be documented, but I always assumed
; v...@dev.mellanox.co.il; e...@openfabrics.org;
e...@mellanox.co.il; Moni Shoua
Subject: Re: [ewg] [PATCH v4] IB Core: RAW ETH support
Alekseys Senin wrote:
This patch adds support to RAW ETH QP in ib core.
are these patches applicable to the mainstream kernel code or would only
apply/function
Hi Vlad
Please apply this patch.
Now, when RAW patches that you accepted today are applied, RAW ETH QP support
is broken for NE driver unless this patch is applied.
thanks
Aleksey Senin wrote:
This patch will fix existing NES code to support RAW_ETH instead of RAW_ETY
type.
I think that Eli was referring to the MLX4 specific patches but we'll
send the common and specific patches together once again
-Original Message-
From: Vladimir Sokolovsky [mailto:v...@dev.mellanox.co.il]
Sent: Sunday, June 13, 2010 5:28 PM
To: Aleksey Senin
Cc: Moni Shoua; ewg
Aleksey Senin wrote:
Those patches add new RAW_ETH QP type to the kernel in order to support
creation of RAW Ethernet packets for iWARP and RDMAOE protocols.
The reason for new type is that RAW_ETY QP already used by Mellanox
drivers for another purpose. Another reason, that there is RAW_ETH
Hi,
Is there an intension to put them on the web for download?
thanks
MoniS
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
That's it
Thanks :)
-Original Message-
From: Rupert Dance [mailto:rsda...@lampreynetworks.com]
Sent: Wednesday, March 31, 2010 17:44
To: Moni Shoua; 'Jeffrey Scott'
Cc: 'Bill Boas'; 'ewg'
Subject: RE: [ewg] A request for Sonoma 2010 presentations
Hi Moni,
I believe
No,
I don't plan any other commits before 1.5.1 GA
-Original Message-
From: Vladimir Sokolovsky [mailto:v...@dev.mellanox.co.il]
Sent: Monday, March 22, 2010 18:48
To: Moni Shoua
Cc: ewg
Subject: Re: [ewg] [PATCH] rdma_cm: Address change event is not sent
underRHEL4.7/8
Moni Shoua wrote
Fix the bug where rdma_cm doesn't send the event RDMA_CM_EVENT_ADDR_CHANG under
RHEL4.7 and RHEL4.8. This only needed to copy the backport of RHEL4.6 to the
backport
directory of RHEL4.7 and RHEL4.8.
This commit closes https://bugs.openfabrics.org/show_bug.cgi?id=1998
Signed-off-by: Moni Shoua
Hi,
The issue of ditributing OFED as a container of bimary packages was raised in
the EWG
conf. call yesterday. We have some experiecne with it in Voltaire that we would
like to share.
Today, OFED is distributed as a container of source packages (SRPMS) along with
an
install script. Container
.
Patches are to the install script and build scripts and were split into three
parts.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
--- install.pl.orig 2010-02-21 11:17:26.0 +0200
+++ install.pl 2010-02-21 13:00:36.0 +0200
@@ -1679,7 +1679,7 @@
# Iser
build and install iser without iscsi modules.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
--- ofa_kernel.spec.orig2010-02-21 14:52:55.0 +0200
+++ ofa_kernel.spec 2010-02-21 14:54:47.0 +0200
@@ -51,6 +51,7 @@
%define build_srp %(if ( echo %{configure_options
Add parsing of option bonding_force_all_os and passing
it to rpmbuild when building ib-bonding. The meaning of this new
option is: build ib-bonding for any OS and override the judgment
in the spec file.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
--- install.pl.orig 2010-02-16 18:41
Hi,
I've put the SRPM in the public folder in my account and it should be available
in the next OFED build
The only change in this version is a fix for bugzilla 1915
(https://bugs.openfabrics.org/show_bug.cgi?id=1915)
To avoid an unneeded compilation under RHEL5.4 I added an option in ofed.conf
Brian J. Murrell wrote:
On Mon, 2010-01-11 at 18:26 +0200, Tziporet Koren wrote:
This is the agenda for the meeting today:
1. OFED 1.5.1 status:
...
Other changes that should be done:
...
- Other - let discuss in the meeting today
How about the ISER/iSCSI on kernels 2.6.30
Limitations:
- SLES10 SP3 on IA64 is not supported yet
In what way is it unsupported?
Our QA say that OFED-1.5 can be compiled on this platform.
___
ewg mailing list
ewg@lists.openfabrics.org
Notes:
RDMAoE and mlx4_en capabilities are supported only for x86 and x86_64
architectures.
Since Voltaire is testing and distributing the RDMAoE release and we wish to
take care of a single package, I'd like to better understand this.
If OFED-RDMAoE is compiled on non x86_64 and used
There is no need to unenslave IPoIB interfaces before stopping the stack.
This can cause a kernel crash as described in
https://bugs.openfabrics.org/show_bug.cgi?id=1821
Signed-off-by: Moni Shoua mo...@voltaire.com
---
diff --git a/ofed_scripts/openibd b/ofed_scripts/openibd
index 8b0c3e6
Give a better explanation to the ib-bonding OS scope
Signed-off-by: Moni Shoua mo...@voltaire.com
---
diff --git a/ipoib_release_notes.txt b/ipoib_release_notes.txt
index f60ca7c..2d1217a 100644
--- a/ipoib_release_notes.txt
+++ b/ipoib_release_notes.txt
@@ -272,10 +272,13 @@ Notes:
no longer
Is there going to be a meeting tomorrow?
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Voltaire intends to test with OFED-RDMAoE release.
The following setup are used in our QA while
each setup consists of 2 machines.
Arch|OS / Kernel | HCA|SM
+-+--+---
x86_64 | RH4 up6 *| ConnectX QDR | OpenSM
Import commit 5ee95120841fd623c48d7d971182cf58e3b0c8de from Roland's tree to
fix bugzilla 1726
Signed-off-by: Moni Shoua mo...@voltaire.com
---
kernel_patches/fixes/ipoib_0570_check_state_before_carrier.patch | 43
++
1 file changed, 43 insertions(+)
diff --git a/kernel_patches/fixes
If gcc is not installed then don't check for gcc version. This prevents error
messages to be printed on the screen.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
openibd |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/ofed_scripts/openibd b/ofed_scripts/openibd
index
Add comment for ib-bonding and distros that use new kernels (i.e. SLES11)
Signed-off-by: Moni Shoua mo...@voltaire.com
---
ipoib_release_notes.txt |5 +
1 file changed, 5 insertions(+)
diff --git a/ipoib_release_notes.txt b/ipoib_release_notes.txt
index adf1304..3c3d70f 100644
Add commit 5e47596bee12597824a3b5b21e20f80b61e58a35 to kernel fixes.
This will fix https://bugs.openfabrics.org/show_bug.cgi?id=1664.
Signed-off-by: Moni Shoua mo...@voltaire.com
---
kernel_patches/fixes/ipoib_0550_check_multicast_address_format.patch | 51
++
1 file changed, 51
Hi Tziporet,
This is a patch for bonding doc
thanks
diff --git a/ib-bonding.txt b/ib-bonding.txt
index c797cde..c6266f4 100644
--- a/ib-bonding.txt
+++ b/ib-bonding.txt
@@ -15,6 +15,8 @@ IB Bonding
---
ib-bonding is a
Hi,
In spec file of openmpi there is a boolean option called ofed.
When it is true, among other things, it makes openmpi to depend
on mpi-selector.
Why not leave it to the user to decide whether to use (and install) mpi-selector
or not. This can also create a conflict between mpi-selector that
Eli Cohen wrote:
On Wed, Mar 25, 2009 at 09:20:19PM +0200, Yossi Etigin wrote:
Wouldn't this patch break the fast bond+SM failover recovery?
Up till now, in such case neigh-dgid was the old gid until path record
query was successful, and that caused to retry the path query until it's
Alexander Schmidt wrote:
Hi all,
with the latest daily build (OFED-1.4.1-20090305-0600), I run into a build
problem with ib-bonding, running on SLES10 SP2 on powerpc. The last daily
build I tested (OFED-1.4-20090208-0600) worked fine, it contained
ib-bonding-0.9.0-36 (new version is
not checked.
Reported-by: Philippe Bernadat philippe.berna...@hp.com
Debugged-by: Philippe Bernadat philippe.berna...@hp.com
Signed-off-by: Moni Shoua mo...@voltaire.com
---
kernel_addons/backport/2.6.16_sles10/include/linux/inetdevice.h | 22
++
kernel_addons/backport
-general] [PATCH] ipoib: Fix loss of connectivity after
bonding failover on both sides
Date: Tue, 18 Nov 2008 13:34:45 +0200
From: Moni Shoua [EMAIL PROTECTED]
To: Yossi Etigin [EMAIL PROTECTED], Roland Dreier [EMAIL PROTECTED]
CC: Olga Shern [EMAIL PROTECTED], general list [EMAIL PROTECTED
Hi Moni,
Let's give people some time for review.
Sure.
Did you checked that replacing the patch in OFED will not require
backport patches update?
Yes. The change doesn't require any change in backports
If not, please check and update backports if required.
Regards,
Vladimir
Thanks
Hi,
I released a new version today. The only change is setting the default number
of grar. ARP to send after failover to three.
This change is an outcome of bug 1335
___
ewg mailing list
ewg@lists.openfabrics.org
This one is already in my directory and should be in next OFED.
Change Log
==
Fix bug 1250
Fix bug 1230
thanks
MoniS
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Vladimir Sokolovsky wrote:
Olga Shern (Voltaire) wrote:
Hi Eli,
We also want to run regression tests with this patch.
Please let me know when OFED daily build will include it.
Thanks
Olga
Hi Olga,
OFED-1.4-20080928-0756.tgz includes this patch.
Regards,
Vladimir
Vladimir Sokolovsky wrote:
On Sun, 2008-09-28 at 18:09 +0300, Moni Shoua wrote:
Vladimir Sokolovsky wrote:
Olga Shern (Voltaire) wrote:
Hi Eli,
We also want to run regression tests with this patch.
Please let me know when OFED daily build will include it.
Thanks
Olga
Hi Olga,
OFED-1.4
Vladimir Sokolovsky wrote:
Thanks.
Did you check that this patch doesn't invalidate backport patch?
Did you have to change any backport patch?
Yes,
See commit 7891514239b97fe3396e92d2fcf24f7ab18e60e9.
Moni, don't you have OFED's git tree to check?
Regards,
Vladimir
interrupts have
been reenabled.
Also, since there is still a race that could end up with a call to skb_orphan()
for an
already freed skb, a module parameter allows to disable it totally.
Signed-off-by: Arthur Kepner [EMAIL PROTECTED]
Signed-off-by: Moni Shoua [EMAIL PROTECTED]
---
Index: ofa_kernel
Eli Cohen wrote:
Commit ee1e2c82c245a5fb2864e9dbcdaab3390fde3fcc introduced an
optimization on path flushing. This caused a new possible scenario in
which unicast_arp_send triggers path query which could fail, causing
path-ah to become NULL. A successive successfull path query will then
Eli Cohen wrote:
On Tue, Sep 23, 2008 at 02:01:00PM +0300, Moni Shoua wrote:
Eli Cohen wrote:
Commit ee1e2c82c245a5fb2864e9dbcdaab3390fde3fcc introduced an
optimization on path flushing. This caused a new possible scenario in
which unicast_arp_send triggers path query which could fail
Eli Cohen wrote:
On Tue, Sep 23, 2008 at 02:15:47PM +0300, Moni Shoua wrote:
I agree
I gave a thought here.
It's possible, when path_rec_completion() is called with nonzero status, to
do nothing with ah.
Only when path query finishes with success do the replacement.
This is good for cases
Hi,
This release fixes the problem of loud output when bringing up a bond interface
on Redhat 5
It also changes a bit the way to configure bond interfaces with standard OS
tools. Please read the doc file for details.
Thanks
MoniS
___
ewg mailing
Moni Shoua wrote:
Vlad,
Please hold on with this patch.
I might have a little change for it.
thanks
MoniS
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Vlad,
You can apply
outside ofaed modules suite) doesn't set
this flag and as a result event are filtered in a wrong way.
Getting rid of this filter is safe. In rare cases it may save a little time
but
even than, in the slow path.
Signed-off-by: Moni Shoua [EMAIL PROTECTED]
---
kernel_patches/backport/2.6.16
Vlad,
Please hold on with this patch.
I might have a little change for it.
thanks
MoniS
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
Hi,
SRPM is in the usual place for OFED-1.4 (the file latest.txt was updaated)
Change Log
--
1. Fix bug 1141
2. Add support for Redhat 4 Up 7
3. Add support for Redhat 5 Up 2
thanks
Vlad
We still need to add support in install.pl. Do you want me to do that?
Hi,
SRPM is in OFA server and should be taken during the build of OFED tarball.
Vlad,
Please note that I've put it under ~monis/public_html/ofed_1_4 as well as under
~monis/public_html/ofed_1_3
Changes since last release
* Take some fixes that were added from linux-2.6.24 to linux-2.6.26
Hi,
This is a question regarding preconditions for openibd script before it
starts. The question assumes that OS is SuSE but the only difference
from other OS is just in the mechanism.
openibd script is installed with Required-Start: $network directive.
This forces /etc/init.d/network to run
Hi,
I have put a new release of the ib-bonding RPM (file latest.txt is
up-to-date)
Release content
1) Bug fix with bonding interface and multicast traffic
2) Improve failover mechanism.
thanks
MoniS
___
ewg mailing list
Moni Shoua wrote:
move a little up the code that checks for a situation where the remote
GID stored in the ipoib_neigh is
different than the one present in the neighbour (handle Gratuitous ARP)
or that a bonding fail over has
happened but the neighbour still has a pointer
- bond0 appears as constant instead of $BOND_NAME in one
place
5. Fix: Destroy bonding master only if it exists
Vlad,
Please apply the patch below to ofed_1_3_scripts to complete the support for
RH4 Up6
Add 2.6.9-67 to the list of kernels that are supported by ib-bonding
Signed-off-by: Moni
Scott Weitzenkamp (sweitzen) wrote:
Or,
I don't see /sbin/call_ifenslave in my OFED-1.3-20080115-0600 ib-bonding
package.
Also, please run ib-bond --version. /sbin/call_ifenslave should be there only
from release 16 and higher.
___
ewg mailing
Scott Weitzenkamp (sweitzen) wrote:
Or,
I don't see /sbin/call_ifenslave in my OFED-1.3-20080115-0600 ib-bonding
package.
[EMAIL PROTECTED] ~]# uname -a
Linux svbu-qa1850-1 2.6.9-55.ELsmp #1 SMP Fri Apr 20 16:36:54 EDT 2007
x86_64 x8
6_64 x86_64 GNU/Linux
[EMAIL PROTECTED] ~]# rpm -ql
Moni,
Please send a patch to OFED_tips.txt explaining how to use this new
feature (or do you think it should go to the user guide?
Tziporet
Hi Tziporet,
After giving it a second thought, OFED_tips.txt look the right place. Please
see the patch below.
Thanks
diff --git a/OFED_tips.txt
Applied.
Regards,
Vladimir
Thanks. Please apply the following addition to the patch.
This will allow the users to add lines to ofed.conf which will not be parsed by
install.pl
and can be used by the vendor scripts. Such line should start with the string:
vendor_config
diff --git
This is a modification to the section in the patch for the file install.pl
The pupose of the change is to allow ofed.conf to contain lines of the form
vendor_script* (e.g. vendor_config_fix_something=yes).
When install.pl parses the configuration file it silently ignores lines that
start with
Moni Shoua wrote:
Vendors that distribute OFED sometimes need to add vendor specific
actions to the install and/or the uninstall process. This patch adds this
capability of adding scripts that would be executed before and after OFED
install and uninstall. Names of the scripts are given
71 matches
Mail list logo