Hi Hal,
Good catch!
So I guess the osm_lid_mgr is broken when the given lid it out of
range.
My comment is nit picking but I think that a function that validate and modify
LID it got from the HW should have a more meaningful name to clarify the
"modify".
I would propose:
osm_physp_fix_out_of_r
Title: RE: osm_port_info_rcv.c Error Numbering
No problem.
-Original Message-
From: Hal Rosenstock [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, September 27, 2005 10:57 PM
To: Yael Kalka
Cc: Eitan Zahavi; openib-general@openib.org
Subject: osm_port_info_rcv.c Error Numbering
Hi Yael
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:
rsync://rsync.kernel.org/pub/scm/linux/kernel/git/roland/infiniband.git
for-linus
This will pull the following changes (full patch
Quoting r. Grant Grundler <[EMAIL PROTECTED]>:
> Subject: netperf over SDP bug
>
> Hi Michael,
> I'm trying to collect a full set of netperf TCP_STREAM over SDP for
> SVN r3547 on 2.6.13 kernel. But some netperf runs get no throughput.
> Usually when sending 1k to 4k messages. The same netperf p
OpenSM: Protect against bad LIDs returned from faulty hardware/SMA
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: include/opensm/osm_port.h
===
--- include/opensm/osm_port.h (revision 3590)
+++ include/opensm/osm_port.h
VBabu> I am not sure I can post the whole code. But here is the
VBabu> part of it.
It's a little hard to debug without being able to run your code and
reproduce the error. The only things I see obviously wrong are that
you never seem to set elem -> u.txwr.opcode to IB_WR_SEND, so you may
Title: RE: [openib-general] Local QP operation err while sending packet over UD transport
VBabu> Thanks for your quick response.
I'm not sure if it matters, but firmware version 5.0.1 is old -- you
might want to upgrade to 5.1.0.
VBabu > I will check with the vendor and findout how to up
VBabu> I am using verbs layer in OpenIB of linux kernel 2.6.13.1
VBabu> and Mellanox HCA (CA type: MT25208, Number of ports: 2,
VBabu> Firmware version: 5.0.1, Hardware version: a0)
I'm not sure if it matters, but firmware version 5.0.1 is old -- you
might want to upgrade to 5.1.0. Al
Title: Local QP operation err while sending packet over UD transport
Hi,
I am using verbs layer in OpenIB of linux kernel 2.6.13.1 and Mellanox HCA (CA type: MT25208, Number of ports: 2, Firmware version: 5.0.1, Hardware version: a0). I have created a QP with Unreliable Datagram transpo
Hi Michael,
I'm trying to collect a full set of netperf TCP_STREAM over SDP for
SVN r3547 on 2.6.13 kernel. But some netperf runs get no throughput.
Usually when sending 1k to 4k messages. The same netperf parameters
sing IPoIB seem to be working fine - just alot slower of course.
Summary of all
Hal,
Thanks.. works like a charm...
-Viswa
On 27 Sep 2005 16:13:01 -0400, Hal Rosenstock <[EMAIL PROTECTED]> wrote:
On Tue, 2005-09-27 at 16:00, Viswanath Krishnamurthy wrote:> Hal,>> I added a hack now to get around the problem. There needs to be a> proper fix later..Can you try this instead ? T
Sean Hefty wrote:
I noticed while compiling against an up-to-date kernel tree that SDP
and IBAT both use the function ip_dev_find(). The EXPORT_SYMBOL for
this function was removed during the 2.6.14 devel cycle.
I'm calling using this routine in the rdam_cma code as well. I call it
to get t
Federico,
> Federico> I might have done something wrong, but tried to build
> Federico> using a plain source from the openib gen2 svn tree and
> Federico> Pete's patches (those that were not rejected).
>
> For whatever it's worth, basic MVAPICH tests like osu_bw work fine for
> me w
Federico> I might have done something wrong, but tried to build
Federico> using a plain source from the openib gen2 svn tree and
Federico> Pete's patches (those that were not rejected).
For whatever it's worth, basic MVAPICH tests like osu_bw work fine for
me with two and even four pr
I had such high hopes for using openib gen2 when I got ibv_uc_pingpong
to pass packets on our infiniband cluster. However, I cannot get mvapich
to work, even with Pete Wyckoff's patches. A simple program run on two
hosts always segfaults.
I might have done something wrong, but tried to build using
Robert, I just committed the first step towards merging your branch:
pure simplication/fixing, based on your uverbs changes.
- R.
--- infiniband/core/uverbs_cmd.c(revision 3579)
+++ infiniband/core/uverbs_cmd.c(working copy)
@@ -1,6 +1,7 @@
/*
* Copyright (c) 2005 Topspin Comm
Just an FYI, I already checked in... just make two vars only used in a
single file static to appease sparse.
--- infiniband/core/ucm.c (revision 3579)
+++ infiniband/core/ucm.c (working copy)
@@ -113,8 +113,8 @@ static struct ib_client ucm_client = {
.remove = ib_ucm_remove_on
Thanks, committed to svn and queued for 2.6.14.
- R.
___
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
James,
Please review the following uDAPL patch which fixes some ia64
build problems (atomics) with the latest Redhat EL4.0 update and
adds support for SuSe. Feel free to come up with a better solution.
Thanks,
-arlin
Signed-off by: Arlin Davis <[EMAIL PROTECTED]>
Index: dapl/udapl/linux/da
This patch will transition a QP to the INIT state and bind the QP to
the cma_id. It is called after a route has been resolved and
should assist with transport independent code.
Signed-off-by: Sean Hefty <[EMAIL PROTECTED]>
Index: ulp/cma/cma.c
===
On Tue, 2005-09-27 at 16:00, Viswanath Krishnamurthy wrote:
> Hal,
>
> I added a hack now to get around the problem. There needs to be a
> proper fix later..
Can you try this instead ? Thanks.
-- Hal
Index: include/opensm/osm_port.h
==
Michael> Fine with me. Roland?
Yes, seems like a good solution.
- R.
___
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
ScanMail for Microsoft Exchange has blocked an attachment.
Sender = [EMAIL PROTECTED]
Recipient(s) = openib-general@openib.org
Subject = [openib-general] MEMBERS SUPPORT
Scanning time = 9/27/2005 1:05:57 PM
Action on file blocking:
The attachment email-details.zip matches the file blocking settin
Hi Yael,
It looks to me like the two 4A0x error messages below in
osm_port_info_rcv.c should be changed to something unused in 0x0Fxx
range.
osm_log( p_log, OSM_LOG_ERROR,
"osm_physp_has_pkey: ERR 4A02: "
osm_log( p_log, OSM_LOG_ERROR,
"osm_physp_has_pkey
Hal,
I added a hack now to get around the problem. There needs to be a proper fix later..
[EMAIL PROTECTED] opensm]# svn diff osm_port.h
Index: osm_port.h
===
--- osm_port.h (revision 3549)
+++ osm_port.h (working copy)
@@ -1049,6
Well, I moved the card to slot 5 and things seem to be working...
I have another Power5 and Mellanox card available, so I decided to
retest with them to see what the solution was. I dropped the card
into slot 5 on the second Power 5, and it came right up, even without
the firmware upgrade (though
On Tue, 2005-09-27 at 14:13, Viswanath Krishnamurthy wrote:
> I tracked down the issue to a bug in osm_lid_mgr.c
>
> function: __osm_lid_mgr_init_sweep(...)
>
> The bad hardware was retutning an assigned LID of 0x. In this
> function there is a loop
> as follows where opensm is getting stuc
Hi Viswa,
On Tue, 2005-09-27 at 14:13, Viswanath Krishnamurthy wrote:
> I tracked down the issue to a bug in osm_lid_mgr.c
>
> function: __osm_lid_mgr_init_sweep(...)
>
> The bad hardware was retutning an assigned LID of 0x. In this
> function there is a loop
> as follows where opensm is g
I tracked down the issue to a bug in osm_lid_mgr.c
function: __osm_lid_mgr_init_sweep(...)
The bad hardware was retutning an assigned LID of 0x. In this function there is a loop
as follows where opensm is getting stuck.. (with line number)
392 p_port_guid_tbl = &p_mgr->p_subn->port_g
>I think that we should have
a sync meeting to come up with the best way of sharing/developing the tests/regression
tools.
That's a good idea.
Thanks
Shirley Ma
IBM Linux Technology Center
15300 SW Koll Parkway
Beaverton, OR 97006-6063
Phone(Fax): (503) 578-7638
_
On Tue, 27 Sep 2005, ALEXIS GARCIA RUIZ wrote:
> Hello, all:
> I just got the task to get into the IB technology, I made some
> reading and installed some packages on a Host.
> But still I am lost.
> Could you lead me to some reading?
>
> Regards
>
> Alex
The Wiki is a good place for i
Hello, all:
I just got the task to get into the IB technology, I made some reading and installed some packages on a Host.
But still I am lost.
Could you lead me to some reading?
Regards
Alex
__Do You Yahoo!?Tired of spam? Yahoo! Mail has the
On Tue, 27 Sep 2005, Todd Bowman wrote:
> Never mind, I found the post from Roland discussing this issue:
>
> Roland wrote:
> I didn't try to fix uDAPL, because some thought probably needs to go
> into how to use completion channels most efficiently.
I appologize. I forgot that the current tr
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [openib-general] Re: [PATCH] add cq error events
>
> Michael S. Tsirkin wrote:
> > Okay ... one problem that I've run into adding this, is
> > that IB_EVENT_CQ_ERR is the first item in the ib_event_type enum.
> > And since uverbs seems to j
Just for the benefit of the community, I am forwarding a mail in which Hal
already answered my question (I forgot to hit reply all...).
-Original Message-
From: Hal Rosenstock [mailto:[EMAIL PROTECTED]
Sent: Tuesday, September 27, 2005 12:27 PM
To: Suresh Shelvapille
Subject: RE: [openib
On 9/27/05, Todd Bowman <[EMAIL PROTECTED]> wrote:
On 9/27/05, James Lentini <
[EMAIL PROTECTED]> wrote:
On Tue, 27 Sep 2005, Todd Bowman wrote:> On 9/27/05, James Lentini <[EMAIL PROTECTED]
> wrote:>> What is the output of cat /sys/class/net/*/ifindex?
> >> > cat /sys/class/net/*/ifindex> 1 #eth0>
Log sent off-list...
-Viswa
On 9/27/05, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
Hi Viswa,Please send a full /var/log/osm.log file of opensm -V .You can send us a copy off the list if it is too big:yael and eitan in @mellanox.co.ilEZ
Hal Rosenstock wrote:> On Mon, 2005-09-26 at 19:57, Viswanath Kri
Guy German wrote:
Basically I think that we can definitely agree that if the cma can
implement ib_at intended functionality it should replace it - no need to
have 2 modules doing the same thing.
I think that there will still be a need for a separate address translation
module(s). I have a ca
On 9/27/05, James Lentini <[EMAIL PROTECTED]> wrote:
On Tue, 27 Sep 2005, Todd Bowman wrote:> On 9/27/05, James Lentini <[EMAIL PROTECTED]> wrote:>> What is the output of cat /sys/class/net/*/ifindex?
> >> > cat /sys/class/net/*/ifindex> 1 #eth0> 10 #ib0> 11 #ib1> 2 #lo> 3 #tunl0This looks like the
Hi Shirley
We are also running a nightly regression system on
different platforms/Oss,
I think that we should have a sync meeting to come
up with the best way of sharing/developing the tests/regression
tools
From: Shirley Ma [mailto:[EMAIL PROTECTED]
Sent: Tuesday, September 27, 2005 7:
Dotan> 1) VL (Verification Library) is a library that contains an
Dotan> abstraction to the operating system calls (for example:
Dotan> sleep, gettimeofday ..) to use in tests (something like the
Dotan> MOSAL library in gen1). In order to use the VL, one should
Dotan> execute "
On Tue, 27 Sep 2005, Hal Rosenstock wrote:
> > Since we don't check for a kmalloc failure in DT_Tdep_PT_Printf, this
> > oops occurs:
> >
> > > Sep 26 10:29:30 hal kernel: Unable to handle kernel NULL pointer
> > > dereference at virtual address 0004
> >
> > I've checked in the patch bel
Brian> It's even more annoying to be deluged with SPAM ;). We
Brian> (the LAM developers) used to try to keep our mailing lists
Brian> as open as possible. In the end, SPAM pushed the signal to
Brian> noise ratio way too high and something had to be done.
Brian> Requiring subs
On Tue, 2005-09-27 at 12:40, James Lentini wrote:
> It looks like you have DEBUG_PAGEALLOC turned on. Is that correct?
Yes.
> My
> test systems don't have this turned on. I'll turn it on and see what
> happens.
>
> Since we don't check for a kmalloc failure in DT_Tdep_PT_Printf, this
> oops
Michael S. Tsirkin wrote:
Okay ... one problem that I've run into adding this, is
that IB_EVENT_CQ_ERR is the first item in the ib_event_type enum.
And since uverbs seems to just copy the event over to userspace,
changing all the enum values would break the ABI.
Given that IB_EVENT_CQ_ERR wasnt
On Mon, 26 Sep 2005, Hal Rosenstock wrote:
> Hi James,
>
> I keep getting the following when running kdapltest. This is similar to
> what I saw before and reported a couple of times but now seems more
> consistent in occurring.
>
> -- Hal
Until the point at which the page allocation fails, I
> It sounds ok, but svn diff will produce essentially the same results
> (svn diff --diff-cmd diff -x --brief). To see the files you've added
> you need to do an svn add (don't worry they won't be committed until
> you do an svn commit).
Sure, but 'svn status' is a lot easier to type :-)
--
R
ScanMail for Microsoft Exchange has detected virus-infected attachment(s).
Sender = [EMAIL PROTECTED]
Recipient(s) = openib-general@openib.org
Subject = [openib-general] MEMBERS SUPPORT
Scanning time = 9/27/2005 11:20:21 AM
Engine/Pattern = 7.510-1002/2.859.00
Action on virus found:
The attachmen
That's a great news. We have started
posting pure nightly build results for both mainline and openIB Gen2 stacks.
We would like to integrate these test results into our automation framework
and test Gen2 stack along with the nightly build.
Thanks
Shirley Ma
IBM Linux Technology Center
15300 SW Ko
Hi,
I am building mvapich from the gen2 source in svn from 9-22. I must use
the mpd job launcher in my cluster but although I see the mpid/mpd dir
and compile mvapich with
-DUSE_MPD_BASIC
I do not have any mpd binaries after the build completes. Here is my
configure command:
CFLAGS="-D_X86_6
Hi Sean,
Basically I think that we can definitely agree that if the cma can implement
ib_at intended functionality it should replace it - no need to have 2 modules
doing the same thing.
So the points that need to be considered are:
1. Caching
sean> generic SA caching module should be a part
Title: Mellanox verification team is starting to check in to the SVN tests
Hi everyone.
Mellanox is starting to develop tests to the openIB (over the gen2 stack and the winib stack).
We will start to check in the tests that we will write.
We already checked in 2 components to: https://open
On Tue, 27 Sep 2005, Todd Bowman wrote:
> > I am having a different problem in ips_by_gid:
>
> open_hca: Found dev mthca0 f42202c90200
> open_hca: GID subnet 80fe id f52202c90200
> ips_by_gid: ERR ips_by_gid -1 No such device
> open_hca: ERR ib_at_ips_by_gid for mthca0
> dap
On 27 Sep 2005 09:55:02 -0400, Hal Rosenstock <[EMAIL PROTECTED]> wrote:
On Tue, 2005-09-27 at 09:51, James Lentini wrote:> On Mon, 26 Sep 2005, Hal Rosenstock wrote:>> > On Mon, 2005-09-26 at 18:05, Todd Bowman wrote:> > > I am having a problem with uDAPL accessing
> > > /dev/infiniband/{uat,ucm0}
On Mon, 26 Sep 2005, Sean Hefty wrote:
> James Lentini wrote:
> > - move listen declaration closer to accept and reject
>
> Accepted - will be pushed in with next version.
>
> > - add private data and new cma_id fields to event structure
>
> Added private_data_len field to cma_id structure.
Okay I will try that too and let you know.
Thank you.
Jerome
- Original Message -
From: "Michael S. Tsirkin" <[EMAIL PROTECTED]>
To: "Roland Dreier" <[EMAIL PROTECTED]>
Cc: "Jerome Pioux" <[EMAIL PROTECTED]>; "Tom Duffy"
<[EMAIL PROTECTED]>;
Sent: Monday, September 26, 2005 8:05 PM
On Tue, 2005-09-27 at 09:51, James Lentini wrote:
> On Mon, 26 Sep 2005, Hal Rosenstock wrote:
>
> > On Mon, 2005-09-26 at 18:05, Todd Bowman wrote:
> > > I am having a problem with uDAPL accessing
> > > /dev/infiniband/{uat,ucm0}. I am running 3549, 2.6.12 kernel with
> > > backport. Here is a
On Sep 26, 2005, at 4:20 PM, Roland Dreier wrote:
[It's somewhat annoying to have to subscribe to [EMAIL PROTECTED]
just to be able to send patches, but oh well...]
It's even more annoying to be deluged with SPAM ;). We (the LAM
developers) used to try to keep our mailing lists as open as
On Mon, 26 Sep 2005, Hal Rosenstock wrote:
> On Mon, 2005-09-26 at 18:05, Todd Bowman wrote:
> > I am having a problem with uDAPL accessing
> > /dev/infiniband/{uat,ucm0}. I am running 3549, 2.6.12 kernel with
> > backport. Here is a snippet of the uDAPL debug messages running
> > dtest. The
On Tue, 27 Sep 2005, Robert Walsh wrote:
> Hi all,
>
> I've added an svn:ignore property to a bunch of directories in the
> subversion repository. This causes subversion to ignore certain files
> (like .o files and other generated files) in a particular directory.
> This is useful if you've mo
The patch below fixes the incorrect value returned for the
hw_ver field in ib/v_query_device.
Jack
Signed-off-by: Jack Morgenstein <[EMAIL PROTECTED]>
Index: linux-kernel/infiniband/hw/mthca/mthca_provider.c
===
--- linux-kernel/
On Tue, 2005-09-27 at 08:06, Yael Kalka wrote:
> There is a multiple line in the osm_helper.c in osm_dump_path_record.
> Attached is a patch resolving this.
Thanks. Applied.
-- Hal
___
openib-general mailing list
openib-general@openib.org
http://openib
On Tue, 2005-09-27 at 07:59, Michael S. Tsirkin wrote:
> Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> > Subject: Re: [PATCH] SDP: In do_link_path_lookup, make sure device is IPoIB
> > hardware type
> >
> > On Tue, 2005-09-27 at 07:35, Michael S. Tsirkin wrote:
> > > Quoting r. Hal Rosenstock
Hi Hal,
There is a multiple line in the osm_helper.c in osm_dump_path_record.
Attached is a patch resolving this.
Thanks,
Yael
Signed-off-by: Yael Kalka <[EMAIL PROTECTED]>
Index: osm/opensm/osm_helper.c
===
--- osm/opensm/osm_hel
Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] SDP: In do_link_path_lookup, make sure device is IPoIB
> hardware type
>
> On Tue, 2005-09-27 at 07:35, Michael S. Tsirkin wrote:
> > Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> > > Subject: [PATCH] SDP: In do_link_path_lo
On Tue, 2005-09-27 at 07:35, Michael S. Tsirkin wrote:
> Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> > Subject: [PATCH] SDP: In do_link_path_lookup, make sure device is IPoIB
> > hardware type
> >
> > In do_link_path_lookup, make sure device selected is IPoIB hardware
> > type before accessi
Quoting r. Hal Rosenstock <[EMAIL PROTECTED]>:
> Subject: [PATCH] SDP: In do_link_path_lookup, make sure device is IPoIB
> hardware type
>
> In do_link_path_lookup, make sure device selected is IPoIB hardware
> type before accessing its local private data
Hal, does this fix some actual problem
In do_link_path_lookup, make sure device selected is IPoIB hardware
type before accessing its local private data
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: sdp_link.c
===
--- sdp_link.c (revision 3552)
+++ sdp_link.c
On Tue, 2005-09-27 at 00:08, Michael S. Tsirkin wrote:
> I think it would, that would be cleaner than what I do now.
> However, I dont think SDP needs this pointer after the point where it
> sends an arp request. Why does AT need to keep netdev reference for
> longer?
For AT, the routine this cod
On Mon, 2005-09-26 at 10:41, Michael S. Tsirkin wrote:
> If ip_route_output_key resolves to a loopback device,
> sdp uses ip_dev_find to try and locate the actual hardware device
> that the source ip address is for.
Right, and if that device (returned by ip_dev_find) is the loopback
device, it ju
On Tue, 2005-09-27 at 04:11, Abhijit Gadgil wrote:
> Hi All,
>
> I am new to IPoIB. I have a query, as per the IPoIB Architecture
> document, whenever an IPoIB interface is brought up, it needs to do a
> Full Member Join to the "broadcast" Multicast group. Where exactly in
> the code, is this taki
On Tue, 2005-09-27 at 03:09, Robert Walsh wrote:
> Hi all,
>
> I've added an svn:ignore property to a bunch of directories in the
> subversion repository. This causes subversion to ignore certain files
> (like .o files and other generated files) in a particular directory.
> This is useful if you'
Title: RE: [openib-general] IPoIB question
> Further, I am putting SM in testability 'debug' mode (DEBUG=10 in
> /etc/opensm.conf), however I am still not seeing any dump of messages
> about FullMember join whenever I try restarting the IB interfaces. What
> should be log-level to put SM to du
Hi All,
I am new to IPoIB. I have a query, as per the IPoIB Architecture
document, whenever an IPoIB interface is brought up, it needs to do a
Full Member Join to the "broadcast" Multicast group. Where exactly in
the code, is this taking place? I have been able to trace a little bit -
eg. in ipoib
Hi Viswa,
Please send a full /var/log/osm.log file of opensm -V .
You can send us a copy off the list if it is too big:
yael and eitan in @mellanox.co.il
EZ
Hal Rosenstock wrote:
On Mon, 2005-09-26 at 19:57, Viswanath Krishnamurthy wrote:
I have an exerciser in the IB network. The exerciser
Hi all,
I've added an svn:ignore property to a bunch of directories in the
subversion repository. This causes subversion to ignore certain files
(like .o files and other generated files) in a particular directory.
This is useful if you've modified and built a bunch of stuff and only
want 'svn sta
76 matches
Mail list logo