Hi Roland,
Thanks. I guess then the only neat solution is NAPI. I shall keep you posted if we can get NAPI implemented in the meanwhile.
thanks,
harishOn 8/29/06, Roland Dreier <[EMAIL PROTECTED]> wrote:
harish> Hi Roland, I was hoping that the function will take careharish> of that by int
We will have a work-around in OFED 1.1.
Quoting r. zhu shi song <[EMAIL PROTECTED]>:
Subject: Re: why sdp connections cost so much memory
If it's spec bug is it difficult to solve? And how
long will it take you to complete the bugfix? I
really hope SDP can work as stably as TCP.
zhu
--- "Mic
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH ] RFC IB/cm do not track remote QPN in timewait state
>
> Michael S. Tsirkin wrote:
> >>If we completely ignore timewait, what conditions are required to have a
> >>problem
> >>occur?
> >
> > Outstanding packets with PSNs and QP n
If it's spec bug is it difficult to solve? And how
long will it take you to complete the bugfix? I
really hope SDP can work as stably as TCP.
zhu
--- "Michael S. Tsirkin" <[EMAIL PROTECTED]> wrote:
> I did - this is the spec bug we are discussing with
> Sean.
>
> Quoting r. zhu shi song <[EM
Can you please help me to contact your Information
Technology Director? My company has just opened a new
computer datacenter in Franklin, TN for the purpose of
providing disaster recovery, business continuity, and
outsourcing services to small and medium size businesses
in the Nashville area. Th
On 18:54 Tue 29 Aug , Hal Rosenstock wrote:
> On Tue, 2006-08-29 at 18:25, Doug Ledford wrote:
> > On Tue, 2006-08-29 at 15:13 -0400, Hal Rosenstock wrote:
> > > On Tue, 2006-08-29 at 11:21, Greg Lindahl wrote:
> > > > On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
> > > >
> > >
harish> Hi Roland, I was hoping that the function will take care
harish> of that by introducing some kind of timer since last
harish> notification. The logic I had in mind was that the
harish> notification is triggered in either of the two events is
harish> true:
-> We have
Hi Roland,I was hoping that the function will take care of that by introducing some kind of timer since last notification. The logic I had in mind was that the notification is triggered in either of the two events is true:
->We have 10+ completions or->Time since last notification > (some specified
Hi Hal,
On 20:09 Tue 29 Aug , Hal Rosenstock wrote:
> Hi Sasha,
>
> On Fri, 2006-08-25 at 09:17, Sasha Khapyorsky wrote:
> > This provides RPC like API which may work with several ports.
>
> I think you mean "can work" rather "may work" :-)
Yes.
Some limitation we will have from libumad -
harish> Hi Roland, As regards the CQ notification frequency, I
harish> noticed that the function req_ncomp_notif used by
harish> ib_req_ncom_notif is not implemented yet. I was hoping
harish> that if this was implemented, I would just use
harish> ib_req_ncom_notif with a count o
Hi Sasha,
On Fri, 2006-08-25 at 09:17, Sasha Khapyorsky wrote:
> This provides RPC like API which may work with several ports.
I think you mean "can work" rather "may work" :-)
> Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
> ---
>
> libibmad/include/infiniband/mad.h |9 +++
> libib
Hi Roland,As regards the CQ notification frequency, I noticed that the function req_ncomp_notif used by ib_req_ncom_notif is not implemented yet. I was hoping that if this was implemented, I would just use ib_req_ncom_notif with a count of 10 in place of ib_req_com_notif. Please share your comment
On Friday, 25 August 2006 at 13:19, Bryan O'Sullivan wrote:
> On Fri, 2006-08-25 at 12:45 -0700, Roland Dreier wrote:
> > How did you generate these patches?
>
> Using Mercurial.
>
> > because the line
> >
> > diff --git a/drivers/infiniband/hw/ipath/Makefile
> > b/drivers/infiniband/hw/ipath
On Tue, 2006-08-29 at 18:25, Doug Ledford wrote:
> On Tue, 2006-08-29 at 15:13 -0400, Hal Rosenstock wrote:
> > On Tue, 2006-08-29 at 11:21, Greg Lindahl wrote:
> > > On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
> > >
> > > > I would definitely put the option in, and in fact would
harish> Hi Roland, Thanks a lot for the prompt response. Could you
harish> please let me know as to what is the expected time frame
harish> for having NAPI for IPoIB implemented. Also as regards to
harish> the first question, would it make sense to play around
harish> with the
Hi Roland,Thanks a lot for the prompt response. Could you please let me know as to what is the expected time frame for having NAPI for IPoIB implemented. Also as regards to the first question, would it make sense to play around with the cq notification frequency. Will it help to reduce the CPU uti
On Tue, 2006-08-29 at 17:59, Sasha Khapyorsky wrote:
> On 17:03 Tue 29 Aug , Hal Rosenstock wrote:
> > On Tue, 2006-08-29 at 15:01, Sasha Khapyorsky wrote:
> >
> > > > I'd prefer an incremental one off the last patch related to this if that
> > > > isn't too much work as I'm close to committin
harish> Hi, The interruptThresholdRate module parameter allows you
harish> to control the maximum number of interrupts/sec for an
harish> e1000 Intel NIC for example. Is there an equivalent
harish> parameter for Infiniband NICs. I am using a Mellanox
harish> Infiniband NIC. Plea
On Tue, 2006-08-29 at 15:13 -0400, Hal Rosenstock wrote:
> On Tue, 2006-08-29 at 11:21, Greg Lindahl wrote:
> > On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
> >
> > > I would definitely put the option in, and in fact would default it to
> > > *NOT* truncate.
> >
> > I agree. I ha
Hello,
Trying to build OFED-1.0.tgz on a suse 10.1 x86_64 box with kernel
2.6.16.21-0.13. I am using the build.sh script but it stops while
building the kernel modules.
build.sh fails with this message:
error: too few arguments to function 'sk_eat_skb'
From what I can tell there is patch f
On 17:03 Tue 29 Aug , Hal Rosenstock wrote:
> On Tue, 2006-08-29 at 15:01, Sasha Khapyorsky wrote:
>
> > > I'd prefer an incremental one off the last patch related to this if that
> > > isn't too much work as I'm close to committing the previous one now (and
> > > it'd be more work to start ov
Hi,The interruptThresholdRate module parameter allows you to control the maximum number of interrupts/sec for an e1000 Intel NIC for example. Is there an equivalent parameter for Infiniband NICs. I am using a Mellanox Infiniband NIC. Please let me know if you need any more information.
Noticed from
On Tue, 2006-08-29 at 15:01, Sasha Khapyorsky wrote:
> > I'd prefer an incremental one off the last patch related to this if that
> > isn't too much work as I'm close to committing the previous one now (and
> > it'd be more work to start over on this).
>
> Ok. There is:
>
>
> Optional log file
Michael S. Tsirkin wrote:
>>If we completely ignore timewait, what conditions are required to have a
>>problem
>>occur?
>
> Outstanding packets with PSNs and QP numbers coinside between the 2
> connections.
> Look for "Stale packet" in IB spec.
From what I can tell, a QP will receive an incom
On Tue, 2006-08-22 at 17:18, Sasha Khapyorsky wrote:
> Hi Hal,
>
> There is new option which specified max size of OpenSM log file. The
> default is '0' (not-limited). Please note osm_log_init() has new
> parameter now.
>
> We already saw the problems with FS overflowing in real life - we may
> w
> --- l/src/userspace/libmthca/src/mthca.h (revision 7569)
> +++ l/src/userspace/libmthca/src/mthca.h (working copy)
> @@ -36,6 +36,7 @@
> #ifndef MTHCA_H
> #define MTHCA_H
>
> +#include
> #include
> #include
svn blame shows me that mthca.h has included stddef.h since r88
http://openib.org/bugzilla/show_bug.cgi?id=214
[EMAIL PROTECTED] changed:
What|Removed |Added
Status|NEW |RESOLVED
Resolution|
On 19:39 Tue 29 Aug , Michael S. Tsirkin wrote:
> Quoting r. Sasha Khapyorsky <[EMAIL PROTECTED]>:
> > > Assuming I disconnect and re-connect the cable, does not
> > > SM want to know and try bringing the logical link up?
> >
> > AFAIK OpenSM does not detect the local port disconnection explic
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH ] RFC IB/cm do not track remote QPN in timewait state
>
> Michael S. Tsirkin wrote:
> > Hmm. But you need timewait already after you get to RTR, right?
>
> The active side looks fine. The passive side can enter timewait without
>
Hi Sasha,
On Tue, 2006-08-29 at 14:15, Sasha Khapyorsky wrote:
> On 18:28 Sun 27 Aug , Doug Ledford wrote:
> > On Sun, 2006-08-20 at 20:18 +0300, Sasha Khapyorsky wrote:
> > > On 13:01 Sun 20 Aug , Hal Rosenstock wrote:
> > > > Hi Sasha,
> > > >
> > > > On Sun, 2006-08-20 at 12:05, Sasha
On Mon, 2006-08-28 at 04:11, Yevgeny Kliteynik wrote:
> Hi Hal.
>
> I noticed that there are some unused defaults:
> OSM_DEFAULT_MGRP_MTU and OSM_DEFAULT_MGRP_RATE.
> The corresponding values in the code are hadcoded.
>
> Fixed the code to use these defaults, and updated the
> OSM_DEFAULT_MGRP
Michael S. Tsirkin wrote:
> Hmm. But you need timewait already after you get to RTR, right?
The active side looks fine. The passive side can enter timewait without moving
through RTS if it gets an RTU timeout. I'm not sure how much going into
timewait really helps in this case though.
If we c
On Tue, 2006-08-29 at 13:39, Hal Rosenstock wrote:
> On Tue, 2006-08-29 at 10:50, Yevgeny Kliteynik wrote:
> > Hi Hal.
> >
> > This patch is just fixing some error in the OSM usage message.
> > Instead of old '-vf' option, there should be '-D'.
> >
> > Thanks.
> >
> > Yevgeny
> >
> > Signed-off
On Tue, 2006-08-29 at 10:50, Yevgeny Kliteynik wrote:
> Hi Hal.
>
> This patch is just fixing some error in the OSM usage message.
> Instead of old '-vf' option, there should be '-D'.
>
> Thanks.
>
> Yevgeny
>
> Signed-off-by: Yevgeny Kliteynik <[EMAIL PROTECTED]>
Thanks. Applied (to both tru
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH ] RFC IB/cm do not track remote QPN in timewait state
>
> Michael S. Tsirkin wrote:
> >>Verbs gets local_ack_timeout through qp_attr.timeout when modifying the QP
> >>to
> >>RTS.
> >
> >
> > Isn't that RTR?
>
> It's the transiti
Michael S. Tsirkin wrote:
>>Verbs gets local_ack_timeout through qp_attr.timeout when modifying the QP to
>>RTS.
>
>
> Isn't that RTR?
It's the transition from RTR to RTS.
> So it seems we won't need any API changes. This begins to look good.
> I waner what Roland and other low level driver ma
On Tue, 2006-08-29 at 11:21, Greg Lindahl wrote:
> On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
>
> > I would definitely put the option in, and in fact would default it to
> > *NOT* truncate.
>
> I agree. I have never seen any other daemon with a logfile do this,
> why are we out
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH ] RFC IB/cm do not track remote QPN in timewait state
>
> Sean Hefty wrote:
> > How would the driver determine how long the QP should remain in timewait
>
> The spec isn't totally clear to me on this, but here's what I can gather:
>
On 08:21 Tue 29 Aug , Greg Lindahl wrote:
> On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
>
> > I would definitely put the option in, and in fact would default it to
> > *NOT* truncate.
>
> I agree. I have never seen any other daemon with a logfile do this,
OpenSM is not a re
On 14:18 Tue 29 Aug , Hal Rosenstock wrote:
> Hi Sasha,
>
> On Tue, 2006-08-29 at 14:15, Sasha Khapyorsky wrote:
> > On 18:28 Sun 27 Aug , Doug Ledford wrote:
> > > On Sun, 2006-08-20 at 20:18 +0300, Sasha Khapyorsky wrote:
> > > > On 13:01 Sun 20 Aug , Hal Rosenstock wrote:
> > > > >
Sean Hefty wrote:
> How would the driver determine how long the QP should remain in timewait
The spec isn't totally clear to me on this, but here's what I can gather:
timewait = packet lifetime x 2 + remote ack delay
local_ack_timeout (in CM REQ) = packet lifetime x 2 + local ack delay
Verbs get
On Sun, Aug 27, 2006 at 06:28:06PM -0400, Doug Ledford wrote:
> I would definitely put the option in, and in fact would default it to
> *NOT* truncate.
I agree. I have never seen any other daemon with a logfile do this,
why are we out to surprise the admin? The admin might want the start
of the l
On 18:28 Sun 27 Aug , Doug Ledford wrote:
> On Sun, 2006-08-20 at 20:18 +0300, Sasha Khapyorsky wrote:
> > On 13:01 Sun 20 Aug , Hal Rosenstock wrote:
> > > Hi Sasha,
> > >
> > > On Sun, 2006-08-20 at 12:05, Sasha Khapyorsky wrote:
> > > > In case when OpenSM log file overflows filesystem
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [openib-general] [PATCH ] RFC IB/cm do not track remote QPN in
> timewait state
>
> Michael S. Tsirkin wrote:
> >>I've thought about this too, and I think this may end up making the most
> >>sense.
> >>How would the driver determine how l
>Looked into the openIB kernel sources and found that the minor number
>seems to be wrong in the README file. With a minor number "224" and the
>creation like:
>
> "mknod /dev/infiniband/ucm0 c 231 224"
The README file was never updated when the userspace CM added per device
handling. I've
Jack Morgenstein wrote:
> Fix compilation on SLES10:
> cm.c uses offsetof, so it must include stddef.h
Thanks - committed in 9150.
- Sean
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To un
Hal Rosenstock wrote:
> OK. So shouldn't IBV_SA_METHOD_SEND be removed from sa_net.h ?
I was just defining the well known methods. I can remove this.
> By raw access, do you mean SEND_MAD operation ?
>
> How do those applications gain this privilege ?
The kernel module exports two files to pe
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] libibcm: modify API to support multi-threaded event
> processing
>
> Michael S. Tsirkin wrote:
> > I think offsetof is defined in stddef.h, so you must include that.
>
> Dotan,
>
> Can you see if adding this include works for you
On Tue, 2006-08-29 at 11:58, Sean Hefty wrote:
> >Why SEND ? In general, couldn't it be used like SET/DELETE (in addition
> >to being used like the GET method variants) ? Also, the SA doesn't use
> >the SEND method.
>
> The latest version of the patch only allows GET or GET_TABLE for PathRecords
>
Quoting r. Sasha Khapyorsky <[EMAIL PROTECTED]>:
> > Assuming I disconnect and re-connect the cable, does not
> > SM want to know and try bringing the logical link up?
>
> AFAIK OpenSM does not detect the local port disconnection explicitly
> (however will detect this at next periodic sweep if suc
Michael S. Tsirkin wrote:
>>I've thought about this too, and I think this may end up making the most
>>sense.
>>How would the driver determine how long the QP should remain in timewait,
>
>
> Need to look into this - likely we can just add a call for that.
> Roland?
The Intel gen1 code passed t
Michael S. Tsirkin wrote:
> I think offsetof is defined in stddef.h, so you must include that.
Dotan,
Can you see if adding this include works for you? I just re-tested the build
on
my system, and it worked fine without it (gcc 3.3.3). Jack posted a patch for
this earlier if you need one.
-
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH ] RFC IB/cm do not track remote QPN in timewait state
>
> >Here's an idea:
> >how about we move the whole timewait thing to low level driver,
> >starting timer automatically upon QP destroy?
>
> I've thought about this too, and I th
On 21:41 Sat 26 Aug , Michael S. Tsirkin wrote:
> Quoting r. Sasha Khapyorsky <[EMAIL PROTECTED]>:
> > Subject: Re: [openib-general] [PATCH] osm: handle local events
> >
> > On 16:28 Thu 24 Aug , Michael S. Tsirkin wrote:
> > > Quoting r. Yevgeny Kliteynik <[EMAIL PROTECTED]>:
> > > > Inde
>I handled it all myself this time, but in the future it is easier for
>me if each patch is inline in a separate email. A couple of other
>things that would also make my life easier:
That's not a problem. I think in the past I've just referred you to the svn
revision numbers. I was just trying
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] libibcm: modify API to support multi-threaded event
> processing
>
> >There are compilation errors with this patch when using gcc 4.1.0:
>
>
> Hmmm... I will look into this.
I think offsetof is defined in stddef.h, so you must i
>Here's an idea:
>how about we move the whole timewait thing to low level driver,
>starting timer automatically upon QP destroy?
I've thought about this too, and I think this may end up making the most sense.
How would the driver determine how long the QP should remain in timewait, and
how would y
>Why SEND ? In general, couldn't it be used like SET/DELETE (in addition
>to being used like the GET method variants) ? Also, the SA doesn't use
>the SEND method.
The latest version of the patch only allows GET or GET_TABLE for PathRecords
ServiceRecords, and MCMemberRecords, and GET_MULTI for Mul
>There are compilation errors with this patch when using gcc 4.1.0:
Hmmm... I will look into this.
- Sean
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://
On Sun, 2006-08-27 at 06:16, Yevgeny Kliteynik wrote:
> Hi Hal
>
> This patch just makes the error message more informative for user,
> since another instance of running SM is most probably the reason
> why osm_opensm_bind failed.
>
> Yevgeny
>
> Signed-off-by: Yevgeny Kliteynik <[EMAIL PROTECT
Hi Eitan,
> Hi Hal
> These are two trivial fixes for memory leaks in the
> ibumad vendor.
> Thanks
> Eitan
> Signed-off-by: Eitan Zahavi
Thanks. Applied (to both trunk and 1.1).
-- Hal
___
openib-general mailing list
openib-general@openib.org
At 08:56 AM 8/25/2006, Greg Lindahl wrote:
On Fri, Aug 25, 2006 at
10:13:01AM -0500, Tom Tucker wrote:
> He does say this, but his analysis does not support this conclusion.
His
> analysis revolves around MPI send/recv, not the MPI 2.0 get/put
> services.
Nobody uses MPI put/get anyway, so leavin
Fix compilation on SLES10 RC2:
mthca.h uses offsetof so it must include stddef.h
Signed-off-by: Jack Morgenstein <[EMAIL PROTECTED]>
Signed-off-by: Michael S. Tsirkin <[EMAIL PROTECTED]>
Index: l/src/userspace/libmthca/src/mthca.h
==
Hi Hal.
This patch is just fixing some error in the OSM usage message.
Instead of old '-vf' option, there should be '-D'.
Thanks.
Yevgeny
Signed-off-by: Yevgeny Kliteynik <[EMAIL PROTECTED]>
Index: osm/opensm/main.c
===
--- osm/o
Hi All,
In testing today we found that on SLES9 SP3 memory locking as a regular
user fails.
Although I changed /etc/security/limits.conf and added the following two
lines:
* soft memlock
* hard memlock
Note that same change does work in SLES10.
Another change I tried (that worked in gen1) was
Hi All,
RC3 will not be available today.
There are 2 items that are gating this:
1. SDP last issue with CM resolution
2. ipath patches update
So if everything goes well it will be available tomorrow or on Thursday.
Tziporet
___
openib-general mailing l
James Lentini wrote:
If OpenFabrics.org had git repositories, why would we remove the code
from the OpenFabrics.org?
Matt from Sandia is checking this since I also asked him to provide git
repository for OFED release.
Tziporet
___
ope
Dunno if this is the correct fix for all platforms/distros, but it
worked for me on FC5/X86_64... CLK_TCK wasn't getting defined for this
distro...
Signed-off-by: Steve Wise <[EMAIL PROTECTED]>
---
Index: test/dapltest/mdep/linux/dapl_mdep_user.c
On Mon, 28 Aug 2006, Roland Dreier wrote:
> Fabian> Mellanox is currently tracking the MTHCA code base for
> Fabian> Windows, and moving it out of SVN could make that harder,
> Fabian> even impossible if it were to lose the BSD license.
>
> There's no thought of changing the license
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> I believe that this tracking is done, and is reported to the user by the
> timewait exit event. QP transitions are the responsibility of the user.
>
> This is related to a problem that Arlin and I have been discussing. There's
> nothing that the CM
Sean Hefty wrote:
>> Modify the libibcm API to provide better support for multi-threaded
>> event processing. CM devices are no longer tied to verb devices
>> and hidden from the user. This should allow an application to direct
>> events to specific threads for processing.
>>
>> This patch also rem
On Wed, 2006-08-23 at 17:26, Sean Hefty wrote:
> Roland Dreier wrote:
> > What's the plan for how this would be used? We can't let unprivileged
> > userspace processes talk to the SA, because they could cause problems
> > like deleting someone else's multicast group membership. And I don't
> > th
resending the response (after a problem in my mail client):
> john t wrote:
>>
>> What is the meaning of above fields or where can I find the
>> definition of above fields ? Can I change the value of fields like
>> "timeout" or should it be always set to a fixed value.
> You can find the descri
Made some progress today:
The file OFED-1.0.1/SOURCES/openib-1.0.1/src/userspace/libibcm/README
talks about
"mknod /dev/infiniband/ucm0 c 231 255"
This is not working for me. I'm getting the device /dev/infiniband/ucm0
created but the libibcm still could not open the device
Looked into t
I did - this is the spec bug we are discussing with Sean.
Quoting r. zhu shi song <[EMAIL PROTECTED]>:
> Subject: Re: why sdp connections cost so much memory
>
> Have you resolved the problem?
> zhu
>
> --- "Michael S. Tsirkin" <[EMAIL PROTECTED]> wrote:
>
> > Yes, I have reproduced the connect
Have you resolved the problem?
zhu
--- "Michael S. Tsirkin" <[EMAIL PROTECTED]> wrote:
> Yes, I have reproduced the connection refusal
> problem and I am looking into it.
> Thanks!
>
> MST
>
> Quoting r. zhu shi song <[EMAIL PROTECTED]>:
> Subject: Re: why sdp connections cost so much memory
>
Fix compilation on SLES10:
cm.c uses offsetof, so it must include stddef.h
Signed-off-by: Jack Morgenstein <[EMAIL PROTECTED]>
Index: l/src/userspace/libibcm/src/cm.c
===
--- l/src/userspace/libibcm/src/cm.c(revision 9128)
+++ l/
77 matches
Mail list logo