On Tuesday 09 May 2006 19:25, Sean Hefty wrote:
This is fine for route lookup, but not rdma_get_option.
I missed adding rdma_ib.c to my tag file, so missed the rdma_get_option usage
of local_sa.c -- sorry about that.
My preference is to take the most recent rdma_cm, since there are several
Quoting r. Jack Morgenstein [EMAIL PROTECTED]:
Subject: Re: [PATCH 3/3] librdmacm: add ability to get/set transport specific
options
On Tuesday 09 May 2006 19:25, Sean Hefty wrote:
This is fine for route lookup, but not rdma_get_option.
I missed adding rdma_ib.c to my tag file, so
On Wednesday 10 May 2006 10:17, Michael S. Tsirkin wrote:
Maybe just return -ENODATA? Then you don't need to modify any code ...
Userspace rdma_get_option() will then also get -ENODATA. OK.
We can, therefore, do the following:
the dummy procedures in the dummy ib_local_sa.h file will
Michael S. Tsirkin wrote:
Can you try this simple patch and see if it fixes your problem? You will
need to call rdma_accept() or rdma_reject() after receiving a CONNECT_RESPONSE
event. The conn_param to rdma_accept() should be NULL.
Signed-off-by: Sean Hefty [EMAIL PROTECTED]
---
Index: cma.c
Quoting r. Or Gerlitz [EMAIL PROTECTED]:
Hmm, it seems that QP will get modified to RTS on RTU only.
That's bad - I want to move it to RTS on REP, since receive completion
might cross RTU and I want to be able to respond to each send immediately.
No, looking in the code shows that qp will be
Kill some dead code in mthca_eq.c
Signed-off-by: Michael S. Tsirkin [EMAIL PROTECTED]
Index: linux-2.6.16/drivers/infiniband/hw/mthca/mthca_eq.c
===
--- linux-2.6.16.orig/drivers/infiniband/hw/mthca/mthca_eq.c2006-04-06
On 5/9/06, Roland Dreier [EMAIL PROTECTED] wrote:
Why do you need this possibility? Having counters reset locally is
going to confuse any performance manager running on the fabric that
might be reading the counters remotely.
Is it really a must for this to be done by writing to sysfs?
Roland Dreier wrote:
It seems good. I should check it into svn today. I still need to
figure out how I want to put it on the for-2.6.18 git branch, since it
depends on some earlier fixes that Linus hasn't pulled yet.
- R.
Thanks, frankly I do not care about 2.6.18 git for now since OFED
I can't get the latest source from
svn co https://openfabrics.org/svn/gen2; in one whole
day, it's so slow.
Do you think the lastest source solve the problem? Or
can you test sdp for 2000 concurrent connections?
tks
zhu
--- Michael S. Tsirkin [EMAIL PROTECTED] wrote:
Quoting r. zhu shi
At 03:12 PM 5/9/2006, Roland Dreier wrote:
BTW, does Mellanox (or anyone else) have any numbers showing that
using FMRs makes any difference in performance on a semi-realistic benchmark?
Not me. Using the current FMRs to register/deregister windows for
each NFS/RDMA operation yields only a slight
On Tue, 2006-05-09 at 13:08, Hal Rosenstock wrote:
OpenSM: Add SA client API for MultiPathRecord queries
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
Applied to trunk only.
-- Hal
___
openib-general mailing list
openib-general@openib.org
At 11:13 PM 5/9/2006, Shirley Ma wrote:
Have you tried to send payload smaller than 2044? Any difference?
You mean MTU or ULP payload? The default NFS reads and writes are
32KB, and in the addressing mode used in these tests they were
broken into 8 page-sized RDMA ops. So, there were 9 ops from
On Tue, 2006-05-09 at 13:22, Hal Rosenstock wrote:
osmtest: Add rudimentary SA MultiPathRecord tests
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
Applied to trunk only.
-- Hal
___
openib-general mailing list
openib-general@openib.org
Quoting r. zhu shi song [EMAIL PROTECTED]:
Subject: Re: sdp can't support many connections (2000)
I can't get the latest source from
svn co https://openfabrics.org/svn/gen2; in one whole
day, it's so slow.
I use openib.org/svn/gen2 but I expect its just a redirection.
Hmm. We'll be
On Tue, 2006-05-09 at 14:15, Sasha Khapyorsky wrote:
Trivial low-level QoS configuration parameters description, definition
and processing.
Signed-off-by: Sasha Khapyorsky [EMAIL PROTECTED]
Thanks. Applied to trunk only.
-- Hal
___
openib-general
Sean, could you take a look please? I think it makes sense to make
same librdmacm work on as many kernels as possible, so that it's
seamless for people to switch kernels.
And since backport users install kernel modules or patches anyway,
I think it's reasonable to ask them to always install the
On Tue, 2006-05-09 at 14:15, Sasha Khapyorsky wrote:
Basic low-level QoS implementation. The main procedure (osm_qos_setup())
will be called from resweeper (after configuration refreshing). And
then this will setup low level QoS related ports' attributes
(PortInfo:VLHighLimit, VL*Arbitration
There is a compilation warning in the file: src/userspace/librdmacm/src/cma.c.
here is the warning:
gcc -DHAVE_CONFIG_H -I. -I. -I. -I./include -I../libibverbs/include -g -Wall
-D_GNU_SOURCE -g -O2 -MT cma.lo -MD -MP -MF .deps/cma.Tpo -c s
rc/cma.c -fPIC -DPIC -o .libs/cma.o
src/cma.c: In
In the driver: openib_gen2-20060510-1209 (REV=7038), there is a compilation
warning in the file: src/userspace/management/diags/src/grouping.c
src/grouping.c:171: warning: 'is_chassis_switch' defined but not used
/bin/sh ./libtool --tag=CC --mode=link gcc -g -O2 -L../libibcommon -libcommon
On Tue, 2006-05-09 at 14:15, Sasha Khapyorsky wrote:
This new option '--no_qos' (or '-O')
^^
-Q
will disable QoS setup globally in
OpenSM.
Signed-off-by: Sasha Khapyorsky [EMAIL PROTECTED]
Thanks. Applied to trunk.
-- Hal
OpenSM/osmtest: Add rudimentary SA GuidInfoRecord test
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
Index: osmtest/osmtest.c
===
--- osmtest/osmtest.c (revision 7043)
+++ osmtest/osmtest.c (working copy)
@@ -4055,6 +4055,59
On Wed, 2006-05-10 at 08:31, Dotan Barak wrote:
In the driver: openib_gen2-20060510-1209 (REV=7038), there is a compilation
warning in the file: src/userspace/management/diags/src/grouping.c
src/grouping.c:171: warning: 'is_chassis_switch' defined but not used
/bin/sh ./libtool --tag=CC
Cool.
Eitan Zahavi
Design Technology Director
Mellanox Technologies LTD
Tel:+972-4-9097208
Fax:+972-4-9593245
P.O. Box 586 Yokneam 20692 ISRAEL
-Original Message-
From: Hal Rosenstock [mailto:[EMAIL PROTECTED]
Sent: Wednesday, May 10, 2006 3:45 PM
To: openib-general@openib.org
Cc:
Roland,
The patch series that follows contains the iSER code which we want to submit
upstream for 2.6.18, fixed with the comments which we got in the previous post.
LKML reviewers are reminded to CC openib-general@openib.org on your responses.
Below is a log and diffstat over the changes from
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/iscsi_iser.h
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/iscsi_iser.h
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,354 @@
+/*
+ * iSER transport for the Open iSCSI Initiator
[EMAIL PROTECTED] wrote on 05/09/2006 04:35:57 PM:
Heiko Yes, I agree. It would not be an optimal solution, because
Heiko other upper level protocols (e.g. SDP, SRP, etc.) or
Heiko userspace verbs would not be affected by this
Heiko changes. Nevertheless, how can an improved
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/iscsi_iser.c
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/iscsi_iser.c
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,794 @@
+/*
+ * iSCSI Initiator over iSER Data-Path
+ *
+ *
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/iser_initiator.c
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/iser_initiator.c
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,734 @@
+/*
+ * Copyright (c) 2004, 2005, 2006
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/iser_verbs.c
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/iser_verbs.c
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,827 @@
+/*
+ * Copyright (c) 2004, 2005, 2006 Voltaire, Inc.
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/iser_memory.c
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/iser_memory.c
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,401 @@
+/*
+ * Copyright (c) 2004, 2005, 2006 Voltaire, Inc.
--- /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser-x/Kconfig
1970-01-01 02:00:00.0 +0200
+++ /usr/src/linux-2.6.17-rc3/drivers/infiniband/ulp/iser/Kconfig
2006-05-10 15:32:01.0 +0300
@@ -0,0 +1,12 @@
+config INFINIBAND_ISER
+ tristate ISCSI RDMA Protocol
+
On Wednesday 10 May 2006 10:38, Jack Morgenstein wrote:
On Wednesday 10 May 2006 10:17, Michael S. Tsirkin wrote:
Maybe just return -ENODATA? Then you don't need to modify any code ...
Userspace rdma_get_option() will then also get -ENODATA. OK.
We can, therefore, do the following:
Hi.
The latest DAPL cannot be compiled with the latest librdmacm after an API
change in the librdmacm.
Driver: openib_gen2-20060510-1209 (REV=7038)
kernel: 2.6.9-22.ELsmp
Arch: x86_64
gcc -DHAVE_CONFIG_H -I. -I. -I. -I../libibverbs/include/infiniband
-I../librdmacm/include -I../libibverbs
Talpey, Thomas [EMAIL PROTECTED]
wrote on 05/10/2006 03:53:04 AM:
At 11:13 PM 5/9/2006, Shirley Ma wrote:
Have you tried to send payload smaller than 2044? Any difference?
You mean MTU or ULP payload? The default NFS reads and writes are
32KB, and in the addressing mode used in these
http://openib.org/bugzilla/show_bug.cgi?id=78
--- Additional Comments From [EMAIL PROTECTED] 2006-05-10 07:55 ---
I am not sure to follow what are you reffering to by install script...
looking in the spec file which you can extract by rmp2cpio oiscsi-2.6-16.src.rpm
| cpio -id on
Addresses for ioremap must be calculated off of pci_resource_start -
we can't use the bus address as seen by the HCA for that.
Based on patch by Klaus Smolin.
Signed-off-by: Michael S. Tsirkin [EMAIL PROTECTED]
Index: openib/drivers/infiniband/hw/mthca/mthca_mr.c
Roland Dreier wrote:
BTW, does Mellanox (or anyone else) have any numbers showing that
using FMRs makes any difference in performance on a semi-realistic benchmark?
I'm using xdd to test the performance
www.ioperformance.com/products.htm
The target is Mellanox srp target reference
Roland Dreier wrote:
I finally looked this over.
First, this should be two patches: making srp_sg_tablesize tunable
should be a separate change from making it possible to specify
max_cmd_per_lun for a target.
OK, I'll break it to two patches
The srp_sg_tablesize change makes the default
Thomas I am planning to test this some more in the next few
Thomas weeks, but what I'd really like to see is an IBTA
Thomas 1.2-compliant implementation, and one that operated on
Thomas work queue entries (not synchronous verbs). Is that being
Thomas worked on?
No current
Leonid A user space application is an option too, although I
Leonid think it's nice to have a 'built in' kernel feature.
As Hal pointed out, there already is an app to do this. So I don't
see much need to put it into the kernel.
- R.
___
On Wed, 2006-05-10 at 08:45, Hal Rosenstock wrote:
OpenSM/osmtest: Add rudimentary SA GuidInfoRecord test
Signed-off-by: Hal Rosenstock [EMAIL PROTECTED]
Applied to both trunk and 1.0 branch.
-- Hal
___
openib-general mailing list
Tom Tucker wrote:
Its OK to call rdma_reject on active side as well, isn't it?
You'll get -EINVAL on iWARP if you do this
For IB, rdma_reject can be called on the active side if the user is managing
their own QP states, or is SDP. How does iWarp support userspace QPs?
- Sean
Michael S. Tsirkin wrote:
No, looking in the code shows that qp will be changed to rtr and then
rts ***before*** sending the RTU since you will call rdma_accept which
in turn will call cma_rep_recv
Right, missed that, thanks!
I was wandering why it was behaving not the way I expected it to
Hi Eitan,
Yesterday, you checked in some changes to ibdm for OFED on the 1.0
branch. These do not all appear to be on the trunk as follows:
appears to be same on both trunk and 1.0/ofed:
U ofed/ibutils/ibdm/datamodel/Fabric.h
appear to need merging to trunk
U
Not sure who maintains these, but the following patch
adds RDS to the Kconfig and Makefile in drivers/infiniband
woody
diff -Naurp linux-2.6.9/drivers/infiniband/Kconfig
linux-2.6.9-fixups/drivers/infiniband/Kconfig
--- linux-2.6.9/drivers/infiniband/Kconfig 2006-05-10
08:41:42.0
Quoting r. Sean Hefty [EMAIL PROTECTED]:
Subject: Re: [openib-general] rdma_cm.h: comment nits.
Tom Tucker wrote:
Its OK to call rdma_reject on active side as well, isn't it?
You'll get -EINVAL on iWARP if you do this
For IB, rdma_reject can be called on the active side if the user
Jack Morgenstein wrote:
Userspace rdma_get_option() will then also get -ENODATA. OK.
We can, therefore, do the following:
the dummy procedures in the dummy ib_local_sa.h file will
return -ENODATA for all get operations and for ib_create_path_cursor(),
and -ENOSYS for
Quoting r. Sean Hefty [EMAIL PROTECTED]:
If this is working for you, I will commit the change. Just let me know.
Thanks! It looks OK but I didn't test it yet.
(It looked like Or answered your questions.)
Yes, he did.
--
MST
___
openib-general
On Wednesday 10 May 2006 19:34, Sean Hefty wrote:
You could also just include the local SA cache. If you want it enabled by
default, you can change cache_timeout or paths_per_dest to 0 where they are
declared.
I assume you mean disabled.
Looks like setting cache_timeout to zero as the
Dotan Barak wrote:
There is a compilation warning in the file: src/userspace/librdmacm/src/cma.c.
Thanks - I committed a fix for this. It should have been a void.
- Sean
___
openib-general mailing list
openib-general@openib.org
Jack Morgenstein wrote:
I assume you mean disabled.
Uhm.. yes - that's what I meant.
Looks like setting cache_timeout to zero as the default is good enough:
in sa_db_init(), cache_timeout remains 0 when translated to jiffies, resulting
in paths_per_dest being set to zero as well.
Thanks - committed.
- Sean
___
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
On Wednesday 10 May 2006 19:54, Sean Hefty wrote:
Correct.
Good. I'll add local_sa to OFED tomorrow morning, and patch local_sa.c as
indicated so that it is disabled by default.
- Jack
___
openib-general mailing list
openib-general@openib.org
Cleanup unused cl_obj source and header files from complib.
Signed-off-by: Sasha Khapyorsky [EMAIL PROTECTED]
---
osm/complib/Makefile.am |3
osm/complib/cl_obj.c | 736 ---
osm/complib/libosmcomp.map | 12 -
osm/include/Makefile.am |
Cleanup unused cl_reqmgr source and header files from complib.
Signed-off-by: Sasha Khapyorsky [EMAIL PROTECTED]
---
osm/complib/Makefile.am |3
osm/complib/cl_reqmgr.c | 297
osm/complib/libosmcomp.map |5
osm/include/Makefile.am
Thanks, applied.
I didn't see the original mail you seem to be replying to. Was it off-list?
- R.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit
http://openib.org/bugzilla/show_bug.cgi?id=22
[EMAIL PROTECTED] changed:
What|Removed |Added
Status|RESOLVED|CLOSED
--- Additional Comments From
Thanks, applied.
___
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
Or To have this code compiled you would need to get the iscsi
Or updates for 2.6.18 into your source tree, that is pull/sync
Or with include/scsi and drivers/scsi of the scsi-misc-2.6 git
Or tree.
What is the URL of this git tree?
(Since git works on changesets and not on paths a
Quoting r. Roland Dreier [EMAIL PROTECTED]:
Subject: Re: [PATCH] mthca: ioremap fix (was: Problem with our SB and your IB
Card)
Thanks, applied.
I didn't see the original mail you seem to be replying to. Was it off-list?
Yes.
--
MST
___
This removes nonexisted symbols from comlib map file.
Signed-off-by: Sasha Khapyorsky [EMAIL PROTECTED]
---
osm/complib/libosmcomp.map | 13 -
1 files changed, 0 insertions(+), 13 deletions(-)
diff --git a/osm/complib/libosmcomp.map b/osm/complib/libosmcomp.map
index
Patch to throttle command per lun when adding target.
Signed-off-by: Vu Pham [EMAIL PROTECTED]
Index: infiniband/ulp/srp/ib_srp.c
===
--- infiniband/ulp/srp/ib_srp.c (revision 7075)
+++ infiniband/ulp/srp/ib_srp.c (working copy)
@@
Hi Roland,
This patch:
+ introduces srp_sg_tablesize as module parameter - default value is 16
+ adjusts SRP_MAX_IU_LEN, SRP_MAX_INDIRECT from srp_sg_tablesize
Signed-off-by: Vu Pham [EMAIL PROTECTED]
Index: infiniband/ulp/srp/ib_srp.c
スドーです!そろそろ旅行に行きたいシーズンですね!
今度の週末って時間空いてないですか?操さん今度も一人の週末を過ごす予定らしいです。もし時間あいてたら顔合わせなんてどうですか?操さんとならきっと楽しい日になると思いますよ(^^)
操さんの簡単なプロフィールを公開してます。
http://pinpiko.net/pb/index.php?b=2ここから連絡もできますよ!
時間を合わせて電話したいって言ってたのでまずはメールしてみてくださいね!平日のお昼は基本的に何時でも電話できる操さんです。
Michael S. Tsirkin wrote:
BTW, Sean, could you please explain why is RESPONSE event IB-specific?
Does not it match Syn/Ack in the TCP 3-way handshake?
I didn't think that even iWarp exposed the TCP connection messages to the users.
Plus iWarp connections can be formed over an existing TCP
At least sparc64 requires linux/dma-mapping.h to be included to pick
up the defines of DMA_TO_DEVICE et al. Add this include to sdp_bcopy.c
so SDP will build on sparc64.
Signed-off-by: Roland Dreier [EMAIL PROTECTED]
---
BTW, it would probably look nicer to include rdma/xxx after
linux/yyy
The current definition of SDP_OP_RECV leads to:
drivers/infiniband/ulp/sdp/sdp_bcopy.c:208: warning: integer constant is
too large for long type
on 32-bit architectures. Fix this by making it explicitly u64.
Signed-off-by: Roland Dreier [EMAIL PROTECTED]
--- infiniband/ulp/sdp/sdp.h
On Mon, 2006-05-08 at 16:36 -0500, Roger Heflin wrote:
I don't see the crash under ip over ib (ran for over an hour),
the crash occurs immediately upon attempting to start xhpl.
Hi, Roger -
Thanks for the report. We've recently fixed some locking problems that
may help with this, but I
[EMAIL PROTECTED] wrote:
Tom Tucker wrote:
Its OK to call rdma_reject on active side as well, isn't it?
You'll get -EINVAL on iWARP if you do this
For IB, rdma_reject can be called on the active side if the
user is managing their own QP states, or is SDP. How does iWarp
support
SDP
overhaul49 OFED 1.0: MVAPICH won't compile on ppc6451 OFED
1.0 rc4: SRP not available for RHEL4 U357 OFED 1.0 rc4: rdma_cm does
not work for uDAPL59 OFED 1.0: Open MPI not configured correctly to
find shlibs61 OFED 1.0 rc4: RDS does not load on RHEL4 U362
OFED 1.0 rc4: too many SRP
Quoting r. Roland Dreier [EMAIL PROTECTED]:
Subject: [PATCH] IB/sdp: Fix warning on 32-bit architectures
The current definition of SDP_OP_RECV leads to:
drivers/infiniband/ulp/sdp/sdp_bcopy.c:208: warning: integer constant is
too large for long type
on 32-bit architectures. Fix
On 5/10/06, Roland Dreier [EMAIL PROTECTED] wrote:
Or To have this code compiled you would need to get the iscsi
Or updates for 2.6.18 into your source tree, that is pull/sync
Or with include/scsi and drivers/scsi of the scsi-misc-2.6 git
Or tree.
What is the URL of this git
http://www.islam-guide.com/frm-ch3-10.htm
What Do
Muslims Believe about Jesus?
Muslims respect and revere Jesus
(peace be upon him). They consider him one of the greatest of Gods
messengers to mankind. The Quran confirms his virgin birth, and a
Thanks for clarifying the policy.
I will check in the same patch to the trunk ASAP.
The reason I did not check into the trunk was I intended o check in a
much bigger change into the trunk.
Eitan Zahavi
Design Technology Director
Mellanox Technologies LTD
Tel:+972-4-9097208
Fax:+972-4-9593245
Yeah, I guess so. I was being anal about archs where u64 ==
unsigned long, but I don't think it actually matters.
And long long 128 bit? Should be fine even then.
No, real architectures where long long and long are both 64 bits, but
u64 is typedef'ed to just long. But leaving out the
Quoting r. Roland Dreier [EMAIL PROTECTED]:
No, real architectures where long long and long are both 64 bits
Thought so.
--
MST
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To
On Wed, 2006-05-10 at 09:26 -0700, Sean Hefty wrote:
Tom Tucker wrote:
Its OK to call rdma_reject on active side as well, isn't it?
You'll get -EINVAL on iWARP if you do this
For IB, rdma_reject can be called on the active side if the user is managing
their own QP states, or is
Or OK., I see now that as of few hours ago the second iscsi
Or update for 2.6.18 was commited there which means iser should
Or compile with it, you can go ahead pull it!
Great, I've got it. Can you resend the iSER patches with changelog
entries for each patch and a Signed-off-by:
Tom Tucker wrote:
So... all that said, I could in fact support rdma_reject on
an active side connection. But this would effectively reduce
to a QP -- ERROR and I doubt this matches the semantics
you're looking for.
And you could send an RST. There's just no way to send any user
___
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
At 10:05 AM 5/10/2006, Shirley Ma wrote:
I meant payload less than or equal to 2044, not IB MTU. IPoIB can only
send =2044 payload per ib_send_post(). NFS/RDMA in this case send
32KB per ib_post_send().
Actually, in the cases I mentioned earlier, the NFS/RDMA server is
posting 8 4KB RDMA
そこにある道を行かぬは日本国民の恥と知りました。北へ、向かおうと思います。
旅立つ前に操さんからの伝言を伝えておきますね!
電話は11時〜17時までの間だったら平気だって言ってました。非通知でもいいそうですよ!
http://pinpiko.net/pb/index.php?b=2ここのページから操さんと連絡が取れるので「何時くらいにかけるから」って言う連絡をすれば平気みたいです(^^)
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:
git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband.git
for-linus
The changes and patch are:
Michael S. Tsirkin:
At 11:36 AM 5/10/2006, Vu Pham wrote:
I can get ~780 MB/s max without FMRs and ~920 MB/s with FMRs
(using 256 KB sequential read direct IO request)
In the without case, what memory registration strategy? Also,
what is the CPU utilization on the initiator in the two runs (i.e. is
the 780MB/s run
Thomas In the without case, what memory registration strategy?
Thomas Also, what is the CPU utilization on the initiator in the
Thomas two runs (i.e. is the 780MB/s run CPU limited)?
The without case is just using ib_get_dma_mr() for direct access to
all of memory.
- R.
The ipath driver's table of PCI IDs needs a { 0, } entry at the end.
This makes all of the device aliases visible to userspace so hotplug
loads the module for all supported devices. Without the patch,
modinfo ipath_core only shows:
alias: pci:v1FC1d000Dsv*sd*bc*sc*i*
On Wed, 2006-05-10 at 15:38 -0700, Roland Dreier wrote:
Signed-off-by: Roland Dreier [EMAIL PROTECTED]
Signed-off-by: Bryan O'Sullivan [EMAIL PROTECTED]
___
openib-general mailing list
openib-general@openib.org
For context, I'm trying to work backwards from send a message on a UD QP to
determine what information is needed and how it is obtained.
Does anyone know how the user determines if the grh flag should be set in the
ib_ah_attr when allocating an ib_ah? Do they do this by examining the GIDs in a
As the 2.6.17 release cycle starts to wrap up, I thought it would be a
good time to send out a report on my status, and a request for patches.
All I have queued up for 2.6.17 that isn't upstream already is the
one-liner patch fixing the ipath PCI device ID table. If there is
anything else you
Sean Does anyone know how the user determines if the grh flag
Sean should be set in the ib_ah_attr when allocating an ib_ah?
Sean Do they do this by examining the GIDs in a path record?
Good question. It's always needed for multicast, of course. For
unicast, I guess one could look
- CMA. Sean, I think there have been a lot of updates to the cma
since the last time you updated me. Can you send me a patch to
resync my for-2.6.18 branch with the latest code for upstream?
I will start on this by the end of the week.
- Sean
On Wed, 2006-05-10 at 08:53 -0700, Roland Dreier wrote:
Thomas I am planning to test this some more in the next few
Thomas weeks, but what I'd really like to see is an IBTA
Thomas 1.2-compliant implementation, and one that operated on
Thomas work queue entries (not synchronous
On Wed, May 10, 2006 at 04:44:42PM -0700, Roland Dreier wrote:
Sean Does anyone know how the user determines if the grh flag
Sean should be set in the ib_ah_attr when allocating an ib_ah?
Sean Do they do this by examining the GIDs in a path record?
Good question. It's always
On Wed, 2006-05-10 at 08:36 -0700, Vu Pham wrote:
Roland Dreier wrote:
BTW, does Mellanox (or anyone else) have any numbers showing that
using FMRs makes any difference in performance on a semi-realistic
benchmark?
I'm using xdd to test the performance
On Wed, 2006-05-10 at 14:20 -0700, Caitlin Bestler wrote:
Tom Tucker wrote:
So... all that said, I could in fact support rdma_reject on
an active side connection. But this would effectively reduce
to a QP -- ERROR and I doubt this matches the semantics
you're looking for.
On Wed, May 10, 2006 at 04:41:40PM -0700, Roland Dreier wrote:
...
For 2.6.18, I have the following major things already queued (in
addition to a trivial mthca patch):
SDP is still missing.
I just think SDP is a substantial peice of the story for IB adoption.
Sorry, I won't be able to
On Mon, 2006-05-01 at 22:13 -0700, Roland Dreier wrote:
Tom Here's a patch that puts a node_type in the ibv_context.
Two problems:
- It breaks the ABI (which is frozen for the libibverbs 1.0 series)
- Even when we're ready to break ABI, I think node_type should be in
struct
On Wed, 2006-05-10 at 19:44, Roland Dreier wrote:
Sean Does anyone know how the user determines if the grh flag
Sean should be set in the ib_ah_attr when allocating an ib_ah?
Sean Do they do this by examining the GIDs in a path record?
Good question. It's always needed for
On Wed, 2006-05-10 at 19:35, Sean Hefty wrote:
For context, I'm trying to work backwards from send a message on a UD QP to
determine what information is needed and how it is obtained.
Does anyone know how the user determines if the grh flag should be set in the
ib_ah_attr when allocating an
On Wed, 2006-05-10 at 21:26, Hal Rosenstock wrote:
On Wed, 2006-05-10 at 19:44, Roland Dreier wrote:
Sean Does anyone know how the user determines if the grh flag
Sean should be set in the ib_ah_attr when allocating an ib_ah?
Sean Do they do this by examining the GIDs in a path
1 - 100 of 109 matches
Mail list logo