Quoting r. Michael S. Tsirkin [EMAIL PROTECTED]:
Subject: [PATCH] support kernel-level sockets in sdp
Hi!
I plan to commit the following. Comments?
---
The following patch adds support for kernel-level sockets in SDP Zcopy
(currently used with AIO).
Signed-off-by: Michael S. Tsirkin
Johann Works for me. I assume the list is null terminated so we
Johann do not have to get num_devices?
Yes, that's right.
- R.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To
Hello, Roland!
I am still seeing IPoIB oopsing about once a week around
ipoib_mcast_join_complete (oops below).
While looking at it, a question occured to me:
what protects the following code in ipoib_mcast_stop_thread
list_for_each_entry(mcast, priv-multicast_list, list) {
How does the following strike you?
I didnt notice any performance impact - could more people test this please?
---
Assuming that a remote node is replaced and its address changes
(e.g. gid change), it seems that the ha field will gets out of sync with
the address handle stored in ipoib_neigh-ah,
Pradeep I am seeking suggestions with respect to the
Pradeep 32-bit/64-bit issues. Is the macro
Pradeep test_thread_flag(TIF_32BIT), to ascertain if we are in
Pradeep the kernel in the context of a 32-bit app or not, generic
Pradeep enought that it will work across all the
Roland Dreier [EMAIL PROTECTED] wrote on 11/14/2005 07:49:14 AM:
Pradeep I am seeking suggestions with respect to the
Pradeep 32-bit/64-bit issues. Is the macro
Pradeep test_thread_flag(TIF_32BIT), to ascertain if we are in
Pradeep the kernel in the context of a 32-bit app or not,
Hi,
Is there some way to compile-out parts of OpenSM so I could fit it inside a
small flash rom?
Maybe I can modify the Makefile to exclude some of the object files? What
functionality isn't mandatory for fabric management?
Thanks,
y
___
Pradeep I am trying to use copy_from_user()/copy_to_user of data
Pradeep structures that contains pointers.
If you are defining a new interface, then the simplest thing is not to
do that: always put pointers in a 64-bit field.
If it is an existing interface that can't be changed, then
This patch provides padding beyond the source and destination addresses
to handle IPv6 address sizes.
Does anyone know of a better way to handle this?
Signed-off-by: Sean Hefty [EMAIL PROTECTED]
Index: include/rdma/rdma_cm.h
===
Hi Yipee,
It would be nice (or I should say just dandy) if we could skip some of
the functionality of OpenSM but actually for Gen2 to work right most of
it is required...
The only parts that are not absolutely required are some SA request
handlers.
Mostly ServiceRecords, NodeInfoRecord,
How small does it need to be ? What processor architecture ?
-- Hal
From: [EMAIL PROTECTED] on behalf of yipee
Sent: Mon 11/14/2005 2:36 PM
To: openib-general@openib.org
Subject: [openib-general] OpenSM size
Hi,
Is there some way to compile-out parts of
Michael Krause wrote:
At 01:01 PM 11/11/2005, Nitin Hande wrote:
Michael Krause wrote:
At 10:28 AM 11/9/2005, Rick Frank wrote:
Yes, the application is responsible for detecting lost msgs at the
application level - the transport can not do this.
RDS does not guarantee that a message has
Just downloaded the SVN 4044, it's boken,
see below compile error:
In file included from drivers/infiniband/include/rdma/ib_sa.h:42,
from drivers/infiniband/core/at.c:53:
drivers/infiniband/include/rdma/ib_mad.h:601:
error: parse error before gfp_t
Thanks
Shirley Ma
IBM Linux Technology
I am seeing the following errors using svn ver 4044 and 2.6.14 kernel.
Have I missed a patch?
drivers/infiniband/core/uat.c: In function `ib_uat_init':
drivers/infiniband/core/uat.c:834: warning: passing arg 2 of
`class_device_create' makes integer from pointer without a
On 14.11.2005 [19:03:20 +0200], Dan Bar Dov wrote:
Hi Nishanth,
I committed fixes to the ppc64 compile warnings in iser:
Committed revision 4044.
Just wanted to confirm that 4044 compiles w/o infiniband warnings now.
Thanks again,
Nish
___
On 14.11.2005 [19:03:20 +0200], Dan Bar Dov wrote:
Hi Nishanth,
I committed fixes to the ppc64 compile warnings in iser:
Committed revision 4044.
Great, thanks!
-Nish
___
openib-general mailing list
openib-general@openib.org
Roland Dreier wrote:
Alfred I am seeing the following errors using svn ver 4044 and
Alfred 2.6.14 kernel. Have I missed a patch?
You have not replaced include/rdma in the kernel with the svn copy.
The easiest thing to do is to rm -rf include/rdma and just link
linux-kernel/infiniband
At 03:47 PM 11/14/2005, Roland Dreier wrote:
Alfred I am seeing the following errors using svn ver 4044 and
Alfred 2.6.14 kernel. Have I missed a patch?
You have not replaced include/rdma in the kernel with the svn copy.
The easiest thing to do is to rm -rf include/rdma and just link
I'll answer your question, anyway (your next post indicated that you're moving
to 2.6.14).
You need to apply patch:
https://openib.org/svn/gen2/branches/backport/2.6.13/verbs_malloc_3926_to_2_6_13.patch
to fix the above problem.
Jack
On Mon, Nov 14, 2005 at 11:17:40PM +0200, Shirley Ma wrote:
Hal Rosenstock halr at voltaire.com writes:
How small does it need to be ? What processor architecture ?
The processor I'm using is a x86_64, but I guess I can compile OpenSM for x86
(32bits).
The binary size I'm seeing is about 2MB. After bzip'ing it shrinks to 670KB. If
I could reduce it to
I think I see the problem. As you said, mthca was incorrectly
checking the responder resources to see if it should enable RDMA
writes on the receive queue. However, responder resources only apply
to RDMA reads and atomics and are never set for UC QPs, so this was
never set.
Does the
21 matches
Mail list logo