Here's a patch to replace explicit cast to sdp_opt with container_of, removing
assumptions on structure layout.
As a side note, wouldnt sk_to_sdp and sdp_to_sk be better names for
sdp_sk and sk_sdp?
---
Use container_of in place of cast to sdp_opt.
Signed-off-by: Michael S. Tsirkin <[EMAIL PROTE
On Mon, Jun 20, 2005 at 08:40:47PM -0700, Roland Dreier wrote:
> Kevin> Maybe somebody could help me understand the proper way to
> Kevin> map between an IP address assigned to a port to the "device
> Kevin> name" and "port number" in the gen2 architecture. If I
> Kevin> have an IP
On Mon, Jun 20, 2005 at 11:06:12AM -0700, Tom Duffy wrote:
> Here is the updated patch based off r2663 and building off of 2.6.12.
>
> Signed-off-by: Tom Duffy <[EMAIL PROTECTED]>
Thanks a lot Tom for tracking the changes this whole time.
Committed revision 2665.
-Libor
_
Kevin> Maybe somebody could help me understand the proper way to
Kevin> map between an IP address assigned to a port to the "device
Kevin> name" and "port number" in the gen2 architecture. If I
Kevin> have an IP address can I map it to a name that i get back
Kevin> from ibv_get
Maybe somebody could help me understand the proper way to map between an IP
address assigned to a port to the
"device name" and "port number" in the gen2 architecture. If I have an IP
address can I map it to a name that i get back
from ibv_get_device_name() or pass to ibv_open_device().
Andrei> In what context CQ completion handler is called? I'm
Andrei> talking about kernel module using ib verbs.
In general you should assume it is called from an unknown context.
The current mthca driver calls it from interrupt context, but future
drivers might use softirq or even process
In what context CQ completion handler is called? I'm talking about
kernel module using ib verbs.
Thanks,
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubscribe, please visit http://open
On Mon, Jun 20, 2005 at 02:19:16PM -0700, Grant Grundler wrote:
> On Mon, Jun 20, 2005 at 01:06:24PM -0700, Andrei Petrov wrote:
> > Application I'm working on needs to receive and process messages
> > from RQ, so I dma_map and ib_post_recv pool of buffers.
> > Then after I receive a message I pla
On Mon, Jun 20, 2005 at 02:19:34PM -0700, Roland Dreier wrote:
> Andrei> Application I'm working on needs to receive and process
> Andrei> messages from RQ, so I dma_map and ib_post_recv pool of
> Andrei> buffers. Then after I receive a message I plan to reuse
> Andrei> already mapp
On Tue, 2005-06-07 at 19:27 -0400, Hal Rosenstock wrote:
> I didn't try broadcast but the limited broadcast (192.168.0.255_ and it
> worked. Does it for you ? I will try this next time. Not sure this is an
> OpenSM issue.
This wasn't an opensm issue. I have apparently gotten used to Linux
where i
Quoting r. Grant Grundler <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] rdma_lat-09 and results
>
> On Thu, Jun 16, 2005 at 12:53:34AM +0300, Michael S. Tsirkin wrote:
> > Quoting r. Grant Grundler <[EMAIL PROTECTED]>:
> > > 2) I working towards making the server side a daemon.
> >
> > I dont see t
Andrei> Application I'm working on needs to receive and process
Andrei> messages from RQ, so I dma_map and ib_post_recv pool of
Andrei> buffers. Then after I receive a message I plan to reuse
Andrei> already mapped buffer for consequent ib_post_recv. Is that
Andrei> safe? Should
On Mon, Jun 20, 2005 at 01:06:24PM -0700, Andrei Petrov wrote:
> Application I'm working on needs to receive and process messages
> from RQ, so I dma_map and ib_post_recv pool of buffers.
> Then after I receive a message I plan to reuse already mapped
> buffer for consequent ib_post_recv. Is that
On Mon, Jun 20, 2005 at 03:52:11PM -0400, William Jordan wrote:
> When receiving UDP traffic on a slow host (no user level end to end
> flow control), you can improve receive throughput by reducing the
> number of outstanding receives. The CPU can be saturated receiving
> packets which are overflow
On Mon, 2005-06-20 at 13:22 -0700, Johann George wrote:
> SDP does not compile on the 2.6.12 kernel. It references sk_debug,
> sk_localroute and sk_rcvtstamp which are no longer defined in sock.h.
> (Using SVN 2663).
Yup. Please see my patch.
-tduffy
signature.asc
Description: This is a digit
Quoting r. William Jordan <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] ipoib tune rx ring size
>
> On 6/20/05, Grant Grundler <[EMAIL PROTECTED]> wrote:
> > On Mon, Jun 20, 2005 at 06:14:04PM +0300, Michael S. Tsirkin wrote:
> > > Reducing the rx ring size gives me bandwidth win of about 1%.
> >
>
SDP does not compile on the 2.6.12 kernel. It references sk_debug,
sk_localroute and sk_rcvtstamp which are no longer defined in sock.h.
(Using SVN 2663).
Johann
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinf
Application I'm working on needs to receive and process messages
from RQ, so I dma_map and ib_post_recv pool of buffers.
Then after I receive a message I plan to reuse already mapped
buffer for consequent ib_post_recv. Is that safe? Should I worry
about dma syncronisation?
Thanks,
__
On 6/20/05, Grant Grundler <[EMAIL PROTECTED]> wrote:
> On Mon, Jun 20, 2005 at 06:14:04PM +0300, Michael S. Tsirkin wrote:
> > Reducing the rx ring size gives me bandwidth win of about 1%.
>
> Michael,
> Doesn't this sound counter-intuitive to you?
>
> Can you share which configuration (chipset,
On Mon, 2005-06-20 at 11:28 -0700, Libor Michalek wrote:
> Looks good, but I do have one question, why did you move the clearing
> of 'struct sock' from 'struct socket' to the end of the function?
> Specifically if we are waiting on linger during the release I don't
> think we want sock to have an
On Mon, Jun 20, 2005 at 11:06:12AM -0700, Tom Duffy wrote:
> Here is the updated patch based off r2663 and building off of 2.6.12.
>
> Index: linux-2.6.12-openib/drivers/infiniband/ulp/sdp/sdp_inet.c
> ===
> --- linux-2.6.12-openib/dri
On Mon, 2005-06-20 at 13:11 -0400, Hal Rosenstock wrote:
> > Nevermind. Just tried it with Solaris Nevada build 16 and it works
> > too!
>
> What's Nevada ?
Sorry, that is the code name for Solaris 11, also what OpenSolaris is
based off of.
> Can you retry with the latest and greatest (both
Here is the updated patch based off r2663 and building off of 2.6.12.
Signed-off-by: Tom Duffy <[EMAIL PROTECTED]>
Index: linux-2.6.12-openib/drivers/infiniband/ulp/sdp/sdp_rcvd.c
===
--- linux-2.6.12-openib/drivers/infiniband/ulp/sd
Hi Tom,
On Tue, 2005-06-07 at 18:37, Tom Duffy wrote:
> On Tue, 2005-06-07 at 15:21 -0700, Tom Duffy wrote:
> > On Tue, 2005-06-07 at 18:12 -0400, Hal Rosenstock wrote:
> > > With this change, able to interoperate with Solaris 10 SA client (which
> > > uses RMPP) with OpenSM
> >
> > Cool. I will
Hi Sean,
On Tue, 2005-06-07 at 17:58, Sean Hefty wrote:
> Hal, can you test this and see if it corrects the problems that you were
> seeing on the SA side?
Just tried this with Solaris 10 and it works fine either way.
The problem I was having appeared related but was separate from this and
has s
On Mon, 2005-06-20 at 20:00 +0300, Itamar Rabenstein wrote:
> Hi Tom,
> I was just copied to original code from dapl_os_strdup() to
> dapl_hca_alloc().
> so if it is a problem is it was there in the code before my patch.
> But why do you think we need strncpy and not strcpy ?
> name param is null t
Hi Tom,
I was just copied to original code from dapl_os_strdup() to
dapl_hca_alloc().
so if it is a problem is it was there in the code before my patch.
But why do you think we need strncpy and not strcpy ?
name param is null terminated string.
Itamar
> -Original Message-
> From: Tom Duf
On Mon, Jun 20, 2005 at 06:14:04PM +0300, Michael S. Tsirkin wrote:
> Reducing the rx ring size gives me bandwidth win of about 1%.
Michael,
Doesn't this sound counter-intuitive to you?
Can you share which configuration (chipset, CPU, IO bus, HCA)
you measured this on?
I've also seen this to be
On Sun, Jun 19, 2005 at 12:13:41PM -0700, Roland Dreier wrote:
> > + * Copyright (c) 2005 Mellanox Technologies (Michael S. Tsirkin).
>
> I believe you want to include the statement "All rights reserved."
> Also, who is the copyright holder -- MST as an individual or Mellanox
> Technologies?
On Thu, Jun 16, 2005 at 12:53:34AM +0300, Michael S. Tsirkin wrote:
> Quoting r. Grant Grundler <[EMAIL PROTECTED]>:
> > 2) I working towards making the server side a daemon.
>
> I dont see this necessarily as an improvement.
> I'm happy running server and client with ssh from a script
> with mult
Reducing the rx ring size gives me bandwidth win of about 1%.
Signed-off-by: Michael S. Tsirkin <[EMAIL PROTECTED]>
Index: ulp/ipoib/ipoib.h
===
--- ulp/ipoib/ipoib.h (revision 2656)
+++ ulp/ipoib/ipoib.h (working copy)
@@ -62,7
On 6/20/05, Itamar Rabenstein <[EMAIL PROTECTED]> wrote:
> Small changes in dapl_hca_alloc/dapl_hca_free function
>
> Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
>
> diff -Nurp -X dontdiff dat-provider_hash/dapl_hca_util.c
> dat-provider/dapl_hca_util.c
> --- dat-provider_hash/dapl_hca_
ntegrate dapl_hca_alloc/dapl_hca_free to dapl_provider.c
(no need for 2 files just for 2 simple function that kmalloc and kfree.
There is not any special logic in this functions that need to separate
them into different files)
Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
diff -Nurp -X d
Integrate dapl_hca_link_ia/dapl_hca_unlink_ia into dapl_ia.c
(no need for functions that just call LIST_ADD and LIST_DEL)
Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
diff -Nurp -X dontdiff dat-provider_simp/dapl_hca_util.c
dat-provider/dapl_hca_util.c
--- dat-provider_simp/dapl_hca_util
Small changes in dapl_hca_alloc/dapl_hca_free function
Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
diff -Nurp -X dontdiff dat-provider_hash/dapl_hca_util.c
dat-provider/dapl_hca_util.c
--- dat-provider_hash/dapl_hca_util.c Sun Jun 19 16:12:55 2005
+++ dat-provider/dapl_hca_util.c
Remove dapl_hash (there is no need for the hash)
Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
diff -Nurp -X dontdiff dat-provider_bind/Makefile dat-provider/Makefile
--- dat-provider_bind/Makefile Sun Jun 19 15:17:54 2005
+++ dat-provider/Makefile Sun Jun 19 15:24:03 2005
@@ -20,7
Update dat_rmr_bind function API (added lmr_handle as input param)
Signed-off-by: Itamar Rabenstein <[EMAIL PROTECTED]>
Index: test/dapltest/test/dapl_bpool.c
===
--- test/dapltest/test/dapl_bpool.c (revision 2656)
+++ test/daplt
O.K
i am resending this patch and spliting it to 5 small patches
1. Update dat_rmr_bind API (added lmr_handle input param)
2. Remove dapl_hash (there is no need for the hash)
3. Small changes in dapl_hca_alloc/dapl_hca_free functions
4. Integrate dapl_hca_link_ia/dapl_hca_unlink_ia into dapl_ia.c
38 matches
Mail list logo