Hello Hal,
In Gnu make-3.81 (unline 3.80) in action lists lines will be concatanated
without leading tab symbol. So lines like
fi\
done
become 'fidone'. There is fix for osm/Makefile.
Sasha.
Fixes for GNU make-3.81
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
Hi,
The cmpost test (under https://openib.org/svn/gen2/utils/src/linux-kernel/infiniband/util/cmpost)
fails on some cases:
1. If I change the local and the remote timeout for
ib_cm_req_param to 40 (instead of 20, the default value) it causes kernel oops.
2. With the following parameters:
Don't wait for pkey tables update responses in partition manager -
we may just continue resweep process.
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
---
osm/opensm/osm_pkey_mgr.c | 66 +---
osm/opensm/osm_state_mgr.c | 41 ++--
Send changed pkey table blocks to ports only after full update and not
after each pkey value change/update.
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
---
osm/include/opensm/osm_pkey.h | 51 +
osm/opensm/osm_pkey.c | 32 ++
osm/opensm/osm_pkey_mgr.c | 233 +
The structure osm_pkey_mgr_t is not used for pkey management -
clean it up.
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
---
osm/include/opensm/osm_pkey_mgr.h | 183 ++--
osm/include/opensm/osm_sm.h|2
osm/include/opensm/osm_state_mgr.h |
When router port is connected directly to CA don't try handle it as
switch external ports (update pkey table and enforce partitions).
Router ports are handled by partition manager as end ports.
Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
---
osm/opensm/osm_pkey_mgr.c | 43
Hello Hal,
There are various improvements and fixes for opensm pkey manager as well
as some cleanups.
All patches are functionally independent, but order will be important for
applying cleanly.
Sasha.
___
openib-general mailing list
openib-general@open
Or Gerlitz wrote:
I am getting the below trace on 2.6.17-rc2 / AMD x86_64 / PCIX HCA
with both the IB sources that come with the kernel and svn trunk 6520.
This happens if i just modprobe -r ib_mthca after fresh reboot, can
anyone reproduce it on her/his system as well? The module does get
mod
Shirley,
some additional information you may be interested:
According to our experience with the Voltaire IPoIB driver,
splitting CQ harmed the throughput (we checked with the iperf
application, UDP mode.) Splitting the the CQ caused more interrupts,
context switches and CQ polls.
Note,
On Thursday 20 April 2006 01:48, Arlin Davis wrote:
> >Here is a back trace of the hanged process:
> >(gdb) bt
> >#0 0x2b31c86a in pthread_cond_wait@@GLIBC_2.3.2 () from
> >/lib64/tls/libpthread.so.0
> >#1 0x2b42ef5b in dapl_os_wait_object_wait (wait_obj=0x516650,
> >timeout_val=)
don't include a header we don't seem to use
Signed-off-by: Michael S. Tsirkin <[EMAIL PROTECTED]>
Index: linux-2.6.16/drivers/infiniband/core/cma.c
===
--- linux-2.6.16.orig/drivers/infiniband/core/cma.c 2006-04-23
15:04:05.
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> Subject: [PATCH] RDMA CM: only allow privileged access to ports below 1024
>
> Restrict access to ports below 1024 to privileged users.
>
> Signed-off-by: Sean Hefty <[EMAIL PROTECTED]>
Seems to work fine here, please checkin.
--
MST
___
Moni Levy wrote:
On 4/23/06, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
Hi Moni,
Sorry it took me a while to get back to you (was out on vacation ...)
Moni Levy wrote:
On 4/10/06, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
Hi Hal,
-Original Message-
From: Hal Rosenstock [mailto:[EM
On 4/23/06, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
> Hi Moni,
>
> Sorry it took me a while to get back to you (was out on vacation ...)
>
> Moni Levy wrote:
> > On 4/10/06, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
> >
> >>Hi Hal,
> >>
> >>
> >>>-Original Message-
> >>>From: Hal Rosenstock [
On Wednesday 19 April 2006 22:17, Arlin Davis wrote:
> >>OpenIB-cma u1.2 nonthreadsafe default /usr/lib/libdaplcma.so
> >>mv_dapl.1.2 "mthca0 1" ""
> >>OpenIB-cma0-1 u1.2 nonthreadsafe default /usr/lib/libdaplcma.so
> >>mv_dapl.1.2 "mthca0 1" ""
> >>OpenIB-cma0-2 u1.2 nonthrea
I am getting the below trace on 2.6.17-rc2 / AMD x86_64 / PCIX HCA
with both the IB sources that come with the kernel and svn trunk 6520.
This happens if i just modprobe -r ib_mthca after fresh reboot, can
anyone reproduce it on her/his system as well? The module does get
modprobed out.
Or.
$
Hi Moni,
Sorry it took me a while to get back to you (was out on vacation ...)
Moni Levy wrote:
On 4/10/06, Eitan Zahavi <[EMAIL PROTECTED]> wrote:
Hi Hal,
-Original Message-
From: Hal Rosenstock [mailto:[EMAIL PROTECTED]
Sent: Monday, April 10, 2006 2:00 PM
To: Eitan Zahavi
Cc: Ro
Quoting r. Sean Hefty <[EMAIL PROTECTED]>:
> rdma_bind_addr() is called for the user in rdma_resolve_addr(), unless the
> user
> has already bound the cm_id.
BTW, Sean, if you intend to emulate socket interface, same is necessary for
listen: listen without bind is equivalent to binding to anyport
; isn't necessary after {} code block.
Signed-off-by: Michael S. Tsirkin <[EMAIL PROTECTED]>
Index: linux-2.6.16/drivers/infiniband/core/cma.c
===
--- linux-2.6.16.orig/drivers/infiniband/core/cma.c 2006-04-20
17:47:05.0
19 matches
Mail list logo
