Or Gerlitz wrote:
Yevgeny Kliteynik wrote:
There was a hand-over problem in OFED 1.4, but later it turned out to
be FW issue. The thing is, FW version 2.6.648 doesn't have this bug
any more...
so things should work fine with the newly released 2.7 firmware?
Yes
if this
is still under
Hi,
I have tried to use hop weighted routing and noticed that it is supported only
for minhop.
Is there a reason that this was not implemented for up-down as well?
Thanks,
Eli
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to
Alright I'm sending this for the 3rd time in plain text because
apparently vger wont' take html mail. Fair 'nuff.
I double checked and firmware version 2.6.648 is in use on the cluster
with the failover issue. The cluster that doesn't have the issue is
running a MT25204 Mellanox HCAs with
On Mon, Oct 12, 2009 at 12:08:43PM +0200, Eli Dorfman (Voltaire) wrote:
Hi,
I have tried to use hop weighted routing and noticed that it is supported
only for minhop.
Is there a reason that this was not implemented for up-down as well?
Thanks,
Eli
No, no particular reason. Just didn't
Use the official names for the supported devices as they are documented in
http://pciids.sourceforge.net/v2.2/pci.ids. The patch also removes device ID
0x6354 from the list of supported devices which is a none existent device.
Signed-off-by: Eli Cohen e...@mellanox.co.il
---
Add MT26438 [ConnectX EN 40GigE PCIe 2.0 5GT/s] to the list of supported
devices.
Signed-off-by: Eli Cohen e...@mellanox.co.il
---
drivers/net/mlx4/main.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index
This patch doesn't apply since e76d0b67 (mlx4_core: Add 40GigE device
ID) which adds ID 0x676e for 40GE.
Also if we're going to churn all these comments, I think it would be a
good idea to at least make the names follow a consistent scheme:
A) choose either PCIe 2.0 or PCIe Gen2 and use that
same comment -- This patch doesn't apply since e76d0b67 (mlx4_core: Add
40GigE device ID) which adds ID 0x676e for 40GE. (and which is
upstream in 2.6.32-rc4)
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More
On Mon, Oct 12, 2009 at 07:43:36AM -0700, Roland Dreier wrote:
same comment -- This patch doesn't apply since e76d0b67 (mlx4_core: Add
40GigE device ID) which adds ID 0x676e for 40GE. (and which is
upstream in 2.6.32-rc4)
Will re-send based on your for-linus branch.
--
To unsubscribe from
Add MT26438, ConnectX EN 40GigE PCIe 2.0 5GT/s to the list of supported
devices.
Signed-off-by: Eli Cohen e...@mellanox.co.il
---
Resending the patch based on Roland's for-linus branch.
I will send another patch later to fix official names of supported
devices.
drivers/net/mlx4/main.c |1 +
Hi Hal,
On 12:05 Sun 04 Oct , Hal Rosenstock wrote:
Heap memory consumption by the unicast and multicast routing tables can be
reduced.
Using valgrind --tool=massif (for heap profiling), there are couple of places
that consume most of the heap memory:
-38.75% (11,206,656B) 0x43267E:
Fix dereferencing type-punned pointer will break strict-aliasing rules
warning catched by gcc-4.4.1.
Signed-off-by: Sasha Khapyorsky sas...@voltaire.com
---
opensm/include/iba/ib_types.h |3 +--
opensm/opensm/osm_helper.c|7 +++
2 files changed, 4 insertions(+), 6 deletions(-)
On 10:47 Fri 09 Oct , Hal Rosenstock wrote:
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
On 10:48 Fri 09 Oct , Hal Rosenstock wrote:
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
On 09:43 Mon 05 Oct , Hal Rosenstock wrote:
On Sat, Oct 3, 2009 at 8:19 PM, Sasha Khapyorsky sas...@voltaire.com wrote:
Move off subnet destination (router address) resolution code to separate
function to improve readability.
Signed-off-by: Sasha Khapyorsky sas...@voltaire.com
---
sizeof unsigned long != sizeof void * in Windows world
Use correct type.
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/include/opensm/st.h b/opensm/include/opensm/st.h
index 30cc308..fc558c8 100644
--- a/opensm/include/opensm/st.h
+++ b/opensm/include/opensm/st.h
@@ -49,7
Hi,
We are using OFED 1.4.2 and trying to simulate the effect of a bonding
failover initiated by a switch failure using echo commands in parallel
to the /sys/class/net/bond0/bonding/active_slave file on a few of the
nodes attached to the switch. Is this an acceptable technique? We are
On 12:38 Mon 05 Oct , Rolf Manderscheid wrote:
Sasha Khapyorsky wrote:
Move off subnet destination (router address) resolution code to separate
function to improve readability.
...
+static ib_net64_t find_router(osm_sa_t *sa, ib_net64_t prefix)
Much better. The *sa argument to
Move off subnet destination (router address) resolution code to separate
function to improve readability.
Signed-off-by: Sasha Khapyorsky sas...@voltaire.com
---
v2 changes:
- log router guid resolution failure on VERBOSE level - it is client's
failure
- make *sa parameter const for
@@ -1523,6 +1521,7 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct
ib_send_wr *wr,
__be32 *lso_wqe;
__be32 uninitialized_var(lso_hdr_sz);
int i;
+__be32 blh = 0;
So isn't this a bug if multiple work requests are passed in? Shouldn't
the blh initialization be
+ *blh = unlikely(halign 64) ? 1 : 0;
This idiom of (boolean condition) ? 1 : 0 looks odd to me... doesn't
(halign 64) already evaluate to 1 or 0 anyway? Does the unlikely()
actually affect code generation here?
True, (halign 64) is the same and is cleaner. As for the
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/main.c b/opensm/osmtest/main.c
index 4bb9f82..287baf3 100644
--- a/opensm/osmtest/main.c
+++ b/opensm/osmtest/main.c
@@ -285,7 +285,7 @@ ib_net64_t get_port_guid(IN osmtest_t * p_osmt, uint64_t
port_guid)
Clarify assignment remove WDK compiler warning.
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/main.c b/opensm/osmtest/main.c
index 4bb9f82..6bfd7a6 100644
--- a/opensm/osmtest/main.c
+++ b/opensm/osmtest/main.c
@@ -433,7 +433,7 @@ int main(int argc, char *argv[])
On Wed, 2009-10-07 at 15:34 -0700, Roland Dreier wrote:
So I looked a little deeper into this, and I don't think (even with the
filtering extensions) that perf events are directly applicable to this
problem. The first issue is that, assuming I'm understanding the
comment in
On Fri, 2009-10-09 at 16:46 -0600, Jason Gunthorpe wrote:
On Fri, Oct 09, 2009 at 02:15:01PM -0700, David J. Wilder wrote:
Yes ping6 will work but I must specify the interface to use:
ping6 fe80::202:c903:1:1925%ib0
Here is the patch to rping to allow the source address to be
On 09:36 Mon 05 Oct , Hal Rosenstock wrote:
Should come from mft_context rather than lft_context
(even though they are currently the same)
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe
On 13:30 Mon 05 Oct , Hal Rosenstock wrote:
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
* Jason Gunthorpe jguntho...@obsidianresearch.com wrote:
On Wed, Sep 30, 2009 at 11:44:56AM +0200, Ingo Molnar wrote:
OK. It would be nice to tie into something more general, but I
think I agree -- perf counters are missing the filtering and the no
lost events that ummunotify
On 10:46 Fri 09 Oct , Hal Rosenstock wrote:
Per published MgtWG errata RefID 4576
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
---
diff --git a/opensm/include/iba/ib_types.h b/opensm/include/iba/ib_types.h
index 25ed35f..c820372 100644
--- a/opensm/include/iba/ib_types.h
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/osmtest.c b/opensm/osmtest/osmtest.c
index c6ec955..423e836 100644
--- a/opensm/osmtest/osmtest.c
+++ b/opensm/osmtest/osmtest.c
@@ -59,10 +59,6 @@
#include complib/cl_debug.h
#include osmtest.h
-#ifndef __WIN__
Remove unused ifdef __WIN__ redundant include.
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/osmtest.c b/opensm/osmtest/osmtest.c
index c6ec955..82a814e 100644
--- a/opensm/osmtest/osmtest.c
+++ b/opensm/osmtest/osmtest.c
@@ -43,19 +43,13 @@
*
*/
-#ifdef
On Mon, Oct 12, 2009 at 08:19:44PM +0200, Ingo Molnar wrote:
After that point the scheme is perfectly lossless.
Well if it can OOM it's not lossless, obviously. You just define event
loss to be equivalent to Destruction of the universe. ;-)
It can't OOM once the ummunotify registration is
On Mon, Oct 12, 2009 at 10:52:59AM -0700, David J. Wilder wrote:
It is not, IPv6 link local addresses must be scoped. rping is
parsing the address with getaddrinfo, that does correctly set the
sin6_scope_id value in the sockaddr.
ping6 is scoping the address (setting sin6_scope_id) by
* Jason Gunthorpe jguntho...@obsidianresearch.com wrote:
On Mon, Oct 12, 2009 at 08:19:44PM +0200, Ingo Molnar wrote:
After that point the scheme is perfectly lossless.
Well if it can OOM it's not lossless, obviously. You just define
event loss to be equivalent to Destruction of
On Mon, Oct 12, 2009 at 10:04:18AM -0700, Roland Dreier wrote:
@@ -1523,6 +1521,7 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct
ib_send_wr *wr,
__be32 *lso_wqe;
__be32 uninitialized_var(lso_hdr_sz);
int i;
+ __be32 blh = 0;
So isn't this a bug if multiple work
On 10:17 Mon 12 Oct , Stan C. Smith wrote:
Clarify assignment remove WDK compiler warning.
Signed-off-by: stan smith stan.sm...@intel.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
On 11:52 Mon 12 Oct , Stan C. Smith wrote:
Signed-off-by: stan smith stan.sm...@intel.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
On 10:10 Mon 12 Oct , Stan C. Smith wrote:
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/main.c b/opensm/osmtest/main.c
index 4bb9f82..287baf3 100644
--- a/opensm/osmtest/main.c
+++ b/opensm/osmtest/main.c
@@ -285,7 +285,7 @@ ib_net64_t get_port_guid(IN
On 09:32 Wed 07 Oct , Stan C. Smith wrote:
Signed-off-by: stan smith stan.sm...@intel.com
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at
Sasha Khapyorsky wrote:
On 10:10 Mon 12 Oct , Stan C. Smith wrote:
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/osmtest/main.c b/opensm/osmtest/main.c
index 4bb9f82..287baf3 100644
--- a/opensm/osmtest/main.c
+++ b/opensm/osmtest/main.c
@@ -285,7 +285,7 @@
Remove redundant strings.h inclusion (and associated ifndef __WIN__).
Signed-off-by: Sasha Khapyorsky sas...@voltaire.com
---
opensm/osmtest/osmtest.c |3 ---
1 files changed, 0 insertions(+), 3 deletions(-)
diff --git a/opensm/osmtest/osmtest.c b/opensm/osmtest/osmtest.c
index
On 14:58 Mon 12 Oct , Smith, Stan wrote:
{
static osmtest_t osm_test;
osmtest_opt_t opt = { 0 };
Sean maintains a separate set of patches applied to IB diags in order to
address issues like the x86 requirement for __cdecl on main().
Since OSM_CDECL was already defined
Sasha Khapyorsky wrote:
On 16:46 Wed 07 Oct , Stan C. Smith wrote:
In osm_db_domain_init() the filename is appended to a path string
'db_dir_name' which contains a trailing '/'. Remove extra '/'.
Signed-off-by: stan smith stan.sm...@intel.com
diff --git a/opensm/opensm/osm_lid_mgr.c
Sasha Khapyorsky wrote:
On 10:16 Thu 08 Oct , Smith, Stan wrote:
Hefty, Sean wrote:
Code cleanup, standardize on a single Windows #define '__WIN__';
WIN32 -- __WIN__. Flip usage of ifndef WIN32 -- ifdef __GNUC__
GNUC indicates a specific compiler, not a platform, which is what
the check
Sasha Khapyorsky wrote:
On 14:58 Mon 12 Oct , Smith, Stan wrote:
{
static osmtest_t osm_test;
osmtest_opt_t opt = { 0 };
Sean maintains a separate set of patches applied to IB diags in
order to address issues like the x86 requirement for __cdecl on
main(). Since OSM_CDECL
---BeginMessage---
Testing srp fail-over with dm-multipath/multipathd/srp_daemon, the
current srp implementation will take average 3-5 minutes to complete the
error recovery before return DID_BAD_TARGET so that dm-multipath can
switch to other paths. During this error recovery, there is no
---BeginMessage---
Disconnect connection without waiting.
Signed-off-by: Vu Pham v...@mellanox.com
---
drivers/infiniband/ulp/srp/ib_srp.c |2 --
1 file changed, 2 deletions(-)
Index: ofed_kernel/drivers/infiniband/ulp/srp/ib_srp.c
---BeginMessage---
Recreating qp, cq at reconnect instead of re-use them. We need this so
that we don't have to wait on disconnect connection
Signed-off-by: Vu Pham v...@mellanox.com
---
drivers/infiniband/ulp/srp/ib_srp.c | 22 +++---
1 file changed, 11 insertions(+), 11
On Mon, Oct 12, 2009 at 10:20:46PM +0200, Ingo Molnar wrote:
It might be more acceptable because the flag-hint mechanism can at most
cause over-flushing - while with perf events we might miss to invalidate
a range altogether.
Right. Overflushing is not important, but missing an event
On Mon, Oct 12, 2009 at 03:29:38PM -0700, Smith, Stan wrote:
If __linux__ doesn't work for you, then please create a Linux Platform define
I can use.
Pretty much all the patches I've seen you make should be guarded by
__WIN__, you shouldn't be using __linux__.
opensm and the other OFA stuff
50 matches
Mail list logo
