Fix memory leak for QOS string parameters.
Signed-off-by: Slava Strebkov sla...@amirm.voltaire.com
---
opensm/opensm/osm_subnet.c | 21 +
1 files changed, 21 insertions(+), 0 deletions(-)
diff --git a/opensm/opensm/osm_subnet.c b/opensm/opensm/osm_subnet.c
index
Add support for PortXmitWait counter
Show PortCounters::PortXmitWait when this capability is supported by the
firmware.
If not supported show this counter as 0.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
infiniband-diags/src/perfquery.c | 10 +-
Nicolas Morey Chaisemartin wrote:
Hello,
I was wondering if there is a doc somewhere with a list of the trap
codes (for generic traps) and what is stored into the associated
ib_mad_notice_attr_t structure?
It is documented in the IB spec 14.2.5.1 NOTICES AND TRAPS.
And in
Add new partition keyword for all switches and hca.
To allow firmware upgrade within managed switches we
want all switch port 0 to have full membership.
'ALL_SWITCH' means all switch end ports in the subnet
'ALL_CA' means all CA end ports in the subnet.
New default partition configuration
set switch end ports to full member in default partition configuration.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_prtn.c |3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/opensm/opensm/osm_prtn.c b/opensm/opensm/osm_prtn.c
index 8b9301e..21c7add
Sasha Khapyorsky wrote:
Hi Eli,
On 14:58 Tue 23 Dec , Eli Dorfman (Voltaire) wrote:
Add support for PortXmitWait counter
Show PortCounters::PortXmitWait when this capability is supported by the
firmware.
If not supported show this counter as 0.
Signed-off-by: Eli Dorfman e
Jason Gunthorpe wrote:
On Mon, Jan 05, 2009 at 02:47:19PM +0200, Eli Dorfman (Voltaire) wrote:
Should XMT_WAIT support be added to output_aggregate_perfcounters(),
reset and other places too?
reset is not supported by the firmware (at the moment).
need to add xmitwait
Hal Rosenstock wrote:
Eli,
On Mon, Jan 5, 2009 at 7:47 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Sasha Khapyorsky wrote:
Hi Eli,
On 14:58 Tue 23 Dec , Eli Dorfman (Voltaire) wrote:
Add support for PortXmitWait counter
Show PortCounters::PortXmitWait when
Sasha Khapyorsky wrote:
On 22:30 Wed 07 Jan , Eli Dorfman wrote:
I am working with ConnectX HCA and latest fw versiong 2.6.0.
The IS4 switch chip does not support portXmitWait reset.
But does it report XmitWait counter support in ClassPortInfo? Anyway it
would be better to implement
Hal Rosenstock wrote:
Eli.
On Thu, Jan 8, 2009 at 7:17 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
I understand that ClassPortInfo specifies whether PortXmitWait is supported.
It does not distinguish between get and set operation.
We have seen that ClassPortInfo to IS4 returns
This patch adds support for PortXmitWait counter get and set (reset)
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit
add PortXmitWait counter to fields.
add CounterSelect2 to fields to allow reset of PortXmitWait(MgtWg comment#4527)
Signed-off-by: Eli Dorfman e...@voltaire.com
---
libibmad/include/infiniband/mad.h |2 ++
libibmad/src/fields.c |2 ++
2 files changed, 4 insertions(+), 0
support PortXmitWait get and set
Signed-off-by: Eli Dorfman e...@voltaire.com
---
infiniband-diags/src/perfquery.c | 14 +-
libibmad/src/gs.c|2 ++
2 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/infiniband-diags/src/perfquery.c
support PortXmitWait get and set
fix syntax error
Signed-off-by: Eli Dorfman e...@voltaire.com
---
infiniband-diags/src/perfquery.c | 14 +-
libibmad/src/gs.c|2 ++
2 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/infiniband-diags/src/perfquery.c
Add new partition keywords for all hca, switches and routers
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit
Add new partition keyword for node type groups.
The following new keywords were added:
'ALL_CAS' means all CA end ports in the subnet.
'ALL_SWITCHES' means all Switch end ports in the subnet
'ALL_ROUTERS' means all Router end ports in the subnet
For example, to allow firmware upgrade within
update documenatation about new partition keywords
'ALL_CAS', 'ALL_SWITCHES', 'ALL_ROUTERS'
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/doc/partition-config.txt |5 -
opensm/man/opensm.8.in |5 -
2 files changed, 8 insertions(+), 2 deletions(-)
diff --git
Sasha Khapyorsky wrote:
Hi Eli,
On 13:54 Sun 21 Dec , Eli Dorfman (Voltaire) wrote:
Fix memory leak for QOS string parameters.
Signed-off-by: Slava Strebkov sla...@amirm.voltaire.com
---
opensm/opensm/osm_subnet.c | 21 +
1 files changed, 21 insertions(+), 0
Sasha Khapyorsky wrote:
Hi Eli,
On 16:56 Tue 13 Jan , Eli Dorfman (Voltaire) wrote:
support PortXmitWait get and set
fix syntax error
Signed-off-by: Eli Dorfman e...@voltaire.com
---
infiniband-diags/src/perfquery.c | 14 +-
libibmad/src/gs.c|2
The following patches are handling subnet configuration update.
Subnet configuration parameters are rescanned every heavy sweep and if possible
are updated.
___
general mailing list
general@lists.openfabrics.org
make setup and destroy routing engines functions global.
change setup_routing_engines() and destroy_routing_engines()
declaration
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_opensm.h | 53
opensm/opensm/osm_opensm.c
rescan subnet configuration after SIGHUP
call osm_subn_rescan_conf_files() after SIGHUP.
this is important when priority is changed and SM is in standby.
in that case it will not send capability mask trap and will not become master.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
put qos options flat below subnet opt
put all qos option parameters (default, ca, sw, router) flat below subnet opt
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_subnet.h | 40 +++-
1 files changed, 30 insertions(+), 10
support new arrangement of qos parameters
change functions to access qos parameters flat below subnet opt
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_qos.c | 46 +++---
1 files changed, 31 insertions(+), 15 deletions(-)
diff
support subnet configuration rescan and update
subnet configuration parameters are rescanned every heavy sweep.
every parameter is defined with an unpack function to parse its value
from opensm configuration file.
some params require special post update operation to apply them.
every
Sasha Khapyorsky wrote:
It follows port order routing load balancer improvements
(implemented using --guid_routing_order_file command line option).
The idea of this patch is about default behavior and it is to balance
routing paths in such order that most loaded links enter balancer first
-
Sasha Khapyorsky wrote:
Hi Eli,
On 16:02 Tue 20 Jan , Eli Dorfman (Voltaire) wrote:
make setup and destroy routing engines functions global.
change setup_routing_engines() and destroy_routing_engines()
declaration
How is it related to configuration update? Is it?
I cannot see
Sasha Khapyorsky wrote:
Hi Eli,
On 16:04 Tue 20 Jan , Eli Dorfman (Voltaire) wrote:
put qos options flat below subnet opt
put all qos option parameters (default, ca, sw, router) flat below subnet
opt
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm
Sasha Khapyorsky wrote:
On 15:56 Tue 20 Jan , Eli Dorfman (Voltaire) wrote:
The following patches are handling subnet configuration update.
Subnet configuration parameters are rescanned every heavy sweep and if
possible are updated.
Patches 3 and 4 doesn't compile. Patch 5 doesn't
The following patches are handling subnet configuration update.
Subnet configuration parameters are rescanned every heavy sweep and if possible
are updated.
___
general mailing list
general@lists.openfabrics.org
make setup and destroy routing engines fucntions global.
change setup_routing_engines() and destroy_routing_engines()
declaration
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_opensm.h | 53
opensm/opensm/osm_opensm.c
rescan subnet configuration after SIGHUP
call osm_subn_rescan_conf_files() after SIGHUP.
this is important when priority is changed and SM is in standby.
in that case it will not send capability mask trap and will not become master.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
save log_max_size in subnet opt in MB
the max_size in the log object is converted to bytes.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/main.c|5 ++---
opensm/opensm/osm_log.c |2 +-
2 files changed, 3 insertions(+), 4 deletions(-)
diff --git
support subnet configuration rescan and update
subnet configuration parameters are rescanned every heavy sweep.
every parameter is parsed by parse function according to its type.
some params require special post update function to setup them.
every parameter has also a flag that specifies
Sasha Khapyorsky wrote:
On 16:32 Mon 26 Jan , Eli Dorfman (Voltaire) wrote:
rescan subnet configuration after SIGHUP
call osm_subn_rescan_conf_files() after SIGHUP.
this is important when priority is changed and SM is in standby.
in that case it will not send capability mask trap
Sasha Khapyorsky wrote:
On 11:28 Tue 03 Feb , Eli Dorfman (Voltaire) wrote:
Sasha Khapyorsky wrote:
On 16:32 Mon 26 Jan , Eli Dorfman (Voltaire) wrote:
rescan subnet configuration after SIGHUP
call osm_subn_rescan_conf_files() after SIGHUP.
this is important when priority
Sasha Khapyorsky wrote:
On 15:43 Tue 03 Feb , Eli Dorfman (Voltaire) wrote:
This probably can work with updn and minhops, but it certainly will be
destructive when LASH routing engine is used. LASH stores internal data
between sweep cycles, it is used to answer correct SL value in SA
Sasha Khapyorsky wrote:
Hi Eli,
On 16:11 Tue 03 Feb , Eli Dorfman (Voltaire) wrote:
rescan configuration as first step on every heavy sweep
this is a must in case of priority change (increase) for standby SM
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm
enable log_max_size opt update
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_subnet.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/opensm/opensm/osm_subnet.c b/opensm/opensm/osm_subnet.c
index f589180..d6d39a6 100644
---
fix parse functions for big endian machines
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_subnet.c | 10 +-
1 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/opensm/opensm/osm_subnet.c b/opensm/opensm/osm_subnet.c
index d6d39a6..7b33659 100644
---
Sasha Khapyorsky wrote:
Hi Eli,
On 21:23 Sun 08 Feb , Eli Dorfman wrote:
yes, but wouldn't it be better to separate between heavy sweep and
config rescan (due to SIGHUP).
SIGHUP main purpose always was to trigger heavy sweep.
I think that user should know when configuration is
Sasha Khapyorsky wrote:
On 15:47 Mon 09 Feb , Eli Dorfman (Voltaire) wrote:
Sasha Khapyorsky wrote:
Hi Eli,
On 21:23 Sun 08 Feb , Eli Dorfman wrote:
yes, but wouldn't it be better to separate between heavy sweep and
config rescan (due to SIGHUP).
SIGHUP main purpose always
Command Line Arguments:
Log File: /var/log/opensm.log
-
OpenSM 3.3.0_c4d9bcf
Entering DISCOVERING state
Using default GUID 0x2c9020022f019
Loading Cached Option:qos_vlarb_high =
0:4,1:0,2:0,3:0,4:0,5:0,6:0,7:0,8:0,9:0,10:0,11:0,12:0,13:0,14:0
add setup function to routing engine.
call it only when we want to use this routing engine.
this will save allocation for routing algorithms that are not used.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_opensm.h |2 ++
1 files changed, 2 insertions(+), 0
setup routing engine when in use and delete when fail
setup routing engine before use.
delete resources when routing algorithm fails
this will save allocation for routing algorithms that are not used.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_opensm.c| 20
support routing engine update.
add prev routing engine list.
save active routing engine list as prev routing engine list.
this is used to cleanup used routing engine allocation if needed
and only after new routing engine was configured.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
support routing engine update.
save active routing engine list as prev routing engine list.
this is used to cleanup used routing engine allocation if needed
and only after new routing engine was configured.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_opensm.c|9
set IS_SM in local capability mask
Set IS_SM in local portinfo capability mask when
setting it on the SMA.
This fixes a problem of SM that enters to standby and has
mismatch between local capability (IS_SM off) and the SMA.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only after local sm
(previous master) is in standby and will not query its port again.
Signed-off-by: Eli Dorfman
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change local smlid only after local sm
(previous master) is in standby and will not query its port again.
Update our smlid with master
send trap144 when local priority is higher than master priority
send trap144 when local(standby) sm priority is higher
than master sm priority.
this will trigger heavy sweep by the master and handover of mastership.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_sminfo_rcv.c
[PATCH] fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only after local sm
(previous master) is in standby and will not query its port again.
Signed-off-by: Eli
Hal Rosenstock wrote:
On Thu, Mar 5, 2009 at 7:59 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only after local sm
(previous master) is in standby and will not query its port again.
Signed-off-by: Eli Dorfman
Hi,
Can we revert commit 2dbb8b95d9dc27423a6fdb85d88ef385ecee0005 ?
remove c99 definitions within the ib_mad_f structure
to enable better portability with WinOF.
Using the definitions makes the code clearer and safer to maintain.
Aren't there better alternative, such as using gcc for example?
Hi Sasha,
Commit 9b8561ac02e5e90c616b58bbbcfe2eb1a25d42fd breaks opensm:
osm_console.o: In function `osm_console':
/tmp/management/opensm/opensm/osm_console.c:1471: undefined reference to
`cio_close'
collect2: ld returned 1 exit status
make[2]: *** [opensm] Error 1
make[2]: Leaving directory
Set IS_SM bit during opensm init
Set IS_SM bit during opensm initalization.
This fixes a problem of SM that enters to standby and has
mismatch between local capability (IS_SM off) and the SMA.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_opensm.c|5 +
Sasha Khapyorsky wrote:
On 17:36 Thu 26 Feb , Eli Dorfman (Voltaire) wrote:
setup routing engine when in use and delete when fail
setup routing engine before use.
delete resources when routing algorithm fails
this will save allocation for routing algorithms that are not used.
Signed
Eli Dorfman (Voltaire) wrote:
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only after local sm
(previous master) is in standby and will not query its port again
Sasha Khapyorsky wrote:
On 16:54 Thu 05 Mar , Eli Dorfman (Voltaire) wrote:
fix local port smlid
when opensm goes from master to standby due to handover its local
port smlid will be wrong (equal to its own lid).
the new master opensm will change smlid only after local sm
(previous
support routing engine update
perform setup of routing engine only before configuration.
save last routing engine before routing engine selection
and cleanup of last if not used any more.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_opensm.h |5 +++
Sasha Khapyorsky wrote:
On 17:58 Thu 12 Mar , Eli Dorfman (Voltaire) wrote:
Instead of touching port_info I would suggest to setup destination lid
using smport which you already have:
madw-mad_addr.dest_lid = smport-p_physp-port_info.base_lid;
Local standby SM is polling the master
add newline to log message
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_opensm.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/opensm/opensm/osm_opensm.c b/opensm/opensm/osm_opensm.c
index 7126658..732f427 100644
--- a/opensm/opensm/osm_opensm.c
Added send trap for trap 129 (local link integrity).
Signed-off-by: Julia Volynsky jul...@voltaire.com
---
infiniband-diags/src/ibsendtrap.c | 45 +++-
1 files changed, 43 insertions(+), 2 deletions(-)
diff --git a/infiniband-diags/src/ibsendtrap.c
This patch fixes a bug in support routing engine update patch.
(Do you want a new patch that includes both?)
It fixes cleanup of last routing engine
Also use routing engine context instead of initialized flag.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/opensm/osm_opensm.h |
Sasha Khapyorsky wrote:
On 10:24 Mon 16 Mar , Eli Dorfman (Voltaire) wrote:
snip...
diff --git a/opensm/include/opensm/osm_opensm.h
b/opensm/include/opensm/osm_opensm.h
index c121be4..d297dd8 100644
--- a/opensm/include/opensm/osm_opensm.h
+++ b/opensm/include/opensm/osm_opensm.h
add c99 definitions to ib_mad_f
Signed-off-by: Eli Dorfman e...@voltaire.com
---
libibmad/src/fields.c | 525 -
1 files changed, 262 insertions(+), 263 deletions(-)
diff --git a/libibmad/src/fields.c b/libibmad/src/fields.c
index 19c8fc1..c142f39
Sean Hefty wrote:
this knowingly breaks the windows build...
So what do you suggest?
Ira replied that he has no problems with it.
___
general mailing list
general@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To
setup routing engine when in use and delete when fail
setup routing engine and allocate resources before use.
delete resources when routing algorithm fails.
this will save allocation for routing algorithms that are not used.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
support routing engine update.
save used routing engine as last routing engine.
destroy old routing engine list and allocate new routing list.
cleanup used routing engine allocation if needed
and only after new routing engine was configured.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
Nicolas Morey Chaisemartin wrote:
Hi,
We've noticed while setting up a new cluster a problem with OpenSM.
As usual, there are some cable problems while plugging the cluster but one of
the cable was changing state over 10 000 thousands times per second (OFF/ON)
and sending each time a 128
Hal Rosenstock wrote:
On Sun, Mar 29, 2009 at 8:18 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
snip...
You can change sweep_on_trap option in opensm.conf to FALSE.
This should stop opensm heavy sweeps.
Yes but is that what is really desired ? If sweep on trap is disabled
Sasha Khapyorsky wrote:
On 15:08 Mon 30 Mar , Doron Shoham wrote:
support xmitwait counters
Signed-off-by: Doron Shoham dor...@voltaire.com
Applied. Thanks.
What are the plans to use it (now unlike other counters XmitWait will
be always zero)?
We have plans to add new commands to
fix wrong padding for SA portinfo record after addition
of max_credit_hint and link_rt_latency to SM portinfo
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/include/iba/ib_types.h |1 -
1 files changed, 0 insertions(+), 1 deletions(-)
diff --git a/opensm/include/iba/ib_types.h
Sasha Khapyorsky wrote:
Hi Eli,
On 15:37 Thu 19 Mar , Eli Dorfman (Voltaire) wrote:
setup routing engine when in use and delete when fail
setup routing engine and allocate resources before use.
delete resources when routing algorithm fails.
this will save allocation for routing
Sasha Khapyorsky wrote:
On 14:01 Sun 26 Apr , Eli Dorfman (Voltaire) wrote:
No.
routing_engine-setup is called by osm_ucast_mgr_process()
and if routing algorithm fails then delete is called (and is already set).
Right. And after all on OpenSM exit destroy_routing_engines() is called
Sasha Khapyorsky wrote:
On 14:48 Sun 26 Apr , Eli Dorfman (Voltaire) wrote:
delete() is called conditionally from destroy_routing_engines(osm_opensm_t
*osm)
if (r-delete)
r-delete(r-context);
Also all re(s) are cleared when created so delete is NULL if setup
Hal Rosenstock wrote:
On Tue, May 5, 2009 at 9:00 AM, Doron Shoham dor...@voltaire.com wrote:
when setting max_op_vls = 0
do not force it to 1.
0 is valid value which means No change
Signed-off-by: Doron Shoham dor...@voltaire.com
---
opensm/opensm/osm_port.c |6 --
Hal Rosenstock wrote:
On Tue, May 5, 2009 at 9:48 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Hal Rosenstock wrote:
On Tue, May 5, 2009 at 9:00 AM, Doron Shoham dor...@voltaire.com wrote:
when setting max_op_vls = 0
do not force it to 1.
0 is valid value which means No change
opensm assigns conflicting LIDs to node after lmc change (e.g. 0 to 1)
when node guid is in the guid2lid cache.
In the following example CA port 1 lid 24 lmc 1
and switch lid is 25 which overlaps with CA port's lid.
This happens because switch port guid is in the guid2lid cache
lid persistent range wrong check
used lids were not properly chekced which
caused duplicate lid assignment in some cases.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_lid_mgr.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git
Hal Rosenstock wrote:
On Sat, May 9, 2009 at 7:31 AM, Eli Dorfman dorfman@gmail.com wrote:
On Sat, May 9, 2009 at 1:41 PM, Hal Rosenstock hal.rosenst...@gmail.com
wrote:
On Sat, May 9, 2009 at 6:32 AM, Eli Dorfman dorfman@gmail.com wrote:
On Fri, May 8, 2009 at 4:57 PM, Hal
Hal Rosenstock wrote:
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
---
diff --git a/opensm/opensm/osm_port.c b/opensm/opensm/osm_port.c
index 2e6c642..17bac73 100644
--- a/opensm/opensm/osm_port.c
+++ b/opensm/opensm/osm_port.c
@@ -381,7 +381,7 @@ uint8_t
Doron Shoham wrote:
check if op_vls = 0 before max_op_vls comparison
Signed-off-by: Doron Shoham dor...@voltaire.com
---
opensm/opensm/osm_port.c |9 +
1 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/opensm/opensm/osm_port.c b/opensm/opensm/osm_port.c
index
Hi,
What could be the reason that open(/dev/infiniband/umad0, O_RDWR|O_NONBLOCK)
blocks and does not return.
I did not find any errors in dmesg.
Thanks,
Eli
___
general mailing list
general@lists.openfabrics.org
MFT tables are not set after non full member re-join
In case of non full member re-join MFT tables are not set.
No need to set or check non full member reference to mlid (port-mcm_list).
This list should be used only for full members for cleanup when port goes down.
A simple scenarion to
Eli Dorfman (Voltaire) wrote:
MFT tables are not set after non full member re-join
In case of non full member re-join MFT tables are not set.
No need to set or check non full member reference to mlid (port-mcm_list).
This list should be used only for full members for cleanup when port goes
Do not change logical state on SubnAdmSet
When changing physical state do not change logical port state.
From the IB spec When writing PortInfo:PortState, only legal transitions are
valid. So if PortState is ACTIVE and we try to set it to ACTIVE this will fail.
This patch allows reset in a
Hal Rosenstock wrote:
On Sun, May 31, 2009 at 10:44 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Do not change logical state on SubnAdmSet
The method is SubnSet r.t. SubnAdmSet. Subject might also indicate
ibportstate.c and it would be clearer saying logical port state r.t.
just
Hal Rosenstock wrote:
On Mon, Jun 1, 2009 at 5:36 PM, Hal Rosenstock hal.rosenst...@gmail.com
wrote:
On Mon, Jun 1, 2009 at 4:27 PM, Sean Hefty sean.he...@intel.com wrote:
Yes, RMPP is an overhead when the response is a single MAD but is this
significant ? Anyhow, how can the spec be changed
Eli Dorfman (Voltaire) wrote:
Hal Rosenstock wrote:
On Mon, Jun 1, 2009 at 5:36 PM, Hal Rosenstock hal.rosenst...@gmail.com
wrote:
On Mon, Jun 1, 2009 at 4:27 PM, Sean Hefty sean.he...@intel.com wrote:
Yes, RMPP is an overhead when the response is a single MAD but is this
significant
Return single PR record for SubnAdmGet
Fix to return single PathRecord for SubnAdmGet when SGID and/or DGID
are 0 in SA component mask (wildcard)
Also avoid iterating beyond requested number of paths.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_sa_path_record.c | 29
Hal Rosenstock wrote:
On Wed, Jun 3, 2009 at 7:01 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Hal Rosenstock wrote:
On Mon, Jun 1, 2009 at 5:36 PM, Hal Rosenstock hal.rosenst...@gmail.com
wrote:
On Mon, Jun 1, 2009 at 4:27 PM, Sean Hefty sean.he...@intel.com wrote:
Yes, RMPP
Hal Rosenstock wrote:
On Wed, Jun 3, 2009 at 3:28 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Hal Rosenstock wrote:
On Sun, May 31, 2009 at 10:44 AM, Eli Dorfman (Voltaire)
dorfman@gmail.com wrote:
Do not change logical state on SubnAdmSet
The method is SubnSet r.t
Do not change port state on SubnSet
When changing physical state do not change logical port state to avoid invalid
transitions in port state machine.
From the IB spec When writing PortInfo:PortState, only legal transitions are
valid. So if PortState is ACTIVE and we try to set it to ACTIVE this
add attribute and component mask to error message
about too many records for SubnAdmGet query
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_sa.c |5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/opensm/opensm/osm_sa.c b/opensm/opensm/osm_sa.c
index
use comp mask to match by LID or GUID
when matching node records use comp mask to filter
queries that use 0 lid or guid while mask is set
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_sa_node_record.c | 11 ++-
1 files changed, 6 insertions(+), 5 deletions(-)
perform reset in a single mad
Perform reset by changing physical state to polling instead of
disable and enable.
This ensures that reset a port will not disconnect it (assuming negotiation
succeeds).
Only disable operation is not allowed on HCA ports since it will irreversibly
disconnect it
Hi Sasha,
Resending. I haven't seen a response to this patch.
Eli
Original Message
Subject: Re: [PATCH ] opensm: MFT tables are not set after non full member
re-join
Date: Tue, 19 May 2009 11:56:36 +0300
From: Eli Dorfman (Voltaire) dorfman@gmail.com
To: Sasha Khapyorsky
Resending.
Original Message
Subject: [PATCH] opensm/osm_sa.c add attribute and component mask to error
message
Date: Thu, 04 Jun 2009 15:30:34 +0300
From: Eli Dorfman (Voltaire) dorfman@gmail.com
To: Sasha Khapyorsky sas...@voltaire.com
CC: OpenIB general
1 - 100 of 114 matches
Mail list logo
