]> git.openfabrics.org - ~shefty/rdma-win.git/log
~shefty/rdma-win.git
14 years agolibibverbs/device: destroy completion channel when closing device
Sean Hefty [Fri, 29 Jan 2010 05:07:01 +0000 (05:07 +0000)]
libibverbs/device: destroy completion channel when closing device

From: Sean Hefty <sean.hefty@intel.com>

ibv_close_device should destroy the completion channel
when closing the device and freeing the memory.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2689 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodapl: use private_data_len for mem copies
Sean Hefty [Fri, 29 Jan 2010 05:06:11 +0000 (05:06 +0000)]
dapl: use private_data_len for mem copies

From: Sean Hefty <sean.hefty@intel.com>

When copying private_data out of rdma_cm events, use the
reported private_data_len for the size, and not IB maximums.
This fixes a bug running over the librdmacm on windows, where
DAPL accessed invalid memory.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2688 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodapl/cma: fix referencing freed address
Sean Hefty [Fri, 29 Jan 2010 05:05:03 +0000 (05:05 +0000)]
dapl/cma: fix referencing freed address

From: Sean Hefty <sean.hefty@intel.com>

DAPL uses a pointer to reference the local and remote addresses
of an endpoint.  It expects that those addresses are located
in memory that is always accessible.  Typically, for the local
address, the pointer references the address stored with the DAPL
HCA device.  However, for the cma provider, it changes this pointer
to reference the address stored with the rdma_cm_id.

This causes a problem when that endpoint is connected on the
passive side of a connection.  When connect requests are given
to DAPL, a new rdma_cm_id is associated with the request.  The
DAPL code replaces the current rdma_cm_id associated with a
user's endpoint with the new rdma_cm_id.  The old rdma_cm_id is
then deleted.  But the endpoint's local address pointer still
references the address stored with the old rdma_cm_id.  The
result is that any reference to the address will access freed
memory.

Fix this by keeping the local address pointer always pointing
to the address associated with the DAPL HCA device.  This is about
the best that can be done given the DAPL interface design.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2687 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agolibrdmacm: set private_data_len
Sean Hefty [Fri, 29 Jan 2010 05:04:08 +0000 (05:04 +0000)]
librdmacm: set private_data_len

From: Sean Hefty <sean.hefty@intel.com>

Set the private_data_len for reported events.  This allows
DAPL to use the value when copying private data.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2686 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodapl: move close device after async thread is done using it
Sean Hefty [Fri, 29 Jan 2010 04:58:32 +0000 (04:58 +0000)]
dapl: move close device after async thread is done using it

Before calling ibv_close_device, wait for the asynchronous
processing thread to finish using the device.  This prevents
a use after free error.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2685 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DAPL2]
Stan Smith [Tue, 26 Jan 2010 20:31:36 +0000 (20:31 +0000)]
[DAPL2]
libibverbs/device: destroy completion channel when closing device
ibv_close_device should destroy the completion channel
when closing the device and freeing the memory.
 trunk/ulp/libibverbs/src/device.cpp

dapl: move close device after async thread is done using it
Before calling ibv_close_device, wait for the asynchronous
processing thread to finish using the device.  This prevents a use after free error.
 trunk/ulp/dapl2/dapl/openib_cma/device.c
 trunk/ulp/dapl2/dapl/openib_scm/device.c

librdmacm: set private_data_len
Set the private_data_len for reported events.  This allows DAPL to use the value when copying private data.
 trunk/ulp/librdmacm/include/rdma/rdma_cma.h
 trunk/ulp/librdmacm/src/cma.cpp

dapl: quick fix for wrong private data size
DAPL expects the private data size to be up to 256 bytes, but
on windows the private data size is limited to 56 bytes.  As
a result, DAPL can access memory beyond the end of what's
allocated.
A more 'correct' fix is being submitted upstream to dapl.  This
is a simpler fix for the Windows 2.2 release only.
 trunk/ulp/dapl2/dapl/openib_cma/cm.c

dapl/cma: fix referencing freed address
DAPL uses a pointer to reference the local and remote addresses
of an endpoint.  It expects that those addresses are located
in memory that is always accessible.  Typically, for the local
address, the pointer references the address stored with the DAPL
HCA device.  However, for the cma provider, it changes this pointer
to reference the address stored with the rdma_cm_id.

This causes a problem when that endpoint is connected on the
passive side of a connection.  When connect requests are given
to DAPL, a new rdma_cm_id is associated with the request.  The
DAPL code replaces the current rdma_cm_id associated with a
user's endpoint with the new rdma_cm_id.  The old rdma_cm_id is
then deleted.  But the endpoint's local address pointer still
references the address stored with the old rdma_cm_id.  The
result is that any reference to the address will access freed
memory.
Fix this by keeping the local address pointer always pointing
to the address associated with the DAPL HCA device.  This is about
the best that can be done given the DAPL interface design.
 trunk/ulp/dapl2/dapl/openib_common/qp.c

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2684 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[core] Improved error message. (mlnx 5511)
Tzachi Dar [Tue, 26 Jan 2010 10:09:53 +0000 (10:09 +0000)]
[core] Improved error message. (mlnx 5511)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2683 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[core] Release the RDMA interface if ib_register_ca fails. (mlnx 5510)
Tzachi Dar [Tue, 26 Jan 2010 10:01:33 +0000 (10:01 +0000)]
[core] Release the RDMA interface if ib_register_ca fails. (mlnx 5510)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2682 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] RC3 prep
Stan Smith [Fri, 22 Jan 2010 23:19:06 +0000 (23:19 +0000)]
[WINOF] RC3 prep

git-svn-id: svn://openib.tc.cornell.edu/gen1@2681 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINVERBS] winverbs/lib: cache pkey table.
Stan Smith [Fri, 22 Jan 2010 23:11:37 +0000 (23:11 +0000)]
[WINVERBS] winverbs/lib: cache pkey table.
Connection establishment rates take a huge performance hit
as a result of using FindPkey, which results in querying the
HCA to lookup current pkey information.  Add a cache to the
userspace library for pkey information.

The cache is updated using the CWVDevice:Notify() routine.
Before any access is made to the cache, the caller checks
for changes to the partition tables.  If a change is found,
then the cache is updated.  Otherwise, the cache is accessed
to map pkey values to indices.

This results in the connection rate over winverbs almost doubling.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2680 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINVERBS} winverbs/ep: handle receiving REQ then DREQ.
Stan Smith [Fri, 22 Jan 2010 23:08:40 +0000 (23:08 +0000)]
[WINVERBS} winverbs/ep: handle receiving REQ then DREQ.
For fast connections, it's possible to receive a DREQ immediately
after receiving a REQ, without an RTU coming in between.  If we've
sent a REP to the REQ, then the DREQ should be treated as if the
connection had been fully established.  (The RTU could be delayed,
and the communication established event is processed asynchronously,
so there's no way to tell for certain.)

This fixes an issue where the passive side Accept() call fails
waiting for the RTU, but receives a DREQ instead.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2679 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[CORE] ib/mad: fix routing of vendor mads.
Stan Smith [Fri, 22 Jan 2010 23:05:49 +0000 (23:05 +0000)]
[CORE] ib/mad: fix routing of vendor mads.
SVN commit 2174 introduced an error that resulted in all
vendor MADs being routed to the local HCA driver.

This results in the ib-diag vendstat failing to receive
a response when trying to gather statistics about a remote
device.

We should only route vendor mads to the local HCA if the
mad is one of the mellanox vendor classes, the mad is not
a response, and the local HCA is the destination for the
mad.  Simplify both the send and receive processing to
be consistent in vendor mad handling.

Problem reported by Mohammad Sawalha.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2678 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[LIBIBUMAD] libibumad: fix freeing wrong memory on failure
Stan Smith [Fri, 22 Jan 2010 23:02:38 +0000 (23:02 +0000)]
[LIBIBUMAD] libibumad: fix freeing wrong memory on failure
umad_get_ca() should free the allocated 'ports' memory on failure,
not the ca structure, which is provided by the user.

Bug reported by Leonid Keller.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2677 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs branch: test author fix
Sean Hefty [Thu, 21 Jan 2010 06:25:34 +0000 (06:25 +0000)]
winverbs branch: test author fix

From: Sean Hefty <sean.hefty@intel.com>

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2676 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoib/mad: fix routing of vendor mads
Sean Hefty [Thu, 21 Jan 2010 06:09:33 +0000 (06:09 +0000)]
ib/mad: fix routing of vendor mads

SVN commit 2174 introduced an error that resulted in all
vendor MADs being routed to the local HCA driver.

This results in the ib-diag vendstat failing to receive
a response when trying to gather statistics about a remote
device.

We should only route vendor mads to the local HCA if the
mad is one of the mellanox vendor classes, the mad is not
a response, and the local HCA is the destination for the
mad.

Problem reported by Mohammad Sawalha.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2675 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/ep: handle receiving REQ then DREQ
Sean Hefty [Thu, 21 Jan 2010 06:09:19 +0000 (06:09 +0000)]
winverbs/ep: handle receiving REQ then DREQ

For fast connections, it's possible to receive a DREQ immediately
after receiving a REQ, without an RTU coming in between.  If we've
sent a REP to the REQ, then the DREQ should be treated as if the
connection had been fully established.  (The RTU could be delayed,
and the communication established event is processed asynchronously,
so there's no way to tell for certain.)

This fixes an issue where the passive side Accept() call fails
waiting for the RTU, but receives a DREQ instead.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2674 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodocs/interfaces.txt: add documentation to describe interface differences
Sean Hefty [Thu, 21 Jan 2010 06:09:06 +0000 (06:09 +0000)]
docs/interfaces.txt: add documentation to describe interface differences

Provide a simple documentation that defines the main userspace
verbs interfaces and the differences between them.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2673 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/lib: cache pkey table
Sean Hefty [Thu, 21 Jan 2010 06:08:53 +0000 (06:08 +0000)]
winverbs/lib: cache pkey table

Connection establishment rates take a huge performance hit
as a result of using FindPkey, which results in querying the
HCA to lookup current pkey information.  Add a cache to the
userspace library for pkey information.

The cache is updated using the CWVDevice:Notify() routine.
Before any access is made to the cache, the caller checks
for changes to the partition tables.  If a change is found,
then the cache is updated.  Otherwise, the cache is accessed
to map pkey values to indices.

This results in the connection rate over winverbs almost
doubling.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2672 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agolibibumad: fix freeing wrong memory on failure
Sean Hefty [Thu, 21 Jan 2010 06:08:38 +0000 (06:08 +0000)]
libibumad: fix freeing wrong memory on failure

umad_get_ca() should free the allocated 'ports' memory on failure,
not the ca structure, which is provided by the user.

Bug reported by Leonid Keller.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2671 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINVERBS,ND,WINMAD,MAD] delete critical sections when no longer needed.
Stan Smith [Fri, 15 Jan 2010 23:13:42 +0000 (23:13 +0000)]
[WINVERBS,ND,WINMAD,MAD] delete critical sections when no longer needed.
To avoid leaking memory, we need to call DeleteCriticalSection
for all critical sections once they are no longer needed.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2670 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs: delete critical sections when no longer needed
Sean Hefty [Fri, 15 Jan 2010 17:14:46 +0000 (17:14 +0000)]
winverbs: delete critical sections when no longer needed

To avoid leaking memory, we need to call DeleteCriticalSection
for all critical sections once they are no longer needed.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2669 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoetc/getopt: allow use of '/' for command line args
Sean Hefty [Fri, 15 Jan 2010 17:14:20 +0000 (17:14 +0000)]
etc/getopt: allow use of '/' for command line args

Windows uses '/' to specify command line arguments.  Change
getopt to allow using '-' or '/'.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2668 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] RC2 prep.
Stan Smith [Wed, 13 Jan 2010 22:18:46 +0000 (22:18 +0000)]
[WINOF] RC2 prep.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2667 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[ND/WINVERBS] winverbs/nd: map completion status to ND value convert flushed completi...
Stan Smith [Wed, 13 Jan 2010 22:12:21 +0000 (22:12 +0000)]
[ND/WINVERBS] winverbs/nd: map completion status to ND value convert flushed completion status values
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2666 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/nd: map completion status to ND values
Sean Hefty [Wed, 13 Jan 2010 19:34:38 +0000 (19:34 +0000)]
winverbs/nd: map completion status to ND values

Convert flushed completion status values

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2665 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/netdirect: add support for inline sends
Sean Hefty [Wed, 13 Jan 2010 19:34:27 +0000 (19:34 +0000)]
winverbs/netdirect: add support for inline sends

By default, use inline sends, and allow user to limit
inline send size through an environment variable.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2664 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoib/cm: poll for CM REQ events
Sean Hefty [Wed, 13 Jan 2010 19:34:17 +0000 (19:34 +0000)]
ib/cm: poll for CM REQ events

Replace the callback mechanism for reporting connection
requests with one that requires the user to poll for the
events.  This allows queuing REQs in the CM until the user
is ready to process the events.

Still provide a callback mechanism to notify the user that
REQ events are ready to be retrieved.

This change improves the connection rate for winverbs when
the user retrieves only a small number of requests at a time.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2663 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoib/cm: separate listen callback into own handler
Sean Hefty [Wed, 13 Jan 2010 19:34:07 +0000 (19:34 +0000)]
ib/cm: separate listen callback into own handler

Extract listen handling in cm_cep_handler into a separate
handler.  This will simplify changes to support queuing
connection requests and allow connection requests to move
to a fetch model, rather than a callback model.

Since connection requests require allocating new resources,
connection rates can be improved by queuing the requests
until the user has allocated the resources, then fetches
the request.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2662 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoib/cm: fix handling of REJ in REQ_RCVD state
Sean Hefty [Wed, 13 Jan 2010 19:33:56 +0000 (19:33 +0000)]
ib/cm: fix handling of REJ in REQ_RCVD state

If a REJ is received immediately after receiving a REQ,
then a callback can be queued to the user for the REJ before
the user has had a chance to see or process the REQ.  The
result is that the user will not have had a chance to set
any context associated with the REQ, which can lead to a
crash processing the REJ.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2661 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[NETDIRECT] [PATCH 4/4] winverbs/netdirect: add support for inline sends
Stan Smith [Wed, 13 Jan 2010 19:04:32 +0000 (19:04 +0000)]
[NETDIRECT] [PATCH 4/4] winverbs/netdirect: add support for inline sends
By default, use inline sends, and allow user to limit
inline send size through an environment variable.

To support existing deployments, winverbs make use of the same
environment setting as the ibal ND provider.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
---
Note: The kernel HCA drivers report 0 for the max inline send
value for user space QPs, even if inline sends are enabled.  I didn't
understand how to fix the kernel drivers, so they could report the
correct value.  The winverbs ND code simply returns the value that it
uses when creating the QP.

 trunk/ulp/netdirect/user/nd_ep.cpp
 trunk/ulp/netdirect/user/nd_ep.h

git-svn-id: svn://openib.tc.cornell.edu/gen1@2660 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[CORE,INC] [PATCH 1/4] ib/cm: fix handling of REJ in REQ_RCVD state
Stan Smith [Wed, 13 Jan 2010 18:59:42 +0000 (18:59 +0000)]
[CORE,INC] [PATCH 1/4] ib/cm: fix handling of REJ in REQ_RCVD state
If a REJ is received immediately after receiving a REQ,
then a callback can be queued to the user for the REJ before
the user has had a chance to see or process the REQ.  The
result is that the user will not have had a chance to set
any context associated with the REQ, which can lead to a
crash processing the REJ.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
 trunk/core/al/kernel/al_cm_cep.c

[PATCH 2/4] ib/cm: separate listen callback into own handler
Extract listen handling in cm_cep_handler into a separate
handler.  This will simplify changes to support queuing
connection requests and allow connection requests to move
to a fetch model, rather than a callback model.

Since connection requests require allocating new resources,
connection rates can be improved by queuing the requests
until the user has allocated the resources, then fetches
the request.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
trunk/core/al/kernel/al_cm.c

[PATCH 3/4] ib/cm: poll for CM REQ events
Replace the callback mechanism for reporting connection
requests with one that requires the user to poll for the
events.  This allows queuing REQs in the CM until the user
is ready to process the events.

Still provide a callback mechanism to notify the user that
REQ events are ready to be retrieved.

This change improves the connection rate for winverbs when
the user retrieves only a small number of requests at a time.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
trunk/core/al/kernel/al_cm.c
 trunk/core/al/kernel/al_cm_cep.c
 trunk/core/winverbs/kernel/wv_ep.c
 trunk/inc/kernel/iba/ib_cm_ifc.h

git-svn-id: svn://openib.tc.cornell.edu/gen1@2659 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB_NDIS6_CM] fix the improper reporting of IPoIB port rate and removes obsolete...
Stan Smith [Wed, 13 Jan 2010 18:40:25 +0000 (18:40 +0000)]
[IPOIB_NDIS6_CM] fix the improper reporting of IPoIB port rate and removes obsolete OID_GEN_LINK_SPEED
signed-off by: Alexander Naslednikov (xalex at mellanox.co.il)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2658 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB/IPOIB_NDIS6_CM] fix the problem when IPoIB increments the receive statistics...
Stan Smith [Wed, 13 Jan 2010 17:39:56 +0000 (17:39 +0000)]
[IPOIB/IPOIB_NDIS6_CM] fix the problem when IPoIB increments the receive statistics twice.
Signed-off by: Tzachi Dar (tzachid at mellanox.co.il)
Signed-off by: Alex Naslednikov (xalex@mellanox.co.il)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2657 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB] install the ND/winverbs (wvndprov.dll) provider along with the ND/ibal provider.
Stan Smith [Wed, 13 Jan 2010 00:39:37 +0000 (00:39 +0000)]
[IPOIB] install the ND/winverbs (wvndprov.dll) provider along with the ND/ibal provider.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2656 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] 2.2 RC1 prep.
Stan Smith [Thu, 7 Jan 2010 22:41:27 +0000 (22:41 +0000)]
[WINOF] 2.2 RC1 prep.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2655 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] whitespace cleanup & remove unused SM startup code.
Stan Smith [Thu, 7 Jan 2010 22:36:45 +0000 (22:36 +0000)]
[WINOF] whitespace cleanup & remove unused SM startup code.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2654 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINVERBS] winverbs: use separate structure to queue async work
Stan Smith [Tue, 5 Jan 2010 19:05:28 +0000 (19:05 +0000)]
[WINVERBS] winverbs: use separate structure to queue async work
KMDF tracks all requests that pass through an IO queue.  Even after a
request has been removed from the queue, the request still maintains a
reference on the queue.  Any attempt to delete the queue will block
until all requests holding references on the queue have completed.

To avoid deadlock conditions during cleanup, we need to ensure that
all requests can complete during cleanup.  Modify the asynchronous
handling code to queue a separate data structure, so that all requests
can remain on the IO queues.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2653 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IBAL] ib/cm: fix handling failed send completions
Stan Smith [Tue, 5 Jan 2010 18:59:21 +0000 (18:59 +0000)]
[IBAL] ib/cm: fix handling failed send completions
__cep_mad_send_cb() assumes that the mad being processed is
associated with the current state of the CEP.  This may not be
the case.
For example, for a short lived connection, it was observed that
a REP mad completed with status canceled.  This is normal.  However,
the user already attempted to disconnect the connection by sending
a DREQ.  This left the cep in the DREQ_SENT state by the time that
the REP mad completed.  Since the REP status was not success, but the
state was DREQ_SENT, the code assumed that the DREQ had failed and
transitioned the cep into TIMEWAIT.  The result is that the DREQ is never
>matched with a DREP or canceled, but holds a reference on the CEP.
Until the DREQ times out (time depends on connection, but easily
up to a minute), attempts to destroy the CEP are blocked.
Fix this by simply discarding any completed sends that were not
sent from the current state of the cep when the completion handler
is invoked.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2652 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB_NDIS6_CM] fix the wrong length calculation of
Stan Smith [Tue, 5 Jan 2010 18:51:23 +0000 (18:51 +0000)]
[IPoIB_NDIS6_CM] fix the wrong length calculation of
first MDL within Net Buffer object.
The actual length of first MDL should be incremented by it's offset
signed-off by: Alexander Naslednikov (xalex at mellanox.co.il)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2651 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoib/cm: fix handling failed send completions
Sean Hefty [Mon, 4 Jan 2010 19:01:09 +0000 (19:01 +0000)]
ib/cm: fix handling failed send completions

__cep_mad_send_cb() assumes that the mad being processed is
associated with the current state of the CEP.  This may not be
the case.

For example, for a short lived connection, it was observed that
a REP mad completed with status canceled.  This is normal.  However,
the user already attempted to disconnect the connection by sending
a DREQ.  This left the cep in the DREQ_SENT state by the time that
the REP mad completed.  Since the REP failed, but the state was
DREQ_SENT, the code assumed that the DREQ had failed and transitioned
the cep into TIMEWAIT.  The result is that the DREQ is never
matched with a DREP or canceled, but holds a reference on the CEP.

Until the DREQ times out (time depends on connection, but easily
up to a minute), attempts to destroy the CEP are blocked.

Fix this by simply discarding any completed sends that were not
sent from the current state of the cep when the completion handler
is invoked.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2650 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoKMDF tracks all requests that pass through an IO queue. Even after a
Sean Hefty [Mon, 4 Jan 2010 19:00:54 +0000 (19:00 +0000)]
KMDF tracks all requests that pass through an IO queue.  Even after a
request has been removed from the queue, the request still maintains a
reference on the queue.  Any attempt to delete the queue will block
until all requests holding references on the queue have completed.

To avoid deadlock conditions during cleanup, we need to ensure that
all requests can complete during cleanup.  Modify the asynchronous
handling code to queue a separate data structure, so that all requests
can remain on the IO queues.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2649 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoetc/bld: allow specifying x32 for x86 builds
shefty [Mon, 4 Jan 2010 19:00:38 +0000 (19:00 +0000)]
etc/bld: allow specifying x32 for x86 builds

To make it easier on the user, allow specifying a 32-bit build
using x32 or x86.  x32 is more natural with x64 being the
64-bit build.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2648 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[mlx4_bus] Add support for RoCEE to the low level driver.
tzachid [Mon, 4 Jan 2010 17:34:19 +0000 (17:34 +0000)]
[mlx4_bus] Add support for RoCEE to the low level driver.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2647 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DOCS] manual.htm - Corrected uDAT/uDAPL discussion as uDAPL v1.1 provider is no...
stansmith [Tue, 29 Dec 2009 21:31:52 +0000 (21:31 +0000)]
[DOCS] manual.htm - Corrected uDAT/uDAPL discussion as uDAPL v1.1 provider is no longer supported. Fixed lingering %ProgramFiles(x86)% install path problems; removed (x86) as WinOF is installed as %ProgramFIles%\ on all systems. Updated SRP discussion to include a how to configure a Linx/OFED SRP target (vdisk+BLOCKIO) accessible from Windows.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2646 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] Temp fix for broken SRP operation; revert back to working code. Real fix...
stansmith [Thu, 24 Dec 2009 00:40:46 +0000 (00:40 +0000)]
[OPENSM] Temp fix for broken SRP operation; revert back to working code. Real fix touches numerous files, will be delayed until after 2.2 release.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2645 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] added syslog() call to record Windows Service control calls (128,129) [clear...
stansmith [Thu, 24 Dec 2009 00:36:08 +0000 (00:36 +0000)]
[OPENSM] added syslog() call to record Windows Service control calls (128,129) [clear log file & start heavy sweep].

git-svn-id: svn://openib.tc.cornell.edu/gen1@2644 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DAPL2] tolerate a disconnect callback racing ahead during EP disconnect(), do not...
stansmith [Thu, 24 Dec 2009 00:26:19 +0000 (00:26 +0000)]
[DAPL2] tolerate a disconnect callback racing ahead during EP disconnect(), do not return an error on invalid handle.
update .rc file to reflect IBAL provider.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2643 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] Install previous OpenSM version 3.0.0 as an emergency SM backup; 2-2 only...
stansmith [Tue, 15 Dec 2009 20:53:37 +0000 (20:53 +0000)]
[WINOF] Install previous OpenSM version 3.0.0 as an emergency SM backup; 2-2 only. Reference/install OpenSM related files from a single file instead of OpenSM_service.inc & tools.inc.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2642 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] reference/install OpenSM related files from a single file instead of OpenSM_s...
stansmith [Tue, 15 Dec 2009 20:48:55 +0000 (20:48 +0000)]
[WINOF] reference/install OpenSM related files from a single file instead of OpenSM_service.inc & tools.inc.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2641 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] added OpenSM version to osmtest.rc file description string for easy ID....
stansmith [Tue, 15 Dec 2009 20:39:53 +0000 (20:39 +0000)]
[OPENSM] added OpenSM version to osmtest.rc file description string for easy ID. osmtest.rc was not being compiled in, added osmtest.rc file to SOURCES file.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2640 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] version update
stansmith [Mon, 14 Dec 2009 19:46:03 +0000 (19:46 +0000)]
[WINOF] version update

git-svn-id: svn://openib.tc.cornell.edu/gen1@2639 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] 2.2 RC0 setup
stansmith [Mon, 14 Dec 2009 19:44:20 +0000 (19:44 +0000)]
[WINOF] 2.2 RC0 setup

git-svn-id: svn://openib.tc.cornell.edu/gen1@2638 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] Wof 2.2 build 3.0.0 version of opensm.exe for failback purposes.
stansmith [Mon, 14 Dec 2009 19:31:42 +0000 (19:31 +0000)]
[OPENSM] Wof 2.2 build 3.0.0 version of opensm.exe for failback purposes.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2637 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DAPL] WinOF 2.2 - removed DAPL V1; not sure how this came into the 2.2 branch as...
stansmith [Mon, 14 Dec 2009 19:27:49 +0000 (19:27 +0000)]
[DAPL] WinOF 2.2 - removed DAPL V1; not sure how this came into the 2.2 branch as the branch was off the HEAD and DAPL was removed from HEAD a few weeks ago?

git-svn-id: svn://openib.tc.cornell.edu/gen1@2636 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] Wof 2.2 build 3.0.0 version of opensm.exe for failback purposes.
stansmith [Mon, 14 Dec 2009 19:26:08 +0000 (19:26 +0000)]
[OPENSM] Wof 2.2 build 3.0.0 version of opensm.exe for failback purposes.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2635 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] branch WinOF 2.2 @ svn.2633
stansmith [Mon, 14 Dec 2009 17:44:50 +0000 (17:44 +0000)]
[WINOF] branch WinOF 2.2 @ svn.2633

git-svn-id: svn://openib.tc.cornell.edu/gen1@2634 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] set opensm cache & config folder as %ProgramFiles%\WinOF\OpenSM; integrated...
stansmith [Sat, 12 Dec 2009 00:57:11 +0000 (00:57 +0000)]
[OPENSM] set opensm cache & config folder as %ProgramFiles%\WinOF\OpenSM; integrated with WinOF installers.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2633 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] introduce new cmds 'allnf path' & 'compnf path' to work around a bug in the...
stansmith [Fri, 11 Dec 2009 21:14:34 +0000 (21:14 +0000)]
[WINOF] introduce new cmds 'allnf path' & 'compnf path' to work around a bug in the WDK build env. If one forces a build (compf ulp\ipoib) for IPoIB or IPoIB_NDIS6_CM then 'all' instances of ipoib.sys are deleted and only the specified path component is rebuild leaving an incomplete build. 'compnf/allnf ulp\ipoib' allows nmake dependancies to make the correct rebuild descision without deleting all instances of ipoib.sys.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2632 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] fully support Windows7 builds; reanme windows7 distribution folder to include...
stansmith [Fri, 11 Dec 2009 21:07:33 +0000 (21:07 +0000)]
[WINOF] fully support Windows7 builds; reanme windows7 distribution folder to include Server 2008 R2

git-svn-id: svn://openib.tc.cornell.edu/gen1@2631 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] optionally sign drivers only for a specific OS & processor architecture.
stansmith [Fri, 11 Dec 2009 21:05:02 +0000 (21:05 +0000)]
[WINOF] optionally sign drivers only for a specific OS & processor architecture.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2630 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] Windows variants of printf/snprintf do not support '%zu', condense to '%u'.
stansmith [Thu, 10 Dec 2009 00:15:41 +0000 (00:15 +0000)]
[OPENSM] Windows variants of printf/snprintf do not support '%zu', condense to '%u'.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2629 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] shed some light on the 'how and why' of having seemingly duplicate files...
stansmith [Mon, 7 Dec 2009 23:13:43 +0000 (23:13 +0000)]
[OPENSM] shed some light on the 'how and why' of having seemingly duplicate files in the opensm src tree.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2628 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OpenSM] OFED version 3.3.3 ported to Windows - vendor == IBAL for now; umad in the...
stansmith [Mon, 7 Dec 2009 23:01:20 +0000 (23:01 +0000)]
[OpenSM] OFED version 3.3.3 ported to Windows - vendor == IBAL for now; umad in the works.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2627 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] remove files no longer used in opensm build.
stansmith [Mon, 7 Dec 2009 22:29:57 +0000 (22:29 +0000)]
[OPENSM] remove files no longer used in opensm build.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2626 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] create a tag on the 'current' OpenSM 3.0.0.
stansmith [Mon, 7 Dec 2009 18:18:27 +0000 (18:18 +0000)]
[OPENSM] create a tag on the 'current' OpenSM 3.0.0.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2625 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[OPENSM] added version number to FileDescription string for easy identification.
stansmith [Mon, 7 Dec 2009 18:06:48 +0000 (18:06 +0000)]
[OPENSM] added version number to FileDescription string for easy identification.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2624 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] (mlnx 5232)
tzachid [Mon, 7 Dec 2009 11:24:16 +0000 (11:24 +0000)]
[IPoIB NDIS 6.0 CM] (mlnx 5232)
Bug at s_buf->p_endpt initialization
Debug prints for Endpoint Communication

git-svn-id: svn://openib.tc.cornell.edu/gen1@2623 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] (mlnx 5220)
tzachid [Mon, 7 Dec 2009 11:23:12 +0000 (11:23 +0000)]
[IPoIB NDIS 6.0 CM] (mlnx 5220)
Access to all CL lists should be lock-protected.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2622 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodapl2: sync with dapl.git tree
shefty [Thu, 3 Dec 2009 20:40:04 +0000 (20:40 +0000)]
dapl2: sync with dapl.git tree

git-svn-id: svn://openib.tc.cornell.edu/gen1@2621 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/ND: fix calculating RDMA read/write address
shefty [Thu, 3 Dec 2009 20:39:03 +0000 (20:39 +0000)]
winverbs/ND: fix calculating RDMA read/write address

The RDMA address should be base + offset, in network byte order.  Rather
than trying to convert the offset into network byte order, then add that
to the base, which is already in network byte order, we need to convert
the base to host order, add the offset, then convert the results back.

This allows the winverbs ND provider to pass the ndmw test.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2620 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs/ND: fix calculating RDMA read/write address
shefty [Thu, 3 Dec 2009 20:32:45 +0000 (20:32 +0000)]
winverbs/ND: fix calculating RDMA read/write address

The RDMA address should be base + offset, in network byte order.  Rather
than trying to convert the offset into network byte order, then add that
to the base, which is already in network byte order, we need to convert
the base to host order, add the offset, then convert the results back.

This allows the winverbs ND provider to pass the ndmw test.

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2619 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[mlx4_bus] Fix for vlan handling. (mlnx 5160)
tzachid [Thu, 3 Dec 2009 16:12:28 +0000 (16:12 +0000)]
[mlx4_bus] Fix for vlan handling. (mlnx 5160)
sign off by: galina

git-svn-id: svn://openib.tc.cornell.edu/gen1@2618 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Use correct branding for inx file.
tzachid [Wed, 2 Dec 2009 09:51:33 +0000 (09:51 +0000)]
[IPoIB NDIS 6.0 CM] Use correct branding for inx file.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2617 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] (mlnx 5191)
tzachid [Wed, 2 Dec 2009 09:41:13 +0000 (09:41 +0000)]
[IPoIB NDIS 6.0 CM] (mlnx 5191)

Fixing 2 bugs:
1. Bug caused by NDIS 6.5 offloadmisc WHQL test: it assumes that the driver version is 6.2 and sends structs with bigger size
2. Internal bug at offload handling.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2616 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB_NDIS6_CM] Build fixes (mlnx 5193)
tzachid [Wed, 2 Dec 2009 09:35:40 +0000 (09:35 +0000)]
[IPOIB_NDIS6_CM] Build fixes (mlnx 5193)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2615 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB_NDIS6_CM] g_stat: added to the adapter structure two previously global counter...
tzachid [Wed, 2 Dec 2009 09:28:59 +0000 (09:28 +0000)]
[IPOIB_NDIS6_CM] g_stat: added to the adapter structure two previously global counters for multi-homed machines. (mlnx 5192)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2614 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Return correct status for an unsupported OIDs. (mlnx 5214)
tzachid [Tue, 1 Dec 2009 16:43:52 +0000 (16:43 +0000)]
[IPoIB NDIS 6.0 CM] Return correct status for an unsupported OIDs. (mlnx 5214)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2613 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Return success value even if NdisReadNetworkAddress fails (the...
tzachid [Tue, 1 Dec 2009 16:42:37 +0000 (16:42 +0000)]
[IPoIB NDIS 6.0 CM] Return success value even if NdisReadNetworkAddress fails (the default)(mlnx 5212)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2612 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Added logging facilities. (mlnx 5211).
tzachid [Tue, 1 Dec 2009 16:40:42 +0000 (16:40 +0000)]
[IPoIB NDIS 6.0 CM] Added logging facilities. (mlnx 5211).

git-svn-id: svn://openib.tc.cornell.edu/gen1@2611 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB] [IPoIB NDIS 6.0 CM] Added support for GUID and fix debug 00:25:90 (mlnx 5210)
tzachid [Tue, 1 Dec 2009 16:39:27 +0000 (16:39 +0000)]
[IPoIB] [IPoIB NDIS 6.0 CM] Added support for GUID and fix debug 00:25:90 (mlnx 5210)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2610 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM]
tzachid [Tue, 1 Dec 2009 16:37:01 +0000 (16:37 +0000)]
[IPoIB NDIS 6.0 CM]
Bug at LSO flow.
When header is contained in 2 first SG Elements, or the total number of the SG List > MAX_SGE_NUM,
we enter send_copy flow. So, the p_send_buf should be at least 64K. (mlnx 5209)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2609 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB] fixing bugs in previous commit. (mlnx 5205)
tzachid [Tue, 1 Dec 2009 16:35:45 +0000 (16:35 +0000)]
[IPOIB] fixing bugs in previous commit. (mlnx 5205)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2608 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Tracing facilities (mlnx 5208)
tzachid [Tue, 1 Dec 2009 16:32:43 +0000 (16:32 +0000)]
[IPoIB NDIS 6.0 CM] Tracing facilities (mlnx 5208)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2607 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[SRP] revert the changes, occasionally made in 2605
leonidk [Tue, 1 Dec 2009 13:08:20 +0000 (13:08 +0000)]
[SRP] revert the changes, occasionally made in 2605

git-svn-id: svn://openib.tc.cornell.edu/gen1@2606 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Fixing a bug when ipoib_get_adapter_params fails reading paramete...
tzachid [Mon, 30 Nov 2009 15:46:11 +0000 (15:46 +0000)]
[IPoIB NDIS 6.0 CM] Fixing a bug when ipoib_get_adapter_params fails reading parameter but returns NDIS_STATUS_SUCCESS. (mlnx 5190)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2605 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPOIB] g_stat: added more fields: prev_port and halt thread.
tzachid [Mon, 30 Nov 2009 15:41:53 +0000 (15:41 +0000)]
[IPOIB] g_stat: added more fields: prev_port and halt thread.

When IPoIB is stuck, it is usually found in the halt thread, waiting for resources reclamation. (mlnx 5188)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2604 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Fixing a bug when IPoIB does not dereference port object during...
tzachid [Mon, 30 Nov 2009 15:39:35 +0000 (15:39 +0000)]
[IPoIB NDIS 6.0 CM] Fixing a bug when IPoIB does not dereference port object during shutdown progress. (mlnx 5185)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2603 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Fixing a possible race when we get PORT_REMOVE before PORT_DOWN...
tzachid [Mon, 30 Nov 2009 15:36:34 +0000 (15:36 +0000)]
[IPoIB NDIS 6.0 CM] Fixing a possible race when we get PORT_REMOVE before PORT_DOWN. (mlnx 5184)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2602 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Fixing the problem when MCast packet stuck the pending list....
tzachid [Mon, 30 Nov 2009 15:34:19 +0000 (15:34 +0000)]
[IPoIB NDIS 6.0 CM] Fixing the problem when MCast packet stuck the pending list. (mlnx 5162)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2601 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM] Fixing a possible race when we can destroy IPoIB without destroyi...
tzachid [Mon, 30 Nov 2009 15:32:24 +0000 (15:32 +0000)]
[IPoIB NDIS 6.0 CM] Fixing a possible race when we can destroy IPoIB without destroying and completing its pending list. (mlnx 5166)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2600 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[IPoIB NDIS 6.0 CM]
tzachid [Mon, 30 Nov 2009 15:22:32 +0000 (15:22 +0000)]
[IPoIB NDIS 6.0 CM]
Upgrade revision to NDIS 6.1 because of bug at WHQL tests:
When running NDIS6.5 test on 6.0 driver, it starts to execute VMQ tests that not supported at IPoIB driver.
(mlnx 5136)

git-svn-id: svn://openib.tc.cornell.edu/gen1@2599 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agodapl: sync with 2.0.25
shefty [Wed, 25 Nov 2009 00:01:00 +0000 (00:01 +0000)]
dapl: sync with 2.0.25

commit 0983c66cbd9511128c1fa221470c4c983903e420

git-svn-id: svn://openib.tc.cornell.edu/gen1@2598 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs branch: update to svn trunk 2596
shefty [Mon, 23 Nov 2009 22:18:52 +0000 (22:18 +0000)]
winverbs branch: update to svn trunk 2596

git-svn-id: svn://openib.tc.cornell.edu/gen1@2597 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DAPL,ETC,WINOF] Remove DAPL 1.2 from WinOF installer: do not copy DAPL 1.2 files...
stansmith [Mon, 23 Nov 2009 20:13:48 +0000 (20:13 +0000)]
[DAPL,ETC,WINOF] Remove DAPL 1.2 from WinOF installer: do not copy DAPL 1.2 files in makebin; remove DAPL 1.2 options from DAT config file.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2596 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[WINOF] switch from WIX 2.0 to use the stable release of WIX 3.0 for building WinOF...
stansmith [Mon, 23 Nov 2009 20:10:16 +0000 (20:10 +0000)]
[WINOF] switch from WIX 2.0 to use the stable release of WIX 3.0 for building WinOF installers.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2595 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years ago[DAPL2] synchronize with OFED DAPL git tree.
stansmith [Mon, 23 Nov 2009 18:59:00 +0000 (18:59 +0000)]
[DAPL2] synchronize with OFED DAPL git tree.

git-svn-id: svn://openib.tc.cornell.edu/gen1@2594 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs branch: update to trunk svn 2592
shefty [Mon, 23 Nov 2009 18:30:20 +0000 (18:30 +0000)]
winverbs branch: update to trunk svn 2592

git-svn-id: svn://openib.tc.cornell.edu/gen1@2593 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs branch: acm: fix build error
shefty [Mon, 23 Nov 2009 18:22:52 +0000 (18:22 +0000)]
winverbs branch: acm: fix build error

one too many ')'

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2592 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agoetc/.bat: update pkg and sign batch files
shefty [Mon, 23 Nov 2009 18:22:11 +0000 (18:22 +0000)]
etc/.bat: update pkg and sign batch files

include dpal pkg and timestamp with signing

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2591 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86

14 years agowinverbs branch: etc/.bat: update pkg and sign batch files
shefty [Mon, 23 Nov 2009 18:21:40 +0000 (18:21 +0000)]
winverbs branch: etc/.bat: update pkg and sign batch files

include dpal pkg and timestamp with signing

Signed-off-by: Sean Hefty <sean.hefty@intel.com>
git-svn-id: svn://openib.tc.cornell.edu/gen1@2590 ad392aa1-c5ef-ae45-8dd8-e69d62a5ef86