On Thu, 2009-11-19 at 16:14 -0800, Roland Dreier wrote:
> Seems all this infrastructure could be profitably ripped out and
> replaced with the more powerful and flexible ftrace events stuff...
>
> > +#ifndef ATOMIC_NOTIFIER_INIT
> > +/*
> > + * Some backports don't have this, so define here for
Update copyright from Intel-NE, Inc. to Intel Corporation.
Use proper branding string in Kconfig and simplify description.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/Kconfig |9 -
drivers/infiniband/hw/nes/nes.c |2 +-
drivers/infiniband/hw/nes/nes.h
Add a check to nes_create_cq() to return -EINVAL if creating
a CQ with depth > max_cqe (32766).
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_hw.c|2 +-
drivers/infiniband/hw/nes/nes_verbs.c |5 -
2 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/drive
Remove unused and not really used variables.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_verbs.c |6 --
drivers/infiniband/hw/nes/nes_verbs.h |9 +
2 files changed, 1 insertions(+), 14 deletions(-)
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c
b/driver
Add IB_SINGAL_ALL_WR support as an iWARP extension. If set, make
sure all WR for the QP is signalled.
Consolidate flags used in nesqp structure.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_verbs.c |6 --
drivers/infiniband/hw/nes/nes_verbs.h | 15 +--
2 fi
Add additional PHY uC status check in case PHY firmware is not running
properly with heartbeat. Add a hard PHY reset if uC status is 0x0
after initial reset.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_hw.c | 28 ++--
1 files changed, 22 insertions(+),
Replace alloc_fmr, unmap_fmr, dealloc_fmr and map_phys_fmr with
alloc_fast_reg_mr, alloc_fast_reg_page_list, free_fast_reg_page_list.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_hw.c|5 +-
drivers/infiniband/hw/nes/nes_hw.h| 27 ++-
drivers/infiniband/hw/nes/nes_us
Add support for IB_WR_SEND_WITH_INV, IB_WR_RDMA_READ_WITH_INV
and IB_WR_LOCAL_INV.
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_verbs.c | 203 ++---
1 files changed, 113 insertions(+), 90 deletions(-)
diff --git a/drivers/infiniband/hw/nes/nes_verbs.
From: Frank Zago
On error set bad_wr in nes_post_recv(). Stop processing ib_wr
queue when an error is detected.
Signed-off-by: Frank Zago
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_verbs.c |7 +--
1 files changed, 5 insertions(+), 2 deletions(-)
diff --git a/drivers
From: Frank Zago
On error set bad_wr in nes_post_send(). Stop processing ib_wr
queue when an error is detected.
Signed-off-by: Frank Zago
Signed-off-by: Chien Tung
---
drivers/infiniband/hw/nes/nes_verbs.c | 10 --
1 files changed, 8 insertions(+), 2 deletions(-)
diff --git a/driv
Roland,
Patch 1 and 2 are Frank Zago's patches for nes, resubmitted per your request.
Hopefully these will show up on patchwork.
Patch 3 adds send with invalidate and local invalidate support.
Patch 4 corrects fast memory registration, we are now using the
proper interface.
Patch 5 adds additio
H5 Sasha,
Sasha Khapyorsky wrote:
On 18:17 Tue 01 Dec , Sasha Khapyorsky wrote:
On 16:45 Mon 30 Nov , Yevgeny Kliteynik wrote:
Fix compilation error of that was introduced with patch
that added configurable transaction retries - osm_vendor_mlx
didn't have a definition of OSM_DEFAULT_RE
On Tue, 1 Dec 2009, Eli Cohen wrote:
> On Mon, Nov 30, 2009 at 09:03:47AM -0500, Jeff Squyres wrote:
> > Per my prior question: is it expected that IBoE will function
> > *exactly* the same as real IB? The addition of the port attribute
> > seems to imply not.
>
> IBoE and IB should work exactly
On Wed, 25 Nov 2009, Jason Gunthorpe wrote:
> If you have a single physical chip with two ports and they are running
> different protocols it seems much cleaner to me to report it to verbs
> apps as two devices.
>
> Doing this avoids creating compatability problems.
Right. Mellanox has some limit
On Thu, Oct 29, 2009 at 9:23 PM, Sasha Khapyorsky wrote:
>> diff --git a/opensm/opensm/osm_slvl_map_rcv.c
>> b/opensm/opensm/osm_slvl_map_rcv.c
>> index 9c37442..67c71bd 100644
>> --- a/opensm/opensm/osm_slvl_map_rcv.c
>> +++ b/opensm/opensm/osm_slvl_map_rcv.c
>> @@ -2,6 +2,7 @@
>> * Copyrigh
Optimized SLtoVLMappingTable programming reduces the number of MADs
needed from O(n**2) to O(n). See IBA 1.2.1 vol 1 p. 843 14.2.5.8
SLtoVLMappingTable.
Signed-off-by: Hal Rosenstock
---
Changes since v2:
Use osm_log rather than OSM_LOG in osm_slvl_rcv_process for optimized
sl2vl mapping program
On Tue, 1 Dec 2009 05:17:22 +0200
Sasha Khapyorsky wrote:
> Hi Ralph,
>
> On 11:56 Thu 05 Nov , Ralph Campbell wrote:
> > This patch adds new commands to ibportstate to support initializing
> > the link for CAs connected back-to-back. It also allows more than
> > one field to be changed at t
On Tue, Dec 01, 2009 at 06:22:06PM +0200, Liran Liss wrote:
> > Dealing with ABI compatability is a different issue, this new scheme
> is API incompatible due to the change in semantics for existing values.
>
> For rdmacm applications, there are no semantic changes between IB and
> RDMAoE.
So? Th
I don't mind introducing a new command. I just thought
that since iportstate already supported some changes
it was logical to extend it. Note that I haven't changed
any of the original commands so scripts which rely on
the old behavior should still work.
You are the maintainer. I'm willing to go w
The issue is that from an rdmacm app's point of view - there is no
visible difference between IB and RDMAoE ports: both support the
complete set of Verbs, just as any IB transport provider.
Therefore, both ports should reside on a node with an IB transport type.
-Original Message-
From: Se
On Mon, Nov 30, 2009 at 02:01:45PM -0600, Todd Rimmer wrote:
>
> If a given architecture rounds sizeof(transport) up to 16 bits or 32 bits,
> then the replacement field should be uint16_t or uint32_t respectively,
> otherwise existing binary applications which fetch transport will fetch
> addit
On 18:17 Tue 01 Dec , Sasha Khapyorsky wrote:
> On 16:45 Mon 30 Nov , Yevgeny Kliteynik wrote:
> > Fix compilation error of that was introduced with patch
> > that added configurable transaction retries - osm_vendor_mlx
> > didn't have a definition of OSM_DEFAULT_RETRY_COUNT.
BTW as far as
>RDMAoE *is* IB transport over Ethernet
RDMAoE carries something that looks a lot like IB L3 or an IPv6 header, so it
isn't exactly IB transport over Ethernet.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More major
> Dealing with ABI compatability is a different issue, this new scheme
is API incompatible due to the change in semantics for existing values.
For rdmacm applications, there are no semantic changes between IB and
RDMAoE.
> Please look at my message regarding using multiple devices, perhaps
you ca
Rather than check msg_id for CL_DISP_MSGID_NONE, move into default case
of switch statement in sm_mad_ctrl_process_get/set/trap
Eliminate dead code in sm_mad_ctrl_process_get_resp
Signed-off-by: Hal Rosenstock
---
diff --git a/opensm/opensm/osm_sm_mad_ctrl.c b/opensm/opensm/osm_sm_mad_ctrl.c
ind
On 16:45 Mon 30 Nov , Yevgeny Kliteynik wrote:
> Fix compilation error of that was introduced with patch
> that added configurable transaction retries - osm_vendor_mlx
> didn't have a definition of OSM_DEFAULT_RETRY_COUNT.
>
> Signed-off-by: Yevgeny Kliteynik
Applied. Thanks.
Sasha
--
To un
On 16:41 Mon 30 Nov , Yevgeny Kliteynik wrote:
> Eliminate bunch of compilation warnings in printing
> 64-bit values to log.
>
> Signed-off-by: Yevgeny Kliteynik
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to major
On 14:45 Fri 27 Nov , Line Holen wrote:
> Added transaction ID to a message
>
> Signed-off-by: Line Holen
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 http://vger
Hi Line,
On 14:39 Fri 27 Nov , Line Holen wrote:
> Added info to some error messages and lowered log level on
> subsequent call to osm_dump_dr_smp() to reduce amount of logging
> with default log level setting.
>
> Signed-off-by: Line Holen
>
> ---
>
> diff --git a/opensm/opensm/osm_sm_mad
On 14:16 Fri 27 Nov , Line Holen wrote:
> Using INFO instead of SYS whenever ftree decides to fallback to another
> algorithm,
> and VERBOSE instead of INFO some places where the same info is displayed at
> INFO
> level in fabric_dump_general_info()
>
> Signed-off-by: Line Holen
Applied. T
On 14:27 Fri 27 Nov , Line Holen wrote:
> Using single line when displaying DR path info in
> osm_dump_dr_path() and osm_dump_smp_dr_path()
>
> Signed-off-by: Line Holen
Applied. Thanks.
Sasha
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message
On Tue, Dec 1, 2009 at 9:58 AM, Sasha Khapyorsky wrote:
> On 14:48 Mon 30 Nov , Hal Rosenstock wrote:
>>
>> The multicast tree (MLID) has one rate, MTU, etc. so these parameters
>> should be enforced on multicast groups (MGIDs) overlaid on the same
>> MLID.
>
> Assuming that Multicast groups (
>Really it looks as if this check should be predicated on the actual QP
>state which we don't seem to have at this time. The CM state also doesn't
>seem to be useful as it is already ESTABLISHED in this case.
>
>Any suggestions?
Given the current code, you could transition to RTS before sending t
On 14:14 Fri 27 Nov , Line Holen wrote:
> Added status_value and print_desc to some error messages
>
> Signed-off-by: Line Holen
Applied with one change (see below). Thanks.
>
> ---
>
> diff --git a/opensm/opensm/osm_state_mgr.c b/opensm/opensm/osm_state_mgr.c
> index 7540adc..05c31f7 100
On 14:48 Mon 30 Nov , Hal Rosenstock wrote:
>
> The multicast tree (MLID) has one rate, MTU, etc. so these parameters
> should be enforced on multicast groups (MGIDs) overlaid on the same
> MLID.
Assuming that Multicast groups (MGIDs) are consistent in terms of rate,
MTU and packet life time
On Mon, Nov 30, 2009 at 10:50:02AM -0800, Roland Dreier wrote:
>
> I was thinking the same thing, although maybe a name like "link_layer"
> would be clearer?
>
Sure, that's cleaer - I'll change that.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a messag
On Mon, Nov 30, 2009 at 09:03:47AM -0500, Jeff Squyres wrote:
> Per my prior question: is it expected that IBoE will function
> *exactly* the same as real IB? The addition of the port attribute
> seems to imply not.
IBoE and IB should work exactly the same from the perspective of a
user level app
Signed-off-by: Hal Rosenstock
---
diff --git a/opensm/libvendor/osm_vendor_ibumad.c
b/opensm/libvendor/osm_vendor_ibumad.c
index 5ea79cb..8f9bbf5 100644
--- a/opensm/libvendor/osm_vendor_ibumad.c
+++ b/opensm/libvendor/osm_vendor_ibumad.c
@@ -334,7 +334,7 @@ static void *umad_receiver(void *p_pt
For trap 144s which indicate NodeDescription changed, there is no need for
a heavy sweep if that is the only change indicated by the trap. Determine that
by checking that there are no other change flags set and that the
capability mask is unchanged. Still sweep if this is not the case (there
is s
Roland Dreier wrote:
> The IBA takes into account this lack of ordering in multiple places --
> defining
> "communication established" async events, etc.
same goes for the IB stack... e.g take a look on the ib_cm_notify and
rdma_notify APIs
Or.
--
To unsubscribe from this list: send the line "u
40 matches
Mail list logo