On 12/7/2014 2:22 PM, Matan Barak wrote:
Applications might want to create a CQ on n different cores
You mean like an IRQ can flush on a mask potentially made of multiple CPUs?
Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c
b/drivers/infiniband/ulp/iser/iscsi_iser.c
index 0c46f4a..a2dbd04 100644
--- a/drivers/infiniband/ulp/iser
-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.c | 22 ++
1 files changed, 18 insertions(+), 4 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c
b/drivers/infiniband/ulp/iser/iscsi_iser.c
index 47bd87a..0b453e5 100644
--- a/drivers
count per connection. We don't need it to be atomic
as we are safe under the iscsi session frwd_lock taken by libiscsi on
the queuecommand path.
Fixes: 6df5a128f0fd ('IB/iser: Suppress scsi command send completions')
Signed-off-by: Max Gurtovoy
Signed-off-by: Sagi Grimberg
Signe
check_mask setting from scsi_cmnd
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.c | 21 +-
drivers/infiniband/ulp/iser/iser_memory.c | 62 +++-
2 files changed, 26 insertions(+), 57 deletions(-)
diff --git a
From: Sagi Grimberg
It is enough to check mem_h pointer assignment, mem_h == NULL
will indicate that buffer is not registered using mr.
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.h |2 --
drivers/infiniband/ulp/iser/iser_memory.c
mpete.
Signed-off-by: Ariel Nahum
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iser_verbs.c | 19 +++
1 files changed, 11 insertions(+), 8 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c
b/drivers/infiniband/ulp
From: Minh Tran
Re-adjust max CQEs per CQ and max send_wr per QP according
to the resource limits supported by underlying hardware.
Signed-off-by: Minh Tran
Signed-off-by: Jayamohan Kallickal
Acked-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.c
From: Sagi Grimberg
Eliminates code duplication.
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iser_memory.c | 32 ++--
1 files changed, 16 insertions(+), 16 deletions(-)
diff --git a/drivers/infiniband/ulp/iser
Alex Tabachnik
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iser_verbs.c |4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c
b/drivers/infiniband/ulp/iser/iser_verbs.c
index 08e51e7..cd4acc5 1
From: Sagi Grimberg
Use likely() for wc.status == IB_WC_SUCCESS
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iser_verbs.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c
b/drivers
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.h |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h
b/drivers/infiniband/ulp/iser/iscsi_iser.h
index cf704da..5ce2681 100644
--- a
d disconnect handlers
IB/iser: Fix possible NULL derefernce ib_conn->device in session_create
Max Gurtovoy (1):
IB/iser: Fix possible SQ overflow
Minh Tran (1):
IB/iser: Re-adjust CQ and QP send ring sizes to HW limits
Or Gerlitz (1):
IB/iser: Bump version to 1.5
Sagi Grimberg (11):
introducing lock contention in the IO path). This is indeed best effort
approach, but sufficient to survive multi targets sudden death while heavy
IO is inflight.
While we are on it, add a nice kernel-doc style documentation.
Reported-by: Ariel Nahum
Signed-off-by: Sagi Grimberg
Signed-off-by: Or
From: Sagi Grimberg
And fix a checkpatch warning.
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.h | 13 +
1 files changed, 5 insertions(+), 8 deletions(-)
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h
b/drivers
From: Sagi Grimberg
Use uintptr_t to handle wr_id casting, which was found by
Kbuild test robot and smatch. Also remove an internal definition
of variable which potentially shadows an external one (and make
sparse happy).
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers
From: Sagi Grimberg
When creating a connection QP we choose the least used CQ and inc
the number of active QPs on that. If we fail to create the QP,
we need to decrement the active QPs counter.
Reported-by: Roi Dayan
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers
From: Sagi Grimberg
No reason to settle with four, can use the min between
device max comp vectors and number of cores.
Signed-off-by: Sagi Grimberg
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/iser/iscsi_iser.h |4 +---
drivers/infiniband/ulp/iser/iser_verbs.c | 12
On 12/7/2014 4:09 PM, Or Gerlitz wrote:
This set applies on top of Linus tree + the 3.18 fixes which you
didn't pick up (http://marc.info/?l=linux-rdma&m=141448976230915&w=2)
Correction, it includes the 3.18 fixes you didn't pick up, too late for
3.18, we'll go through -s
On Mon, Dec 8, 2014 at 2:23 AM, Weiny, Ira wrote:
>> I find it very annoying that upper level drivers replicate in different ways
>> elements from the IB device attributes returned by ib_query_device. I met
>> that
>> in multiple drivers and upcoming designs for which I do code review. Are you
>
On 6/4/2012 9:05 AM, Roland Dreier wrote:
thanks, applied
Happy to hear that! any questions/clarifications you still need on patch #2?
Or.
--
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
From: Shlomo Pongratz
Depending on the mlx4 device capabilities, support the RSS IB device
capability, using Topelitz or XOR hash functions. Support creating QP
groups where all RX and TX QPs have contiguous QP numbers.
Signed-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
---
drivers
ned-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
---
drivers/infiniband/core/uverbs_cmd.c |1 +
drivers/infiniband/core/verbs.c |3 +++
drivers/infiniband/hw/amso1100/c2_provider.c |3 +++
drivers/infiniband/hw/cxgb3/iwch_provider.c |2 ++
drivers/infiniba
de is known to be racy. In order
to avoid the case where multiple TX flows would attempt to delete the
same IPoIB neighbour, the current queue selection implementation binds
all packets (skbs) that use a given IPoIB neighbour to be sent through
a single ring.
Signed-off-by: Shlomo Pongratz
Sign
From: Shlomo Pongratz
Hi Roland, Sean, everyone!
With 3.5-rc1 being out by now, here's V1 of the TSS/RSS patch series,
basically its very similar to V0, with taking out what was already
accepted and few minor fixes for issues we stepped over testing since
the V0 submission.
Still, the 1st and m
On Mon, Jun 4, 2012 at 8:06 PM, Hefty, Sean wrote:
> > Still, the 1st and most thing to handle here is feedback on the QP
> > groups concept suggested by this patch set to support TSS/RSS over
> > verbs. The plan is for this concept to (with little help from a
> > framework for verbs extension) a
On 6/12/2012 10:00 AM, Roland Dreier wrote:
Jack Morgenstein wrote:
several of the patches have notations indicating things that will be fixed in V1
Not sure what you want me to do with this -- it seems you yourself are saying
this series is not ready to merge yet?
Roland,
We observed few
Or Gerlitz wrote:
> Hefty, Sean wrote:
>> Conceptually, it seems that there should be more alignment with XRC, which
>> also
>> ends up with an associated group of QPs.
> Yep, sure, if its possible to align/squeez/unite these QP groups into the XRC
> interfaces
>
On 6/14/2012 7:09 PM, Sridhar Samudrala wrote:
Enabling NETIF_F_SG improves the throughput further by avoiding a
additional kernel memcpy caused by skb_linearize() in dev_queue_xmit().
Hi Sridhar,
If you **only** enable NETIF_F_SG for CM, does this yields any gain? did
you had to patch the co
On 6/14/2012 4:21 PM, Yann Droneaud wrote:
How will interact vHCA/HCA regarding Automatic Path Migration (APM)
and IPoIB bonding with fail-over (HA), with RDMA_CM and IP usages in
mind ?
work on vHCA the same way they do on HCA, please let me know if you have
anything more special to realize
On 7/2/2012 2:44 PM, Klaus Wacker wrote:
[...] So whats the nature of the returned DMA address of ib_dma_map_single and
can we
assume that it is the same value on different devices on one system?
Currently, over all IB drivers/devices except for ipath and qib,
ib_dma_map_single is just plain
> Let me stress this once again...
> 1.) Is it possible/allowed to call ib_dma_map_single() on Linux, same
> buffer address, size and dma_data direction, for two different ib devices?
> For the data transmission we will have controls in place that not both
> devices write at the same time into that
> ibnetdiscover shows FDR10 speeds as QDR due to bug in setting
> default parameters
How old is this bug?
Or.
--
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.kernel.org/majordomo-inf
On Fri, Jul 6, 2012 at 3:09 AM, Roland Dreier wrote:
> On Tue, Jun 19, 2012 at 1:21 AM, Jack Morgenstein
> wrote:
>> 1. Slave may not set the IS_SM capability for the port.
> So I can't run a SM in a guest?
nope, guests are untrusted and we don't let them send nor recv SMPs
>> 2. No DEV_MGR i
On Fri, Jul 6, 2012 at 2:26 AM, Roland Dreier wrote:
> Part of my problem is I don't see anywhere that
> MLX4_RESERVED_QKEY_MASK is actually used in this patch...
see in patch 13/29
@ -1209,8 +1375,24 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
if (attr_mask & IB_QP_RQ_PSN)
On Fri, Jul 6, 2012 at 3:09 AM, Roland Dreier wrote:
>> 2. No DEV_MGR in multifunc mode.
> I can't run an SRP target in a guest?
To be precise, you can run the target, but you would need to discover it not
through the regular mechnism which makes use of the DM bit of the port.
Or.
--
To unsubsc
On 7/6/2012 2:26 AM, Roland Dreier wrote:
Jack Morgenstein wrote:
For security reasons (i.e., to prevent guests from sending MADs to tunnel QPs
belonging to other guests), each proxy-tunnel qp pair is assigned a unique,
reserved, qkey. These qkeys are available only for proxy and tunnel qp's -
On Fri, Jul 6, 2012 at 6:09 AM, Roland Dreier wrote:
>>> So I can't run a SM in a guest?
[...]
>>> I can't run an SRP target in a guest?
> Seems like an unfortunate set of limitations, especially if I want a
> completely virtualized environment. In my case I might want to
> test the same OS imag
a garbage collection task that runs
every
30 seconds and deletes every ipoib_neigh instance that was idle for at least 60
seconds. The deletion is safe since the ipoib_neigh instances are protected
using RCU and reference count mechanisms.
Signed-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
On 7/10/2012 7:57 PM, Roland Dreier wrote:
thanks, applied
thanks, I see now that you've picked also patch #8, so we're making
progress...
any further thoughts/feedback following the responses I provided on your
questions around DM and SM?
Or.
--
To unsubscribe from this list: send the line
Hi Roland,
Here's a batch of small fixes from Dotan for 3.6
Or.
Dotan Barak (6):
IB/mthca: Events received for non-existent QPs should generate a warning
IB/mthca: Fill in the attribute sq_sig_type value in query QP
IB/mlx4: Fill the masked_atomic_cap attribute in query device
net/mlx4_
From: Dotan Barak
The query QP code was didn't fill that attribute, do that.
Signed-off-by: Dotan Barak
Reviewed-by: Jack Morgenstein
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mthca/mthca_qp.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/dr
From: Dotan Barak
In mlx4_init_icm_table, free the allocated table if we failed to allocate
memory to its entries.
Signed-off-by: Dotan Barak
Reviewed-by: Yevgeny Petrilin
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/icm.c |2 ++
1 files changed, 2 insertions(+), 0
From: Dotan Barak
Clean the idr as part of the error flow since it is a resource too.
Signed-off-by: Dotan Barak
Reviewed-by: Sean Hefty
Signed-off-by: Or Gerlitz
---
drivers/infiniband/core/cm.c | 16 ++--
1 files changed, 10 insertions(+), 6 deletions(-)
diff --git a
From: Dotan Barak
Spotted four duplicate declarations in icm.h, remove them.
Signed-off-by: Dotan Barak
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/icm.h |6 --
1 files changed, 0 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4
From: Dotan Barak
When the user queries for device capabilities, fill the masked_atomic_cap
attrbiute according to the support level of atomic capabilities instead of
hard coded value.
Signed-off-by: Dotan Barak
Reviewed-by: Eli Cohen
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mlx4
From: Dotan Barak
Events received for non-existent QPs should generate a warning along
with the event type which was received.
Signed-off-by: Dotan Barak
Reviewed-by: Jack Morgenstein
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mthca/mthca_qp.c |3 ++-
1 files changed, 2
Hi Roland,
Here's a batch of small fixes from Dotan for 3.6
Please ignore the prev posting, it had wrong sign-off address for me.
Or.
Dotan Barak (6):
IB/mthca: Events received for non-existent QPs should generate a warning
IB/mthca: Fill in the attribute sq_sig_type value in query QP
IB
From: Dotan Barak
Events received for non-existent QPs should generate a warning along
with the event type which was received.
Signed-off-by: Dotan Barak
Reviewed-by: Jack Morgenstein
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mthca/mthca_qp.c |3 ++-
1 files changed, 2
From: Dotan Barak
The query QP code was didn't fill that attribute, do that.
Signed-off-by: Dotan Barak
Reviewed-by: Jack Morgenstein
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mthca/mthca_qp.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/dr
From: Dotan Barak
In mlx4_init_icm_table, free the allocated table if we failed to allocate
memory to its entries.
Signed-off-by: Dotan Barak
Reviewed-by: Yevgeny Petrilin
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/icm.c |2 ++
1 files changed, 2 insertions(+), 0
From: Dotan Barak
Spotted four duplicate declarations in icm.h, remove them.
Signed-off-by: Dotan Barak
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/icm.h |6 --
1 files changed, 0 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4
From: Dotan Barak
Clean the idr as part of the error flow since it is a resource too.
Signed-off-by: Dotan Barak
Reviewed-by: Sean Hefty
Signed-off-by: Or Gerlitz
---
drivers/infiniband/core/cm.c | 16 ++--
1 files changed, 10 insertions(+), 6 deletions(-)
diff --git a
From: Dotan Barak
When the user queries for device capabilities, fill the masked_atomic_cap
attrbiute according to the support level of atomic capabilities instead of
hard coded value.
Signed-off-by: Dotan Barak
Reviewed-by: Eli Cohen
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mlx4
On 7/11/2012 6:40 PM, Roland Dreier wrote:
On Wed, Jul 11, 2012 at 8:33 AM, Or Gerlitz wrote:
- mthca_warn(dev, "Async event for bogus QP %08x\n", qpn);
+ mthca_warn(dev, "Async event %d for bogus QP %08x\n",
+ (int) even
any reason not to remove it all together?
Or.
--
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.kernel.org/majordomo-info.html
On 7/10/2012 1:01 PM, Or Gerlitz wrote:
[...] This patch aims to solve the race conditions found in the IPoIB driver.
The patch breaks the connection between the core networking neighbour structure
and the ipoib_neigh structure. Except for avoiding the race, it allows to in
under a setup where
On Fri, Jul 13, 2012 at 3:58 AM, David Miller wrote:
> You don't need a R/W lock for that, we have RCU hash tables that get
> resized dynamically and the lookup still only needs pure RCU protection.
Hi Dave,
Thanks for the feedback, Shlomo will look into this and if needed,
will ask for some co
On Wed, Jul 11, 2012 at 9:37 PM, Roland Dreier wrote:
> thanks, applied
cool, so we have now 8 out of 29 merged... any info from our side
needed to continue with the review/acceptance process?
Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message t
On 7/13/2012 3:58 AM, David Miller wrote:
From: Shlomo Pongratz
Date: Thu, 12 Jul 2012 20:13:28 +
The RCU and reference count protect the individual entries in the hash. The R/W
lock protects the hash table itself. e.g. deleting the hash table itself, or in
the future re-sizing it.
You
e that was idle for at least 60
seconds. The deletion is safe since the ipoib_neigh instances are protected
using RCU and reference count mechanisms.
Signed-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/ipoib/ipoib.h | 59 ++-
drivers/infiniband/ulp/ipoib
It seems that for some reason the cover-letter wasn't sent properly,
here's listing of the changes:
changes from V0:
- following feedback from Mike and Dave, changed the ipoib_neigh hash table
to allow for lock-free read side, the model follows the RCU based
implementation
in net/core/neig
changes from V0:
- following feedback from Mike and Dave, changed the ipoib_neigh hash table
to allow for lock-free read side, the model follows the RCU based
implementation
in net/core/neighbour.c
- since RCU hash table uses unidirectional collision list, now ipoib_neigh_free
needs to
On 7/19/2012 6:40 PM, David Miller wrote:
You should CC: netdev on patches like this.
OK, I was just thinking that cross-posting typically create more damage
than benefit... will do RESEND now to linux-rdma && netdev
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
t
Adding netdev, as of the high relevancy, see V0 @
http://marc.info/?l=linux-rdma&m=134191474831867&w=2
changes from V0:
- following feedback from Mike and Dave, changed the ipoib_neigh hash table
to allow for lock-free read side, the model follows the RCU based
implementation
in net/core/n
e that was idle for at least 60
seconds. The deletion is safe since the ipoib_neigh instances are protected
using RCU and reference count mechanisms.
Signed-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/ipoib/ipoib.h | 59 ++-
drivers/infiniband/ulp/ipoib
On 7/19/2012 6:24 PM, Christoph Lameter wrote:
On Thu, 19 Jul 2012, Shlomo Pongartz wrote:
The garbage collection and stale times follow the default ipv4/6
neigh.default.gc_yyy
sysctl values, for example
net.ipv4.neigh.default.gc_interval = 30
net.ipv4.neigh.default.gc_stale_time = 60
If giv
On Thu, Jul 19, 2012 at 9:26 PM, Roland Dreier wrote:
> I think this is pushing the module parameter facility way past the
> breaking point of usability. We really need to come up with a better
> way to get complex structured config info like this into the drive
IB's netlink infrastructure?
Or
On Thu, Jul 19, 2012 at 4:18 PM, Or Gerlitz wrote:
> From: Shlomo Pongratz
>
> Dave Miller provided a detailed description of why the
> way IPoIB is using neighbours for its own ipoib_neigh struct is buggy:
[...]
> This patch aims to solve the race conditions found in the IPoIB
On Fri, Jul 20, 2012 at 2:40 PM, Marciniszyn, Mike
> If you could give me some code to look at?
include/rdma/rdma_netlink.h
drivers/infiniband/core/netlink.c
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majord
On 7/19/2012 8:08 PM, David Miller wrote:
These numbers come from the IPV6 Neighbour Discovery RFCs. IPV4 replicates the
Neighbour Unreachability Detection schemes of IPV6 in pretty much it's
entirety, and therefore takes on the same timeout et al. parameters.
OK, got it. At this point, I gu
On 7/19/2012 4:04 PM, Mike Marciniszyn wrote:
Added a congestion control agent in the driver that handles
gets and sets from the congestion control manager in the
fabric for the Performance Scale Messagin (PSM) library.
Hi Mike,
Nice doing.
Can you elaborate a little on the role/usage of the
On 22/07/2012 08:44, Or Gerlitz wrote:
On 7/19/2012 4:04 PM, Mike Marciniszyn wrote:
Added a congestion control agent in the driver that handles
gets and sets from the congestion control manager in the
fabric for the Performance Scale Messagin (PSM) library.
Hi Mike,
Nice doing.
Can you
On 20/07/2012 18:49, Or Gerlitz wrote:
On Thu, Jul 19, 2012 at 4:18 PM, Or Gerlitz wrote:
From: Shlomo Pongratz
Dave Miller provided a detailed description of why the
way IPoIB is using neighbours for its own ipoib_neigh struct is buggy:
[...]
This patch aims to solve the race conditions
On Mon, Jul 23, 2012 at 8:17 PM, Eric Dumazet wrote:
> I have no idea of what you are talking about, I have not the patch or a
> copy of it ;)
http://marc.info/?t=13427149132&r=1&w=2
http://marc.info/?t=13427149135&r=1&w=2
--
To unsubscribe from this list: send the line "unsubscribe linu
changes from V1:
- applied feeback from Dave on hash usage and identation
- applied feedback from Christoph, use values fom the exported arp_tbl
for the hash table size and the frequency of garbage collection
changes from V0:
- following feedback from Mike and Dave, changed the ipoib_neigh
that was idle for at least 2*M
seconds. The deletion is safe since the ipoib_neigh instances are protected
using RCU and reference count mechanisms.
The number of buckets (N) and frequency of running the GC thread (M), are taken
from the exported arb_tbl, to allow for flexibility and not u
On 24/07/2012 18:14, Yishai Hadas wrote:
Just encountered a kernel oops in IPoIB on upstream kernel 3.5 [...]
oops happened in ipoib_mcast_join_task.
Roland,
I made a review now on the issue Yishai raised, and took a look on few
related commits to that area, as you wrote in a77a57a1a "IPoIB:
On Mon, Aug 6, 2012 at 6:51 PM, Yishai Hadas wrote:
> In the meanwhile I have found the root cause of the limit of log_num_mtt of
> 28. Plan to
> send in coming days an extra patch that enables value of 31 which match to
> 8TB.
nice doing. Just an ordering comment, for the upstream inclusion, i
On 10/08/2012 21:25, Kleber Sacilotto de Souza wrote:
This patch fixes the problem by checking for a valid netdev pointer
before using it to get the port MAC address.
just curious, how did you actually stepped on this, code inspection or
any actual race?
Or.
--
To unsubscribe from this list:
On 03/08/2012 11:40, Jack Morgenstein wrote:
This patch set adds SRIOV support for IB interfaces.
Patches 1-4 are "precondition" patches.
Patches 5-22 actually implement the feature.
NOTE: Patch 18 depends on patch IB/mlx4: fix possible deadlock with sm_lock
spinlock
(a separately-sub
On 07/08/2012 15:34, yish...@dev.mellanox.co.il wrote:
From: Yishai Hadas
This patch solves below issue:
Fix the mlx4 core limitation of log num mtt higher than 28.
- There were some int overflows which were fixed to enable using 31.
- When we auto scaling number of MTTs with the size of the s
higher, and
is a step in the direction of allowing to register large amounts
of memory.
Signed-off-by: Yishai Hadas
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/mr.c | 21 +++--
1 files changed, 15 insertions(+), 6 deletions(-)
diff --git a/drivers/net/ethernet
b5a7a65c05 "mlx4_core: Scale size of MTT table with system RAM"
to function properly. Now, when the number of MTTs is scaled with the
size of the RAM we can map up to 8TB.
Signed-off-by: Yishai Hadas
Signed-off-by: Jack Morgenstein
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mel
Hi Roland,
Here are Yishai's fixes which added few issues that come into play
for systems with large RAM, with these fixes, your commit db5a7a65
"mlx4_core: Scale size of MTT table with system RAM" can really play,
where without them, things don't work.
These are fixes, should be ok to 3.6 and t
ned-off-by: Or Gerlitz
---
drivers/infiniband/ulp/ipoib/ipoib_cm.c |3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
index 95ecf4e..24683fd 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_
Hi Roland,
Commit b63b70d8 "IPoIB: Use a private hash table for path lookup
in xmit path" introduced bugs found when further stress tests were
applied to IPoIB, this patch set contains the fixes.
Or.
Shlomo Pongratz (2):
IB/ipoib: Add missing locking when CM object is deleted
IB/ipoib: Fix
ned-off-by: Shlomo Pongratz
Signed-off-by: Or Gerlitz
---
drivers/infiniband/ulp/ipoib/ipoib_main.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c
b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 97920b7..3e2085a 100644
---
On Tue, Aug 14, 2012 at 7:20 PM, Roland Dreier wrote:
> It's not in patchwork because I already applied it.
I don't see it in none of the branches of your kernel.org tree, is
that on a local clone?
Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a mes
Currently, for an app to open a raw QP from user space, we (verbs)
require admin permission, for which we (Mellanox) got customer feedback
saying this is problematic on some of the environments.
Suppose we allow to user to provide source mac+vlan when creating the QP
or when modifying its stat
On 15/08/2012 16:40, Steve Wise wrote:
On 8/15/2012 8:28 AM, Or Gerlitz wrote:
Currently, for an app to open a raw QP from user space, we (verbs)
require admin permission, for which we (Mellanox) got customer
feedback saying this is problematic on some of the environments.
Suppose we allow
On 15/08/2012 17:06, Christoph Lameter wrote:
On Wed, 15 Aug 2012, Or Gerlitz wrote:
Currently, for an app to open a raw QP from user space, we (verbs) require
admin permission, for which we (Mellanox) got customer feedback saying this is
problematic on some of the environments.
Well yes it
Jason Gunthorpe wrote:
> Can you fix this by elevating the process with SELinux?
Chirstoph, do you think this would valid option from users standpoint?
Or.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majordo
On 16/08/2012 07:07, Roland Dreier wrote:
[...] should be fixed in tomorrow's for-next (added include of
).
thanks for taking care of that.
See Rule 1 in Documentation/SubmitChecklist.
Heh, but it compiles fine on x86!
I wonder if there is a way to easily catch such errors when building o
Hi Roland,
Here's a batch with few simple patches, please apply for 3.7
Or.
Dotan Barak (3):
IB/core: Remove unused variables in ucm/ucma
IB/mlx4: Fill in sq_sig_type in query QP
net/mlx4_core: Fix wrong offset in query device caps
Or Gerlitz (1):
net/mlx4_core: Remove annoying
This innocent print makes it very hard to actually use the mlx4 core
debug messages -- for example, the module load sequence of a device with
two VFs yielded 3200 debug prints, with 2800 of them being this
exact one, remove it.
Signed-off-by: Or Gerlitz
---
.../net/ethernet/mellanox/mlx4
From: Dotan Barak
The wrong offset was used when parsing the number of XRCs in
mlx4_QUERY_DEV_CAP(), fix that.
Signed-off-by: Dotan Barak
Signed-off-by: Or Gerlitz
---
drivers/net/ethernet/mellanox/mlx4/fw.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/drivers
From: Dotan Barak
Remove unused wait objects from ucm/ucma events flow.
Signed-off-by: Dotan Barak
Signed-off-by: Or Gerlitz
---
drivers/infiniband/core/ucm.c |1 -
drivers/infiniband/core/ucma.c |1 -
2 files changed, 0 insertions(+), 2 deletions(-)
diff --git a/drivers/infiniband
From: Dotan Barak
The query QP code was didn't fill that attribute, do that.
Signed-off-by: Dotan Barak
Signed-off-by: Or Gerlitz
---
drivers/infiniband/hw/mlx4/qp.c |4
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/dr
-off-by: Or Gerlitz
---
Trying to actually reproduce the problem without this patch, I used an IPoIB
partition for which the MTU is 4k, that is set by the following in
partitions.conf
# 5 = 4k 0x3 --> 0x8003 --> ib.8003
pkey2=0x3, ipoib, mtu=5, defmember=full : ALL, SELF=full;
and invoked
1001 - 1100 of 2427 matches
Mail list logo