From: Daniel Klein
Bug description:
When PerfMgr starts discovery, it resets the discovery_count of ports and
node, but it doesn't reset the physp_discoved buffers.
As a result, the drop manager frees osm_port_t and osm_physp_t objects
of nodes without removing the nodes themselves.
On the next d
As the locking fixes have missed two releases now, I'm resending them.
The other two items are minor (the pr_* function one at least makes
the system more usable by making output message point to the module
making all of the dmesg noise, the other is one I contend is the right
thing to do until we
Commit a9c8ba5884 (IPoIB: Fix usage of uninitialized multicast objects)
added a new flag MCAST_JOIN_STARTED, but was not very strict in how it
was used. We didn't always initialize the completion struct before we
set the flag, and we didn't always call complete on the completion
struct from all pa
From: Bart Van Assche
This change slightly reduces the time needed to log in.
Signed-off-by: Bart Van Assche
Reviewed-by: Sagi Grimberg
Reviewed-by: David Dillow
Cc: Sebastian Parschauer
Signed-off-by: Doug Ledford
---
drivers/infiniband/ulp/srp/ib_srp.c | 9 +
1 file changed, 5 in
The driver already defined the pr_format, it just hadn't
been converted to use pr_info, pr_warn, and pr_err instead
of the equivalent printks. Convert so that messages from
the driver are now properly tagged with their driver name
and can be more easily debugged.
In addition, a number of these pr
Our mcast_dev_flush routine and our mcast_restart_task can race against
each other. In particular, they both hold the priv->lock while
manipulating the rbtree and while removing mcast entries from the
multicast_list and while adding entries to the remove_list, but they
also both drop their locks p
Setting the mtu can safely be moved to the carrier_on_task, which keeps
us from needing to take the rtnl lock in the join_finish section.
Signed-off-by: Doug Ledford
---
drivers/infiniband/ulp/ipoib/ipoib_multicast.c | 8 ++--
1 file changed, 2 insertions(+), 6 deletions(-)
diff --git a/dri
In preparation for using per device work queues, we need to move the
start of the neighbor thread task to after ipoib_ib_dev_init and move
the destruction of the neighbor task to before ipoib_ib_dev_cleanup.
Otherwise we will end up freeing our workqueue with work possibly still
on it.
Signed-off-
During my recent work on the rtnl lock deadlock in the IPoIB driver, I
saw that even once I fixed the apparent races for a single device, as
soon as that device had any children, new races popped up. It turns out
that this is because no matter how well we protect against races on a
single device,
Various places in the IPoIB code had a deadlock related to flushing the
ipoib workqueue. Now that we have per device workqueues and a specific
flush workqueue, there is no longer a deadlock issue with flushing the
device specific workqueues and we can do so unilaterally.
Signed-off-by: Doug Ledfo
We blindly assume that we can just take the rtnl lock and that will
prevent races with downing this interface. Unfortunately, that's not
the case. In ipoib_mcast_stop_thread() we will call flush_workqueue()
in an attempt to clear out all remaining instances of ipoib_join_task.
But, since this tas
We used to pass a flush variable to mcast_stop_thread to indicate if we
should flush the workqueue or not. This was due to some code trying to
flush a workqueue that it was currently running on which is a no-no.
Now that we have per-device work queues, and now that
ipoib_mcast_restart_task has tak
Hi Roland and Yishai,
Yishai Hadas wrote on 12.11.2014 13:38:49:
[...]
> On 11/12/2014 1:22 AM, Roland Dreier wrote:
> > On Wed, Nov 5, 2014 at 7:04 AM, Utz Bacher
wrote:
> >> (B) Status of patches
> >> 1. kernel code -- the new system call: reviewed, acked and accepted by
the
> >> s390x maintai
Hi,
Le mardi 11 novembre 2014 à 18:36 +0200, Haggai Eran a écrit :
> In some drivers there's a need to read data from a user space area that
> was pinned using ib_umem, when running from a different process context.
>
> The ib_umem_copy_from function allows reading data from the physical pages
>
On 10/12/2014 15:04, Yann Droneaud wrote:
> Le mardi 11 novembre 2014 à 18:36 +0200, Haggai Eran a écrit :
>> From: Eli Cohen
>>
>> Add extensible query device capabilities verb to allow adding new features.
>> ib_uverbs_ex_query_device is added and copy_query_dev_fields is used to copy
>> capabil
Le mardi 11 novembre 2014 à 18:36 +0200, Haggai Eran a écrit :
> From: Eli Cohen
>
> Add extensible query device capabilities verb to allow adding new features.
> ib_uverbs_ex_query_device is added and copy_query_dev_fields is used to copy
> capability fields to be used by both ib_uverbs_query_de
Hi Roland,
Le mardi 09 décembre 2014 à 13:43 -0800, Roland Dreier a écrit :
> I was getting ready to apply the ODP series, but then I noticed:
>
> On Tue, Nov 11, 2014 at 8:36 AM, Haggai Eran wrote:
> > diff --git a/drivers/infiniband/core/uverbs_main.c
> > b/drivers/infiniband/core/uverbs_main
On Tue, Dec 09, 2014 at 07:47:27PM +0200, Or Gerlitz wrote:
> On Sun, Dec 7, 2014 at 7:05 PM, Yuval Shaia wrote:
> > This patch is merely makes code more nice and readable.
> > Instead of checking for DPDP on every loop cycle the check moves out of the
> > loop.
>
> Few short comments:
>
> 1. a
On 12/7/2014 7:05 PM, Yuval Shaia wrote:
This patch is merely makes code more nice and readable.
Instead of checking for DPDP on every loop cycle the check moves out of the
loop.
Signed-off-by: Yuval Shaia
---
drivers/net/ethernet/mellanox/mlx4/main.c |7 +++
1 files changed, 3 in
On 12/8/2014 1:48 AM, Yuval Shaia wrote:
1. Add indication whether feature is supported or not.
2. Add descriptions of all features.
Without this fix there is no way to tell if feature is not supported or that
description is not exist.
Signed-off-by: Yuval Shaia
---
drivers/net/ethernet/me
From: Li RongQing
The neigh returned from neigh_lookup has its ref added, not need
to hold RCU lock
and move neigh_release to avoid to check if neigh is NULL again.
Signed-off-by: Li RongQing
---
drivers/infiniband/hw/nes/nes_cm.c | 6 +-
1 file changed, 1 insertion(+), 5 deletions(-)
di
On Tue, Dec 09, 2014 at 07:47:27PM +0200, Or Gerlitz wrote:
> On Sun, Dec 7, 2014 at 7:05 PM, Yuval Shaia wrote:
> > This patch is merely makes code more nice and readable.
> > Instead of checking for DPDP on every loop cycle the check moves out of the
> > loop.
>
> Few short comments:
>
> 1. a
On Tue, Dec 09, 2014 at 07:55:07PM +0200, Or Gerlitz wrote:
> On Mon, Dec 8, 2014 at 1:48 AM, Yuval Shaia wrote:
> > 1. Add indication whether feature is supported or not.
> > 2. Add descriptions of all features.
> > Without this fix there is no way to tell if feature is not supported or
> > that
23 matches
Mail list logo