On Tue, Aug 4, 2015 at 6:10 AM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Mon, Aug 03, 2015 at 04:09:01PM +0300, Matan Barak wrote:
The release function is called after the device was put.
Although vendor drivers aren't expected to use IB cache in their
removal process, we
Fix the reference counting usage to be handled in the event file
creation/destruction function, instead of being done by the caller.
This is done for both async/non-async event files.
Based on Jason Gunthorpe report at https://www.mail-archive.com/
linux-rdma@vger.kernel.org/msg24680.html:
The
Done in preparation for deploying RCU for the device removal
flow. Allows isolating the RCU handling to the uverb_main layer and
keeping the uverbs_cmd code as is.
Signed-off-by: Yishai Hadas yish...@mellanox.com
Signed-off-by: Shachar Raindel rain...@mellanox.com
---
Currently, if there is any user space application using an IB device,
it is impossible to unload the HW device driver for this device.
Similarly, if the device is hot-unplugged or reset, the device driver
hardware removal flow blocks until all user contexts are destroyed.
This patchset removes
Fixes: 2a72f212263701b927559f6850446421d5906c41 (IB/uverbs: Remove dev_table)
Before this commit there was a device look-up table that was protected
by a spin_lock used by ib_uverbs_open and by ib_uverbs_remove_one. When
it was dropped and container_of was used instead, it enabled the race
with
Enables the uverbs_remove_one to succeed despite the fact that there are
running IB applications working with the given ib device. This
functionality enables a HW device to be unbind/reset despite the fact that
there are running user space applications using it.
It exposes a new IB kernel API
Implements the IB core disassociate_ucontext API. The driver detaches the HW
resources for a given user context to prevent a dependency between application
termination and device disconnecting. This is done by managing the VMAs that
were mapped to the HW bars such as door bell and blueflame. When
Hi all,
please take a look at my RFC patch here:
http://git.infradead.org/users/hch/scsi.git/commitdiff/751774250b71da83a26ba8584cff70f5e7bb7b1e
the commit contains my explanation, but apparently the patch is too
large for the list limit and didn't make it through.
--
To unsubscribe
On Tue, Aug 04, 2015 at 08:48:31PM -0400, ira.weiny wrote:
We are using Netlink multicast. I believe that netlink_bind only allows root
to bind to multicast.
That is a good start...
That said I have not tested the ability to change the timeout settings if one
were to bind without multicast
On Mon, Aug 03, 2015 at 09:15:34PM -0600, Jason Gunthorpe wrote:
On Thu, Jul 09, 2015 at 01:34:26PM -0400, kaike@intel.com wrote:
From: Kaike Wan kaike@intel.com
This patch adds a function to check if listeners for a netlink multicast
group are present. It also adds a function to
On Tue, Aug 04, 2015 at 05:03:24PM +0300, Yishai Hadas wrote:
Signed-off-by: Yishai Hadas yish...@mellanox.com
Signed-off-by: Shachar Raindel rain...@mellanox.com
drivers/infiniband/core/uverbs.h |3 +-
drivers/infiniband/core/uverbs_main.c | 43
+++--
On Tue, Aug 4, 2015 at 8:10 PM, Sagi Grimberg sa...@dev.mellanox.co.il wrote:
Why SIZE_4K and not PAGE_SIZE?
Yes, I'll change that to PAGE_SIZE.
Thanks.
Would non-4KB pages (e.g. PPC 64KB) be an issue? Would this work between
hosts with different page sizes?
iser was always using 4K
On Tue, Aug 04, 2015 at 05:03:28PM +0300, Yishai Hadas wrote:
Currently, IB/cma remove_one flow blocks until all user descriptor managed by
IB/ucma are released. This prevents hot-removal of IB devices. This patch
allows IB/cma to remove devices regardless of user space activity. Upon
getting
Something like this:
CPU A CPU B
ucma_destroy_id()
wait_for_completion()
.. anything
ucma_put_ctx()
complete()
On Tue, Aug 04, 2015 at 09:55:03PM +0300, Matan Barak wrote:
If it fails in ib_device_register_sysfs, the device release function
isn't called and the device pointer isn't freed.
Ugh, yes, the abuse in ib_dealloc_device needs to go too.
Attached is a compile tested patch that fixes that up.
On Tue, Aug 04, 2015 at 05:03:28PM +0300, Yishai Hadas wrote:
Currently, IB/cma remove_one flow blocks until all user descriptor managed by
IB/ucma are released. This prevents hot-removal of IB devices. This patch
allows IB/cma to remove devices regardless of user space activity. Upon
getting
On Tue, Aug 04, 2015 at 04:07:42PM +, Hefty, Sean wrote:
This looks like a reasonable start. It may help with feedback if you
could just post the changes to ib_verbs.h.
Not sure it's all that useful, but here we go:
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index
On Tue, Aug 04, 2015 at 03:09:39PM +0300, Matan Barak wrote:
Correct, I'll change this comment to:
The release function is called after the device was put.
This is in order to avoid use-after-free errors if the vendor
driver's teardown code uses IB cache.
.. the vendor driver uses
On 08/04/2015 09:29 AM, Christoph Hellwig wrote:
On Tue, Aug 04, 2015 at 04:07:42PM +, Hefty, Sean wrote:
This looks like a reasonable start. It may help with feedback if you
could just post the changes to ib_verbs.h.
Not sure it's all that useful, but here we go:
diff --git
-Original Message-
From: linux-rdma-ow...@vger.kernel.org
[mailto:linux-rdma-ow...@vger.kernel.org] On Behalf Of Marciniszyn, Mike
Sent: Tuesday, August 04, 2015 11:33 AM
To: Doug Ledford; Steve Wise
Cc: linux-rdma@vger.kernel.org; t...@opengridcomputing.com
Subject: RE: [PATCH]
On Mon, Aug 03, 2015 at 11:33:51AM -0700, Bart Van Assche wrote:
Bart, do you know what hardware this workaround is for?
I hope the HW vendors can comment on this. Sorry but I'm not sure which HCA
models and/or firmware versions do not support FMR mapping with a non-zero
offset.
Perhaps
please take a look at my RFC patch here:
http://git.infradead.org/users/hch/scsi.git/commitdiff/751774250b71d
a83a26ba8584cff70f5e7bb7b1e
the commit contains my explanation, but apparently the patch is too
large for the list limit and didn't make it through.
This looks like a
From: Jubin John jubin.j...@intel.com
The hfi1 driver depends on the infiniband core stack.
Building a kernel without the infiniband core but with the
hfi1 driver causes undefined reference errors to the ib_* functions.
Reported by 0-day build:
From: Dennis Dalessandro dennis.dalessan...@intel.com
Moving the ipath driver to staging now requires a guard on the INFINIBAND
subsytem in order to prevent the driver from being built without infiniband
enabled. This will lead to a broken build.
Reviewed-by: Mike Marciniszyn
Why SIZE_4K and not PAGE_SIZE?
Yes, I'll change that to PAGE_SIZE.
Thanks.
Would non-4KB pages (e.g. PPC 64KB) be an issue? Would this work between hosts
with different page sizes?
iser was always using 4K segments for reasons I don't perfectly
understand. Maybe Or can comment on this?
On Tue, Aug 04, 2015 at 08:06:16PM +0300, Sagi Grimberg wrote:
Question though, a ULP may want to keep a couple of WRs around instead
of having each allocated in the stack and handled one by one. We need
to provide it with a hint of what is the size it needs.
Note that with the drastic shrink
On Tue, Aug 04, 2015 at 08:44:26PM +0300, Sagi Grimberg wrote:
I do agree that the size on the stack is less of an issue now. What
still can matter is handling each wr one by one vs. doing a collective
post.
But if structured correctly you can still do that with on-stack WRs.
I can
This might run into the same issue as in
https://lists.01.org/pipermail/kbuild-all/2015-August/011216.html.
Hey Mike, what is the issue exactly?
The problem is when CONFIG_INFINIBAND=n and CONFIG_INFINIBAND_HFI1=y. Here is
what I did for ipath which had the same issue:
On 8/4/2015 8:47 PM, Christoph Hellwig wrote:
On Tue, Aug 04, 2015 at 08:44:26PM +0300, Sagi Grimberg wrote:
I do agree that the size on the stack is less of an issue now. What
still can matter is handling each wr one by one vs. doing a collective
post.
But if structured correctly you can
On Tue, Aug 04, 2015 at 09:36:49AM -0700, Bart Van Assche wrote:
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
[ ... ]
struct ib_recv_wr {
+struct ib_send_wr wr;
struct ib_recv_wr *next;
u64 wr_id;
struct ib_sge
On 8/3/2015 8:01 PM, Bart Van Assche wrote:
Move the __attribute_const__ declarations such that sparse understands
that these apply to the function itself and not to the return type.
This avoids that sparse reports error messages like the following:
drivers/infiniband/core/verbs.c:73:12: error:
On 8/4/2015 10:16 AM, Christoph Hellwig wrote:
This patch split up struct ib_send_wr so that all non-trivial verbs
use their own structure which embedds struct ib_send_wr. This dramaticly
shrinks the size of a WR for most common operations.
Hey Christoph,
I think this looks good in general.
Moving the amso1100 driver to staging now requires a guard on the
INFINIBAND subsytem in order to prevent the driver from being built
without infiniband enabled. This will lead to a broken build.
Signed-off-by: Steve Wise sw...@opengridcomputing.com
---
drivers/staging/amso1100/Kconfig |2
-Original Message-
From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
ow...@vger.kernel.org] On Behalf Of Steve Wise
Sent: Tuesday, August 4, 2015 1:13 PM
To: Marciniszyn, Mike; 'Doug Ledford'
Cc: linux-rdma@vger.kernel.org; t...@opengridcomputing.com
Subject: RE: [PATCH]
Subject: Re: [PATCH] RDMA/amso1100: deprecate the amso1100 provider
On 07/29/2015 10:44 AM, Steve Wise wrote:
The HW hasn't been sold since 2005, and the SW has definite bit rot.
Its time to remove it. So move it to staging for a few releases and
then remove it after that.
On 8/4/2015 8:23 PM, Christoph Hellwig wrote:
On Tue, Aug 04, 2015 at 08:06:16PM +0300, Sagi Grimberg wrote:
Question though, a ULP may want to keep a couple of WRs around instead
of having each allocated in the stack and handled one by one. We need
to provide it with a hint of what is the size
-Original Message-
From: Sagi Grimberg [mailto:sa...@dev.mellanox.co.il]
Sent: Tuesday, August 04, 2015 12:26 PM
To: Steve Wise; dledf...@redhat.com
Cc: infinip...@intel.com; sa...@mellanox.com; ogerl...@mellanox.com;
r...@mellanox.com; linux-rdma@vger.kernel.org;
On Tue, Aug 4, 2015 at 7:46 PM, Jason Gunthorpe
jguntho...@obsidianresearch.com wrote:
On Tue, Aug 04, 2015 at 03:09:39PM +0300, Matan Barak wrote:
Correct, I'll change this comment to:
The release function is called after the device was put.
This is in order to avoid use-after-free
The ULP only needs this if the insecure register_always performance
optimization is enabled, or if FRWR/FMR is not supported in the driver.
This is a WIP for this functionality, there are several WARN_ONs in
this patch that can be hit under certain work loads. Additional patches
will be needed to
On Tue, Aug 04, 2015 at 05:03:26PM +0300, Yishai Hadas wrote:
Enables the uverbs_remove_one to succeed despite the fact that there are
running IB applications working with the given ib device. This
functionality enables a HW device to be unbind/reset despite the fact that
there are running
On Tue, Aug 04, 2015 at 04:21:30PM -0600, Jason Gunthorpe wrote:
Every single ULP requires a local_dma_lkey to do anything with
a QP, so let us ensure one exists for every PD created.
If the driver can supply a global local_dma_lkey then use that, otherwise
ask the driver to create a local
41 matches
Mail list logo
