Previously, cma_match_net_dev called cma_protocol_roce which
tried to verify that the IB device uses RoCE protocol. However,
if rdma_id wasn't bound to a port, then the check would occur
against the first port of the device without regard to whether
that port was even of the same type as the type of port the
incoming packet was received on.
Fix this by passing the port of the request and only checking
against the same port of the device.
Reported-by: Or Gerlitz <gerlitz.or@gmail.com>
Fixes: b8cab5dab1 ('IB/cma: Accept connection without a valid netdev on RoCE')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The copy of the attributes present on the device is now used by all consumers
except for uverbs in case of serving user-space query, where dev->query_device
is called.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use the cached copy of the attributes present on the device, except for
the case of a query originating from user-space, where we have to invoke
the driver query_device entry, so they can fill in their udata.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This way both the IB core and upper level drivers can access these cached
device attributes rather than querying or caching them on their own.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This adds an abstraction that allows ULPs to simply pass a completion
object and completion callback with each submitted WR and let the RDMA
core handle the nitty gritty details of how to handle completion
interrupts and poll the CQ.
In detail there is a new ib_cqe structure which just contains the
completion callback, and which can be used to get at the containing
object using container_of. It is pointed to by the WR and WC as an
alternative to the wr_id field, similar to how many ULPs already use
the field to store a pointer using casts.
A driver using the new completion callbacks allocates it's CQs using
the new ib_create_cq API, which in addition to the number of CQEs and
the completion vectors also takes a mode on how we poll for CQEs.
Three modes are available: direct for drivers that never take CQ
interrupts and just poll for them, softirq to poll from softirq context
using the to be renamed blk-iopoll infrastructure which takes care of
rearming and budgeting, or a workqueue for consumer who want to be
called from user context.
Thanks a lot to Sagi Grimberg who helped reviewing the API, wrote
the current version of the workqueue code because my two previous
attempts sucked too much and converted the iSER initiator to the new
API.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Receipt of CM MAD with other than the Send method for an attribute
other than the ClassPortInfo attribute is invalid.
CM attributes other than ClassPortInfo only use the send method.
The SRP initiator does not maintain a timeout policy for CM connect
requests relies on the CM layer to do that. The result was that
the SRP initiator hung as the connect request never completed.
A new SRP target has been observed to respond to Send CM REQ
with GetResp of CM REQ with bad status. This is non conformant
with IBA spec but exposes a vulnerability in the current MAD/CM
code which will respond to the incoming GetResp of CM REQ as if
it was a valid incoming Send of CM REQ rather than tossing
this on the floor. It also causes the MAD layer not to
retransmit the original REQ even though it has not received a REP.
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Hal Rosenstock <hal@mellanox.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
On 12/03/2015 01:18 AM, Christoph Hellwig wrote:
> The patch looks good to me, but while we touch this area, how about
> throwing in a few cosmetic fixes as well?
How about the patch below ? In that version of the ib_sg_to_pages() fix
these concerns have been addressed and additionally to more bugs have been fixed.
------------
[PATCH] IB core: Fix ib_sg_to_pages()
Fix the code for detecting gaps. A gap occurs not only if the
second or later scatterlist element is not aligned but also if
any scatterlist element other than the last does not end at a
page boundary.
In the code for coalescing contiguous elements, ensure that
mr->length is correct and that last_page_addr is up-to-date.
Ensure that this function returns a negative
error code instead of zero if the first set_page() call fails.
Fixes: commit 4c67e2bfc8 ("IB/core: Introduce new fast registration API")
Reported-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>
It was found by Saurabh Sengar that the netlink code tried to allocate
memory with GFP_KERNEL while holding a spinlock. While it is possible
to fix the issue by replacing GFP_KERNEL with GFP_ATOMIC, it is better
to get rid of the spinlock while sending the packet. However, in order
to protect against a race condition that a quick response may be received
before the request is put on the request list, we need to put the request
on the list first.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reported-by: Saurabh Sengar <saurabh.truth@gmail.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The current implementation gets a spin_lock, and at any scale with
qib and hfi1 post send, the lock contention grows exponentially
with the number of QPs.
idr_find() is RCU compatibile, so read doesn't need the lock.
Change to use rcu_read_lock() and rcu_read_unlock() in
__idr_get_uobj().
kfree_rcu() is used to insure a grace period between the
idr removal and actual free.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Commit e622f2f4ad ("IB: split struct ib_send_wr")
introduced a regression for HCAs whose user mode post
sends go through ib_uverbs_post_send().
The code didn't account for the fact that the first sge is
offset by an operation dependent length. The allocation did,
but the pointer to the destination sge list is computed without
that knowledge. The sge list copy_from_user() then corrupts
fields in the work request
Store the operation dependent length in a local variable and
compute the sge list copy_from_user() destination using that length.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Merge second patch-bomb from Andrew Morton:
- most of the rest of MM
- procfs
- lib/ updates
- printk updates
- bitops infrastructure tweaks
- checkpatch updates
- nilfs2 update
- signals
- various other misc bits: coredump, seqfile, kexec, pidns, zlib, ipc,
dma-debug, dma-mapping, ...
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (102 commits)
ipc,msg: drop dst nil validation in copy_msg
include/linux/zutil.h: fix usage example of zlib_adler32()
panic: release stale console lock to always get the logbuf printed out
dma-debug: check nents in dma_sync_sg*
dma-mapping: tidy up dma_parms default handling
pidns: fix set/getpriority and ioprio_set/get in PRIO_USER mode
kexec: use file name as the output message prefix
fs, seqfile: always allow oom killer
seq_file: reuse string_escape_str()
fs/seq_file: use seq_* helpers in seq_hex_dump()
coredump: change zap_threads() and zap_process() to use for_each_thread()
coredump: ensure all coredumping tasks have SIGNAL_GROUP_COREDUMP
signal: remove jffs2_garbage_collect_thread()->allow_signal(SIGCONT)
signal: introduce kernel_signal_stop() to fix jffs2_garbage_collect_thread()
signal: turn dequeue_signal_lock() into kernel_dequeue_signal()
signals: kill block_all_signals() and unblock_all_signals()
nilfs2: fix gcc uninitialized-variable warnings in powerpc build
nilfs2: fix gcc unused-but-set-variable warnings
MAINTAINERS: nilfs2: add header file for tracing
nilfs2: add tracepoints for analyzing reading and writing metadata files
...
__GFP_WAIT has been used to identify atomic context in callers that hold
spinlocks or are in interrupts. They are expected to be high priority and
have access one of two watermarks lower than "min" which can be referred
to as the "atomic reserve". __GFP_HIGH users get access to the first
lower watermark and can be called the "high priority reserve".
Over time, callers had a requirement to not block when fallback options
were available. Some have abused __GFP_WAIT leading to a situation where
an optimisitic allocation with a fallback option can access atomic
reserves.
This patch uses __GFP_ATOMIC to identify callers that are truely atomic,
cannot sleep and have no alternative. High priority users continue to use
__GFP_HIGH. __GFP_DIRECT_RECLAIM identifies callers that can sleep and
are willing to enter direct reclaim. __GFP_KSWAPD_RECLAIM to identify
callers that want to wake kswapd for background reclaim. __GFP_WAIT is
redefined as a caller that is willing to enter direct reclaim and wake
kswapd for background reclaim.
This patch then converts a number of sites
o __GFP_ATOMIC is used by callers that are high priority and have memory
pools for those requests. GFP_ATOMIC uses this flag.
o Callers that have a limited mempool to guarantee forward progress clear
__GFP_DIRECT_RECLAIM but keep __GFP_KSWAPD_RECLAIM. bio allocations fall
into this category where kswapd will still be woken but atomic reserves
are not used as there is a one-entry mempool to guarantee progress.
o Callers that are checking if they are non-blocking should use the
helper gfpflags_allow_blocking() where possible. This is because
checking for __GFP_WAIT as was done historically now can trigger false
positives. Some exceptions like dm-crypt.c exist where the code intent
is clearer if __GFP_DIRECT_RECLAIM is used instead of the helper due to
flag manipulations.
o Callers that built their own GFP flags instead of starting with GFP_KERNEL
and friends now also need to specify __GFP_KSWAPD_RECLAIM.
The first key hazard to watch out for is callers that removed __GFP_WAIT
and was depending on access to atomic reserves for inconspicuous reasons.
In some cases it may be appropriate for them to use __GFP_HIGH.
The second key hazard is callers that assembled their own combination of
GFP flags instead of starting with something like GFP_KERNEL. They may
now wish to specify __GFP_KSWAPD_RECLAIM. It's almost certainly harmless
if it's missed in most cases as other activity will wake kswapd.
Signed-off-by: Mel Gorman <mgorman@techsingularity.net>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Christoph Lameter <cl@linux.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Vitaly Wool <vitalywool@gmail.com>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
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: symbol 'ib_event_msg' redeclared with different type (originally declared at include/rdma/ib_verbs.h:470) - different modifiers
Fixes: 2b1b5b6012 ("IB/core, cma: Nice log-friendly string helpers")
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
No callers and no providers left, go ahead and remove it.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The new fast registration verb ib_map_mr_sg receives a scatterlist
and converts it to a page list under the verbs API thus hiding
the specific HW mapping details away from the consumer.
The provider drivers are provided with a generic helper ib_sg_to_pages
that converts a scatterlist into a vector of page addresses. The
drivers can still perform any HW specific page address setting
by passing a set_page function pointer which will be invoked for
each page address. This allows drivers to avoid keeping a shadow
page vectors and convert them to HW specific translations by doing
extra copies.
This API will allow ULPs to remove the duplicated code of constructing
a page vector from a given sg list.
The send work request ib_reg_wr also shrinks as it will contain only
mr, key and access flags in addition.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Tested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add support for network namespaces from user space. This is done by passing
the network namespace of the process instead of init_net.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add support for network namespaces in the ib_cma module. This is
accomplished by:
1. Adding network namespace parameter for rdma_create_id. This parameter is
used to populate the network namespace field in rdma_id_private.
rdma_create_id keeps a reference on the network namespace.
2. Using the network namespace from the rdma_id instead of init_net inside
of ib_cma, when listening on an ID and when looking for an ID for an
incoming request.
3. Decrementing the reference count for the appropriate network namespace
when calling rdma_destroy_id.
In order to preserve the current behavior init_net is passed when calling
from other modules.
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Keep a struct for each network namespace containing the IDRs for the RDMA
CM port spaces. The struct is created dynamically using the generic_net
mechanism.
This patch is internal infrastructure work for the following patches. In
this patch, init_net is statically used as the network namespace for
the new port-space API.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add network namespace support to the ib_addr module. For that, all the
address resolution and matching should be done using the appropriate
namespace instead of init_net.
This is achieved by:
1. Adding an explicit network namespace argument to exported function that
require a namespace.
2. Saving the namespace in the rdma_addr_client structure.
3. Using it when calling networking functions.
In order to preserve the behavior of calling modules, &init_net is
passed as the parameter in calls from other modules. This is modified as
namespace support is added on more levels.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The GID cache accompanies every GID with attributes.
The GID attributes link the GID with its netdevice, which could be
resolved to smac and vlan id easily. Since we've added the netdevice
(ifindex and net) to the path record, storing the L2 attributes is
duplicated data and hence these attributes are removed.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Smac and vlan id could be resolved from the GID attribute, and thus
these attributes aren't needed anymore. Removing them.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The cm and cma don't need to explicitly handle vlan and smac,
as they are resolved from the GID index now. Removing this
portion of code.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Previously, vlan id and source MAC were used from QP attributes. Since
the net device is now stored in the GID attributes, they could be used
instead of getting this information from the QP attributes.
IB_QP_SMAC, IB_QP_ALT_SMAC, IB_QP_VID and IB_QP_ALT_VID were removed
because there is no known libibverbs that uses them.
This commit also modifies the vendors (mlx4, ocrdma) drivers in order
to use the new approach.
ocrdma driver changes were done by Somnath Kotur <Somnath.Kotur@Avagotech.Com>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
GID cache API users might want to search for GIDs with specific
attributes rather than just specifying GID, net device and port.
This is used in a later patch, where we find the sgid index by
L2 Ethernet attributes.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Previously, cma_validate_port searched for GIDs in IB cache and then
tried to verify the found port. This could fail when there are
identical GIDs on both ports. In addition, netdevice should be taken
into account when searching the GID table.
Fixing cma_validate_port to search only the relevant port's cache
and netdevice.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Previously, the CM has searched the cache for any sgid_index whose
GID matches the path's GID. Since the path record stores the net
device, the CM should now search only for GIDs which originated from
this net device.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In order to find the sgid_index, one could just query the IB cache
with the correct GID and netdevice. Therefore, instead of storing
the L2 attributes directly in the path, we only store the
ifindex and net and use them later to get the sgid_index.
The vlan_id and smac L2 attributes are removed in a later patch.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Sometime consumers might want to search for a GID in a specific port.
For example, when a WC arrives and we want to search the GID
that matches that port - it's better to search only the relevant
port.
Exposing and renaming ib_cache_gid_find_by_port in order to match
the naming convention of the module.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Adding an ability to query the IB cache by a netdev and get the
attributes of a GID. These parameters are necessary in order to
successfully resolve the required GID (when the netdevice is known)
and get the Ethernet L2 attributes from a GID.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Allow setting IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK at create_flags in
ib_uverbs_create_qp_ex.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
ib_uverbs_ex_create_qp follows the extension verbs
mechanism. New features (for example, QP creation flags
field which is added in a downstream patch) could used
via user-space libraries without breaking the ABI.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The INIT_UDATA() macro requires a pointer or unsigned long argument for
both input and output buffer, and all callers had a cast from when
the code was merged until a recent restructuring, so now we get
core/uverbs_cmd.c: In function 'ib_uverbs_create_cq':
core/uverbs_cmd.c:1481:66: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
This makes the code behave as before by adding back the cast to
unsigned long.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 565197dd8f ("IB/core: Extend ib_uverbs_create_cq")
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
ib_send_cm_sidr_rep could sometimes erase the node from the sidr
(depending on errors in the process). Since ib_send_cm_sidr_rep is
called both from cm_sidr_req_handler and cm_destroy_id, cm_id_priv
could be either erased from the rb_tree twice or not erased at all.
Fixing that by making sure it's erased only once before freeing
cm_id_priv.
Fixes: a977049dac ('[PATCH] IB: Add the kernel CM implementation')
Signed-off-by: Doron Tsur <doront@mellanox.com>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When discussing the patches to demux ids in rdma_cm instead of ib_cm, it
was decided that it is best to use the P_Key value in the packet headers.
However, the mlx5 and ipath drivers are currently unable to send correct
P_Key values in GMP headers. They always send using a single P_Key that is
set during the GSI QP initialization.
Change the rdma_cm code to look at the P_Key value that is part of the
packet payload as a workaround. Once the drivers are fixed this patch can
be reverted.
Fixes: 4c21b5bcef ("IB/cma: Add net_dev and private data checks to
RDMA CM")
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Allocating a workqueue might fail, which wasn't checked so far and would
lead to NULL ptr derefs when an attempt to use it was made.
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
If the lookup of a listening ID failed for an AF_IB request, the code
would try to call dev_put() on a NULL net_dev.
Fixes: be688195bd ("IB/cma: Fix net_dev reference leak with failed
requests")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When using ifup/ifdown while executing enum_netdev_ipv4_ips,
ifa could become invalid and cause use after free error.
Fixing it by protecting with RCU lock.
Fixes: 03db3a2d81 ('IB/core: Add RoCE GID table management')
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When ib_cache_gid_set_default_gid is called from several threads,
updating the table could make find_gid fail, therefore a negative
index will be retruned and an invalid table entry will be used.
Locking find_gid as well fixes this problem.
Fixes: 03db3a2d81 ('IB/core: Add RoCE GID table management')
Signed-off-by: Doron Tsur <doront@mellanox.com>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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:
sizeof(struct ib_send_wr) (old): 96
sizeof(struct ib_send_wr): 48
sizeof(struct ib_rdma_wr): 64
sizeof(struct ib_atomic_wr): 96
sizeof(struct ib_ud_wr): 88
sizeof(struct ib_fast_reg_wr): 88
sizeof(struct ib_bind_mw_wr): 96
sizeof(struct ib_sig_handover_wr): 80
And with Sagi's pending MR rework the fast registration WR will also be
down to a reasonable size:
sizeof(struct ib_fastreg_wr): 64
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com> [srp, srpt]
Reviewed-by: Chuck Lever <chuck.lever@oracle.com> [sunrpc]
Tested-by: Haggai Eran <haggaie@mellanox.com>
Tested-by: Sagi Grimberg <sagig@mellanox.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
The netdev checks recently added to RDMA CM expect a valid netdev to be
found for both InfiniBand and RoCE, but the code that find a netdev is
only implemented for InfiniBand.
Currently RoCE doesn't provide an API to find the netdev matching a
given set of parameters, so this patch just disables the netdev enforcement
for each incoming connections when the link layer is RoCE.
Fixes: 4c21b5bcef ("IB/cma: Add net_dev and private data checks to RDMA CM")
Reported-by: Kamal Heib <kamalh@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
- Create drivers/staging/rdma
- Move amso1100 driver to staging/rdma and schedule for deletion
- Move ipath driver to staging/rdma and schedule for deletion
- Add hfi1 driver to staging/rdma and set TODO for move to regular tree
- Initial support for namespaces to be used on RDMA devices
- Add RoCE GID table handling to the RDMA core caching code
- Infrastructure to support handling of devices with differing
read and write scatter gather capabilities
- Various iSER updates
- Kill off unsafe usage of global mr registrations
- Update SRP driver
- Misc. mlx4 driver updates
- Support for the mr_alloc verb
- Support for a netlink interface between kernel and user space cache
daemon to speed path record queries and route resolution
- Ininitial support for safe hot removal of verbs devices
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=EHLg
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Pull inifiniband/rdma updates from Doug Ledford:
"This is a fairly sizeable set of changes. I've put them through a
decent amount of testing prior to sending the pull request due to
that.
There are still a few fixups that I know are coming, but I wanted to
go ahead and get the big, sizable chunk into your hands sooner rather
than waiting for those last few fixups.
Of note is the fact that this creates what is intended to be a
temporary area in the drivers/staging tree specifically for some
cleanups and additions that are coming for the RDMA stack. We
deprecated two drivers (ipath and amso1100) and are waiting to hear
back if we can deprecate another one (ehca). We also put Intel's new
hfi1 driver into this area because it needs to be refactored and a
transfer library created out of the factored out code, and then it and
the qib driver and the soft-roce driver should all be modified to use
that library.
I expect drivers/staging/rdma to be around for three or four kernel
releases and then to go away as all of the work is completed and final
deletions of deprecated drivers are done.
Summary of changes for 4.3:
- Create drivers/staging/rdma
- Move amso1100 driver to staging/rdma and schedule for deletion
- Move ipath driver to staging/rdma and schedule for deletion
- Add hfi1 driver to staging/rdma and set TODO for move to regular
tree
- Initial support for namespaces to be used on RDMA devices
- Add RoCE GID table handling to the RDMA core caching code
- Infrastructure to support handling of devices with differing read
and write scatter gather capabilities
- Various iSER updates
- Kill off unsafe usage of global mr registrations
- Update SRP driver
- Misc mlx4 driver updates
- Support for the mr_alloc verb
- Support for a netlink interface between kernel and user space cache
daemon to speed path record queries and route resolution
- Ininitial support for safe hot removal of verbs devices"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (136 commits)
IB/ipoib: Suppress warning for send only join failures
IB/ipoib: Clean up send-only multicast joins
IB/srp: Fix possible protection fault
IB/core: Move SM class defines from ib_mad.h to ib_smi.h
IB/core: Remove unnecessary defines from ib_mad.h
IB/hfi1: Add PSM2 user space header to header_install
IB/hfi1: Add CSRs for CONFIG_SDMA_VERBOSITY
mlx5: Fix incorrect wc pkey_index assignment for GSI messages
IB/mlx5: avoid destroying a NULL mr in reg_user_mr error flow
IB/uverbs: reject invalid or unknown opcodes
IB/cxgb4: Fix if statement in pick_local_ip6adddrs
IB/sa: Fix rdma netlink message flags
IB/ucma: HW Device hot-removal support
IB/mlx4_ib: Disassociate support
IB/uverbs: Enable device removal when there are active user space applications
IB/uverbs: Explicitly pass ib_dev to uverbs commands
IB/uverbs: Fix race between ib_uverbs_open and remove_one
IB/uverbs: Fix reference counting usage of event files
IB/core: Make ib_dealloc_pd return void
IB/srp: Create an insecure all physical rkey only if needed
...
We have many WR opcodes that are only supported in kernel space
and/or require optional information to be copied into the WR
structure. Reject all those not explicitly handled so that we
can't pass invalid information to drivers.
Cc: stable@vger.kernel.org
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The flags to ibnl_put_msg should be NLM_F_REQUEST instead of GFP_KERNEL.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: John Fleck <john.fleck@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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
the RDMA_CM_EVENT_DEVICE_REMOVAL event we close all the underlying HW resources
for the given ucontext. The ucontext itself is still alive till its explicit
destroying by its creator.
Running applications at that time will have some zombie device, further
operations may fail.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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 named 'disassociate_ucontext' which lets
a driver detaching its HW resources from a given user context without
crashing/terminating the application. In case a driver implemented the
above API and registered with ib_uverb there will be no dependency between its
device to its uverbs_device. Upon calling remove_one of ib_uverbs the call
should return after disassociating the open HW resources without waiting to
clients disconnecting. In case driver didn't implement this API there will be no
change to current behaviour and uverbs_remove_one will return only when last
client has disconnected and reference count on uverbs device became 0.
In case the lower driver device was removed any application will
continue working over some zombie HCA, further calls will ended with an
immediate error.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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 <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Fixes: 2a72f21226 ("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 remove_one as dev might be freed just after:
dev = container_of(inode->i_cdev, struct ib_uverbs_device, cdev) but
before the kref_get.
In addition, this buggy patch added some dead code as
container_of(x,y,z) can never be NULL and so dev can never be NULL.
As a result the comment above ib_uverbs_open saying "the open method
will either immediately run -ENXIO" is wrong as it can never happen.
The solution follows Jason Gunthorpe suggestion from below URL:
https://www.mail-archive.com/linux-rdma@vger.kernel.org/msg25692.html
cdev will hold a kref on the parent (the containing structure,
ib_uverbs_device) and only when that kref is released it is
guaranteed that open will never be called again.
In addition, fixes the active count scheme to use an atomic
not a kref to prevent WARN_ON as pointed by above comment
from Jason.
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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 existing code for this is broken, in ib_uverbs_get_context all
the error paths between ib_uverbs_alloc_event_file and the
kref_get(file->ref) are wrong - this will result in fput() which will
call ib_uverbs_event_close, which will try to do kref_put and
ib_unregister_event_handler - which are no longer paired."
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The majority of callers never check the return value, and even if they
did, they can't do anything about a failure.
All possible failure cases represent a bug in the caller, so just
WARN_ON inside the function instead.
This fixes a few random errors:
net/rd/iw.c infinite loops while it fails. (racing with EBUSY?)
This also lays the ground work to get rid of error return from the
drivers. Most drivers do not error, the few that do are broken since
it cannot be handled.
Since uverbs can legitimately make use of EBUSY, open code the
check.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The pd now has a local_dma_lkey member which completely replaces
ib_get_dma_mr, use it instead.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
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 use all physical memory MR associated
with the new PD.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sagi Grimberg <sagig@dev.mellanox.co.il>
Acked-by: Christoph Hellwig <hch@infradead.org>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This patch routes a SA pathrecord query to netlink first and processes the
response appropriately. If a failure is returned, the request will be sent
through IB. The decision whether to route the request to netlink first is
determined by the presence of a listener for the local service netlink
multicast group. If the user-space local service netlink multicast group
listener is not present, the request will be sent through IB, just like
what is currently being done.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: John Fleck <john.fleck@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Replace kmalloc with kzalloc so that all uninitialized fields in SA query
will be zero-ed out to avoid unintentional consequence. This prepares the
SA query structure to accept new fields in the future.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: John Fleck <john.fleck@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This patch adds a function to check if listeners for a netlink multicast
group are present. It also adds a function to receive netlink response
messages.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Signed-off-by: John Fleck <john.fleck@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
get_netdev: get the net_device on the physical port of the IB transport port. In
port aggregation mode it is required to return the netdev of the active port.
modify_gid: note for a change in the RoCE gid cache. Handle this by writing to
the harsware GID table. It is possible that indexes in cahce and hardware tables
won't match so a translation is required when modifying a QP or creating an
address handle.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Handling bonding and other devices require us to all all GIDs of the
net-devices which are upper-devices of the RoCE port related
net-device.
Active-backup configurations imposes even more challenges as the
default GID should only be set on the active devices (this is
necessary as otherwise the same MAC could be used for several
slaves and thus several slaves will have identical GIDs).
Managing these configurations are done by listening to:
(a) NETDEV_CHANGEUPPER event
(1) if a related net-device is linked, delete all inactive
slaves default GIDs and add the upper device GIDs.
(2) if a related net-device is unlinked, delete all upper GIDs
and add the default GIDs.
(b) NETDEV_BONDING_FAILOVER:
(1) delete the bond GIDs from inactive slaves
(2) delete the inactive slave's default GIDs
(3) Add the bond GIDs to the active slave.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Smatch says that, based on the indenting, we should probably add curly
braces here.
Fixes: 03db3a2d81 ('IB/core: Add RoCE GID table management')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
RoCE GIDs are based on IP addresses configured on Ethernet net-devices
which relate to the RDMA (RoCE) device port.
Currently, each of the low-level drivers that support RoCE (ocrdma,
mlx4) manages its own RoCE port GID table. As there's nothing which is
essentially vendor specific, we generalize that, and enhance the RDMA
core GID cache to do this job.
In order to populate the GID table, we listen for events:
(a) netdev up/down/change_addr events - if a netdev is built onto
our RoCE device, we need to add/delete its IPs. This involves
adding all GIDs related to this ndev, add default GIDs, etc.
(b) inet events - add new GIDs (according to the IP addresses)
to the table.
For programming the port RoCE GID table, providers must implement
the add_gid and del_gid callbacks.
RoCE GID management requires us to state the associated net_device
alongside the GID. This information is necessary in order to manage
the GID table. For example, when a net_device is removed, its
associated GIDs need to be removed as well.
RoCE mandates generating a default GID for each port, based on the
related net-device's IPv6 link local. In contrast to the GID based on
the regular IPv6 link-local (as we generate GID per IP address),
the default GID is also available when the net device is down (in
order to support loopback).
Locking is done as follows:
The patch modify the GID table code both for new RoCE drivers
implementing the add_gid/del_gid callbacks and for current RoCE and
IB drivers that do not. The flows for updating the table are
different, so the locking requirements are too.
While updating RoCE GID table, protection against multiple writers is
achieved via mutex_lock(&table->lock). Since writing to a table
requires us to find an entry (possible a free entry) in the table and
then modify it, this mutex protects both the find_gid and write_gid
ensuring the atomicity of the action.
Each entry in the GID cache is protected by rwlock. In RoCE, writing
(usually results from netdev notifier) involves invoking the vendor's
add_gid and del_gid callbacks, which could sleep.
Therefore, an invalid flag is added for each entry. Updates for RoCE are
done via a workqueue, thus sleeping is permitted.
In IB, updates are done in write_lock_irq(&device->cache.lock), thus
write_gid isn't allowed to sleep and add_gid/del_gid are not called.
When passing net-device into/out-of the GID cache, the device
is always passed held (dev_hold).
The code uses a single work item for updating all RDMA devices,
following a netdev or inet notifier.
The patch moves the cache from being a client (which was incorrect,
as the cache is part of the IB infrastructure) to being explicitly
initialized/freed when a device is registered/removed.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This gets rid of the weird in-between state where struct ib_device
was allocated but the kobject didn't work.
Consequently ib_device_release is now guaranteed to be called in
all situations and we needn't duplicate its kfrees on error paths.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Fully replaced by a more generic and suitable
ib_alloc_mr.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use ib_alloc_mr with specific parameters.
Change the existing callers.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This was added in a thought of uniting all mr allocation
and deallocation routines but the fact is we have a single
deallocation routine already, ib_dereg_mr.
And, move mlx5_ib_destroy_mr specific logic into mlx5_ib_dereg_mr
(includes only signature stuff for now).
And, fixup the only callers (iser/isert) accordingly.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When no matching listening ID is found for a given request, the net_dev
that was used to find the request isn't released.
Fixes: 0b3ca768fc ("IB/cma: Use found net_dev for passive connections")
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Now that there are no ib_cm clients using the compare_data feature for
matching IB CM requests' private data, remove the compare_data parameter of
ib_cm_listen and remove the code implementing the feature.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use ib_cm_insert_listen to create listening IB CM IDs or share existing
ones if needed. When given a request on a specific CM ID, the code now
matches the request to the RDMA CM ID based on the request parameters, so
it no longer needs to rely on the ib_cm's private data matching
capabilities.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When receiving a new connection in cma_req_handler, we actually already
know the net_dev that is used for the connection's creation. Instead of
calling cma_translate_addr to resolve the new connection id's source
address, just use the net_dev that was found.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Pass incoming request parameters through the relevant IPv4/IPv6 routing
tables and make sure the network stack is configured to handle such
requests.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Instead of relying on a the ib_cm module to check an incoming CM request's
private data header, add these checks to the RDMA CM module. This allows a
following patch to to clean up the ib_cm interface and remove the code that
looks into the private headers. It will also allow supporting namespaces in
RDMA CM by making these checks namespace aware later on.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The rdma_cm module will later use the P_Key from the BTH to de-mux
requests.
See discussion at:
http://www.spinics.net/lists/netdev/msg336067.html
Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Cc: Liran Liss <liranl@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add helper functions to access the IDRs by port-space and port number.
Pass around the port-space enum in cma.c instead of using pointers to
port-space IDRs.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When receiving a connection request, rdma_cm needs to associate the request
with a network device, in order to disambiguate requests. To do this, it
needs to know the request's destination IP. For this the module needs to
allow getting this information from the private data in the request packet,
instead of relying on the information already being in the listening RDMA
CM ID.
When creating a new incoming connection ID, the code in
cma_save_ip{4,6}_info can no longer rely on the listener's private data to
find the port number, so it reads it from the requested service ID.
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Enabling network namespaces for RDMA CM will allow processes on different
namespaces to listen on the same port. In order to leave namespace support
out of the CM layer, this requires that multiple RDMA CM IDs will be able
to share a single CM ID.
This patch adds infrastructure to retrieve an existing listening ib_cm_id,
based on its device and service ID, or create a new one if one does not
already exist. It also adds a reference count for such instances
(cm_id_private.listen_sharecount), and prevents cm_destroy_id from
destroying a CM if it is still shared. See the relevant discussion [1].
[1] Re: [PATCH v3 for-next 05/13] IB/cm: Reference count ib_cm_ids
http://www.spinics.net/lists/netdev/msg328860.html
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Expose the service ID on an incoming CM or SIDR request to the event
handler. This will allow the RDMA CM module to de-multiplex connection
requests based on the information encoded in the service ID.
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In the case of IPoIB, and maybe in other cases, the network device is
managed by an upper-layer protocol (ULP). In order to expose this
network device to other users of the IB device, let ULPs implement
a callback that returns network device according to connection parameters.
The IB device and port, together with the P_Key and the GID should
be enough to uniquely identify the ULP net device. However, in current
kernels there can be multiple IPoIB interfaces created with the same GID.
Furthermore, such configuration may be desireable to support ipvlan-like
configurations for RDMA CM with IPoIB. To resolve the device in these
cases the code will also take the IP address as an additional input.
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
An ib_client callback that is called with the lists_rwsem locked only for
read is protected from changes to the IB client lists, but not from
ib_unregister_device() freeing its client data. This is because
ib_unregister_device() will remove the device from the device list with
lists_rwsem locked for write, but perform the rest of the cleanup,
including the call to remove() without that lock.
Mark client data that is undergoing de-registration with a new going_down
flag in the client data context. Lock the client data list with lists_rwsem
for write in addition to using the spinlock, so that functions calling the
callback would be able to lock only lists_rwsem for read and let callbacks
sleep.
Since ib_unregister_client() now marks the client data context, no need for
remove() to search the context again, so pass the client data directly to
remove() callbacks.
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Currently the RDMA subsystem's device list and client list are protected by
a single mutex. This prevents adding user-facing APIs that iterate these
lists, since using them may cause a deadlock. The patch attempts to solve
this problem by adding a read-write semaphore to protect the lists. Readers
now don't need the mutex, and are safe just by read-locking the semaphore.
The ib_register_device, ib_register_client, ib_unregister_device, and
ib_unregister_client functions are modified to lock the semaphore for write
during their respective list modification. Also, in order to make sure
client callbacks are called only between add() and remove() calls, the code
is changed to only add items to the lists after the add() calls and remove
from the lists before the remove() calls.
This patch attempts to solve a similar need [1] that was seen in the RoCE
v2 patch series.
[1] http://www.spinics.net/lists/linux-rdma/msg24733.html
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Cc: Matan Barak <matanb@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Resolving a link-local IPv6 address with an unspecified source address
was broken by commit 5462eddd7a, which prevented the IPv6 stack from
learning the scope id of the link-local IPv6 address, causing random
failures as the IP stack chose a random link to resolve the address on.
This commit 5462eddd7a made us bail out of cma_check_linklocal early if
the address passed in was not an IPv6 link-local address. On the address
resolution path, the address passed in is the source address; if the
source address is the unspecified address, which is not link-local, we
will bail out early.
This is mostly correct, but if the destination address is a link-local
address, then we will be following a link-local route, and we'll need to
tell the IPv6 stack what the scope id of the destination address is.
This used to be done by last line of cma_check_linklocal, which is
skipped when bailing out early:
dev_addr->bound_dev_if = sin6->sin6_scope_id;
(In cma_bind_addr, the sin6_scope_id of the source address is set to the
sin6_scope_id of the destination address, so this is correct)
This line is required in turn for the following line, L279 of
addr6_resolve, to actually inform the IPv6 stack of the scope id:
fl6.flowi6_oif = addr->bound_dev_if;
Since we can only know we are in this failure case when we have access
to both the source IPv6 address and destination IPv6 address, we have to
deal with this further up the stack. So detect this failure case in
cma_bind_addr, and set bound_dev_if to the destination address scope id
to correct it.
Signed-off-by: Spencer Baugh <sbaugh@catern.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Something like this:
CPU A CPU B
Acked-by: Sean Hefty <sean.hefty@intel.com>
======================== ================================
ucma_destroy_id()
wait_for_completion()
.. anything
ucma_put_ctx()
complete()
.. continues ...
ucma_leave_multicast()
mutex_lock(mut)
atomic_inc(ctx->ref)
mutex_unlock(mut)
ucma_free_ctx()
ucma_cleanup_multicast()
mutex_lock(mut)
kfree(mc)
rdma_leave_multicast(mc->ctx->cm_id,..
Fix it by latching the ref at 0. Once it goes to 0 mc and ctx cannot
leave the mutex(mut) protection.
The other atomic_inc in ucma_get_ctx is OK because mutex(mut) protects
it from racing with ucma_destroy_id.
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The verbs are obsolete. The ib_rereg_phys_mr() verb is not used by
kernel ULPs, and the last ib_reg_phys_mr() call site in the kernel
tree has now been removed.
Two staging tree call sites remain in the Lustre client. The Lustre
team has been notified of the deprecation of reg_phys_mr.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Acked-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
ib_ucm_release_dev clears the wrong bit if devnum is greater
than IB_UCM_MAX_DEVICES.
Signed-off-by: Carol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Fixes to allow clients to make remove mapping requests, after
they have provided the user space service with the mapping
information, they are using when the service is restarted.
1) Adding IWPM_REG_VALID, IWPM_REG_INCOMPL and IWPM_REG_UNDEF
registration types for the port mapper clients and functions
to set/check the registration type.
2) If the port mapper user space service is not available to register
the client, then its registration stays IWPM_REG_UNDEF and the
registration isn't checked until the service becomes available
(no mappings are possible, if the user space service isn't running).
3) After the service is restarted, the user space port mapper pid is set
to valid and the client registration is set to IWPM_REG_INCOMPL
to allow the client to make remove mapping requests.
Signed-off-by: Tatyana Nikolova <Tatyana.E.Nikolova@intel.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Tested-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Whenever ib_cm gets remove_one call, like when there is a hot-unplug
event, the driver should mark itself as going_down and confirm that no
new works are going to be queued for that device.
so, the order of the actions are:
1. mark the going_down bit.
2. flush the wq.
3. [make sure no new works for that device.]
4. unregister mad agent.
otherwise, works that are already queued can be scheduled after the mad
agent was freed.
Signed-off-by: Erez Shitrit <erezsh@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The define OPA_LID_PERMISSIVE is big endian and was compared to the
cpu endian variable opa_drslid.
Problem caught by 0-day build infrastructure.
Fixes: 8e4349d13f (IB/mad: Add final OPA MAD processing)
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: John, Jubin <jubin.john@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Persuant to Liran's comments on node_type on linux-rdma
mailing list:
In an effort to reform the RDMA core and ULPs to minimize use of
node_type in struct ib_device, an additional bit is added to
struct ib_device for is_switch (IB switch). This is needed
to be initialized by any IB switch device driver. This is a
NEW requirement on such device drivers which are all
"out of tree".
In addition, an ib_switch helper was added to ib_verbs.h
based on the is_switch device bit rather than node_type
(although those should be consistent).
The RDMA core (MAD, SMI, agent, sa_query, multicast, sysfs)
as well as (IPoIB and SRP) ULPs are updated where
appropriate to use this new helper. In some cases,
the helper is now used under the covers of using
rdma_[start end]_port rather than the open coding
previously used.
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Hal Rosenstock <hal@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
- A large cleanup of how device capabilities are checked for various
features
- Additional cleanups in the MAD processing
- Update to the srp driver
- Creation and use of centralized log message helpers
- Add const to a number of args to calls and clean up call chain
- Add support for extended cq create verb
- Add support for timestamps on cq completion
- Add support for processing OPA MAD packets
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=yISf
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma
Pull rdma updates from Doug Ledford:
- a large cleanup of how device capabilities are checked for various
features
- additional cleanups in the MAD processing
- update to the srp driver
- creation and use of centralized log message helpers
- add const to a number of args to calls and clean up call chain
- add support for extended cq create verb
- add support for timestamps on cq completion
- add support for processing OPA MAD packets
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dledford/rdma: (92 commits)
IB/mad: Add final OPA MAD processing
IB/mad: Add partial Intel OPA MAD support
IB/mad: Add partial Intel OPA MAD support
IB/core: Add OPA MAD core capability flag
IB/mad: Add support for additional MAD info to/from drivers
IB/mad: Convert allocations from kmem_cache to kzalloc
IB/core: Add ability for drivers to report an alternate MAD size.
IB/mad: Support alternate Base Versions when creating MADs
IB/mad: Create a generic helper for DR forwarding checks
IB/mad: Create a generic helper for DR SMP Recv processing
IB/mad: Create a generic helper for DR SMP Send processing
IB/mad: Split IB SMI handling from MAD Recv handler
IB/mad cleanup: Generalize processing of MAD data
IB/mad cleanup: Clean up function params -- find_mad_agent
IB/mlx4: Add support for CQ time-stamping
IB/mlx4: Add mmap call to map the hardware clock
IB/core: Pass hardware specific data in query_device
IB/core: Add timestamp_mask and hca_core_clock to query_device
IB/core: Extend ib_uverbs_create_cq
IB/core: Add CQ creation time-stamping flag
...
For devices which support OPA MADs
1) Use previously defined SMP support functions.
2) Pass correct base version to ib_create_send_mad when processing OPA MADs.
3) Process out_mad_key_index returned by agents for a response. This is
necessary because OPA SMP packets must carry a valid pkey.
4) Carry the correct segment size (OPA vs IBTA) of RMPP messages within
ib_mad_recv_wc.
5) Handle variable length OPA MADs by:
* Adjusting the 'fake' WC for locally routed SMP's to represent the
proper incoming byte_len
* out_mad_size is used from the local HCA agents
1) when sending agent responses on the wire
2) when passing responses through the local_completions
function
NOTE: wc.byte_len includes the GRH length and therefore is different
from the in_mad_size specified to the local HCA agents.
out_mad_size should _not_ include the GRH length as it is added
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add OPA SMP processing functionality.
Define the new OPA SMP format, create support functions for this format using
the previously defined helper functions as appropriate.
These functions are defined in this patch and used in the final OPA MAD support
patch.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This patch is the first of 3 which adds processing of OPA MADs
1) Add Intel Omni-Path Architecture defines
2) Increase max management version to accommodate OPA
3) update ib_create_send_mad
If the device supports OPA MADs and the MAD being sent is the OPA base
version alter the MAD size and sg lengths as appropriate
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In order to support alternate sized MADs (and variable sized MADs on OPA
devices) add in/out MAD size parameters to the process_mad core call.
In addition, add an out_mad_pkey_index to communicate the pkey index the driver
wishes the MAD stack to use when sending OPA MAD responses.
The out MAD size and the out MAD PKey index are required by the MAD
stack to generate responses on OPA devices.
Furthermore, the in and out MAD parameters are made generic by specifying them
as ib_mad_hdr rather than ib_mad.
Drivers are modified as needed and are protected by BUG_ON flags if the MAD
sizes passed to them is incorrect.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
This patch implements allocating alternate receive MAD buffers within the MAD
stack. Support for OPA to send/recv variable sized MADs is implemented later.
1) Convert MAD allocations from kmem_cache to kzalloc
kzalloc is more flexible to support devices with different sized MADs
and research and testing showed that the current use of kmem_cache does
not provide performance benefits over kzalloc.
2) Change struct ib_mad_private to use a flex array for the mad data
3) Allocate ib_mad_private based on the size specified by devices in
rdma_max_mad_size.
4) Carry the allocated size in ib_mad_private to be used when processing
ib_mad_private objects.
5) Alter DMA mappings based on the mad_size of ib_mad_private.
6) Replace the use of sizeof and static defines as appropriate
7) Add appropriate casts for the MAD data when calling processing
functions.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add max MAD size to the device immutable data set and have all drivers that
support MADs report the current IB MAD size (IB_MGMT_MAD_SIZE) to the core.
Verify MAD size data in both the MAD core and when reading the immutable data.
OPA drivers will report alternate MAD sizes in subsequent patches.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In preparation to support the new OPA MAD Base version, add a base version
parameter to ib_create_send_mad and set it to IB_MGMT_BASE_VERSION for current
users.
Definition of the new base version and it's processing will occur in later
patches.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
IB and OPA SMPs share the same processing algorithm but have different header
formats and permissive LID detection.
Add a helper function which is generic to processing the DR forwarding checks which
can be used by both IB and OPA SMP code.
Use this function in the current IB function smi_check_forward_dr_smp.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
IB and OPA SMPs share the same processing algorithm but have different header
formats and permissive LID detection.
Add a helper function which is generic to processing DR SMP Recv messages which
can be used by both IB and OPA SMP code.
Use this function in the current IB function smi_handle_dr_smp_recv.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
IB and OPA SMPs share the same processing algorithm but have different header
formats and permissive LID detection.
Add a helper function which is generic to processing DR SMP Send messages which
can be used by both IB and OPA SMP code.
Use this function in the current IB function smi_handle_dr_smp_send.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Make a helper function to process Directed Route SMPs to be called by the IB
MAD Recv Handler, ib_mad_recv_done_handler.
This cleans up the MAD receive handler code a bit and allows for us to better
share the SMP processing code between IB and OPA SMPs.
IB and OPA SMPs share the same processing algorithm but have different header
formats and permissive LID detection. Therefore this and subsequent patches
split the common processing code from the IB specific code in anticipation of
sharing those algorithms with the OPA code.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
ib_find_send_mad only needs access to the MAD header not the full IB MAD.
Change the local variable to ib_mad_hdr and change the corresponding cast.
This allows for clean usage of this function with both IB and OPA MADs because
OPA MADs carry the same header as IB MADs.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
find_mad_agent only needs read only access to the MAD header. Update the
ib_mad pointer to be const ib_mad_hdr. Adjust call tree.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Vendors should be able to pass vendor specific data to/from
user-space via query_device uverb. In order to do this,
we need to pass the vendors' specific udata.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In order to expose timestamp we need to expose two new attributes in
query_device to be used for CQ completion time-stamping:
timestamp_mask - how many bits are valid in the timestamp, where timestamp
values could be 64bits the most.
hca_core_clock - timestamp is given in HW cycles, the frequency in KHZ units
of the HCA, necessary in order to convert cycles to seconds.
This is added both to ib_query_device and its respective uverbs counterpart.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
ib_uverbs_ex_create_cq follows the extension verbs
mechanism. New features (for example, CQ creation flags
field which is added in a downstream patch) could used
via user-space libraries without breaking the ABI.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Currently, ib_create_cq uses cqe and comp_vecotr instead
of the extendible ib_cq_init_attr struct.
Earlier patches already changed the vendors to work with
ib_cq_init_attr. This patch changes the consumers too.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add a new ib_cq_init_attr structure which contains the
previous cqe (minimum number of CQ entries) and comp_vector
(completion vector) in addition to a new flags field.
All vendors' create_cq callbacks are changed in order
to work with the new API.
This commit does not change any functionality.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> to patch #2
Signed-off-by: Doug Ledford <dledford@redhat.com>
Registering an event handler is done for a device. This device may have
one RoCE port (no SA cap) and one InfiniBand port (has SA cap).
Therefore, warning from the event handler about a specific port that
doesn't have SA cap is correct but pollutes the kernel log without a
need.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In order to support constant callers of agent_send_response we add const
specifiers to the its pointer arguments.
Adjust the call tree accordingly.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Hal Rosenstock <hal@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
rdma-cma/iw_cm: Export tos field to iwarp providers
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Tatyana Nikolova <Tatyana.E.Nikolova@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Support for using UD and AF_IB is currently broken. The
IB_CM_SIDR_REQ_RECEIVED message is not handled properly in
cma_save_net_info() and we end up falling into code that will try and
process the request as ipv4/ipv6, which will end up failing.
The resolution is to add a check for the SIDR_REQ and call
cma_save_ib_info() with a NULL path record. Change cma_save_ib_info()
to copy the src sib info from the listen_id when the path record is NULL.
Reported-by: Hari Shankar <Hari.Shankar@netapp.com>
Signed-off-by: Matt Finlay <matt@mellanox.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
After discussion upstream, it was agreed to transition the usage of iboe
in the kernel to roce. This keeps our terminology consistent with what
was finalized in the IBTA Annex 16 and IBTA Annex 17 publications.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Problem reported by: Ted Kim <ted.h.kim@oracle.com>:
We have a case where a Linux system and a non-Linux system are
trying to interoperate. The Linux host is the active side and
starts the connection establishment, but later decides to not go
through with the connection setup and does rdma_destroy_id().
The rdma_destroy_id() eventually works its way down to cm_destroy_id()
in core/cm.c, where a REJ is sent. The non-Linux system
has some trouble recognizing the REJ because of:
A. CM states which can't receive the REJ
B. Some issues about REJ formatting (missing comm ID)
ISSUE A: That part of the spec says, a Consumer Reject REJ can be
sent for a connection abort, but it goes further
and says: can send a REJ message with a "Consumer Reject"
Reason code if they are in a CM state (i.e. REP
Rcvd, MRA(REP) Sent, REQ Rcvd, MRA Sent) that allows
a REJ to be sent (lines 35-38).
Of the states listed there in that sentence, it would
seem to limit the active side to using the Consumer Reject
(for the abort case) in just the REP-Rcvd and MRA-REP-Sent
states. That is basically only after the active side
sees a REP (or alternatively goes down the state transitions
to timeout in which case a Timeout REJ is sent).
As a fix, in cm-destroy-id() move the IB-CM-MRA-REQ-RCVD case
to the same as REQ-SENT. Essentially, make a REJ sent after
getting an MRA on active side a timeout rather than Consumer-
Reject, which is arguably more correct with the CM state
diagrams previous to getting a REP.
Signed-off-by: Ted Kim <ted.h.kim@oracle.com>
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
As of commit 5eb620c81c "IB/core: Add helpers for uncached GID and P_Key
searches"; pkey_tbl_len and gid_tbl_len are immutable data which are stored in
the ib_device.
The per port core capability flags to be added later are also immutable data to
be stored in the ib_device object.
In preparation for this create a structure for per port immutable data and
place the pkey and gid table lengths within this structure.
"get_port_immutable" is added as a mandatory device function to allow the
drivers to fill in this data.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The addition of the rdma_cap_ib_mad is technically broken in ib_umad_remove_one
because the loop "i" value is not a port value.
This bug resulted in the ib_umad failing to properly remove its resources when
the core capability functions were converted to bit fields.
NOTE: e17371d73908 did not result in broken behavior on its own. It was only
an issue when the implementation of rdma_cap_ib_mad was changed.
Pass the port value to rdma_cap_ib_mad.
Fixes: e17371d73908 ("IB/Verbs: Use management helper rdma_cap_ib_mad()")
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use the new common rdma_[start|end]_port functions instead of using
local variables and figuring it out on the fly.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The following functions only need read access to the data passed to them.
ib_mad_kernel_rmpp_agent
is_rmpp_data_mad
rcv_has_same_gid
ib_find_send_mad
Clarify with const specifiers
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
rcv_has_same_class only needs access to the MAD header
specify WR and Receive WC as const
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
ib_response_mad only needs read access to the MAD header, not write access
to the entire mad struct, so replace struct ib_mad with const struct
ib_mad_hdr
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
validate_mad only needs read access to the MAD header, not write access
to the entire mad struct, so replace struct ib_mad with const struct
ib_mad_hdr
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Some of us keep revisiting the code to decode enumerations that
appear in out logs. Let's borrow the nice logging helpers that
exists in xprtrdma and rds for CMA events, IB events and WC statuses.
Reviewd-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Return values of 0 do not make sense for functions which return enum
smi_action
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
is_rmpp_data_mad is more descriptive for this function.
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Previously start_port and end_port were defined in 2 places, cache.c and
device.c and this prevented their use in other modules.
Make these common functions, change the name to reflect the rdma
name space, and update existing users.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_eth_ah() to help us check if the port of an
IB device support Ethernet Address Handler.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_af_ib() to help us check if the port of an
IB device support Native Infiniband Address.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_ib_mcast() to help us check if the port of an
IB device support Infiniband Multicast.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_ib_sa() to help us check if the port of an
IB device support Infiniband Subnet Administration.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_iw_cm() to help us check if the port of an
IB device support IWARP Communication Manager.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_ib_cm() to help us check if the port of an
IB device support Infiniband Communication Manager.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_ib_smi() to help us check if the port of an
IB device support Infiniband Subnet Management Interface.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Introduce helper rdma_cap_ib_mad() to help us check if the port of an
IB device support Infiniband Management Datagrams.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform rest part in IB-core cma.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Reform cma_acquire_dev() with management helpers, introduce
cma_validate_port() to make the code more clean.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform mcast related part in IB-core cma.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform route related part in IB-core cma.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform cm related part in IB-core cma/ucm.
Few checks focus on the device cm type rather than the port capability,
directly pass port 1 works currently, but can't support mixing cm type
device in future.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform IB-core verbs
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform IB-core multicast.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform IB-core sa_query.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform IB-core cm.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Use raw management helpers to reform IB-core mad/agent/user_mad.
Signed-off-by: Michael Wang <yun.wang@profitbricks.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Tested-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Sean Hefty <sean.hefty@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
The string iwpm_ulib_name is recorded in a nlmsg as a netlink attribute.
Without this fix parsing of the nlmsg by the userspace port mapper service fails
because of unknown attribute length, causing the port mapper service not to
register the client, which has sent the nlmsg.
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Cc: <stable@vger.kernel.org> #v3.16
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Addresses the following kernel logs seen during boot of sparc systems:
Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
Signed-off-by: David Ahern <david.ahern@oracle.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
While unmapping an ODP writable page, the dirty bit of the page is set. In
order to do so, the head of the compound page is found.
Currently, the compound head is found even on non-writable pages, where it is
never used, leading to unnecessary cpu barrier that impacts performance.
This patch moves the search for the compound head to be done only when needed.
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Acked-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Currently, while mapping or unmapping pages for ODP, the umem mutex is locked
and unlocked once for each page. Such lock/unlock operation take few tens to
hundreds of nsecs. This makes a significant impact when mapping or unmapping few
MBs of memory.
To avoid this, the mutex should be locked only once per operation, and not per
page.
Signed-off-by: Guy Shapiro <guysh@mellanox.com>
Acked-by: Shachar Raindel <raindel@mellanox.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Add functionality to enable the port mapper on the passive side to provide to its
clients the actual (non-mapped) ip/tcp address information of the connecting peer
1) Adding remote_info_cb() to process the address info of the connecting peer
The address info is provided by the user space port mapper service when
the connection is initiated by the peer
2) Adding a hash list to store the remote address info
3) Adding functionality to add/remove the remote address info
After the info has been provided to the port mapper client,
it is removed from the hash list
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
When accepting a new IPv4 connect to an IPv6 socket, the CMA tries to
canonize the address family to IPv4, but does not properly process
the listening sockaddr to get the listening port, and does not properly
set the address family of the canonized sockaddr.
Fixes: e51060f08a ("IB: IP address based RDMA connection manager")
Cc: <stable@vger.kernel.org>
Reported-By: Yotam Kenneth <yotamke@mellanox.com>
Signed-off-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Tested-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Hello,
When an application using XRCs abruptly terminates, the mmaped pages
of the CQ buffers are leaked.
This comes from the fact that when resources are released in
ib_uverbs_cleanup_ucontext(), we fail to release the CQs because their
refcount is not 0.
When creating an XRC SRQ, we increment the associated CQ refcount.
This refcount is only decremented when the SRQ is released.
Therefore we need to release the SRQs prior to the CQs to make sure
that all references to the CQs are gone before trying to release these.
Signed-off-by: Sebastien Dugue <sebastien.dugue@bull.net>
Signed-off-by: Doug Ledford <dledford@redhat.com>
In a call to ib_umem_get(), if address is 0x0 and size is
already page aligned, check added in commit 8494057ab5
("IB/uverbs: Prevent integer overflow in ib_umem_get address
arithmetic") will refuse to register a memory region that
could otherwise be valid (provided vm.mmap_min_addr sysctl
and mmap_low_allowed SELinux knobs allow userspace to map
something at address 0x0).
This patch allows back such registration: ib_umem_get()
should probably don't care of the base address provided it
can be pinned with get_user_pages().
There's two possible overflows, in (addr + size) and in
PAGE_ALIGN(addr + size), this patch keep ensuring none
of them happen while allowing to pin memory at address
0x0. Anyway, the case of size equal 0 is no more (partially)
handled as 0-length memory region are disallowed by an
earlier check.
Link: http://mid.gmane.org/cover.1428929103.git.ydroneaud@opteya.com
Cc: <stable@vger.kernel.org> # 8494057ab5 ("IB/uverbs: Prevent integer overflow in ib_umem_get address arithmetic")
Cc: Shachar Raindel <raindel@mellanox.com>
Cc: Jack Morgenstein <jackm@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
If ib_umem_get() is called with a size equal to 0 and an
non-page aligned address, one page will be pinned and a
0-sized umem will be returned to the caller.
This should not be allowed: it's not expected for a memory
region to have a size equal to 0.
This patch adds a check to explicitly refuse to register
a 0-sized region.
Link: http://mid.gmane.org/cover.1428929103.git.ydroneaud@opteya.com
Cc: <stable@vger.kernel.org>
Cc: Shachar Raindel <raindel@mellanox.com>
Cc: Jack Morgenstein <jackm@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>
Properly verify that the resulting page aligned end address is larger
than both the start address and the length of the memory area requested.
Both the start and length arguments for ib_umem_get are controlled by
the user. A misbehaving user can provide values which will cause an
integer overflow when calculating the page aligned end address.
This overflow can cause also miscalculation of the number of pages
mapped, and additional logic issues.
Addresses: CVE-2014-8159
Cc: <stable@vger.kernel.org>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Add on-demand paging capabilities reporting to the extended query device verb.
Yann Droneaud writes:
Note: as offsetof() is used to retrieve the size of the lower chunk
of the response, beware that it only works if the upper chunk
is right after, without any implicit padding. And, as the size of
the latter chunk is added to the base size, implicit padding at the
end of the structure is not taken in account. Both point must be
taken in account when extending the uverbs functionalities.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
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_device and
ib_uverbs_ex_query_device.
Following the discussion about this patch [1], the code now validates
the command's comp_mask is zero, returning -EINVAL for unknown values,
in order to allow extending the verb in the future.
The verb also checks the user-space provided response buffer size and
only fills in capabilities that will fit in the buffer. In attempt to
follow the spirit of presentation [2] by Tzahi Oved that was presented
during OpenFabrics Alliance International Developer Workshop 2013, the
comp_mask bits will only describe which fields are valid. Furthermore,
fields that can simply be cleared when they are not supported, do not
require a comp_mask bit at all. The verb returns a response_length
field containing the actual number of bytes written by the kernel, so
that a newer version running on an older kernel can tell which fields
were actually returned.
[1] [PATCH v1 0/5] IB/core: extended query device caps cleanup for v3.19
http://thread.gmane.org/gmane.linux.kernel.api/7889/
[2] https://www.openfabrics.org/images/docs/2013_Dev_Workshop/Tues_0423/2013_Workshop_Tues_0830_Tzahi_Oved-verbs_extensions_ofa_2013-tzahio.pdf
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
When the last on-demand paging MR is released the notifier count is
left non-zero so that concurrent page faults will have to abort. If a
new MR is then registered, the counter is reset. However, the decision
is made to put the new MR in the list waiting for the notifier count
to reach zero, before the counter is reset. An invalidation or another
MR registration can release the MR to handle page faults, but without
such an event the MR can wait forever.
The patch fixes this issue by adding a check whether the MR is the
first on-demand paging MR when deciding whether it is ready to handle
page faults. If it is the first MR, we know that there are no mmu
notifiers running in parallel to the registration.
Fixes: 882214e2b1 ("IB/core: Implement support for MMU notifiers regarding on demand paging regions")
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The deadlock occurs in __uverbs_modify_qp: we take a lock (idr_read_qp)
and in case of failure in ib_resolve_eth_l2_attrs we don't release
it (put_qp_read). Fix that.
Fixes: ed4c54e5b4 ("IB/core: Resolve Ethernet L2 addresses when modifying QP")
Signed-off-by: Moshe Lazer <moshel@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
When marshaling a user path to the kernel struct ib_sa_path, we need
to zero smac and dmac and set the vlan id to the "no vlan" value.
This is to ensure that Ethernet attributes are not used with
InfiniBand QPs.
Fixes: dd5f03beb4 ("IB/core: Ethernet L2 attributes in verbs/cm structures")
Signed-off-by: Ilya Nelkenbaum <ilyan@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
While commit 7e36ef8205 ("IB/core: Temporarily disable
ex_query_device uverb") is correct as it makes the extended
QUERY_DEVICE uverb (which came as part of commit 5a77abf9a9
("IB/core: Add support for extended query device caps") and commit
860f10a799 ("IB/core: Add flags for on demand paging support")) not
available to userspace, it doesn't address the initial issue regarding
ib_copy_to_udata() [1][2].
Additionally, further discussions around this new uverb seems to
conclude it would require a different data structure than the one
currently described in <rdma/ib_user_verbs.h> [3].
Both of these issues require a revert of the changes, so this patch
partially reverts commit 8cdd312cfe ("IB/mlx5: Implement the ODP
capability query verb") and commit 860f10a799 ("IB/core: Add flags
for on demand paging support") and fully reverts commit 5a77abf9a9
("IB/core: Add support for extended query device caps").
[1] "Re: [PATCH v3 06/17] IB/core: Add support for extended query device caps"
http://mid.gmane.org/1418733236.2779.26.camel@opteya.com
[2] "Re: [PATCH] IB/core: Temporarily disable ex_query_device uverb"
http://mid.gmane.org/1423067503.3030.83.camel@opteya.com
[3] "RE: [PATCH v1 1/5] IB/uverbs: ex_query_device: answer must not depend on request's comp_mask"
http://mid.gmane.org/2807E5FD2F6FDA4886F6618EAC48510E0CC12C30@CRSMSX101.amr.corp.intel.com
Cc: Eli Cohen <eli@mellanox.com>
Cc: Haggai Eran <haggaie@mellanox.com>
Cc: Ira Weiny <ira.weiny@intel.com>
Cc: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Cc: Sagi Grimberg <sagig@mellanox.com>
Cc: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit 5a77abf9a9 ("IB/core: Add support for extended query device caps")
added a new extended verb to query the capabilities of RDMA devices, but the
semantics of this verb are still under debate [1].
Don't expose this verb to userspace until the ABI is nailed down.
[1] [PATCH v1 0/5] IB/core: extended query device caps cleanup for v3.19
http://www.spinics.net/lists/linux-rdma/msg22904.html
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
* Add an interval tree implementation for ODP umems. Create an
interval tree for each ucontext (including a count of the number of
ODP MRs in this context, semaphore, etc.), and register ODP umems in
the interval tree.
* Add MMU notifiers handling functions, using the interval tree to
notify only the relevant umems and underlying MRs.
* Register to receive MMU notifier events from the MM subsystem upon
ODP MR registration (and unregister accordingly).
* Add a completion object to synchronize the destruction of ODP umems.
* Add mechanism to abort page faults when there's a concurrent invalidation.
The way we synchronize between concurrent invalidations and page
faults is by keeping a counter of currently running invalidations, and
a sequence number that is incremented whenever an invalidation is
caught. The page fault code checks the counter and also verifies that
the sequence number hasn't progressed before it updates the umem's
page tables. This is similar to what the kvm module does.
In order to prevent the case where we register a umem in the middle of
an ongoing notifier, we also keep a per ucontext counter of the total
number of active mmu notifiers. We only enable new umems when all the
running notifiers complete.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Yuval Dagan <yuvalda@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
* Extend the umem struct to keep the ODP related data.
* Allocate and initialize the ODP related information in the umem
(page_list, dma_list) and freeing as needed in the end of the run.
* Store a reference to the process PID struct in the ucontext. Used to
safely obtain the task_struct and the mm during fault handling,
without preventing the task destruction if needed.
* Add 2 helper functions: ib_umem_odp_map_dma_pages and
ib_umem_odp_unmap_dma_pages. These functions get the DMA addresses
of specific pages of the umem (and, currently, pin them).
* Support for page faults only - IB core will keep the reference on
the pages used and call put_page when freeing an ODP umem
area. Invalidations support will be added in a later patch.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Majd Dibbiny <majd@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
* Add a configuration option for enable on-demand paging support in
the infiniband subsystem (CONFIG_INFINIBAND_ON_DEMAND_PAGING). In a
later patch, this configuration option will select the MMU_NOTIFIER
configuration option to enable mmu notifiers.
* Add a flag for on demand paging (ODP) support in the IB device capabilities.
* Add a flag to request ODP MR in the access flags to reg_mr.
* Fail registrations done with the ODP flag when the low-level driver
doesn't support this.
* Change the conditions in which an MR will be writable to explicitly
specify the access flags. This is to avoid making an MR writable just
because it is an ODP MR.
* Add a ODP capabilities to the extended query device verb.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
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_device and
ib_uverbs_ex_query_device.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
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 pinned in the ib_umem struct.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
In order to allow umems that do not pin memory, we need the umem to
keep track of its region's address.
This makes the offset field redundant, and so this patch removes it.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Address resolution always does a context switch to a work-queue to
deliver the address resolution event. When the IP address is already
cached in the system ARP table, we're going through the following:
chain:
rdma_resolve_ip --> addr_resolve (cache hit) -->
which ends up with:
queue_req --> set_timeout (now) --> mod_delayed_work(,, delay=1)
We actually do realize that the timeout should be zero, but the code
forces it to a minimum of one jiffie.
Using one jiffie as the minimum delay value results in sub-optimal
scheduling of executing this work item by the workqueue, which on the
below testbed costs about 3-4ms out of 12ms total time.
To fix that, we let the minimum delay to be zero. Note that the
connect step times change too, as there are address resolution calls
from that flow.
The results were taken from running both client and server on the
same node, over mlx4 RoCE port.
before -->
step total ms max ms min us us / conn
create id : 0.01 0.01 6.00 6.00
resolve addr : 4.02 4.01 4013.00 4016.00
resolve route: 0.18 0.18 182.00 183.00
create qp : 1.15 1.15 1150.00 1150.00
connect : 6.73 6.73 6730.00 6731.00
disconnect : 0.55 0.55 549.00 550.00
destroy : 0.01 0.01 9.00 9.00
after -->
step total ms max ms min us us / conn
create id : 0.01 0.01 6.00 6.00
resolve addr : 0.05 0.05 49.00 52.00
resolve route: 0.21 0.21 207.00 208.00
create qp : 1.10 1.10 1104.00 1104.00
connect : 1.22 1.22 1220.00 1221.00
disconnect : 0.71 0.71 713.00 713.00
destroy : 0.01 0.01 9.00 9.00
Signed-off-by: Or Kehati <ork@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Applications can request that the SM assign an MGID by passing a mcast
member request containing MGID = 0. When the SM responds by sending
the allocated MGID, this MGID replaces the 0-MGID in the multicast group.
However, the MGID field in the group is also the key field in the IB
core multicast code rbtree containing the multicast groups for the
port.
Since this is a key field, correct handling requires that the group
entry be deleted from the rbtree and then re-inserted with the new
key, so that the table structure is properly maintained.
The current code does not do this correctly. Correct operation
requires that if the key-field gid has changed at all, it should be
deleted and re-inserted.
Note that when inserting, if the new MGID is zero (not the case here
but the code should handle this correctly), we allow duplicate entries
for 0-MGIDs.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
For RoCE, resolution of layer 2 address attributes forces no VLAN if
link-local GIDs are used. This patch allows applications to choose
the VLAN ID for link-local based RoCE GIDs by setting IB_QP_VID in
their QP attribute mask, and prevents the core from overriding this
choice.
Cc: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
In ib_uverbs_open_qp, the sharable xrc target qp is created as a
"pseudo" qp and added to a list of qp's sharing the same physical
QP. This is done before the "pseudo" qp is assigned a uobject.
There is a race condition here if an async event arrives at the
physical qp. If the event is handled after the pseudo qp is added to
the list, but before it is assigned a uobject, the kernel crashes in
ib_uverbs_qp_event_handler, due to trying to dereference a NULL
uobject pointer.
Note that simply checking for non-NULL is not enough, due to error
flows in ib_uverbs_open_qp. If the failure is after assigning the
uobject, but before the qp has fully been created, we still have a
problem.
Thus, in ib_uverbs_qp_event_handler, we test that the uobject is
present, and also that it is live.
Reported-by: Matthew Finlay <matt@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
During create-ah from userspace, uverbs is sending garbage data in
attr.dmac and attr.vlan_id. This patch sets attr.dmac and
attr.vlan_id to zero.
Fixes: dd5f03beb4 ("IB/core: Ethernet L2 attributes in verbs/cm structures")
Signed-off-by: Devesh Sharma <devesh.sharma@emulex.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Clear the reserved field of struct ib_uverbs_async_event_desc which is
copied to user space.
Signed-off-by: Eli Cohen <eli@mellanox.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
When marsheling a user path to the kernel struct ib_sa_path, need
to zero smac, dmac and set the vlan id to the "no vlan" value.
Fixes: dd5f03beb4 ("IB/core: Ethernet L2 attributes in verbs/cm structures")
Reported-by: Aleksey Senin <alekseys@mellanox.com>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
In debugging an application that receives -ENOMEM from ib_reg_mr(), I
found that ib_umem_get() can fail because the pinned_vm count has
wrapped causing it to always be larger than the lock limit even with
RLIMIT_MEMLOCK set to RLIM_INFINITY.
The wrapping of pinned_vm occurs because the process that calls
ib_reg_mr() will have its mm->pinned_vm count incremented. Later a
different process with a different mm_struct than the one that
allocated the ib_umem struct ends up releasing it which results in
decrementing the new processes mm->pinned_vm count past zero and
wrapping.
I'm not entirely sure what circumstances cause a different process to
release the ib_umem than the one that allocated it but the kernel
stack trace of the freeing process from my situation looks like the
following:
Call Trace:
[<ffffffff814d64b1>] dump_stack+0x19/0x1b
[<ffffffffa0b522a5>] ib_umem_release+0x1f5/0x200 [ib_core]
[<ffffffffa0b90681>] mlx4_ib_destroy_qp+0x241/0x440 [mlx4_ib]
[<ffffffffa0b4d93c>] ib_destroy_qp+0x12c/0x170 [ib_core]
[<ffffffffa0cc7129>] ib_uverbs_close+0x259/0x4e0 [ib_uverbs]
[<ffffffff81141cba>] __fput+0xba/0x240
[<ffffffff81141e4e>] ____fput+0xe/0x10
[<ffffffff81060894>] task_work_run+0xc4/0xe0
[<ffffffff810029e5>] do_notify_resume+0x95/0xa0
[<ffffffff814e3dd0>] int_signal+0x12/0x17
The following patch fixes the issue by storing the pid struct of the
process that calls ib_umem_get() so that ib_umem_release and/or
ib_umem_account() can properly decrement the pinned_vm count of the
correct mm_struct.
Signed-off-by: Shawn Bohrer <sbohrer@rgmadvisors.com>
Reviewed-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Using the new registration mechanism, define a flag that indicates the
user wishes to process RMPP messages in user space rather than have
the kernel process them.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Registrations options are specified through flags. Definitions of flags will
be in subsequent patches.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Registration failures can be difficult to debug from userspace. This
gives more visibility.
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Use dev_* style print when struct device is available.
Also combine previously line broken user-visible strings as per
Documentation/CodingStyle:
"However, never break user-visible strings such as printk messages,
because that breaks the ability to grep for them."
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
[ Remove PFX so the patch actually builds. - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
Use dev_* style print when struct device is available.
Also combine previously line broken user-visible strings as per
Documentation/CodingStyle:
"However, never break user-visible strings such as printk messages,
because that breaks the ability to grep for them."
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
If the user creates a listening cm_id with backlog of 0 the IWCM ends
up not allowing any connection requests at all. The correct behavior
is for the IWCM to pick a default value if the user backlog parameter
is zero.
Lustre from version 1.8.8 onward uses a backlog of 0, which breaks
iwarp support without this fix.
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Memory re-registration is a feature that enables changing the
attributes of a memory region registered by user-space, including PD,
translation (address and length) and access flags.
Add the required support in uverbs and the kernel verbs API.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds iWARP Port Mapper (IWPM) Version 2 support. The iWARP
Port Mapper implementation is based on the port mapper specification
section in the Sockets Direct Protocol paper -
http://www.rdmaconsortium.org/home/draft-pinkerton-iwarp-sdp-v1.0.pdf
Existing iWARP RDMA providers use the same IP address as the native
TCP/IP stack when creating RDMA connections. They need a mechanism to
claim the TCP ports used for RDMA connections to prevent TCP port
collisions when other host applications use TCP ports. The iWARP Port
Mapper provides a standard mechanism to accomplish this. Without this
service it is possible for RDMA application to bind/listen on the same
port which is already being used by native TCP host application. If
that happens the incoming TCP connection data can be passed to the
RDMA stack with error.
The iWARP Port Mapper solution doesn't contain any changes to the
existing network stack in the kernel space. All the changes are
contained with the infiniband tree and also in user space.
The iWARP Port Mapper service is implemented as a user space daemon
process. Source for the IWPM service is located at
http://git.openfabrics.org/git?p=~tnikolova/libiwpm-1.0.0/.git;a=summary
The iWARP driver (port mapper client) sends to the IWPM service the
local IP address and TCP port it has received from the RDMA
application, when starting a connection. The IWPM service performs a
socket bind from user space to get an available TCP port, called a
mapped port, and communicates it back to the client. In that sense,
the IWPM service is used to map the TCP port, which the RDMA
application uses to any port available from the host TCP port
space. The mapped ports are used in iWARP RDMA connections to avoid
collisions with native TCP stack which is aware that these ports are
taken. When an RDMA connection using a mapped port is terminated, the
client notifies the IWPM service, which then releases the TCP port.
The message exchange between the IWPM service and the iWARP drivers
(between user space and kernel space) is implemented using netlink
sockets.
1) Netlink interface functions are added: ibnl_unicast() and
ibnl_mulitcast() for sending netlink messages to user space
2) The signature of the existing ibnl_put_msg() is changed to be more
generic
3) Two netlink clients are added: RDMA_NL_NES, RDMA_NL_C4IW
corresponding to the two iWarp drivers - nes and cxgb4 which use
the IWPM service
4) Enums are added to enumerate the attributes in the netlink
messages, which are exchanged between the user space IWPM service
and the iWARP drivers
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Reviewed-by: PJ Waskiewicz <pj.waskiewicz@solidfire.com>
[ Fold in range checking fixes and nlh_next removal as suggested by Dan
Carpenter and Steve Wise. Fix sparse endianness in hash. - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
Avoid that closing /dev/infiniband/umad<n> or /dev/infiniband/issm<n>
triggers a use-after-free. __fput() invokes f_op->release() before it
invokes cdev_put(). Make sure that the ib_umad_device structure is
freed by the cdev_put() call instead of f_op->release(). This avoids
that changing the port mode from IB into Ethernet and back to IB
followed by restarting opensmd triggers the following kernel oops:
general protection fault: 0000 [#1] PREEMPT SMP
RIP: 0010:[<ffffffff810cc65c>] [<ffffffff810cc65c>] module_put+0x2c/0x170
Call Trace:
[<ffffffff81190f20>] cdev_put+0x20/0x30
[<ffffffff8118e2ce>] __fput+0x1ae/0x1f0
[<ffffffff8118e35e>] ____fput+0xe/0x10
[<ffffffff810723bc>] task_work_run+0xac/0xe0
[<ffffffff81002a9f>] do_notify_resume+0x9f/0xc0
[<ffffffff814b8398>] int_signal+0x12/0x17
Reference: https://bugzilla.kernel.org/show_bug.cgi?id=75051
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Cc: <stable@vger.kernel.org> # 3.x: 8ec0a0e6b5: IB/umad: Fix error handling
Signed-off-by: Roland Dreier <roland@purestorage.com>
The ports kobject isn't being released during error flow in device
registration. This patch refactors the ports kobject cleanup into a
single function called from both the error flow in device registration
and from the unregistration function.
A couple of attributes aren't being deleted (iw_stats_group, and
ib_class_attributes). While this may be handled implicitly by the
destruction of their kobjects, it seems better to handle all the
attributes the same way.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
[ Make free_port_list_attributes() static. - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
When encountering an error during the add_port function, adding a port
to sysfs, the port kobject is freed without being deleted from sysfs.
Instead of freeing it directly, the patch uses kobject_put to release
the kobject and delete it.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The ib_core module will call kobject_get on the parent object of each
kobject it creates. This is redundant since kobject_add does that
anyway.
As a side effect, this patch should fix leaking the ports kobject and
the device kobject during unregister flow, since the previous code
didn't seem to take into account the kobject_get calls on behalf of
the child kobjects.
Signed-off-by: Haggai Eran <haggaie@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Fix a few functions that are declared with __attribute_const__ in the
ib_verbs.h header file but defined without it in verbs.c. This gets rid
of the following sparse warnings:
drivers/infiniband/core/verbs.c:51:5: error: symbol 'ib_rate_to_mult' redeclared with different type (originally declared at include/rdma/ib_verbs.h:469) - different modifiers
drivers/infiniband/core/verbs.c:68:14: error: symbol 'mult_to_ib_rate' redeclared with different type (originally declared at include/rdma/ib_verbs.h:607) - different modifiers
drivers/infiniband/core/verbs.c:85:5: error: symbol 'ib_rate_to_mbps' redeclared with different type (originally declared at include/rdma/ib_verbs.h:476) - different modifiers
drivers/infiniband/core/verbs.c:111:1: error: symbol 'rdma_node_get_transport' redeclared with different type (originally declared at include/rdma/ib_verbs.h:84) - different modifiers
Signed-off-by: Roland Dreier <roland@purestorage.com>
Properly convert gfp_t & result to bool to fix:
drivers/infiniband/core/sa_query.c:621:33: warning: incorrect type in initializer (different base types)
drivers/infiniband/core/sa_query.c:621:33: expected bool [unsigned] [usertype] preload
drivers/infiniband/core/sa_query.c:621:33: got restricted gfp_t
Signed-off-by: Roland Dreier <roland@purestorage.com>
Avoid leaking a kref count in ib_umad_open() if port->ib_dev == NULL
or if nonseekable_open() fails.
Avoid leaking a kref count, that sm_sem is kept down and also that the
IB_PORT_SM capability mask is not cleared in ib_umad_sm_open() if
nonseekable_open() fails.
Since container_of() never returns NULL, remove the code that tests
whether container_of() returns NULL.
Moving the kref_get() call from the start of ib_umad_*open() to the
end is safe since it is the responsibility of the caller of these
functions to ensure that the cdev pointer remains valid until at least
when these functions return.
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Cc: <stable@vger.kernel.org>
[ydroneaud@opteya.com: rework a bit to reduce the amount of code changed]
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
[ nonseekable_open() can't actually fail, but.... - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
The code that resolves the passive side source MAC within the rdma_cm
connection request handler was both redundant and buggy, so remove it.
It was redundant since later, when an RC QP is modified to RTR state,
the resolution will take place in the ib_core module. It was buggy
because this callback also deals with UD SIDR exchange, for which we
incorrectly looked at the REQ member of the CM event and dereferenced
a random value.
Fixes: dd5f03beb4 ("IB/core: Ethernet L2 attributes in verbs/cm structures")
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Running with DMA_API_DEBUG enabled and not checking for DMA mapping
errors triggers a kernel stack trace with "DMA-API: device driver
failed to check map error" message. Add these checks to the MAD
module, both to be be more robust and also eliminate these
false-positive stack traces.
Signed-off-by: Yan Burman <yanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Introduce a verbs interface for signature-related operations. A
signature handover operation configures the layouts of data and
protection attributes both in memory and wire domains.
Signature operations are:
- INSERT:
Generate and insert protection information when handing over
data from input space to output space.
- validate and STRIP:
Validate protection information and remove it when handing over
data from input space to output space.
- validate and PASS:
Validate protection information and pass it when handing over
data from input space to output space.
Once the signature handover opration is done, the HCA will offload
data integrity generation/validation while performing the actual data
transfer.
Additions:
1. HCA signature capabilities in device attributes
Verbs provider supporting signature handover operations fills
relevant fields in device attributes structure returned by
ib_query_device.
2. QP creation flag IB_QP_CREATE_SIGNATURE_EN
Creating a QP that will carry signature handover operations may
require some special preparations from the verbs provider. So we
add QP creation flag IB_QP_CREATE_SIGNATURE_EN to declare that the
created QP may carry out signature handover operations. Expose
signature support to verbs layer (no support for now).
3. New send work request IB_WR_REG_SIG_MR
Signature handover work request. This WR will define the signature
handover properties of the memory/wire domains as well as the
domains layout. The purpose of this work request is to bind all
the needed information for the signature operation:
- data to be transferred: wr->sg_list (ib_sge).
* The raw data, pre-registered to a single MR (normally, before
signature, this MR would have been used directly for the data
transfer)
- data protection guards: sig_handover.prot (ib_sge).
* The data protection buffer, pre-registered to a single MR, which
contains the data integrity guards of the raw data blocks.
Note that it may not always exist, only in cases where the user is
interested in storing protection guards in memory.
- signature operation attributes: sig_handover.sig_attrs.
* Tells the HCA how to validate/generate the protection information.
Once the work request is executed, the memory region that will
describe the signature transaction will be the sig_mr. The
application can now go ahead and send the sig_mr.rkey or use the
sig_mr.lkey for data transfer.
4. New Verb ib_check_mr_status
check_mr_status verb checks the status of the memory region post
transaction. The first check that may be used is
IB_MR_CHECK_SIG_STATUS, which will indicate if any signature
errors are pending for a specific signature-enabled ib_mr. This
verb is a lightwight check and is allowed to be taken from
interrupt context. An application must call this verb after it is
known that the actual data transfer has finished.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This commit introduces verbs for creating/destoying memory
regions which will allow new types of memory key operations such
as protected memory registration.
Indirect memory registration is registering several (one
of more) pre-registered memory regions in a specific layout.
The Indirect region may potentialy describe several regions
and some repitition format between them.
Protected Memory registration is registering a memory region
with various data integrity attributes which will describe protection
schemes that will be handled by the HCA in an offloaded manner.
These memory regions will be applicable for a new REG_SIG_MR
work request introduced later in this patchset.
In the future these routines may replace or implement current memory
regions creation routines existing today:
- ib_reg_user_mr
- ib_alloc_fast_reg_mr
- ib_get_dma_mr
- ib_dereg_mr
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch refactors the IB core umem code and vendor drivers to use a
linear (chained) SG table instead of chunk list. With this change the
relevant code becomes clearer—no need for nested loops to build and
use umem.
Signed-off-by: Shachar Raindel <raindel@mellanox.com>
Signed-off-by: Yishai Hadas <yishaih@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Pull networking updates from David Miller:
1) BPF debugger and asm tool by Daniel Borkmann.
2) Speed up create/bind in AF_PACKET, also from Daniel Borkmann.
3) Correct reciprocal_divide and update users, from Hannes Frederic
Sowa and Daniel Borkmann.
4) Currently we only have a "set" operation for the hw timestamp socket
ioctl, add a "get" operation to match. From Ben Hutchings.
5) Add better trace events for debugging driver datapath problems, also
from Ben Hutchings.
6) Implement auto corking in TCP, from Eric Dumazet. Basically, if we
have a small send and a previous packet is already in the qdisc or
device queue, defer until TX completion or we get more data.
7) Allow userspace to manage ipv6 temporary addresses, from Jiri Pirko.
8) Add a qdisc bypass option for AF_PACKET sockets, from Daniel
Borkmann.
9) Share IP header compression code between Bluetooth and IEEE802154
layers, from Jukka Rissanen.
10) Fix ipv6 router reachability probing, from Jiri Benc.
11) Allow packets to be captured on macvtap devices, from Vlad Yasevich.
12) Support tunneling in GRO layer, from Jerry Chu.
13) Allow bonding to be configured fully using netlink, from Scott
Feldman.
14) Allow AF_PACKET users to obtain the VLAN TPID, just like they can
already get the TCI. From Atzm Watanabe.
15) New "Heavy Hitter" qdisc, from Terry Lam.
16) Significantly improve the IPSEC support in pktgen, from Fan Du.
17) Allow ipv4 tunnels to cache routes, just like sockets. From Tom
Herbert.
18) Add Proportional Integral Enhanced packet scheduler, from Vijay
Subramanian.
19) Allow openvswitch to mmap'd netlink, from Thomas Graf.
20) Key TCP metrics blobs also by source address, not just destination
address. From Christoph Paasch.
21) Support 10G in generic phylib. From Andy Fleming.
22) Try to short-circuit GRO flow compares using device provided RX
hash, if provided. From Tom Herbert.
The wireless and netfilter folks have been busy little bees too.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (2064 commits)
net/cxgb4: Fix referencing freed adapter
ipv6: reallocate addrconf router for ipv6 address when lo device up
fib_frontend: fix possible NULL pointer dereference
rtnetlink: remove IFLA_BOND_SLAVE definition
rtnetlink: remove check for fill_slave_info in rtnl_have_link_slave_info
qlcnic: update version to 5.3.55
qlcnic: Enhance logic to calculate msix vectors.
qlcnic: Refactor interrupt coalescing code for all adapters.
qlcnic: Update poll controller code path
qlcnic: Interrupt code cleanup
qlcnic: Enhance Tx timeout debugging.
qlcnic: Use bool for rx_mac_learn.
bonding: fix u64 division
rtnetlink: add missing IFLA_BOND_AD_INFO_UNSPEC
sfc: Use the correct maximum TX DMA ring size for SFC9100
Add Shradha Shah as the sfc driver maintainer.
net/vxlan: Share RX skb de-marking and checksum checks with ovs
tulip: cleanup by using ARRAY_SIZE()
ip_tunnel: clear IPCB in ip_tunnel_xmit() in case dst_link_failure() is called
net/cxgb4: Don't retrieve stats during recovery
...
Fix the below "make W=1" build warning:
drivers/infiniband/core/iwcm.c: In function ‘destroy_cm_id’:
drivers/infiniband/core/iwcm.c:330: warning: variable ‘ret’ set but not used
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
If addr is not a linklocal address, the code incorrectly fails to
return and ends up assigning the scope ID to the scope id of the
address, which is wrong. Fix by checking if it's a link local address
first, and immediately return 0 if not.
Signed-off-by: Somnath Kotur <somnath.kotur@emulex.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Add the missing unlock before return from function cm_init_qp_rtr_attr()
in the error handling case.
Fixes: dd5f03beb4 ("IB/core: Ethernet L2 attributes in verbs/cm structures")
Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: Roland Dreier <roland@purestorage.com>
IP based addressing introduces the usage of rdma_addr_find_dmac_by_grh()
within ib_core. Since this function is declared in ib_addr, ib_addr
should be a part of the core INFINIBAND modules, rather than
INFINIBAND_ADDR_TRANS.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Existing user space applications provide only IBoE L3 address
attributes to the kernel when they issue a modify QP modify. To work
with them and let such apps (plus kernel consumers which don't use the
RDMA-CM) keep working transparently under the IBoE GID IP addressing
changes, add an Eth L2 address resolution helper.
Signed-off-by: Moni Shoua <monis@mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Currently, the IB core and specifically the RDMA-CM assumes that IBoE
(RoCE) gids encode related Ethernet netdevice interface MAC address
and possibly VLAN id.
Change GIDs to be treated as they encode interface IP address.
Since Ethernet layer 2 address parameters are not longer encoded
within gids, we have to extend the Infiniband address structures (e.g.
ib_ah_attr) with layer 2 address parameters, namely mac and vlan.
Signed-off-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Add the complementary RDMA_NODE_USNIC_UDP for RDMA_TRANSPORT_USNIC_UDP.
Signed-off-by: Upinder Malhi <umalhi@cisco.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch removes the net_random and net_srandom macros and replaces
them with direct calls to the prandom ones. As new commits only seem to
use prandom_u32 there is no use to keep them around.
This change makes it easier to grep for users of prandom_u32.
Signed-off-by: Aruna-Hewapathirane <aruna.hewapathirane@gmail.com>
Suggested-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch add the support for Ethernet L2 attributes in the
verbs/cm/cma structures.
When dealing with L2 Ethernet, we should use smac, dmac, vlan ID and priority
in a similar manner that the IB L2 (and the L4 PKEY) attributes are used.
Thus, those attributes were added to the following structures:
* ib_ah_attr - added dmac
* ib_qp_attr - added smac and vlan_id, (sl remains vlan priority)
* ib_wc - added smac, vlan_id
* ib_sa_path_rec - added smac, dmac, vlan_id
* cm_av - added smac and vlan_id
For the path record structure, extra care was taken to avoid the new
fields when packing it into wire format, so we don't break the IB CM
and SA wire protocol.
On the active side, the CM fills. its internal structures from the
path provided by the ULP. We add there taking the ETH L2 attributes
and placing them into the CM Address Handle (struct cm_av).
On the passive side, the CM fills its internal structures from the WC
associated with the REQ message. We add there taking the ETH L2
attributes from the WC.
When the HW driver provides the required ETH L2 attributes in the WC,
they set the IB_WC_WITH_SMAC and IB_WC_WITH_VLAN flags. The IB core
code checks for the presence of these flags, and in their absence does
address resolution from the ib_init_ah_from_wc() helper function.
ib_modify_qp_is_ok is also updated to consider the link layer. Some
parameters are mandatory for Ethernet link layer, while they are
irrelevant for IB. Vendor drivers are modified to support the new
function signature.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Add RDMA_TRANSPORT_USNIC_UDP which will be used by usNIC.
Signed-off-by: Upinder Malhi <umalhi@cisco.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds a check on the output buffer with access_ok(VERIFY_WRITE, ...)
to ensure the whole buffer is in userspace memory before using the
pointer in uverbs functions. If the buffer or a subset of it is not
valid, returns -EFAULT to the caller.
This will also catch invalid buffer before the final call to
copy_to_user() which happen late in most uverb functions.
Just like the check in read(2) syscall, it's a sanity check to detect
invalid parameters provided by userspace. This particular check was added
in vfs_read() by Linus Torvalds for v2.6.12 with following commit message:
https://git.kernel.org/cgit/linux/kernel/git/tglx/history.git/commit/?id=fd770e66c9a65b14ce114e171266cf6f393df502
Make read/write always do the full "access_ok()" tests.
The actual user copy will do them too, but only for the
range that ends up being actually copied. That hides
bugs when the range has been clamped by file size or other
issues.
Note: there's no need to check input buffer since vfs_write() already does
access_ok(VERIFY_READ, ...) as part of write() syscall.
Link: http://marc.info/?i=cover.1387273677.git.ydroneaud@opteya.com
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Since ib_copy_from_udata() doesn't check yet the available input data
length before accessing userspace memory, an explicit check of this
length is required to prevent:
- reading past the user provided buffer,
- underflow when subtracting the expected command size from the input
length.
This will ensure the newly added flow steering uverbs don't try to
process truncated commands.
Link: http://marc.info/?i=cover.1386798254.git.ydroneaud@opteya.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
If the flow_spec items parsed count does not match the number of items
declared in the flow_attr command, or if not all bytes are used for
flow_spec items (eg. trailing garbage), a log message is reported and
the function leave through the error path. Unfortunately the error
code is currently not set.
This patch set error code to -EINVAL in such cases, so that the error
is reported to userspace instead of silently fail.
Link: http://marc.info/?i=cover.1386798254.git.ydroneaud@opteya.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
As noted by Daniel Vetter in its article "Botching up ioctls"[1]
"Check *all* unused fields and flags and all the padding for whether
it's 0, and reject the ioctl if that's not the case. Otherwise
your nice plan for future extensions is going right down the
gutters since someone *will* submit an ioctl struct with random
stack garbage in the yet unused parts. Which then bakes in the ABI
that those fields can never be used for anything else but garbage."
It's important to ensure that reserved fields are set to known value,
so that it will be possible to use them latter to extend the ABI.
The same reasonning apply to comp_mask field present in newer uverbs
command: per commit 22878dbc91 ("IB/core: Better checking of
userspace values for receive flow steering"), unsupported values in
comp_mask are rejected.
[1] http://blog.ffwll.ch/2013/11/botching-up-ioctls.html
Link: http://marc.info/?i=cover.1386798254.git.ydroneaud@opteya.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
As noted by Daniel Vetter in its article "Botching up ioctls"[1]
"Check *all* unused fields and flags and all the padding for whether
it's 0, and reject the ioctl if that's not the case. Otherwise
your nice plan for future extensions is going right down the
gutters since someone *will* submit an ioctl struct with random
stack garbage in the yet unused parts. Which then bakes in the ABI
that those fields can never be used for anything else but garbage."
It's important to ensure that reserved fields are set to known value,
so that it will be possible to use them latter to extend the ABI.
The same reasonning apply to comp_mask field present in newer uverbs
command: per commit 22878dbc91 ("IB/core: Better checking of
userspace values for receive flow steering"), unsupported values in
comp_mask are rejected.
[1] http://blog.ffwll.ch/2013/11/botching-up-ioctls.html
Link: http://marc.info/?i=cover.1386798254.git.ydroneaud@opteya.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Trying to have a ternary operator to choose between NULL (or 0) and the
real pointer value in invocations leads to an impossible choice between
a sparse error about a literal 0 used as a NULL pointer, and a gcc
warning about "pointer/integer type mismatch in conditional expression."
Rather than clutter the source with more casts, move the ternary
operator into a new INIT_UDATA_BUF_OR_NULL() macro, which makes it
easier to use and simplifies its callers.
Reported-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Userspace input buffer is not modified by kernel, so it can be 'const'.
This is also a prerequisite to remove the implicit cast
from INIT_UDATA().
Link: http://marc.info/?i=cover.1386798254.git.ydroneaud@opteya.com>
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
rem_ref() calls iwcm_deref_id(), which will wake up any blockers on
cm_id_priv->destroy_comp if the refcnt hits 0. That will unblock
someone in iw_destroy_cm_id() which will free the cmid. If that
happens before rem_ref() calls test_bit(IWCM_F_CALLBACK_DESTROY,
&cm_id_priv->flags), then the test_bit() will touch freed memory.
The fix is to read the bit first, then deref. We should never be in
iw_destroy_cm_id() with IWCM_F_CALLBACK_DESTROY set, and there is a
BUG_ON() to make sure of that.
Signed-off-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
- Re-enable flow steering verbs with new improved userspace ABI
- Fixes for slow connection due to GID lookup scalability
- IPoIB fixes
- Many fixes to HW drivers including mlx4, mlx5, ocrdma and qib
- Further improvements to SRP error handling
- Add new transport type for Cisco usNIC
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.14 (GNU/Linux)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=xxC+
-----END PGP SIGNATURE-----
Merge tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband
Pull infiniband/rdma updates from Roland Dreier:
- Re-enable flow steering verbs with new improved userspace ABI
- Fixes for slow connection due to GID lookup scalability
- IPoIB fixes
- Many fixes to HW drivers including mlx4, mlx5, ocrdma and qib
- Further improvements to SRP error handling
- Add new transport type for Cisco usNIC
* tag 'rdma-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (66 commits)
IB/core: Re-enable create_flow/destroy_flow uverbs
IB/core: extended command: an improved infrastructure for uverbs commands
IB/core: Remove ib_uverbs_flow_spec structure from userspace
IB/core: Use a common header for uverbs flow_specs
IB/core: Make uverbs flow structure use names like verbs ones
IB/core: Rename 'flow' structs to match other uverbs structs
IB/core: clarify overflow/underflow checks on ib_create/destroy_flow
IB/ucma: Convert use of typedef ctl_table to struct ctl_table
IB/cm: Convert to using idr_alloc_cyclic()
IB/mlx5: Fix page shift in create CQ for userspace
IB/mlx4: Fix device max capabilities check
IB/mlx5: Fix list_del of empty list
IB/mlx5: Remove dead code
IB/core: Encorce MR access rights rules on kernel consumers
IB/mlx4: Fix endless loop in resize CQ
RDMA/cma: Remove unused argument and minor dead code
RDMA/ucma: Discard events for IDs not yet claimed by user space
IB/core: Add Cisco usNIC rdma node and transport types
RDMA/nes: Remove self-assignment from nes_query_qp()
IB/srp: Report receive errors correctly
...
This commit reverts commit 7afbddfae9 ("IB/core: Temporarily disable
create_flow/destroy_flow uverbs"). Since the uverbs extensions
functionality was experimental for v3.12, this patch re-enables the
support for them and flow-steering for v3.13.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit 400dbc9658 ("IB/core: Infrastructure for extensible uverbs
commands") added an infrastructure for extensible uverbs commands
while later commit 436f2ad05a ("IB/core: Export ib_create/destroy_flow
through uverbs") exported ib_create_flow()/ib_destroy_flow() functions
using this new infrastructure.
According to the commit 400dbc9658, the purpose of this
infrastructure is to support passing around provider (eg. hardware)
specific buffers when userspace issue commands to the kernel, so that
it would be possible to extend uverbs (eg. core) buffers independently
from the provider buffers.
But the new kernel command function prototypes were not modified to
take advantage of this extension. This issue was exposed by Roland
Dreier in a previous review[1].
So the following patch is an attempt to a revised extensible command
infrastructure.
This improved extensible command infrastructure distinguish between
core (eg. legacy)'s command/response buffers from provider
(eg. hardware)'s command/response buffers: each extended command
implementing function is given a struct ib_udata to hold core
(eg. uverbs) input and output buffers, and another struct ib_udata to
hold the hw (eg. provider) input and output buffers.
Having those buffers identified separately make it easier to increase
one buffer to support extension without having to add some code to
guess the exact size of each command/response parts: This should make
the extended functions more reliable.
Additionally, instead of relying on command identifier being greater
than IB_USER_VERBS_CMD_THRESHOLD, the proposed infrastructure rely on
unused bits in command field: on the 32 bits provided by command
field, only 6 bits are really needed to encode the identifier of
commands currently supported by the kernel. (Even using only 6 bits
leaves room for about 23 new commands).
So this patch makes use of some high order bits in command field to
store flags, leaving enough room for more command identifiers than one
will ever need (eg. 256).
The new flags are used to specify if the command should be processed
as an extended one or a legacy one. While designing the new command
format, care was taken to make usage of flags itself extensible.
Using high order bits of the commands field ensure that newer
libibverbs on older kernel will properly fail when trying to call
extended commands. On the other hand, older libibverbs on newer kernel
will never be able to issue calls to extended commands.
The extended command header includes the optional response pointer so
that output buffer length and output buffer pointer are located
together in the command, allowing proper parameters checking. This
should make implementing functions easier and safer.
Additionally the extended header ensure 64bits alignment, while making
all sizes multiple of 8 bytes, extending the maximum buffer size:
legacy extended
Maximum command buffer: 256KBytes 1024KBytes (512KBytes + 512KBytes)
Maximum response buffer: 256KBytes 1024KBytes (512KBytes + 512KBytes)
For the purpose of doing proper buffer size accounting, the headers
size are no more taken in account in "in_words".
One of the odds of the current extensible infrastructure, reading
twice the "legacy" command header, is fixed by removing the "legacy"
command header from the extended command header: they are processed as
two different parts of the command: memory is read once and
information are not duplicated: it's making clear that's an extended
command scheme and not a different command scheme.
The proposed scheme will format input (command) and output (response)
buffers this way:
- command:
legacy header +
extended header +
command data (core + hw):
+----------------------------------------+
| flags | 00 00 | command |
| in_words | out_words |
+----------------------------------------+
| response |
| response |
| provider_in_words | provider_out_words |
| padding |
+----------------------------------------+
| |
. <uverbs input> .
. (in_words * 8) .
| |
+----------------------------------------+
| |
. <provider input> .
. (provider_in_words * 8) .
| |
+----------------------------------------+
- response, if present:
+----------------------------------------+
| |
. <uverbs output space> .
. (out_words * 8) .
| |
+----------------------------------------+
| |
. <provider output space> .
. (provider_out_words * 8) .
| |
+----------------------------------------+
The overall design is to ensure that the extensible infrastructure is
itself extensible while begin more reliable with more input and bound
checking.
Note:
The unused field in the extended header would be perfect candidate to
hold the command "comp_mask" (eg. bit field used to handle
compatibility). This was suggested by Roland Dreier in a previous
review[2]. But "comp_mask" field is likely to be present in the uverb
input and/or provider input, likewise for the response, as noted by
Matan Barak[3], so it doesn't make sense to put "comp_mask" in the
header.
[1]:
http://marc.info/?i=CAL1RGDWxmM17W2o_era24A-TTDeKyoL6u3NRu_=t_dhV_ZA9MA@mail.gmail.com
[2]:
http://marc.info/?i=CAL1RGDXJtrc849M6_XNZT5xO1+ybKtLWGq6yg6LhoSsKpsmkYA@mail.gmail.com
[3]:
http://marc.info/?i=525C1149.6000701@mellanox.com
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Link: http://marc.info/?i=cover.1383773832.git.ydroneaud@opteya.com
[ Convert "ret ? ret : 0" to the equivalent "ret". - Roland ]
Signed-off-by: Roland Dreier <roland@purestorage.com>
The structure holding any types of flow_spec is of no use to
userspace. It would be wrong for userspace to do:
struct ib_uverbs_flow_spec flow_spec;
flow_spec.type = IB_FLOW_SPEC_TCP;
flow_spec.size = sizeof(flow_spec);
Instead, userspace should use the dedicated flow_spec structure for
- Ethernet : struct ib_uverbs_flow_spec_eth,
- IPv4 : struct ib_uverbs_flow_spec_ipv4,
- TCP/UDP : struct ib_uverbs_flow_spec_tcp_udp.
In other words, struct ib_uverbs_flow_spec is a "virtual" data
structure that can only be use by the kernel as an alias to the other.
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Link: http://marc.info/?i=cover.1383773832.git.ydroneaud@opteya.com
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds "flow" prefix to most of data structure added as part
of commit 436f2ad05a ("IB/core: Export ib_create/destroy_flow through
uverbs") to keep those names in sync with the data structures added in
commit 319a441d13 ("IB/core: Add receive flow steering support").
It's just a matter of translating 'ib_flow' to 'ib_uverbs_flow'.
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Link: http://marc.info/?i=cover.1383773832.git.ydroneaud@opteya.com
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit 436f2ad05a ("IB/core: Export ib_create/destroy_flow through
uverbs") added public data structures to support receive flow
steering. The new structs are not following the 'uverbs' pattern:
they're lacking the common prefix 'ib_uverbs'.
This patch replaces ib_kern prefix by ib_uverbs.
Signed-off-by: Yann Droneaud <ydroneaud@opteya.com>
Link: http://marc.info/?i=cover.1383773832.git.ydroneaud@opteya.com
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch fixes the following issues:
1. Unneeded checks were removed
2. Removed the fixed size out of flow_attr.size, thus simplifying the checks.
3. Remove a 32bit hole on 64bit systems with strict alignment in
struct ib_kern_flow_att by adding a reserved field.
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This typedef is unnecessary and should just be removed.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit 3e6628c4b3 ("idr: introduce idr_alloc_cyclic()") adds a new
idr_alloc_cyclic() routine and converts several of these users to it.
This is just a missed one - add it.
Signed-off-by: Zhao Hongjiang <zhaohongjiang@huawei.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Enforce the rule that when requesting remote write or atomic permissions, local
write must be indicated as well. See IB spec 11.2.8.2.
Spotted by: Hagay Abramovsky <hagaya@mellanox.com>
Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The dev variable is never assigned after being initialised.
Signed-off-by: Michal Nazarewicz <mina86@mina86.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Problem reported by Avneesh Pant <avneesh.pant@oracle.com>:
It looks like we are triggering a bug in RDMA CM/UCM interaction.
The bug specifically hits when we have an incoming connection
request and the connecting process dies BEFORE the passive end of
the connection can process the request i.e. it does not call
rdma_get_cm_event() to retrieve the initial connection event. We
were able to triage this further and have some additional
information now.
In the example below when P1 dies after issuing a connect request
as the CM id is being destroyed all outstanding connects (to P2)
are sent a reject message. We see this reject message being
received on the passive end and the appropriate CM ID created for
the initial connection message being retrieved in cm_match_req().
The problem is in the ucma_event_handler() code when this reject
message is delivered to it and the initial connect message itself
HAS NOT been delivered to the client. In fact the client has not
even called rdma_cm_get_event() at this stage so we haven't
allocated a new ctx in ucma_get_event() and updated the new
connection CM_ID to point to the new UCMA context.
This results in the reject message not being dropped in
ucma_event_handler() for the new connection request as the
(if (!ctx->uid)) block is skipped since the ctx it refers to is
the listen CM id context which does have a valid UID associated
with it (I believe the new CMID for the connection initially
uses the listen CMID -> context when it is created in
cma_new_conn_id). Thus the assumption that new events for a
connection can get dropped in ucma_event_handler() is incorrect
IF the initial connect request has not been retrieved in the
first case. We end up getting a CM Reject event on the listen CM
ID and our upper layer code asserts (in fact this event does not
even have the listen_id set as that only gets set up librdmacm
for connect requests).
The solution is to verify that the cm_id being reported in the event
is the same as the cm_id referenced by the ucma context. A mismatch
indicates that the ucma context corresponds to the listen. This fix
was validated by using a modified version of librdmacm that was able
to verify the problem and see that the reject message was indeed
dropped after this patch was applied.
Signed-off-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch adds new rdma node and new rdma transport, and supporting
code used by Cisco's low latency driver called usNIC. usNIC uses its
own transport, distinct from IB and iWARP.
Signed-off-by: Upinder Malhi <umalhi@cisco.com>
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
'op' is the already RDMA_NL_GET_OP() masked 'type'. No need to mask it again.
Signed-off-by: Mathias Krause <minipli@googlemail.com>
Reviewed-by: Yann Droneaud <ydroneaud@opteya.com>
Acked-by: Sean Hefty <sean.hefty@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Currently, we don't copy the immediate data from the userspace struct
to the kernel one when UD messages are being sent.
This patch makes sure that the immediate data is set correctly.
Signed-off-by: Latchesar Ionkov <lucho@ionkov.net>
Signed-off-by: Roland Dreier <roland@purestorage.com>
As a simple optimization that should speed up the vast majority of
connect attemps on IB devices, when we are searching for the GID of an
incoming connection in the cached GID lists of devices, search the
device that received the incoming connection request first. If we
don't find it there, then move on to other devices.
This reduces the time to perform 10,000 connections considerably.
Prior to this patch, a bad run of cmtime would look like this:
connect : 12399.26 12351.10 8609.00 1239.93
With this patch, it looks more like this:
connect : 5864.86 5799.80 8876.00 586.49
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The cma_acquire_dev function was changed by commit 3c86aa70bf
("RDMA/cm: Add RDMA CM support for IBoE devices") to use find_gid_port()
because multiport devices might have either IB or IBoE formatted gids.
The old function assumed that all ports on the same device used the
same GID format.
However, when it was changed to use find_gid_port(), we inadvertently
lost usage of the GID cache. This turned out to be a very costly
change. In our testing, each iteration through each index of the GID
table takes roughly 35us. When you have multiple devices in a system,
and the GID you are looking for is on one of the later devices, the
code loops through all of the GID indexes on all of the early devices
before it finally succeeds on the target device. This pathological
search behavior combined with 35us per GID table index retrieval
results in results such as the following from the cmtime application
that's part of the latest librdmacm git repo:
ib1:
step total ms max ms min us us / conn
create id : 29.42 0.04 1.00 2.94
bind addr : 186705.66 19.00 18556.00 18670.57
resolve addr : 41.93 9.68 619.00 4.19
resolve route: 486.93 0.48 101.00 48.69
create qp : 4021.95 6.18 330.00 402.20
connect : 68350.39 68588.17 24632.00 6835.04
disconnect : 1460.43 252.65-1862269.00 146.04
destroy : 41.16 0.04 2.00 4.12
ib0:
step total ms max ms min us us / conn
create id : 28.61 0.68 1.00 2.86
bind addr : 2178.86 2.95 201.00 217.89
resolve addr : 51.26 16.85 845.00 5.13
resolve route: 620.08 0.43 92.00 62.01
create qp : 3344.40 6.36 273.00 334.44
connect : 6435.99 6368.53 7844.00 643.60
disconnect : 5095.38 321.90 757.00 509.54
destroy : 37.13 0.02 2.00 3.71
Clearly, both the bind address and connect operations suffer
a huge penalty for being anything other than the default
GID on the first port in the system.
After applying this patch, the numbers now look like this:
ib1:
step total ms max ms min us us / conn
create id : 30.15 0.03 1.00 3.01
bind addr : 80.27 0.04 7.00 8.03
resolve addr : 43.02 13.53 589.00 4.30
resolve route: 482.90 0.45 100.00 48.29
create qp : 3986.55 5.80 330.00 398.66
connect : 7141.53 7051.29 5005.00 714.15
disconnect : 5038.85 193.63 918.00 503.88
destroy : 37.02 0.04 2.00 3.70
ib0:
step total ms max ms min us us / conn
create id : 34.27 0.05 1.00 3.43
bind addr : 26.45 0.04 1.00 2.64
resolve addr : 38.25 10.54 760.00 3.82
resolve route: 604.79 0.43 97.00 60.48
create qp : 3314.95 6.34 273.00 331.49
connect : 12399.26 12351.10 8609.00 1239.93
disconnect : 5096.76 270.72 1015.00 509.68
destroy : 37.10 0.03 2.00 3.71
It's worth noting that we still suffer a bit of a penalty on
connect to the wrong device, but the penalty is much less than
it used to be. Follow on patches deal with this penalty.
Many thanks to Neil Horman for helping to track the source of
slow function that allowed us to track down the fact that
the original patch I mentioned above backed out cache usage
and identify just how much that impacted the system.
Signed-off-by: Doug Ledford <dledford@redhat.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
