Arnd Bergmann [Tue, 14 Feb 2017 21:23:07 +0000 (22:23 +0100)]
IB/hfi1: use size_t for passing array length
gcc-7 produces a mysterious warning about the size argument being potentially out
of range:
drivers/infiniband/hw/hfi1/verbs.c: In function 'init_cntr_names':
drivers/infiniband/hw/hfi1/verbs.c:1644:2: error: 'memcpy': specified size between 18446744071562067968 and 18446744073709551615 exceeds maximum object size 9223372036854775807 [-Werror=stringop-overflow=]
This seems to refer to a the case where an 64-bit size_t gets truncated
into a negative 'int' and subsequently turned into a high 64-bit number
again.
The fix is clearly to use size_t here, which matches the type that gets
used for this value elsewhere.
Fixes: b7481944b06e ("IB/hfi1: Show statistics counters under IB stats interface") Signed-off-by: Arnd Bergmann <arnd@arndb.de> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Zhu Yanjun [Tue, 14 Feb 2017 08:43:12 +0000 (03:43 -0500)]
IB/ipoib: Remove redudant label
There are 2 labels to mark the same statements. Replace the 2 labels
with one versatile labe.
Cc: Joe Jin <joe.jin@oracle.com> Cc: Junxiao Bi <junxiao.bi@oracle.com> Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Zhu Yanjun [Fri, 10 Feb 2017 06:48:42 +0000 (01:48 -0500)]
IB/ipoib: remove the unnecessary memory free
In the function ipoib_cm_nonsrq_init_rx, the memory is not
allocated successfully. It is not necessary to free it.
CC: Joe Jin <joe.jin@oracle.com> CC: Junxiao Bi <junxiao.bi@oracle.com> Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Michael J. Ruhl [Wed, 8 Feb 2017 13:28:31 +0000 (05:28 -0800)]
IB/hfi1: Code reuse with memdup_copy
Update several usages of kmalloc/user_copy to memdup_copy and
memdup_copy_nul.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Michael J. Ruhl [Wed, 8 Feb 2017 13:28:13 +0000 (05:28 -0800)]
IB/hfi1: Do not set physical link state if DC is in the shutdown state
If the DC is in shutdown state, the set link state function will return
an error. Since this is not a failure in this state, make sure to
only call set link state if the DC is on.
Reviewed-by: Easwar Hariharan <easwar.hariharan@intel.com> Signed-off-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Use rate-limit state to limit number of messages logged
to kernel message buffer for DCC errors. Add new macro
dd_dev_info_ratelimited for that propose. Replace all
dd_dev_info calls in handle_dcc_err function with
rate-limited version.
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jakub Byczkowski <jakub.byczkowski@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Brian Welty [Wed, 8 Feb 2017 13:27:31 +0000 (05:27 -0800)]
IB/hfi1, rdmavt: Update copy_sge to use boolean arguments
Convert copy_sge and related SGE state functions to use boolean.
For determining if QP is in user mode, add helper function in rdmavt_qp.h.
This is used to determine if QP needs the last byte ordering.
While here, change rvt_pd.user to a boolean.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Reviewed-by: Dean Luick <dean.luick@intel.com> Signed-off-by: Brian Welty <brian.welty@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
The EFI variable that provides board ID is named
by the PCI address of the device, which is published
in upper-case, while the HFI1 driver reads the EFI
variable in lower-case.
This prevents returning the correct board id when
queried through sysfs. Read EFI variables in
upper-case if the lower-case read fails.
Reviewed-by: Easwar Hariharan <easwar.hariharan@intel.com> Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
There are some memory allocation calls in hfi1_create_ctxtdata()
that do not use the numa function parameter. This
can cause cache lines to be filled over QPI.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Having per-CPU reference count for each MR prevents
cache-line bouncing across the system. Thus, it
prevents bottlenecks. Use per-CPU reference counts
per MR.
The per-CPU reference count for FMRs is used in
atomic mode to allow accurate testing of the busy
state. Other MR types run in per-CPU mode MR until
they're freed.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Receive code paths use the QP's device and port
number to access the struct hfi1_ibport. When an
instance of struct hfi1_ctxtdata is present, it can
be used to access struct hfi1_ibport through a pointer.
This makes struct hfi1_ibport lookup time faster as an
array doesn't have to be indexed and access fields in
other cache-lines.
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Sebastian Sanchez <sebastian.sanchez@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
IB/hfi1: Use static CTLE with Preset 6 for integrated HFIs
After extended testing, it was found that the previous PCIe Gen
3 recipe, which used adaptive CTLE with Preset 4, could cause an
NMI/Surprise Link Down in about 1 in 100 to 1 in 1000 power cycles on
some platforms. New EV data combined with extensive empirical data
indicates that the new recipe should use static CTLE with Preset 6 for
all integrated silicon SKUs.
IB/hfi1: Process qp wait list in IRQ thread periodically
In the event that the IRQ thread is extremely busy, the
processing of an rcd wait list can be delayed by quite
a bit until the IRQ thread completes its work.
The QP reset reference count wait can then appear to be stuck, thus
causing up a QP destroy to emit the hung task diagnostic.
Fix by processing the qp wait list periodically from the thread. The
interval is a multiple (currently 4) of the MAX_PKT_RECV.
Also, reduce some of the excessive inlining. The guidelines
are per packet is ok inline, otherwise the choice is based on
likelyhood of execution.
Reviewed-by: Sebastian Sanchez <sebastian.sanchez@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Moni Shoua [Tue, 14 Feb 2017 05:24:08 +0000 (07:24 +0200)]
IB/cma: Destination and source addr families must match
The destination address in a listening rdma_id does not have an address
family. Since address family in both sides of a connection must be the
same in rdma_bind_addr() we set the address family of the destination to
the address family of the source.
This patch serves the logic in cma_port_is_unique() which requires to
know if destination address that is associated with a rdma_id is any address
(cma_zero_addr() and cma_loopback_addr()).
This can happen when port reuse is checked for a port number
that is being listened to.
Fixes: 19b752a19dce ("IB/cma: Allow port reuse for rdma_id") Signed-off-by: Moni Shoua <monis@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Majd Dibbiny [Tue, 14 Feb 2017 05:21:52 +0000 (07:21 +0200)]
IB/cma: Add default RoCE TOS to CMA configfs
Add new entry to the RDMA-CM configfs that allows users
to select default TOS for RDMA-CM QPs.
This is useful for users that want to control the TOS for legacy
applications without changing their code.
Application that sets the TOS explicitly using the rdma_set_option
API will continue to work as expected, meaning overriding the configfs
value.
CC: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Moni Shoua <monis@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Erez Shitrit [Wed, 1 Feb 2017 17:10:05 +0000 (19:10 +0200)]
IB/IPoIB: Add destination address when re-queue packet
When sending packet to destination that was not resolved yet
via path query, the driver keeps the skb and tries to re-send it
again when the path is resolved.
But when re-sending via dev_queue_xmit the kernel doesn't call
to dev_hard_header, so IPoIB needs to keep 20 bytes in the skb
and to put the destination address inside them.
In that way the dev_start_xmit will have the correct destination,
and the driver won't take the destination from the skb->data, while
nothing exists there, which causes to packet be be dropped.
The test flow is:
1. Run the SM on remote node,
2. Restart the driver.
4. Ping some destination,
3. Observe that first ICMP request will be dropped.
Fixes: fc791b633515 ("IB/ipoib: move back IB LL address into the hard header") Cc: <stable@vger.kernel.org> # v4.8+ Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Signed-off-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Tested-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Selvin Xavier [Fri, 10 Feb 2017 11:19:33 +0000 (03:19 -0800)]
RDMA/bnxt_re: Add bnxt_re RoCE driver
This patch introduces the RoCE driver for the Broadcom
NetXtreme-E 10/25/40/50G RoCE HCAs.
The RoCE driver is a two part driver that relies on the parent
bnxt_en NIC driver to operate. The changes needed in the bnxt_en
driver have already been incorporated via Dave Miller's net tree
into the mainline kernel.
The vendor official git repository for this driver is available
on github as:
https://github.com/Broadcom/linux-rdma-nxt/
Leon Romanovsky [Wed, 8 Feb 2017 15:04:09 +0000 (17:04 +0200)]
RDMA: Don't reference kernel private header from UAPI header
Remove references to private kernel header and defines from exported
ib_user_verb.h file.
The code snippet below is used to reproduce the issue:
#include <stdio.h>
#include <rdma/ib_user_verb.h>
int main(void)
{
printf("IB_USER_VERBS_ABI_VERSION = %d\n", IB_USER_VERBS_ABI_VERSION);
return 0;
}
It fails during compilation phase with an error:
➜ /tmp gcc main.c
main.c:2:31: fatal error: rdma/ib_user_verb.h: No such file or directory
#include <rdma/ib_user_verb.h>
^
compilation terminated.
Fixes: 189aba99e700 ("IB/uverbs: Extend modify_qp and support packet pacing") CC: Bodong Wang <bodong@mellanox.com> CC: Matan Barak <matanb@mellanox.com> CC: Christoph Hellwig <hch@infradead.org> Tested-by: Slava Shwartsman <slavash@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Bart Van Assche [Fri, 9 Dec 2016 19:00:33 +0000 (11:00 -0800)]
IB/srpt: Accept GUIDs as port names
Port and ACL information must be configured before an initiator
logs in. Make it possible to configure this information before
a subnet prefix has been assigned to a port by not only accepting
GIDs as target port and initiator port names but by also accepting
port GUIDs.
Add a 'priv' member to struct se_wwn to allow target drivers to
associate their own data with struct se_wwn.
Reported-by: Doug Ledford <dledford@redhat.com>
References: http://www.spinics.net/lists/linux-rdma/msg39505.html Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com> Cc: Christoph Hellwig <hch@lst.de> Cc: Nicholas Bellinger <nab@linux-iscsi.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Zhu Yanjun [Thu, 19 Jan 2017 04:16:06 +0000 (23:16 -0500)]
IB/ipoib: Remove the unnecessary error check
The function ipoib_mcast_start_thread/ipoib_ib_dev_up always return zero.
As such, in the function ipoib_open, err_stop will never be reached.
So remove this err_stop and change the return type of the function
ipoib_mcast_start_thread/ipoib_ib_dev_up to void.
Shiraz Saleem [Wed, 18 Jan 2017 17:48:29 +0000 (11:48 -0600)]
i40iw: Set maj_err and min_err in i40iw_sc_cqp_create
Set maj_err and min_err in i40iw_sc_cqp_create so that it
returns correct values for all return cases. This also
addresses an uninitialized variable warning for maj_err and
min_err in i40iw_create_cqp.
Jack Wang [Tue, 17 Jan 2017 09:11:12 +0000 (10:11 +0100)]
RDMA/core: create struct ib_port_cache
As Jason suggested, we have 4 elements for per port arrays,
it's better to have a separate structure to represent them.
It simplifies code a bit, ~ 30 lines of code less :)
Signed-off-by: Jack Wang <jinpu.wang@profitbricks.com> Reviewed-by: Michael Wang <yun.wang@profitbricks.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Zhu Yanjun [Thu, 12 Jan 2017 07:39:01 +0000 (02:39 -0500)]
IB/ipoib: function interface change
The ipoib_ib_dev_down/ipoib_ib_dev_stop return zero unconditionally
and the callers never check the returned values,
change the return type to void and remove the redundant return values.
Reviewed-by: Shan Hai <shan.hai@oracle.com> Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Moni Shoua [Thu, 5 Jan 2017 14:16:36 +0000 (16:16 +0200)]
IB/cma: Allow port reuse for rdma_id
When allocating a port number for binding to a rdma_id, assuming the
allocation is not for a specific port, the rule is to allow only ports
that were not in use before by any other rdma_id.
This condition is too strong to achieve the goal of a unique 5 tuple
rdma_id. Instead, we can compare current rdma_id with other rdma_id for
difference in one of destination port, source address and destination
address to allow port reuse.
Signed-off-by: Moni Shoua <monis@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Acked-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Zhu Yanjun [Thu, 5 Jan 2017 08:56:08 +0000 (03:56 -0500)]
IB/ipoib: Remove unnecessary returned value check
In the function ipoib_set_dev_features, the returned value is always 0.
As such, it is not necessary to check the returned value.
This is not a bug. It is a trivial problem.
Reviewed-by: Guanglei Li <guanglei.li@oracle.com> Signed-off-by: Zhu Yanjun <yanjun.zhu@oracle.com> Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Trivial fix to spelling mistake in isert_warn message
Signed-off-by: Colin Ian King <colin.king@canonical.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Maor Gottlieb [Thu, 19 Jan 2017 13:25:58 +0000 (15:25 +0200)]
IB/rxe: Fix rxe dev insertion to rxe_dev_list
The first argument of list_add_tail is the new item and the second
is the head of the list. Fix the code to pass arguments in the
right order, otherwise not all the rxe devices will be removed
during teardown.
Fixes: 8700e3e7c4857 ('Soft RoCE driver') Signed-off-by: Maor Gottlieb <maorg@mellanox.com> Reviewed-by: Moni Shoua <monis@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Kenneth Lee [Thu, 5 Jan 2017 07:00:05 +0000 (15:00 +0800)]
IB/umem: Release pid in error and ODP flow
1. Release pid before enter odp flow
2. Release pid when fail to allocate memory
Fixes: 87773dd56d54 ("IB: ib_umem_release() should decrement mm->pinned_vm from ib_umem_get") Fixes: 8ada2c1c0c1d ("IB/core: Add support for on demand paging regions") Signed-off-by: Kenneth Lee <liguozhu@hisilicon.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Ram Amrani [Tue, 24 Jan 2017 11:51:43 +0000 (13:51 +0200)]
RDMA/qedr: Dispatch port active event from qedr_add
Relying on qede to trigger qedr on startup is problematic. When probing
both if qedr loads slowly then qede can assume qedr is missing and not
trigger it. This patch adds a triggering from qedr and protects against
a race via an atomic bit.
Ram Amrani [Tue, 24 Jan 2017 11:50:36 +0000 (13:50 +0200)]
RDMA/qedr: Remove CQ spinlock from CM completion handlers
There is only a single event queue that triggers the completion
events for the RDMA CM and it is being processed serially. This means
that inherently there can no parallelism of CQ completion handler
callbacks, hence the lock is redundant.
Signed-off-by: Ram Amrani <Ram.Amrani@cavium.com> Signed-off-by: Michal Kalderon <Michal.Kalderon@cavium.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Max Gurtovoy [Tue, 17 Jan 2017 22:40:40 +0000 (00:40 +0200)]
IB/iser: remove unused variable from iser_conn struct
max_sectors calculation was fixed in commit: 9c674815d346 ("IB/iser: Fix max_sectors calculation").
Thus, iser_conn variable scsi_max_sectors is not needed anymore.
Max Gurtovoy [Tue, 17 Jan 2017 22:40:39 +0000 (00:40 +0200)]
IB/iser: Fix sg_tablesize calculation
For devices that can register page list that is bigger than
USHRT_MAX, we actually take the wrong value for sg_tablesize.
E.g: for CX4 max_fast_reg_page_list_len is 65536 (bigger than USHRT_MAX)
so we set sg_tablesize to 0 by mistake. Therefore, each IO that is
bigger than 4k splitted to "< 4k" chunks that cause performance degredation.
Remove wrong sg_tablesize assignment, and use the value that was set during
address resolution handler with the needed casting.
Israel Rukshin [Wed, 4 Jan 2017 13:59:37 +0000 (15:59 +0200)]
IB/srp: fix invalid indirect_sg_entries parameter value
After setting indirect_sg_entries module_param to huge value (e.g 500,000),
srp_alloc_req_data() fails to allocate indirect descriptors for the request
ring (kmalloc fails). This commit enforces the maximum value of
indirect_sg_entries to be SG_MAX_SEGMENTS as signified in module param
description.
Fixes: 65e8617fba17 (scsi: rename SCSI_MAX_{SG, SG_CHAIN}_SEGMENTS) Fixes: c07d424d6118 (IB/srp: add support for indirect tables that don't fit in SRP_CMD) Cc: stable@vger.kernel.org # 4.7+ Signed-off-by: Israel Rukshin <israelr@mellanox.com> Signed-off-by: Max Gurtovoy <maxg@mellanox.com> Reviewed-by: Laurence Oberman <loberman@redhat.com> Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>-- Signed-off-by: Doug Ledford <dledford@redhat.com>
Jack Wang [Mon, 2 Jan 2017 12:22:05 +0000 (13:22 +0100)]
RDMA/cma: use cached port state when bind loopback
Signed-off-by: Jack Wang <jinpu.wang@profitbricks.com> Reviewed-by: Michael Wang <yun.wang@profitbricks.com> Acked-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Jack Wang [Mon, 2 Jan 2017 12:20:44 +0000 (13:20 +0100)]
RDMA/cma: resolve to first active ib port
When we try to resolve a dest addr, if we don't give src addr,
cma core will try to resolve to our source ib device automatically.
The current logic only checks if a given port has the same
subnet_prefix as our dest, which is not enough if we use default
well known subnet_prefix on our active port, as it will be the same
as the subnet_prefix on inactive ports and we might match against
an inactive port by accident. To resolve this, we should also check
if port is active before we resolve it as a suitable src address for
a given dest.
Signed-off-by: Jack Wang <jinpu.wang@profitbricks.com> Reviewed-by: Michael Wang <yun.wang@profitbricks.com> Acked-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Jack Wang [Mon, 2 Jan 2017 12:19:24 +0000 (13:19 +0100)]
RDMA/core: export ib_get_cached_port_state
Export function for rdma_cm, patch for rdma_cm to follow.
Signed-off-by: Jack Wang <jinpu.wang@profitbricks.com> Reviewed-by: Michael Wang <yun.wang@profitbricks.com> Acked-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Jack Wang [Mon, 2 Jan 2017 12:17:36 +0000 (13:17 +0100)]
RDMA/core: add port state cache
We need a port state cache in ib_core, later we will use in rdma_cm.
Signed-off-by: Jack Wang <jinpu.wang@profitbricks.com> Reviewed-by: Michael Wang <yun.wang@profitbricks.com> Acked-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Feras Daoud [Wed, 28 Dec 2016 12:47:28 +0000 (14:47 +0200)]
IB/ipoib: Change list_del to list_del_init in the tx object
Since ipoib_cm_tx_start function and ipoib_cm_tx_reap function
belong to different work queues, they can run in parallel.
In this case if ipoib_cm_tx_reap calls list_del and release the
lock, ipoib_cm_tx_start may acquire it and call list_del_init
on the already deleted object.
Changing list_del to list_del_init in ipoib_cm_tx_reap fixes the problem.
Feras Daoud [Wed, 28 Dec 2016 12:47:27 +0000 (14:47 +0200)]
IB/ipoib: Replace list_del of the neigh->list with list_del_init
In order to resolve a situation where a few process delete
the same list element in sequence and cause panic, list_del
is replaced with list_del_init. In this case if the first
process that calls list_del releases the lock before acquiring
it again, other processes who can acquire the lock will call
list_del_init.
Fixes: b63b70d87741 ("IPoIB: Use a private hash table for path lookup") Signed-off-by: Feras Daoud <ferasda@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Alex Vesker <valex@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
Feras Daoud [Wed, 28 Dec 2016 12:47:26 +0000 (14:47 +0200)]
IB/ipoib: Use debug prints instead of warnings in RNR WC status
If a receive request has not been posted to the work queue, the incoming
message is rejected and the peer will receive a receiver-not-ready (RNR)
error. In IPoIB, IB_WC_RNR_RETRY_EXC_ERR error is part of the life cycle
therefore ipoib_cm_handle_tx_wc function will print to debug instead
of warnings.
Feras Daoud [Wed, 28 Dec 2016 12:47:24 +0000 (14:47 +0200)]
IB/ipoib: rtnl_unlock can not come after free_netdev
The ipoib_vlan_add function calls rtnl_unlock after free_netdev,
rtnl_unlock not only releases the lock, but also calls netdev_run_todo.
The latter function browses the net_todo_list array and completes the
unregistration of all its net_device instances. If we call free_netdev
before rtnl_unlock, then netdev_run_todo call over the freed device causes
panic.
To fix, move rtnl_unlock call before free_netdev call.
Feras Daoud [Wed, 28 Dec 2016 12:47:23 +0000 (14:47 +0200)]
IB/ipoib: Fix deadlock between rmmod and set_mode
When calling set_mode from sys/fs, the call flow locks the sys/fs lock
first and then tries to lock rtnl_lock (when calling ipoib_set_mod).
On the other hand, the rmmod call flow takes the rtnl_lock first
(when calling unregister_netdev) and then tries to take the sys/fs
lock. Deadlock a->b, b->a.
The problem starts when ipoib_set_mod frees it's rtnl_lck and tries
to get it after that.
Feras Daoud [Wed, 28 Dec 2016 12:47:22 +0000 (14:47 +0200)]
IB/ipoib: Fix deadlock over vlan_mutex
This patch fixes Deadlock while executing ipoib_vlan_delete.
The function takes the vlan_rwsem semaphore and calls
unregister_netdevice. The later function calls
ipoib_mcast_stop_thread that cause workqueue flush.
When the queue has one of the ipoib_ib_dev_flush_xxx events,
a deadlock occur because these events also tries to catch the
same vlan_rwsem semaphore.
To fix, unregister_netdevice should be called after releasing
the semaphore.
Fixes: cbbe1efa4972 ("IPoIB: Fix deadlock between ipoib_open() and child interface create") Signed-off-by: Feras Daoud <ferasda@mellanox.com> Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Alex Vesker <valex@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
Feras Daoud [Wed, 28 Dec 2016 12:47:21 +0000 (14:47 +0200)]
IB/ipoib: Set device connection mode only when needed
When changing the connection mode, the ipoib_set_mode function
did not check if the previous connection mode equals to the
new one. This commit adds the required check and return 0 if the new
mode equals to the previous one.
Feras Daoud [Wed, 28 Dec 2016 12:47:20 +0000 (14:47 +0200)]
IB/ipoib: When given an invalid UD MTU, give debug msg
In datagram mode, the IB UD (Unreliable Datagram) transport is used
so the MTU of the interface is equal to the IB L2 MTU minus the
IPoIB encapsulation header. Any request to change the MTU value
above the maximum range will change the MTU to the max allowed, but
will not show any warning message. An ipoib_warn is issued in such
cases, letting the user know that even though the value is legal,
it can't be currently applied.
ssh10 [Sat, 24 Dec 2016 01:44:35 +0000 (07:14 +0530)]
RDMA/cxgb4: Use AF_INET for sin_family field
Elsewhere the sin_family field holds a value with a name of the form
AF_..., so it seems reasonable to do so here as well. Also the values
of PF_INET and AF_INET are the same.
The semantic patch that makes this change is as follows:
Jason Gunthorpe [Fri, 23 Dec 2016 01:07:52 +0000 (18:07 -0700)]
RDMA/core: Fix incorrect structure packing for booleans
The RDMA core uses ib_pack() to convert from unpacked CPU structs
to on-the-wire bitpacked structs.
This process requires that 1 bit fields are declared as u8 in the
unpacked struct, otherwise the packing process does not read the
value properly and the packed result is wired to 0. Several
places wrongly used int.
Crucially this means the kernel has never, set reversible
correctly in the path record request. It has always asked for
irreversible paths even if the ULP requests otherwise.
When the kernel is used with a SM that supports this feature, it
completely breaks communication management if reversible paths are
not properly requested.
The only reason this ever worked is because opensm ignores the
reversible bit.
projects / karo-tx-linux.git / log