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>
According to RCU_INIT_POINTER()'s block comment 3.a, it can be used if
"1. This use of RCU_INIT_POINTER() is NULLing out the pointer"
it is better to use it instead of rcu_assign_pointer() because it has a
smaller overhead.
"3. The referenced data structure has already been exposed to readers either
at compile time or via rcu_assign_pointer() -and-
a. You have not made -any- reader-visible changes to this structure since
then".
These cases fulfill the conditions above because between the
rcu_dereference_protected() call and the rcu_assign_pointer() call
there is no update of that value. Therefore, this patch makes the
replacement.
The following Coccinelle semantic patch was used:
@@
@@
- rcu_assign_pointer
+ RCU_INIT_POINTER
(...,
(
rtnl_dereference(...)
|
rcu_dereference_protected(...)
) )
[consolidated from http://marc.info/?l=linux-rdma&m=140836578119485&w=2 and
http://marc.info/?l=linux-rdma&m=140906361403047&w=2]
Tested-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Andreea-Cristina Bernat <bernat.ada@gmail.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This particular reference count is not needed with the rcu protection,
and the current code leaks a reference count, causing a hang in
qib_qp_destroy().
Cc: <stable@vger.kernel.org>
Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Fix the usnic and thw qib drivers to err when QP creation flags that
they don't understand are provided.
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
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>
This adds a seq_file iterator for reporting the QP hash table when the
qp_stats file is read.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Follow Documentation/RCU/rcuref.txt guidance in removing
atomic_inc_not_zero() from QP RCU implementation.
This patch also removes an unneeded synchronize_rcu() in the add path.
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
remove_qp() can execute concurrently with a qib_lookup_qpn() on
another CPU, which in of itself, is ok, given the RCU locking.
The issue is that remove_qp() NULLs out the qp->next field so that a
qib_lookup_qpn() might fail to find a qp if it occurs after the one
that is being deleted. This is a momentary issue and subsequent
qib_lookup_qpn() calls would find the qp's since the search restarts
from the bucket head. At scale, the issue might causes dropped
packets and unnecessary retransmissions.
The fix just deletes the qp->next NULL assignment to prevent the
remove_qp() from hiding qp's from qib_lookup_qpn().
Reviewed-by: Dean Luick <dean.luick@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit 1fb9fed6d4 ("IB/qib: Fix QP RCU sparse warning") broke QP
hash list deletion in qp_remove() badly.
This patch restores the former for loop behavior, while still fixing
the sparse warnings.
Cc: <stable@vger.kernel.org>
Reviewed-by: Gary Leshner <gary.s.leshner@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Commit af061a644a ("IB/qib: Use RCU for qpn lookup") introduced sparse
warnings.
This patch corrects those issues.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
A timing issue can occur where qib_mr_dereg can return -EBUSY if the
MR use count is not zero.
This can occur if the MR is de-registered while RDMA read response
packets are being progressed from the SDMA ring. The suspicion is
that the peer sent an RDMA read request, which has already been copied
across to the peer. The peer sees the completion of his request and
then communicates to the responder that the MR is not needed any
longer. The responder tries to de-register the MR, catching some
responses remaining in the SDMA ring holding the MR use count.
The code now uses a get/put paradigm to track MR use counts and
coordinates with the MR de-registration process using a completion
when the count has reached zero. A timeout on the delay is in place
to catch other EBUSY issues.
The reference count protocol is as follows:
- The return to the user counts as 1
- A reference from the lk_table or the qib_ibdev counts as 1.
- Transient I/O operations increase/decrease as necessary
A lot of code duplication has been folded into the new routines
init_qib_mregion() and deinit_qib_mregion(). Additionally, explicit
initialization of fields to zero is now handled by kzalloc().
Also, duplicated code 'while.*num_sge' that decrements reference
counts have been consolidated in qib_put_ss().
Reviewed-by: Ramkrishna Vepa <ramkrishna.vepa@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
This patch reorganizes the QP and devdata files to be more cache line aware.
qib_qp fields in particular are split into read-mostly, send, and receive fields.
qib_devdata fields are split into read-mostly and read/write fields
Testing has show that bidirectional tests improve by as much as 100%
with this patch.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
A new field is added to qib_qp called timeout_jiffies. It is
initialized upon create and modify.
The field is now used instead of a computation based on qp->timeout.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
The heavy weight spinlock in qib_lookup_qpn() is replaced with RCU.
The hash list itself is now accessed via jhash functions instead of mod.
The changes should benefit multiple receive contexts in different
processors by not contending for the lock just to read the hash
structures.
The patch also adds a lookaside_qp (pointer) and a lookaside_qpn in
the context. The interrupt handler will test the current packet's qpn
against lookaside_qpn if the lookaside_qp pointer is non-NULL. The
pointer is NULL'ed when the interrupt handler exits.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Store both the encoded and decoded MTU in the QP structure as a minor
optimization for UC/RC receive routines.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Changing from +1 to +2 allows for better QP distribution across
receive contexts.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
The basic idea is that on SusieQ, the difficult part of mapping QPN to
context is handled by the mapping registers so the generic QPN
allocation doesn't need to worry about chip specifics. For Monty and
Linda, there is no mapping table so the qpt->mask (same as
dd->qpn_mask), is used to see if the QPN to context falls within
[zero..dd->n_krcv_queues).
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
If these flags are set when the QP is transitioned to the error state,
it will wait until the flags are cleared, which may never happen if
the error transition is due to a link going down.
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
When transitioning a QP to the error state, in progress RWQEs need to
be marked complete. This also involves releasing the reference count
to the memory regions referenced in the SGEs. The locking in the
receive packet processing wasn't sufficient to prevent qib_error_qp()
from modifying the r_sge state at the same time, thus leading to
kernel panics.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
Add a low-level IB driver for QLogic PCIe adapters.
Signed-off-by: Ralph Campbell <ralph.campbell@qlogic.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
