Start the reconnect timer, fast_io_fail timer and dev_loss timers if a
transport layer error occurs.
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Enable fast_io_fail_tmo and dev_loss_tmo functionality for the IB SRP
initiator.  Add kernel module parameters that allow to specify default
values for these parameters.
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Keep the rport data structure around after srp_remove_host() has
finished until cleanup of the IB transport layer has finished
completely. This is necessary because later patches use the rport
pointer inside the queuecommand callback. Without this patch
accessing the rport from inside a queuecommand callback is racy
because srp_remove_host() must be invoked before scsi_remove_host()
and because the queuecommand callback could get invoked after
srp_remove_host() has finished. In other words, without this patch
the queuecommand callback can get invoked after the rport data
structure has been freed.
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Allow the InfiniBand RC retry count to be configured by the user as an
option in the target login string.  Reducing this retry count allows to
reduce the path failover time.
Signed-off-by: NVu Pham <vu@mellanox.com>

[ bvanassche: Rewrote patch description / changed default retry count ]
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This patch changes isert_connect_release() to correctly check for
the existence struct isert_device *device before checking for
isert_device->use_frwr.
Signed-off-by: NVu Pham <vu@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

The SRP specification requires:

  "Response data shall be provided in any SRP_RSP response that is sent in
   response to an SRP_TSK_MGMT request (see 6.7). The information in the
   RSP_CODE field (see table 24) shall indicate the completion status of
   the task management function."

So fix this to avoid the SRP initiator interprets task management functions
that succeeded as failed.
Signed-off-by: NJack Wang <jinpu.wang@profitbricks.com>
Cc: stable@vger.kernel.org # 3.3+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a bug where ib_destroy_cm_id() was incorrectly being called
after srpt_destroy_ch_ib() had destroyed the active QP.

This would result in the following failed SRP_LOGIN_REQ messages:

Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff1762bd, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 1 (guid=0xfe80000000000000:0x2c903009f8f41)
Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff1758f9, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 2 (guid=0xfe80000000000000:0x2c903009f8f42)
Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff175941, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 2 (guid=0xfe80000000000000:0x2c90300a3cfb2)
Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff176299, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 1 (guid=0xfe80000000000000:0x2c90300a3cfb1)
mlx4_core 0000:84:00.0: command 0x19 failed: fw status = 0x9
rejected SRP_LOGIN_REQ because creating a new RDMA channel failed.
Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff176299, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 1 (guid=0xfe80000000000000:0x2c90300a3cfb1)
mlx4_core 0000:84:00.0: command 0x19 failed: fw status = 0x9
rejected SRP_LOGIN_REQ because creating a new RDMA channel failed.
Received SRP_LOGIN_REQ with i_port_id 0x0:0x2590ffff176299, t_port_id 0x2c903009f8f40:0x2c903009f8f40 and it_iu_len 260 on port 1 (guid=0xfe80000000000000:0x2c90300a3cfb1)
Reported-by: NNavin Ahuja <navin.ahuja@saratoga-speed.com>
Cc: stable@vger.kernel.org # 3.3+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Update copyright ownership/year information for target-core,
loopback, iscsi-target, tcm_qla2xx, vhost and iser-target.
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This model was introduced in 00f7ec36 "RDMA/core: Add memory
management extensions support" and works when the IB device
supports the IB_DEVICE_MEM_MGT_EXTENSIONS capability.

Upon creating the isert device, ib_isert will test whether the HCA
supports FRWR. If supported then set the flag and assign
function pointers that handle fast registration and deregistration
of appropriate resources (fast_reg descriptors).

When new connection coming in, ib_isert will check frwr flag and
create frwr resouces, if fail to do it will switch back to
old model of using global dma key and turn off the frwr support.

Registration is done using posting IB_WR_FAST_REG_MR to the QP and
invalidations using posting IB_WR_LOCAL_INV.
Signed-off-by: NVu Pham <vu@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Current driver uses global dma key to register the memory pointed
by sg list provided by the target core.

This is the preparation step for adding more methods like fast
path memory registration, make the reg/unreg calls be function
pointers.
Signed-off-by: NVu Pham <vu@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

isert_put_datain() and isert_get_dataout() share a lot of code
in rdma wr processing, move this common code to a shared function.

Use isert_unmap_cmd to cleanup for RDMA_READ completion.
Remove duplicate field in isert_cmd and isert_rdma_wr structs
Change misc debug messages to track isert_cmd
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NVu Pham <vu@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This command converts iscsi/isert-target to use allocations based on
iscsit_transport->priv_size within iscsit_allocate_cmd(), instead of
using an embedded isert_cmd->iscsi_cmd.

This includes removing iscsit_transport->alloc_cmd() usage, along
with updating isert-target code to use iscsit_priv_cmd().

Also, remove left-over iscsit_transport->release_cmd() usage for
direct calls to iscsit_release_cmd(), and drop the now unused
lio_cmd_cache and isert_cmd_cache.

Cc: Or Gerlitz <ogerlitz@mellanox.com>
Cc: Kent Overstreet <kmo@daterainc.com>
Signed-off-by: NNicholas Bellinger <nab@daterainc.com>

This patch updates iser-target code to support login negotiation
multi-plexing.  This includes only using isert_conn->conn_login_comp
for the first login request PDU, pushing the subsequent processing
to iscsi_conn->login_work -> iscsi_target_do_login_rx(), and turning
isert_get_login_rx() into a NOP.

v3 changes:
   - Drop unnecessary LOGIN_FLAGS_READ_ACTIVE bit set in
     isert_rx_login_req()
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This bug was discovered by Smatch static checker run by Dan Carpenter.
If in free_rx_descriptors(), rx_descs are not NULL then the iser
device is definately not NULL, so no need to check it before
dereferencing it.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Fix leak where desc is not being freed in error flows.
Signed-off-by: NRoi Dayan <roid@mellanox.com>
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

To run discovery over iSER we need to advertize the CAP_TEXT_NEGO capability
towards user space. Also need to make sure the login RX buffer is posted when
SendTargets TEXT PDUs are sent. For that end, we use a setting of the
ISCSI_PARAM_DISCOVERY_SESS iscsi param as an indication that this is
discovery session.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: NMike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NJames Bottomley <JBottomley@Parallels.com>

In several places, this snippet is used when removing neigh entries:

	list_del(&neigh->list);
	ipoib_neigh_free(neigh);

The list_del() removes neigh from the associated struct ipoib_path, while
ipoib_neigh_free() removes neigh from the device's neigh entry lookup
table.  Both of these operations are protected by the priv->lock
spinlock.  The table however is also protected via RCU, and so naturally
the lock is not held when doing reads.

This leads to a race condition, in which a thread may successfully look
up a neigh entry that has already been deleted from neigh->list.  Since
the previous deletion will have marked the entry with poison, a second
list_del() on the object will cause a panic:

  #5 [ffff8802338c3c70] general_protection at ffffffff815108c5
     [exception RIP: list_del+16]
     RIP: ffffffff81289020  RSP: ffff8802338c3d20  RFLAGS: 00010082
     RAX: dead000000200200  RBX: ffff880433e60c88  RCX: 0000000000009e6c
     RDX: 0000000000000246  RSI: ffff8806012ca298  RDI: ffff880433e60c88
     RBP: ffff8802338c3d30   R8: ffff8806012ca2e8   R9: 00000000ffffffff
     R10: 0000000000000001  R11: 0000000000000000  R12: ffff8804346b2020
     R13: ffff88032a3e7540  R14: ffff8804346b26e0  R15: 0000000000000246
     ORIG_RAX: ffffffffffffffff  CS: 0010  SS: 0000
  #6 [ffff8802338c3d38] ipoib_cm_tx_handler at ffffffffa066fe0a [ib_ipoib]
  #7 [ffff8802338c3d98] cm_process_work at ffffffffa05149a7 [ib_cm]
  #8 [ffff8802338c3de8] cm_work_handler at ffffffffa05161aa [ib_cm]
  #9 [ffff8802338c3e38] worker_thread at ffffffff81090e10
 #10 [ffff8802338c3ee8] kthread at ffffffff81096c66
 #11 [ffff8802338c3f48] kernel_thread at ffffffff8100c0ca

We move the list_del() into ipoib_neigh_free(), so that deletion happens
only once, after the entry has been successfully removed from the lookup
table.  This same behavior is already used in ipoib_del_neighs_by_gid()
and __ipoib_reap_neigh().
Signed-off-by: NJim Foraker <foraker1@llnl.gov>
Reviewed-by: NOr Gerlitz <ogerlitz@mellanox.com>
Reviewed-by: NJack Wang <jinpu.wang@profitbricks.com>
Reviewed-by: NShlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Newer HCAs and Virtual functions may not support FMRs but rather a fast
registration model, which we call FRWR - "Fast Registration Work Requests".

This model was introduced in 00f7ec36 ("RDMA/core: Add memory management
extensions support") and works when the IB device supports the
IB_DEVICE_MEM_MGT_EXTENSIONS capability.

Upon creating the iser device iser will test whether the HCA supports
FMRs.  If no support for FMRs, check if IB_DEVICE_MEM_MGT_EXTENSIONS
is supported and assign function pointers that handle fast
registration and allocation of appropriate resources (fast_reg
descriptors).

Registration is done using posting IB_WR_FAST_REG_MR to the QP and
invalidations using posting IB_WR_LOCAL_INV.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This is preparation step for other memory registration methods to be
added.  In addition, change reg/unreg routines signature to indicate
they use FMRs.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This routine will be shared with other rdma management schemes.  The
bounce buffer solution for unaligned SG-lists and the sg_to_page_vec
routine are likely to be used for other registration schemes and not
just FMR.

Move them out of the FMR specific code, and call them from there.
Later they will be called also from other reg methods code.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Currently the driver uses FMRs as the only means to register the
memory pointed by SG provided by the SCSI mid-layer with the RDMA
device.

As preparation step for adding more methods for fast path memory
registration, make the alloc/free and reg/unreg calls function
pointers, which are for now just set to the existing FMR ones.
Signed-off-by: NSagi Grimberg <sagig@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Use cmds_max passed from user space to be the number of PDUs to be
supported for the session instead of hard-coded ISCSI_DEF_XMIT_CMDS_MAX.
This allow controlling the max number of SCSI commands for the session.
Also don't ignore the qdepth passed from user space.

Derive from session->cmds_max the actual number of RX buffers and FMR
pool size to allocate during the connection bind phase.

Since the iser transport connection is established before the iscsi
session/connection are created and bound, we still use one hard-coded
quantity ISER_DEF_XMIT_CMDS_MAX to compute the maximum number of
work-requests to be supported by the RC QP used for the connection.

The above quantity is made to be a power of two between ISCSI_TOTAL_CMDS_MIN
(16) and ISER_DEF_XMIT_CMDS_MAX (512) inclusive.
Signed-off-by: NShlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This is a preparation step to a patch that accepts the number of max
SCSI commands to be supported a session from user space iSCSI tools.

Move the allocation of the login buffer, FMR pool and its associated
page vector from iser_create_ib_conn_res() (which is called prior when
we actually know how many commands should be supported) to
iser_alloc_rx_descriptors() (which is called during the iscsi
connection bind step where this quantity is known).

Also do small refactoring around the deallocation to make that path
similar to the allocation one.
Signed-off-by: NShlomo Pongratz <shlomop@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Commit 4f363882 ("IB/iser: Move informational messages from error
to info level") set info prints to be emitted at a lower debug level
than warning prints, which is a bit odd.  Fix that.

Also move the prints on unaligned SG from warning to debug level.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

IPoIB's required behaviour w.r.t to the pkey used by the device is the following:

- For "parent" interfaces (e.g ib0, ib1, etc) who are created
  automatically as a result of hot-plug events from the IB core, the
  driver needs to take whatever pkey vlaue it finds in index 0, and
  stick to that index.

- For child interfaces (e.g ib0.8001, etc) created by admin directive,
  the driver needs to use and stick to the value provided during its
  creation.

In SR-IOV environment its possible for the VF probe to take place
before the cloud management software provisions the suitable pkey for
the VF in the paravirtualed PKEY table index 0. When this is the case,
the VF IB stack will find in index 0 an invalide pkey, which is all
zeros.

Moreover, the cloud managment can assign the pkey value at index 0 at
any time of the guest life cycle.

The correct behavior for IPoIB to address these requirements for
parent interfaces is to use PKEY_CHANGE event as trigger to optionally
re-init the device pkey value and re-create all the relevant resources
accordingly, if the value of the pkey in index 0 has changed (from
invalid to valid or from valid value X to invalid value Y).

This patch enhances the heavy flushing code which is triggered by pkey
change event, to behave correctly for parent devices. For child
devices, the code remains the same, namely chases pkey value and not
index.
Signed-off-by: NErez Shitrit <erezsh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Make sure that the IB invalid pkey (0x0000 or 0x8000) isn't used for
child devices.

Also, make sure to always set the full membership bit for the pkey of
devices created by rtnl link ops.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

If the transport layer is offline it is more appropriate to let
srp_abort() return FAST_IO_FAIL instead of SUCCESS.
Reported-by: NSebastian Riemer <sebastian.riemer@profitbricks.com>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

This patch adds a check in isert_rx_opcode() to ignore non TEXT + LOGOUT
opcodes when SessionType=Discovery has been negotiated.

Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

The return value wasn't checked by any of the callers.  Assuming this is
correct behaviour, we can simplify some code by not bothering to
generate it.

nab: Add srpt_queue_data_in() + srpt_queue_tm_rsp() nops around
     srpt_queue_response() void return
Signed-off-by: NJoern Engel <joern@logfs.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch adds isert_handle_text_cmd() to handle incoming
ISCSI_OP_TEXT PDU processing, along with isert_put_text_rsp()
for posting ISCSI_OP_TEXT_RSP ib_send_wr response.

It copies ISCSI_OP_TEXT payload using unsolicited payload at
&iser_rx_desc->data[0] into iscsi_cmd->text_in_ptr for usage
with outgoing isert_put_text_rsp() -> iscsit_build_text_rsp()

v2 changes:
  - Let iscsit_build_text_rsp() determine any extra padding
Reported-by: NOr Gerlitz <ogerlitz@mellanox.com>
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Now that these two variables are used for REJECT payloads as well
as SCSI response sense payloads, rename them to something that
makes more sense.
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Add output for ib_wc.vendor_err in isert_cq_[t,r]x_work(), which
is useful for debugging future issues.
Reported-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch addresses a bug where RDMA_CM_EVENT_DISCONNECTED may occur
before the connection shutdown has been completed by rx/tx threads,
that causes isert_free_conn() to wait indefinately on ->conn_wait.

This patch allows isert_disconnect_work code to invoke rdma_disconnect
when isert_disconnect_work() process context is started by client
session reset before isert_free_conn() code has been reached.

It also adds isert_conn->conn_mutex protection for ->state within
isert_disconnect_work(), isert_cq_comp_err() and isert_free_conn()
code, along with isert_check_state() for wait_event usage.

(v2: Add explicit iscsit_cause_connection_reinstatement call
     during isert_disconnect_work() to force conn reset)

Cc: stable@vger.kernel.org  # 3.10+
Cc: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch adds target_get_sess_cmd reference counting for
iscsit_handle_task_mgt_cmd(), and adds a target_put_sess_cmd()
for the failure case.

It also fixes a bug where ISCSI_OP_SCSI_TMFUNC type commands
where leaking iscsi_cmd->i_conn_node and eventually triggering
an OOPs during struct isert_conn shutdown.

Cc: stable@vger.kernel.org  # 3.10+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch moves ISCSI_OP_REJECT failures into iscsit_sequence_cmd()
in order to avoid external iscsit_reject_cmd() reject usage for all
PDU types.

It also updates PDU specific handlers for traditional iscsi-target
code to not reset the session after posting a ISCSI_OP_REJECT during
setup.

(v2: Fix CMDSN_LOWER_THAN_EXP for ISCSI_OP_SCSI to call
     target_put_sess_cmd() after iscsit_sequence_cmd() failure)

Cc: Or Gerlitz <ogerlitz@mellanox.com>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: stable@vger.kernel.org  # 3.10+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch changes iscsit_add_reject() + iscsit_add_reject_from_cmd()
usage to not sleep on iscsi_cmd->reject_comp to address a free-after-use
usage bug in v3.10 with iser-target code.

It saves ->reject_reason for use within iscsit_build_reject() so the
correct value for both transport cases.  It also drops the legacy
fail_conn parameter usage throughput iscsi-target code and adds
two iscsit_add_reject_cmd() and iscsit_reject_cmd helper functions,
along with various small cleanups.

(v2: Re-enable target_put_sess_cmd() to be called from
     iscsit_add_reject_from_cmd() for rejects invoked after
     target_get_sess_cmd() has been called)

Cc: Or Gerlitz <ogerlitz@mellanox.com>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: stable@vger.kernel.org  # 3.10+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch adds the missing isert_put_reject() logic to post
a outgoing payload buffer to hold the 48 bytes of original PDU
header request payload for the rejected cmd.

It also fixes ISTATE_SEND_REJECT handling in isert_response_completion()
-> isert_do_control_comp() code, and drops incorrect iscsi_cmd_t->reject_comp
usage.

Cc: Or Gerlitz <ogerlitz@mellanox.com>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: stable@vger.kernel.org  # 3.10+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

Signed-off-by: NVu Pham <vu@mellanox.com>
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

Several InfiniBand HCAs allow configuring the completion vector per
CQ.  This allows spreading the workload created by IB completion
interrupts over multiple MSI-X vectors and hence over multiple CPU
cores.  In other words, configuring the completion vector properly not
only allows reducing latency on an initiator connected to multiple
SRP targets but also allows improving throughput.
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>

An SRP target is required to maintain a single connection between
initiator and target.  This means that if the 'add_target' attribute
is used to create a second connection to a target, the first
connection will be logged out and that the SCSI error handler will
kick in.  The SCSI error handler will cause the SRP initiator to
reconnect, which will cause I/O over the second connection to fail.
Avoid such ping-pong behavior by disabling relogins.

If reconnecting manually is necessary, that is possible by deleting
and recreating an rport via sysfs.
Signed-off-by: NBart Van Assche <bvanassche@acm.org>
Signed-off-by: NSebastian Riemer <sebastian.riemer@profitbricks.com>
Acked-by: NDavid Dillow <dillowda@ornl.gov>
Signed-off-by: NRoland Dreier <roland@purestorage.com>