linux-stable.git/drivers/target, branch v3.18.9

iscsi-target: Drop problematic active_ts_list usage

2015-03-06T22:52:59+00:00

commit 3fd7b60f2c7418239d586e359e0c6d8503e10646 upstream.

This patch drops legacy active_ts_list usage within iscsi_target_tq.c
code.  It was originally used to track the active thread sets during
iscsi-target shutdown, and is no longer used by modern upstream code.

Two people have reported list corruption using traditional iscsi-target
and iser-target with the following backtrace, that appears to be related
to iscsi_thread_set->ts_list being used across both active_ts_list and
inactive_ts_list.

[   60.782534] ------------[ cut here ]------------
[   60.782543] WARNING: CPU: 0 PID: 9430 at lib/list_debug.c:53 __list_del_entry+0x63/0xd0()
[   60.782545] list_del corruption, ffff88045b00d180->next is LIST_POISON1 (dead000000100100)
[   60.782546] Modules linked in: ib_srpt tcm_qla2xxx qla2xxx tcm_loop tcm_fc libfc scsi_transport_fc scsi_tgt ib_isert rdma_cm iw_cm ib_addr iscsi_target_mod target_core_pscsi target_core_file target_core_iblock target_core_mod configfs ebtable_nat ebtables ipt_MASQUERADE iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 ipt_REJECT xt_CHECKSUM iptable_mangle iptable_filter ip_tables bridge stp llc autofs4 sunrpc ip6t_REJECT nf_conntrack_ipv6 nf_defrag_ipv6 xt_state nf_conntrack ip6table_filter ip6_tables ipv6 ib_ipoib ib_cm ib_uverbs ib_umad mlx4_en mlx4_ib ib_sa ib_mad ib_core mlx4_core dm_mirror dm_region_hash dm_log dm_mod vhost_net macvtap macvlan vhost tun kvm_intel kvm uinput iTCO_wdt iTCO_vendor_support microcode serio_raw pcspkr sb_edac edac_core sg i2c_i801 lpc_ich mfd_core mtip32xx igb i2c_algo_bit i2c_core ptp pps_core ioatdma dca wmi ext3(F) jbd(F) mbcache(F) sd_mod(F) crc_t10dif(F) crct10dif_common(F) ahci(F) libahci(F) isci(F) libsas(F) scsi_transport_sas(F) [last unloaded: speedstep_lib]
[   60.782597] CPU: 0 PID: 9430 Comm: iscsi_ttx Tainted: GF 3.12.19+ #2
[   60.782598] Hardware name: Supermicro X9DRX+-F/X9DRX+-F, BIOS 3.00 07/09/2013
[   60.782599]  0000000000000035 ffff88044de31d08 ffffffff81553ae7 0000000000000035
[   60.782602]  ffff88044de31d58 ffff88044de31d48 ffffffff8104d1cc 0000000000000002
[   60.782605]  ffff88045b00d180 ffff88045b00d0c0 ffff88045b00d0c0 ffff88044de31e58
[   60.782607] Call Trace:
[   60.782611]  [] dump_stack+0x49/0x62
[   60.782615]  [] warn_slowpath_common+0x8c/0xc0
[   60.782618]  [] warn_slowpath_fmt+0x46/0x50
[   60.782620]  [] __list_del_entry+0x63/0xd0
[   60.782622]  [] list_del+0x11/0x40
[   60.782630]  [] iscsi_del_ts_from_active_list+0x29/0x50 [iscsi_target_mod]
[   60.782635]  [] iscsi_tx_thread_pre_handler+0xa1/0x180 [iscsi_target_mod]
[   60.782642]  [] iscsi_target_tx_thread+0x4e/0x220 [iscsi_target_mod]
[   60.782647]  [] ? iscsit_handle_snack+0x190/0x190 [iscsi_target_mod]
[   60.782652]  [] ? iscsit_handle_snack+0x190/0x190 [iscsi_target_mod]
[   60.782655]  [] kthread+0xce/0xe0
[   60.782657]  [] ? kthread_freezable_should_stop+0x70/0x70
[   60.782660]  [] ret_from_fork+0x7c/0xb0
[   60.782662]  [] ? kthread_freezable_should_stop+0x70/0x70
[   60.782663] ---[ end trace 9662f4a661d33965 ]---

Since this code is no longer used, go ahead and drop the problematic usage
all-together.

Reported-by: Gavin Guo 
Reported-by: Moussa Ba 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

target: Drop arbitrary maximum I/O size limit

2015-01-27T16:29:45+00:00

commit 046ba64285a4389ae5e9a7dfa253c6bff3d7c341 upstream.

This patch drops the arbitrary maximum I/O size limit in sbc_parse_cdb(),
which currently for fabric_max_sectors is hardcoded to 8192 (4 MB for 512
byte sector devices), and for hw_max_sectors is a backend driver dependent
value.

This limit is problematic because Linux initiators have only recently
started to honor block limits MAXIMUM TRANSFER LENGTH, and other non-Linux
based initiators (eg: MSFT Fibre Channel) can also generate I/Os larger
than 4 MB in size.

Currently when this happens, the following message will appear on the
target resulting in I/Os being returned with non recoverable status:

  SCSI OP 28h with too big sectors 16384 exceeds fabric_max_sectors: 8192

Instead, drop both [fabric,hw]_max_sector checks in sbc_parse_cdb(),
and convert the existing hw_max_sectors into a purely informational
attribute used to represent the granuality that backend driver and/or
subsystem code is splitting I/Os upon.

Also, update FILEIO with an explicit FD_MAX_BYTES check in fd_execute_rw()
to deal with the one special iovec limitiation case.

v2 changes:
  - Drop hw_max_sectors check in sbc_parse_cdb()

Reported-by: Lance Gropper 
Reported-by: Stefan Priebe 
Cc: Christoph Hellwig 
Cc: Martin K. Petersen 
Cc: Roland Dreier 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi,iser-target: Expose supported protection ops according to t10_pi

2015-01-27T16:29:37+00:00

commit 23a548ee656c8ba6da8cb2412070edcd62e2ac5d upstream.

iSER will report supported protection operations based on
the tpg attribute t10_pi settings and HCA PI offload capabilities.
If the HCA does not support PI offload or tpg attribute t10_pi is
not set, we fall to SW PI mode.

In order to do that, we move iscsit_get_sup_prot_ops after connection
tpg assignment.

Signed-off-by: Sagi Grimberg 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi,iser-target: Initiate termination only once

2015-01-27T16:29:37+00:00

commit 954f23722b5753305be490330cf2680b7a25f4a3 upstream.

Since commit 0fc4ea701fcf ("Target/iser: Don't put isert_conn inside
disconnected handler") we put the conn kref in isert_wait_conn, so we
need .wait_conn to be invoked also in the error path.

Introduce call to isert_conn_terminate (called under lock)
which transitions the connection state to TERMINATING and calls
rdma_disconnect. If the state is already teminating, just bail
out back (temination started).

Also, make sure to destroy the connection when getting a connect
error event if didn't get to connected (state UP). Same for the
handling of REJECTED and UNREACHABLE cma events.

Squashed:

iscsi-target: Add call to wait_conn in establishment error flow

Reported-by: Slava Shwartsman 
Signed-off-by: Sagi Grimberg 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi-target: Fail connection on short sendmsg writes

2015-01-27T16:29:37+00:00

commit 6bf6ca7515c1df06f5c03737537f5e0eb191e29e upstream.

This patch changes iscsit_do_tx_data() to fail on short writes
when kernel_sendmsg() returns a value different than requested
transfer length, returning -EPIPE and thus causing a connection
reset to occur.

This avoids a potential bug in the original code where a short
write would result in kernel_sendmsg() being called again with
the original iovec base + length.

In practice this has not been an issue because iscsit_do_tx_data()
is only used for transferring 48 byte headers + 4 byte digests,
along with seldom used control payloads from NOPIN + TEXT_RSP +
REJECT with less than 32k of data.

So following Al's audit of iovec consumers, go ahead and fail
the connection on short writes for now, and remove the bogus
logic ahead of his proper upstream fix.

Reported-by: Al Viro 
Cc: David S. Miller 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

tcm_loop: Fix wrong I_T nexus association

2015-01-27T16:29:37+00:00

commit 506787a2c7daed45f0a213674ca706cbc83a9089 upstream.

tcm_loop has the I_T nexus associated with the HBA. This causes
commands to become misdirected if the HBA has more than one
target portal group; any command is then being sent to the
first target portal group instead of the correct one.

The nexus needs to be associated with the target portal group
instead.

Signed-off-by: Hannes Reinecke 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

target: return CONFLICT only when SA key unmatched

2014-11-03T06:04:57+00:00

PREEMPT (and PREEMPT AND ABORT) should return CONFLICT iff a specified
SERVICE ACTION RESERVATION KEY is specified and matches no existing
persistent reservation.

Without this patch, a PREEMPT will return CONFLICT if either all
reservations are held by the initiator (self preemption) or there is
nothing to preempt. According to the spec, both of these cases should
succeed.

Signed-off-by: Steven Allen 
Signed-off-by: Nicholas Bellinger

iscsi-target: return the correct port in SendTargets

2014-10-28T20:54:18+00:00

The fact that a target is published on the any address has no bearing on
which port(s) it is published. SendTargets should always send the
portal's port, not the port used for discovery.

Signed-off-by: Steven Allen 
Signed-off-by: Nicholas Bellinger

target: Don't call TFO->write_pending if data_length == 0

2014-10-28T20:54:15+00:00

If an initiator sends a zero-length command (e.g. TEST UNIT READY) but
sets the transfer direction in the transport layer to indicate a
data-out phase, we still shouldn't try to transfer data.  At best it's
a NOP, and depending on the transport, we might crash on an
uninitialized sg list.

Reported-by: Craig Watson 
Signed-off-by: Roland Dreier 
Cc:  # 3.1
Signed-off-by: Nicholas Bellinger

Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending

2014-10-21T20:06:38+00:00

Pull SCSI target updates from Nicholas Bellinger:
 "Here are the target updates for v3.18-rc2 code.  These where
  originally destined for -rc1, but due to the combination of travel
  last week for KVM Forum and my mistake of taking the three week merge
  window literally, the pull request slipped..  Apologies for that.

  Things where reasonably quiet this round.  The highlights include:

   - New userspace backend driver (target_core_user.ko) by Shaohua Li
     and Andy Grover
   - A number of cleanups in target, iscsi-taret and qla_target code
     from Joern Engel
   - Fix an OOPs related to queue full handling with CHECK_CONDITION
     status from Quinn Tran
   - Fix to disable TX completion interrupt coalescing in iser-target,
     that was causing problems on some hardware
   - Fix for PR APTPL metadata handling with demo-mode ACLs

  I'm most excited about the new backend driver that uses UIO + shared
  memory ring to dispatch I/O and control commands into user-space.
  This was probably the most requested feature by users over the last
  couple of years, and opens up a new area of development + porting of
  existing user-space storage applications to LIO.  Thanks to Shaohua +
  Andy for making this happen.

  Also another honorable mention, a new Xen PV SCSI driver was merged
  via the xen/tip.git tree recently, which puts us now at 10 target
  drivers in upstream! Thanks to David Vrabel + Juergen Gross for their
  work to get this code merged"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/nab/target-pending: (40 commits)
  target/file: fix inclusive vfs_fsync_range() end
  iser-target: Disable TX completion interrupt coalescing
  target: Add force_pr_aptpl device attribute
  target: Fix APTPL metadata handling for dynamic MappedLUNs
  qla_target: don't delete changed nacls
  target/user: Recalculate pad size inside is_ring_space_avail()
  tcm_loop: Fixup tag handling
  iser-target: Fix smatch warning
  target/user: Fix up smatch warnings in tcmu_netlink_event
  target: Add a user-passthrough backstore
  target: Add documentation on the target userspace pass-through driver
  uio: Export definition of struct uio_device
  target: Remove unneeded check in sbc_parse_cdb
  target: Fix queue full status NULL pointer for SCF_TRANSPORT_TASK_SENSE
  qla_target: rearrange struct qla_tgt_prm
  qla_target: improve qlt_unmap_sg()
  qla_target: make some global functions static
  qla_target: remove unused parameter
  target: simplify core_tmr_abort_task
  target: encapsulate smp_mb__after_atomic()
  ...