linux-stable.git/drivers/target, branch v3.4.89

target/tcm_fc: Fix use-after-free of ft_tpg

2014-05-06T14:51:44+00:00

commit 2c42be2dd4f6586728dba5c4e197afd5cfaded78 upstream.

ft_del_tpg checks tpg->tport is set before unlinking the tpg from the
tport when the tpg is being removed. Set this pointer in ft_tport_create,
or the unlinking won't happen in ft_del_tpg and tport->tpg will reference
a deleted object.

This patch sets tpg->tport in ft_tport_create, because that's what
ft_del_tpg checks, and is the only way to get back to the tport to
clear tport->tpg.

The bug was occuring when:

- lport created, tport (our per-lport, per-provider context) is
  allocated.
  tport->tpg = NULL
- tpg created
- a PRLI is received. ft_tport_create is called, tpg is found and
  tport->tpg is set
- tpg removed. ft_tpg is freed in ft_del_tpg. Since tpg->tport was not
  set, tport->tpg is not cleared and points at freed memory
- Future calls to ft_tport_create return tport via first conditional,
  instead of searching for new tpg by calling ft_lport_find_tpg.
  tport->tpg is still invalid, and will access freed memory.

see https://bugzilla.redhat.com/show_bug.cgi?id=1071340

Signed-off-by: Andy Grover 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi-target: Fix ERL=2 ASYNC_EVENT connection pointer bug

2014-05-06T14:51:43+00:00

commit d444edc679e7713412f243b792b1f964e5cff1e1 upstream.

This patch fixes a long-standing bug in iscsit_build_conn_drop_async_message()
where during ERL=2 connection recovery, a bogus conn_p pointer could
end up being used to send the ISCSI_OP_ASYNC_EVENT + DROPPING_CONNECTION
notifying the initiator that cmd->logout_cid has failed.

The bug was manifesting itself as an OOPs in iscsit_allocate_cmd() with
a bogus conn_p pointer in iscsit_build_conn_drop_async_message().

Reported-by: Arshad Hussain 
Reported-by: santosh kulkarni 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

target/file: Re-enable optional fd_buffered_io=1 operation

2014-02-20T18:45:33+00:00

commit b32f4c7ed85c5cee2a21a55c9f59ebc9d57a2463 upstream.

This patch re-adds the ability to optionally run in buffered FILEIO mode
(eg: w/o O_DSYNC) for device backends in order to once again use the
Linux buffered cache as a write-back storage mechanism.

This logic was originally dropped with mainline v3.5-rc commit:

commit a4dff3043c231d57f982af635c9d2192ee40e5ae
Author: Nicholas Bellinger 
Date:   Wed May 30 16:25:41 2012 -0700

    target/file: Use O_DSYNC by default for FILEIO backends

This difference with this patch is that fd_create_virtdevice() now
forces the explicit setting of emulate_write_cache=1 when buffered FILEIO
operation has been enabled.

(v2: Switch to FDBD_HAS_BUFFERED_IO_WCE + add more detailed
     comment as requested by hch)

Reported-by: Ferry 
Cc: Christoph Hellwig 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Ben Hutchings 
Cc: Li Zefan 
Signed-off-by: Greg Kroah-Hartman

target/file: Use O_DSYNC by default for FILEIO backends

2014-02-20T18:45:33+00:00

commit a4dff3043c231d57f982af635c9d2192ee40e5ae upstream.

Convert to use O_DSYNC for all cases at FILEIO backend creation time to
avoid the extra syncing of pure timestamp updates with legacy O_SYNC during
default operation as recommended by hch.  Continue to do this independently of
Write Cache Enable (WCE) bit, as WCE=0 is currently the default for all backend
devices and enabled by user on per device basis via attrib/emulate_write_cache.

This patch drops the now unnecessary fd_buffered_io= token usage that was
originally signalling when to explictly disable O_SYNC at backend creation
time for buffered I/O operation.  This can end up being dangerous for a number
of reasons during physical node failure, so go ahead and drop this option
for now when O_DSYNC is used as the default.

Also allow explict FUA WRITEs -> vfs_fsync_range() call to function in
fd_execute_cmd() independently of WCE bit setting.

Reported-by: Christoph Hellwig 
Cc: Linus Torvalds 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Ben Hutchings 
[bwh: Backported to 3.2:
 - We have fd_do_task() and not fd_execute_cmd()
 - Various fields are in struct se_task rather than struct se_cmd
 - fd_create_virtdevice() flags initialisation hasn't been cleaned up]
Cc: Li Zefan 
Signed-off-by: Greg Kroah-Hartman

target/iscsi: Fix network portal creation race

2014-02-06T19:05:48+00:00

commit ee291e63293146db64668e8d65eb35c97e8324f4 upstream.

When creating network portals rapidly, such as when restoring a
configuration, LIO's code to reuse existing portals can return a false
negative if the thread hasn't run yet and set np_thread_state to
ISCSI_NP_THREAD_ACTIVE. This causes an error in the network stack
when attempting to bind to the same address/port.

This patch sets NP_THREAD_ACTIVE before the np is placed on g_np_list,
so even if the thread hasn't run yet, iscsit_get_np will return the
existing np.

Also, convert np_lock -> np_mutex + hold across adding new net portal
to g_np_list to prevent a race where two threads may attempt to create
the same network portal, resulting in one of them failing.

(nab: Add missing mutex_unlocks in iscsit_add_np failure paths)
(DanC: Fix incorrect spin_unlock -> spin_unlock_bh)

Signed-off-by: Andy Grover 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi-target: chap auth shouldn't match username with trailing garbage

2013-12-04T18:50:28+00:00

commit 86784c6bdeeef78eed94d298be7a8879f6a97ee2 upstream.

In iSCSI negotiations with initiator CHAP enabled, usernames with
trailing garbage are permitted, because the string comparison only
checks the strlen of the configured username.

e.g. "usernameXXXXX" will be permitted to match "username".

Just check one more byte so the trailing null char is also matched.

Signed-off-by: Eric Seppanen 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi-target: fix extract_param to handle buffer length corner case

2013-12-04T18:50:27+00:00

commit 369653e4fb511928511b0ce81f41c812ff1f28b6 upstream.

extract_param() is called with max_length set to the total size of the
output buffer.  It's not safe to allow a parameter length equal to the
buffer size as the terminating null would be written one byte past the
end of the output buffer.

Signed-off-by: Eric Seppanen 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

target: Fix trailing ASCII space usage in INQUIRY vendor+model

2013-09-08T04:58:16+00:00

commit ee60bddba5a5f23e39598195d944aa0eb2d455e5 upstream.

This patch fixes spc_emulate_inquiry_std() to add trailing ASCII
spaces for INQUIRY vendor + model fields following SPC-4 text:

  "ASCII data fields described as being left-aligned shall have any
   unused bytes at the end of the field (i.e., highest offset) and
   the unused bytes shall be filled with ASCII space characters (20h)."

This addresses a problem with Falconstor NSS multipathing.

Reported-by: Tomas Molota 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

iscsi-target: Fix tfc_tpg_nacl_auth_cit configfs length overflow

2013-08-04T08:25:41+00:00

commit 0fbfc46fb0b2f543a8b539e94c6c293ebc0b05a6 upstream.

This patch fixes a potential buffer overflow while processing
iscsi_node_auth input for configfs attributes within NodeACL
tfc_tpg_nacl_auth_cit context.

Signed-off-by: Joern Engel 
Signed-off-by: Nicholas Bellinger 
Signed-off-by: Greg Kroah-Hartman

target/iscsi: don't corrupt bh_count in iscsit_stop_time2retain_timer()

2013-06-27T18:27:33+00:00

commit 574780fd5e6ec52bd43e0bdb777a19e4c4c6aa9c upstream.

Here is a fun one.  Bug seems to have been introduced by commit 140854cb,
almost two years ago.  I have no idea why we only started seeing it now,
but we did.

Rough callgraph:
core_tpg_set_initiator_node_queue_depth()
`-> spin_lock_irqsave(&tpg->session_lock, flags);
`-> lio_tpg_shutdown_session()
    `-> iscsit_stop_time2retain_timer()
        `-> spin_unlock_bh(&se_tpg->session_lock);
        `-> spin_lock_bh(&se_tpg->session_lock);
`-> spin_unlock_irqrestore(&tpg->session_lock, flags);

core_tpg_set_initiator_node_queue_depth() used to call spin_lock_bh(),
but 140854cb changed that to spin_lock_irqsave().  However,
lio_tpg_shutdown_session() still claims to be called with spin_lock_bh()
held, as does iscsit_stop_time2retain_timer():
 *      Called with spin_lock_bh(&struct se_portal_group->session_lock) held

Stale documentation is mostly annoying, but in this case the dropping
the lock with the _bh variant is plain wrong.  It is also wrong to drop
locks two functions below the lock-holder, but I will ignore that bit
for now.

After some more locking and unlocking we eventually hit this backtrace:
------------[ cut here ]------------
WARNING: at kernel/softirq.c:159 local_bh_enable_ip+0xe8/0x100()
Pid: 24645, comm: lio_helper.py Tainted: G           O 3.6.11+
Call Trace:
 [] warn_slowpath_common+0x7f/0xc0
 [] ? iscsit_inc_conn_usage_count+0x37/0x50 [iscsi_target_mod]
 [] warn_slowpath_null+0x1a/0x20
 [] local_bh_enable_ip+0xe8/0x100
 [] _raw_spin_unlock_bh+0x15/0x20
 [] iscsit_inc_conn_usage_count+0x37/0x50 [iscsi_target_mod]
 [] iscsit_stop_session+0xfa/0x1c0 [iscsi_target_mod]
 [] lio_tpg_shutdown_session+0x7b/0x90 [iscsi_target_mod]
 [] core_tpg_set_initiator_node_queue_depth+0xe4/0x290 [target_core_mod]
 [] iscsit_tpg_set_initiator_node_queue_depth+0x12/0x20 [iscsi_target_mod]
 [] lio_target_nacl_store_cmdsn_depth+0xa9/0x180 [iscsi_target_mod]
 [] target_fabric_nacl_base_attr_store+0x39/0x40 [target_core_mod]
 [] configfs_write_file+0xbd/0x120
 [] vfs_write+0xc6/0x180
 [] sys_write+0x51/0x90
 [] system_call_fastpath+0x16/0x1b
---[ end trace 3747632b9b164652 ]---

As a pure band-aid, this patch drops the _bh.

Signed-off-by: Joern Engel 
Signed-off-by: Nicholas Bellinger