Using this debugfs knob fastmap self checks can be controlled.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>

If UBI is unable to write the fastmap to the device
we have make sure that upon next attach UBI will fall
back to scanning mode.
In case we cannot ensure that they only thing we can do
is falling back to read-only mode.

The current error handling code is not powercut proof.
It could happen that a powercut while invalidating would
lead to a state where an too old fastmap could be used upon
attach.
This patch addresses the issue by writing a fake fastmap
super block to a fresh PEB instead of reerasing the existing one.
The fake fastmap super block will UBI case to do a full scan.
Signed-off-by: NRichard Weinberger <richard@nod.at>

The current code assumes that each fastmap has the same amount of PEBs.
So far this is true but will change soon.
Signed-off-by: NRichard Weinberger <richard@nod.at>

a) Rename ubi->fm_sem to ubi->fm_eba_sem as this semaphore
protects EBA changes.
b) Turn ubi->fm_mutex into a rw semaphore. It will still serialize
fastmap writes but also ensures that ubi_wl_put_peb() is not
interrupted by a fastmap write. We use a rw semaphore to allow
ubi_wl_put_peb() still to be executed in parallel if no fastmap
write is happening.
Signed-off-by: NRichard Weinberger <richard@nod.at>

If we set it for dynamic ones we might confuse various self checks.
Signed-off-by: NRichard Weinberger <richard@nod.at>

ubi_wl_get_peb() returns a fresh PEB which can be used by
user of UBI. Due to the pool logic fastmap will correctly
map this PEB upon attach time because it will be scanned.

If a new fastmap is written (due to heavy parallel io)
while the before the fresh PEB is assigned to the EBA table
it will not be scanned as it is no longer in the pool.
So, the race window exists between ubi_wl_get_peb()
and the EBA table assignment.
We have to make sure that no new fastmap can be written
while that.

To ensure that ubi_wl_get_peb() will grab ubi->fm_sem in read mode
and the user of ubi_wl_get_peb() has to release it after the PEB
got assigned.
Signed-off-by: NRichard Weinberger <richard@nod.at>

If a LEB is unmapped we have to decrement leb_count as well.
Signed-off-by: NRichard Weinberger <richard@nod.at>

Don't use a fixed size for the WL pool.
Make it instead 50% of the user pool.
We don't make it 100% as it is not as heavily used as the user pool.
Signed-off-by: NRichard Weinberger <richard@nod.at>

We have to switch to ro mode to guarantee that upon next UBI attach
all data is consistent.
Signed-off-by: NRichard Weinberger <richard@nod.at>

This logic is in vain as we treat protected PEBs also as used, so this
case must not happen.
If a PEB is found which is in the EBA table but not known as used
has to be issued as fatal error.
Signed-off-by: NRichard Weinberger <richard@nod.at>

It is legal to have PEBs left in the used list.
This can happen if UBI copies a PEB and a powercut happens
between writing a new fastmap and adding this PEB into the EBA table.
In this case the old PEB will be used.
Signed-off-by: NRichard Weinberger <richard@nod.at>

This function a) requests a new PEB, b) writes data to it,
c) returns the old PEB and d) registers the new PEB in the EBA table.

For the non-fastmap case this works perfectly fine and is powercut safe.
Is fastmap enabled this can lead to issues.
If a new fastmap is written between a) and c) the freshly requested PEB
is no longer in a pool and will not be scanned upon attaching.
If now a powercut happens between c) and d) the freshly requested PEB
will not be scanned and the old one got already scheduled for erase.
After attaching the EBA table will point to a erased PEB.

Fix this issue by swapping steps c) and d).
Signed-off-by: NRichard Weinberger <richard@nod.at>

There is always exactly one ubi_attach_info object allocated,
therefore we don't have to care about the name.
Signed-off-by: NRichard Weinberger <richard@nod.at>

Currently we leak a few ubi_ainf_pebs while attaching.
Signed-off-by: NRichard Weinberger <richard@nod.at>

There is no need to switch to ro mode if ubi_update_fastmap() fails.
Also get rid of the ifdef.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>

...such that we can implement NOP variants of some functions.
This will help to reduce fastmap specific ifdefs in other c files.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>

If ubi_update_fastmap() fails notify the user.
This is not a hard error as ubi_update_fastmap() makes sure that upon failure
the current on-flash fastmap will no be used upon next UBI attach.
Signed-off-by: NRichard Weinberger <richard@nod.at>

Add a ubi_fastmap_close() to free all resources used by fastmap
at WL shutdown.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Tested-by: NGuido Martínez <guido@vanguardiasur.com.ar>
Reviewed-by: NGuido Martínez <guido@vanguardiasur.com.ar>

There is no need to allocate new ones every time, we can reuse
the existing ones.
This makes the code cleaner and more easy to follow.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>
Reviewed-by: NGuido Martínez <guido@vanguardiasur.com.ar>

Currently ubi_refill_pools() first fills the first and then
the second one.
If only very few free PEBs are available the second pool can get
zero PEBs.
Change ubi_refill_pools() to distribute free PEBs fair between
all pools.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NGuido Martínez <guido@vanguardiasur.com.ar>

Make it two functions, wl_get_wle() and wl_get_peb().
wl_get_peb() works exactly like __wl_get_peb() but wl_get_wle()
does not call produce_free_peb().
While refilling the fastmap user pool we cannot release ubi->wl_lock
as produce_free_peb() does.
Hence the fastmap logic uses now wl_get_wle().
Signed-off-by: NRichard Weinberger <richard@nod.at>

ubi_wl_get_peb() has two problems, it reads the pool
size and usage counters without any protection.
While reading one value would be perfectly fine it reads multiple
values and compares them. This is racy and can lead to incorrect
pool handling.
Furthermore ubi_update_fastmap() is called without wl_lock held,
before incrementing the used counter it needs to be checked again.
It could happen that another thread consumed all PEBs from the
pool and the counter goes beyond ->size.
Signed-off-by: NRichard Weinberger <richard@nod.at>

...otherwise the deferred work might run after datastructures
got freed and corrupt memory.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NGuido Martínez <guido@vanguardiasur.com.ar>

If the WL pool runs out of PEBs we schedule a fastmap write
to refill it as soon as possible.
Ensure that only one at a time is scheduled otherwise we might end in
a fastmap write storm because writing the fastmap can schedule another
write if bitflips are detected.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>
Reviewed-by: NGuido Martínez <guido@vanguardiasur.com.ar>

The kerneldoc for @vid_hdr_aloffset continues onto a second line, but
this is not obvious, because the second line isn't indented, and it
begins with '@'.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NRichard Weinberger <richard@nod.at>

The comparison from the previous line seems to have been erroneously
(partially) copied-and-pasted onto the next. The second line should be
checking req.bytes, not req.lnum.

Coverity CID #139400

Cc: stable <stable@vger.kernel.org>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
[rw: Fixed comparison]
Signed-off-by: NRichard Weinberger <richard@nod.at>

In some of the 'out_not_moved' error paths, lnum may be used
uninitialized. Don't ignore the warning; let's fix it.

This uninitialized variable doesn't have much visible effect in the end,
since we just schedule the PEB for erasure, and its LEB number doesn't
really matter (it just gets printed in debug messages). But let's get it
straight anyway.

Coverity CID #113449

Cc: stable <stable@vger.kernel.org>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NRichard Weinberger <richard@nod.at>

If aeb->len >= vol->reserved_pebs, we should not be writing aeb into the
PEB->LEB mapping.

Caught by Coverity, CID #711212.

Cc: stable <stable@vger.kernel.org>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NRichard Weinberger <richard@nod.at>

We are completely discarding the earlier value of 'bitflips', which
could reflect a bitflip found in ubi_io_read_vid_hdr(). Let's use the
bitwise OR of header and data 'bitflip' statuses instead.

Coverity CID #1226856

Cc: stable <stable@vger.kernel.org>
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NRichard Weinberger <richard@nod.at>

If ->run() fails, it can either free the data structures it
allocated, or leave that task to ->free() which will be called
on failures.

However:
  md.c calls ->free() even if ->private_data is NULL, which
     causes problems in some personalities.
  raid0.c frees the data, but doesn't clear ->private_data,
     which will become a problem when we fix md.c

So better fix both these issues at once.
Reported-by: NRichard W.M. Jones <rjones@redhat.com>
Fixes: 5aa61f42
URL: https://bugzilla.kernel.org/show_bug.cgi?id=94381Signed-off-by: NNeilBrown <neilb@suse.de>

Commit b4b55cda (Refine the way to release PCI IRQ resources)
introduced a regression in the PCI IRQ resource management by causing
the IRQ resource of a device, established when pci_enabled_device()
is called on a fully disabled device, to be released when the driver
is unbound from the device, regardless of the enable_cnt.

This leads to the situation that an ill-behaved driver can now make a
device unusable to subsequent drivers by an imbalance in their use of
pci_enable/disable_device().  That is a serious problem for secondary
drivers like vfio-pci, which are innocent of the transgressions of
the previous driver.

Since the solution of this problem is not immediate and requires
further discussion, revert commit b4b55cda and the issue it was
supposed to address (a bug related to xen-pciback) will be taken
care of in a different way going forward.
Reported-by: NAlex Williamson <alex.williamson@redhat.com>
Signed-off-by: NRafael J. Wysocki <rafael.j.wysocki@intel.com>

A check that rejects a CDB with FUA bit set if no write cache is
emulated was added by the following commit:

  fde9f50f target: Add sanity checks for DPO/FUA bit usage

The condition is as follows:

  if (!dev->dev_attrib.emulate_fua_write ||
      !dev->dev_attrib.emulate_write_cache)

However, this check is wrong if the backend device supports WCE but
"emulate_write_cache" is disabled.

This patch uses se_dev_check_wce() (previously named
spc_check_dev_wce) to invoke transport->get_write_cache() if the
device has a write cache or check the "emulate_write_cache" attribute
otherwise.
Reported-by: NChristoph Hellwig <hch@lst.de>
Signed-off-by: NChristophe Vu-Brugier <cvubrugier@fastmail.fm>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a NULL pointer dereference triggered by a late
target_configure_device() -> alloc_workqueue() failure that results
in target_free_device() being called with DF_CONFIGURED already set,
which subsequently OOPses in destroy_workqueue() code.

Currently this only happens at modprobe target_core_mod time when
core_dev_setup_virtual_lun0() -> target_configure_device() fails,
and the explicit target_free_device() gets called.

To address this bug originally introduced by commit 0fd97ccf, go
ahead and move DF_CONFIGURED to end of target_configure_device()
code to handle this special failure case.
Reported-by: NClaudio Fleiner <cmf@daterainc.com>
Cc: Claudio Fleiner <cmf@daterainc.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: <stable@vger.kernel.org> # v3.7+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a NULL pointer dereference OOPs with pSCSI backends
within target_core_stat.c code.  The bug is caused by a configfs attr
read if no pscsi_dev_virt->pdv_sd has been configured.
Reported-by: NOlaf Hering <olaf@aepfle.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch adds a missing set of conditional check braces in
ft_invl_hw_context() originally introduced by commit dcd998cc
when handling DDP failures in ft_recv_write_data() code.

 commit dcd998cc
 Author: Kiran Patil <kiran.patil@intel.com>
 Date:   Wed Aug 3 09:20:01 2011 +0000

    tcm_fc: Handle DDP/SW fc_frame_payload_get failures in ft_recv_write_data
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Cc: Kiran Patil <kiran.patil@intel.com>
Cc: <stable@vger.kernel.org> # 3.1+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a se_cmd->cmd_kref leak buf when se_sess->sess_tearing_down
is true within target_get_sess_cmd() submission path code.

This se_cmd reference leak can occur during active session shutdown when
ack_kref=1 is passed by target_submit_cmd_[map_sgls,tmr]() callers.
Signed-off-by: NBart Van Assche <bart.vanassche@sandisk.com>
Cc: <stable@vger.kernel.org> # 3.6+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch changes loopback, usb-gadget, vhost-scsi and xen-scsiback
fabric code to invoke transport_register_session() instead of the
unprotected flavour, to ensure se_tpg->session_lock is taken when
adding new session list nodes to se_tpg->tpg_sess_list.

Note that since these four fabric drivers already hold their own
internal TPG mutexes when accessing se_tpg->tpg_sess_list, and
consist of a single se_session created through configfs attribute
access, no list corruption can currently occur.

So for correctness sake, go ahead and use the se_tpg->session_lock
protected version for these four fabric drivers.
Signed-off-by: NBart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes the incorrect use of __transport_register_session()
in tcm_qla2xxx_check_initiator_node_acl() code, that does not perform
explicit se_tpg->session_lock when accessing se_tpg->tpg_sess_list
to add new se_sess nodes.

Given that tcm_qla2xxx_check_initiator_node_acl() is not called with
qla_hw->hardware_lock held for all accesses of ->tpg_sess_list, the
code should be using transport_register_session() instead.
Signed-off-by: NBart Van Assche <bart.vanassche@sandisk.com>
Cc: Giridhar Malavali <giridhar.malavali@qlogic.com>
Cc: Quinn Tran <quinn.tran@qlogic.com>
Cc: <stable@vger.kernel.org> # 3.5+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This patch fixes a iser specific logout bug where early complete()
of conn->conn_logout_comp in iscsit_close_connection() was causing
isert_wait4logout() to complete too soon, triggering a use after
free NULL pointer dereference of iscsi_conn memory.

The complete() was originally added for traditional iscsi-target
when a ISCSI_LOGOUT_OP failed in iscsi_target_rx_opcode(), but given
iser-target does not wait in logout failure, this special case needs
to be avoided.
Reported-by: NSagi Grimberg <sagig@mellanox.com>
Cc: Sagi Grimberg <sagig@mellanox.com>
Cc: Slava Shwartsman <valyushash@gmail.com>
Cc: <stable@vger.kernel.org> # v3.10+
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>

This reverts commit 72859d91.

The original patch was wrong, iscsit_close_connection() still needs
to release iscsi_conn during both normal + exception IN_LOGOUT status
with ib_isert enabled.

The original OOPs is due to completing conn_logout_comp early within
iscsit_close_connection(), causing isert_wait4logout() to complete
instead of waiting for iscsit_logout_post_handler_*() to be called.
Reported-by: NSagi Grimberg <sagig@mellanox.com>
Cc: Sagi Grimberg <sagig@mellanox.com>
Cc: Slava Shwartsman <valyushash@gmail.com>
Signed-off-by: NNicholas Bellinger <nab@linux-iscsi.org>