If the erase worker is unable to erase a PEB it will
free the ubi_wl_entry itself.
The failing ubi_wl_entry must not free()'d again after
do_sync_erase() returns.

Cc: <stable@vger.kernel.org>
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

This is more a cosmetic change than a fix.
By using ubi_eba_atomic_leb_change()
we can guarantee that the first VTBL record is always
correct and we don't really need the second one anymore.
But we have to keep the second one to not break anything.

Artem: add a comment
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

If there is more then one UBI device mounted, there is no way to
distinguish between messages from different UBI devices.
Add device number to all ubi layer message types.

The R/O block driver messages were replaced by pr_* since
ubi_device structure is not used by it.

Amended a bit by Artem.
Signed-off-by: NTanya Brokhman <tlinder@codeaurora.org>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

We need to add fm_sb too.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Reviewed-by: NTanya Brokhman <tlinder@codeaurora.org>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

s/of/if/
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

It confused me more than once that the cancel flag of the
work function does not indicate the cancellation of a single work.
In fact it indicates the WL sub-system shutdown and therefore
worker functions have to free their wl_entries too.
That's why you cannot cancel a single work, you can only shutdown
all works.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

There is no need to set err, it will be overwritten in any case
later at:
        if (scrub)
                err = ubi_wl_scrub_peb(ubi, pnum);
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The while loop in produce_free_peb() assumes that each work will produce a
free PEB. This is not true.
If ubi->works_count is 1 and the only scheduled work is the
wear_leveling_worker() produce_free_peb() can loop forever in case
nobody schedules an erase work.
Fix this issue by checking in the while loop whether work is scheduled.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

I noticed this during a code review.  We are checking that the strlen()
of ->name is not less than the ->name_len which the user gave us.  I
believe this bug is harmless but clearly we meant to return here instead
of setting an error code and then not using it.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Make clear what work_sem really does.
Suggested-by: NArtem Bityutskiy <dedekind1@gmail.com>
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The UBI_IOCVOLUP ioctl is used to start an update and also to
truncate a volume. In the first case, a "volume updated" notification
is dispatched when the update is done.

This commit adds the "volume updated" notification to be also sent when
the volume is truncated. This is required for UBI block and gluebi to get
notified about the new volume size.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Cc: stable@vger.kernel.org # v3.15+

Static volumes can change its 'used_bytes' when they get updated,
and so the block interface must listen to the UBI_VOLUME_UPDATED
notification to resize the block device accordingly.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Cc: stable@vger.kernel.org # v3.15+

We are currently taking the block device size from the ubi_volume_info.size
field. However, this is not the amount of data in the volume, but the
number of reserved physical eraseblocks, and hence leads to an incorrect
representation of the volume.

In particular, this produces I/O errors on static volumes as the block
interface may attempt to read unmapped PEBs:

$ cat /dev/ubiblock0_0 > /dev/null
UBI error: ubiblock_read_to_buf: ubiblock0_0 ubi_read error -22
end_request: I/O error, dev ubiblock0_0, sector 9536
Buffer I/O error on device ubiblock0_0, logical block 2384
[snip]

Fix this by using the ubi_volume_info.used_bytes field which is set to the
actual number of data bytes for both static and dynamic volumes.

While here, improve the error message to be less stupid and more useful:
UBI error: ubiblock_read_to_buf: ubiblock0_1 ubi_read error -9 on LEB=0, off=15872, len=512

It's worth noticing that the 512-byte sector representation of the volume
is only correct if the volume size is multiple of 512-bytes. This is true for
virtually any NAND device, given eraseblocks and pages are 512-byte multiple
and hence so is the LEB size.

Artem: tweak the error message and make it look more like other UBI error
messages.

Fixes: 9d54c8a3 ("UBI: R/O block driver on top of UBI volumes")
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Cc: stable@vger.kernel.org # v3.15+

commit 4df38926 ("UBI: block: Avoid disk size integer overflow")
introduced a dereference on dev (which is not initialized at that
point) when printing a warning message.  Re-order disk_capacity check
after the dev is found.

Found by cppcheck:
 [drivers/mtd/ubi/block.c:509]: (error) Uninitialized variable: dev

Artem: tweak the error message a bit
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Acked-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

I ran into this error after a ubiupdatevol, because I forgot to backport
e9110361 UBI: fix the volumes tree sorting criteria.

UBI error: process_pool_aeb: orphaned volume in fastmap pool
UBI error: ubi_scan_fastmap: Attach by fastmap failed, doing a full scan!
kmem_cache_destroy ubi_ainf_peb_slab: Slab cache still has objects
CPU: 0 PID: 1 Comm: swapper Not tainted 3.14.18-00053-gf05cac8dbf85 #1
[<c000d298>] (unwind_backtrace) from [<c000baa8>] (show_stack+0x10/0x14)
[<c000baa8>] (show_stack) from [<c01b7a68>] (destroy_ai+0x230/0x244)
[<c01b7a68>] (destroy_ai) from [<c01b8fd4>] (ubi_attach+0x98/0x1ec)
[<c01b8fd4>] (ubi_attach) from [<c01ade90>] (ubi_attach_mtd_dev+0x2b8/0x868)
[<c01ade90>] (ubi_attach_mtd_dev) from [<c038b510>] (ubi_init+0x1dc/0x2ac)
[<c038b510>] (ubi_init) from [<c0008860>] (do_one_initcall+0x94/0x140)
[<c0008860>] (do_one_initcall) from [<c037aadc>] (kernel_init_freeable+0xe8/0x1b0)
[<c037aadc>] (kernel_init_freeable) from [<c02730ac>] (kernel_init+0x8/0xe4)
[<c02730ac>] (kernel_init) from [<c00093f0>] (ret_from_fork+0x14/0x24)
UBI: scanning is finished

Freeing the cache in the error path fixes the Slab error.

Tested on at91sam9g35 (3.14.18+fastmap backports)
Signed-off-by: NRichard Genoud <richard.genoud@gmail.com>
Cc: stable <stable@vger.kernel.org> # 3.10+

Use the _safe variant because we're iterating over a list where items get
deleted and freed.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

This patch fixes the issue that on very large UBI volumes
UBI block does not work correctly.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

There's no need to set the disk capacity with the mutex held, so this
commit takes the variable setting out of the mutex. This simplifies
the disk capacity fix for very large volumes in a follow up commit.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Currently, ubiblock_resize() can fail if the device is not found
in the list. This commit changes the return type, so the function can
return something meaningful on error paths.
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

UBI assumes that ubi_attach_info will only contain ubi_ainf_volume
structures for volumes with at least one LEB.
In scanning mode this is true because UBI can nicely create a ubi_ainf_volume
on demand while creating the EBA table.

For fastmap this is not true, the fastmap on-flash structure has a list of
all volumes, the ubi_ainf_volume structures are created from this list.
So it can happen that an empty volume ends up in init_volumes().

We can easely deal with that by looking into ->leb_count too.
Signed-off-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The return value from 'ubi_io_read_ec_hdr()' was stored in 'err', not in 'ret'.
This fix makes sure Fastmap-enabled UBI does not miss bit-flip while reading EC
headers, events and scrubs the affected PEBs.

This issue was reported by Coverity Scan.

Artem: improved the commit message.
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>
Acked-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Commig "604b592e UBI: fix rb_tree node comparison in add_map"
broke fastmap backward compatibility and older fastmap images
cannot be mounted anymore. The reason is that it changes the
volumes RB-tree sorting criteria. This patch fixes the problem.

Artem: re-write the commit message
Signed-off-by: NHeiko Schocher <hs@denx.de>
Acked-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Otherwise we'd return a random value if allocation of the workqueue fails.
Signed-off-by: NHelmut Schaa <helmut.schaa@googlemail.com>
Acked-by: NBrian Norris <computersforpeace@gmail.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The comparisons used in add_vol() shouldn't be identical.  Pretty sure
the following is correct but it is completely untested.
Signed-off-by: NMike Snitzer <snitzer@redhat.com>
Acked-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The UBI volume rename ioctl (UBI_IOCRNVOL) open the volumes in exclusive
mode. The volumes are opened for two reasons: to build a volume rename list,
and a volume remove list.

However, the first open constraint is excessive and can be replaced by
a 'read-write' open mode. The second open constraint is properly set as
'exclusive' given the volume is opened for removal and we don't want any
users around.

By weakening the former 'exclusive' mode, we allow 'read-only' users to keep
the volume open, while a rename is taking place. This is useful to perform
an atomic rename, in a firmware upgrade scenario, while keeping the volume
in read-only use (for instance, if a ubiblock is mounted as rootfs).

It's worth mention this is not the case of UBIFS, which keeps the volume
opened as 'read-write' despite mounted as read-write or read-only mode.

This change was suggested at least twice by Artem:
http://lists.infradead.org/pipermail/linux-mtd/2012-September/044175.html
http://permalink.gmane.org/gmane.linux.drivers.mtd/39866Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

When building the name for the workqueue thread, make sure a format
string cannot leak in from the disk name.
Signed-off-by: NKees Cook <keescook@chromium.org>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The ubi->free_count should be updated with every insert/remove to/from
the ubi->free list.
Signed-off-by: NTanya Brokhman <tlinder@codeaurora.org>
Reviewed-by: NDolev Raviv <draviv@codeaurora.org>
Acked-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

In case of an error (if there are not free PEB's for example),
__wl_get_peb will return a negative value. In order to prevent access
violation we need to test the returned value prior to using it later on.
Signed-off-by: NTatyana Brokhman <tlinder@codeaurora.org>
Reviewed-by: NDolev Raviv <draviv@codeaurora.org>
Acked-by: NRichard Weinberger <richard@nod.at>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

This was used in the olden days, back when onions were proper
yellow. Basically it mapped to the current buffer to be
transferred. With highmem being added more than a decade ago,
most drivers map pages out of a bio, and rq->buffer isn't
pointing at anything valid.

Convert old style drivers to just use bio_data().

For the discard payload use case, just reference the page
in the bio.
Signed-off-by: NJens Axboe <axboe@fb.com>

You cannot mark these parameters as __initdata.
Otherwise the data is gone upon module exit.

Fixes:
[  172.045465] BUG: unable to handle kernel paging request at ffffffffa001db38
[  172.046020] IP: [<ffffffff81067aa4>] destroy_params+0x24/0x50
Signed-off-by: NRichard Weinberger <richard@nod.at>
Acked-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.  Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.

Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Brian Norris <computersforpeace@gmail.com>
Cc: linux-mtd@lists.infradead.org
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
[Brian: dropped one incorrect hunk]
Signed-off-by: NBrian Norris <computersforpeace@gmail.com>

Rename the UBI_IOCVOLATTBLK and UBI_IOCVOLDETBLK to UBI_IOCVOLCRBLK and
UBI_IOCVOLRMBLK, because we do not use terms "attach" and "detach" for the R/O
block devices on top of UBI volumes. Instead, we use terms "create" and
"remove". This patch also amends the related commentaries.
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>
Acked-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>

In order to have a way of distinguishing an invalid ioctl from a
not supported (but otherwise valid) ioctl, this commit changes the
return value of the ioctl stubs from ENOTTY to ENOSYS.

This will be useful to report more accurate error messages from
userspace tools.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

The UBI block driver depends on the block infrastructure. Add the
proper dependency and fix a build error when CONFIG_BLOCK is not selected.
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Fixes the following warning on ARCH=avr32:

  drivers/mtd/ubi/block.c: In function 'ubiblock_read':
  drivers/mtd/ubi/block.c:207: warning: comparison of distinct pointer types lacks a cast
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

ubiblock_param_ops should be marked as __init as it's only used to set
a driver parameter on insertion time. This commit fixes the following:

  WARNING: drivers/mtd/built-in.o(.text+0x653ac): Section mismatch in
  reference from the variable ubiblock_param_ops to the function
  .init.text:ubiblock_set_param()

  The function ubiblock_param_ops() references the function __init
  ubiblock_set_param(). This is often because ubiblock_param_ops lacks a
  __init annotation or the annotation of ubiblock_set_param is wrong.

Given gcc errors if the struct is marked const __initdata, this commit
drops the const mark from it.
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

We already use term attach/detach for UBI->MTD relations, let's not use this
for UBI->ubiblock relations to avoid confusion. Just use 'create' and 'remove'
instead. E.g., "create a R/O block device on top of a UBI volume".
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

This commit introduces read-only block device emulation on top of UBI volumes.

Given UBI takes care of wear leveling and bad block management it's possible
to add a thin layer to enable block device access to UBI volumes.
This allows to use a block-oriented filesystem on a flash device.

The UBI block devices are meant to be used in conjunction with any
regular, block-oriented file system (e.g. ext4), although it's primarily
targeted at read-only file systems, such as squashfs.

Block devices are created upon user request through new ioctls:
UBI_IOCVOLATTBLK to attach and UBI_IOCVOLDETBLK to detach.
Also, a new UBI module parameter is added 'ubi.block'. This parameter is
needed in order to attach a block device on boot-up time, allowing to
mount the rootfs on a ubiblock device.
For instance, you could have these kernel parameters:

  ubi.mtd=5 ubi.block=0,0 root=/dev/ubiblock0_0

Or, if you compile ubi as a module:

  $ modprobe ubi mtd=/dev/mtd5 block=/dev/ubi0_0

Artem: amend commentaries and massage the patch a little bit.
Signed-off-by: NEzequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

Move the kmem_cache_free() calls down a couple lines.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>

nor_erase_prepare() will be called before erase a NOR flash, it will program '0'
into a block to mark this block. But program data into a erasure interrupted block
can cause program timtout(several minutes at most) error, could impact other
operation on NOR flash. So UBIFS can read this block first to avoid unneeded
program operation.

This patch try to put read operation at head of write operation in
nor_erase_prepare(), read out the data.
If the data is already corrupt, then no need to program any data into this block,
just go to erase this block.

This patch is validated on Micron NOR flash, part number is：JS28F512M29EWHA
Signed-off-by: NQi Wang <qiwang@micron.com>
Signed-off-by: NArtem Bityutskiy <artem.bityutskiy@linux.intel.com>