Some architectures provides architecture-specific, optimized version of
clear_page()/copy_page(), which may have better performance than
memset()/memcpy(). So use clear_page()/copy_page() to optimize zram
performance if possible.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Now there's no caller of zram_get_num_devices(), so kill it.
And change zram_devices to static because it's only used in zram_drv.c.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Simplify and optimize dev_to_zram() without walking the zram_devices
array.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use zram->init_lock to protect access to zram->meta, otherwise it
may cause invalid memory access if zram->meta has been freed by
zram_reset_device().

This issue may be triggered by:
Thread 1:
while true; do cat mem_used_total; done
Thread 2:
while true; do echo 8M > disksize; echo 1 > reset; done
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Acked-by: NMinchan Kim <minchan@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Function valid_io_request() should verify the entire request are within
the zram device address range. Otherwise it may cause invalid memory
access when accessing/modifying zram->meta->table[index] because the
'index' is out of range. Then it may access non-exist memory, randomly
modify memory belong to other subsystems, which is hard to track down.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

When doing a patial write and the whole page is filled with zero,
zram_bvec_write() will free uncmem twice.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Acked-by: NMinchan Kim <minchan@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

On error recovery path of zram_init(), it leaks the zram device object
causing the failure. So change create_device() to free allocated
resources on error path.
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Acked-by: NMinchan Kim <minchan@kernel.org>
Acked-by: NJerome Marchand <jmarchan@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

zram_slot_free_notify() is free-running without any protection from
concurrent operations. So there are race conditions between
zram_bvec_read()/zram_bvec_write() and zram_slot_free_notify(),
and possible consequences include:
1) Trigger BUG_ON(!handle) on zram_bvec_write() side.
2) Access to freed pages on zram_bvec_read() side.
3) Break some fields (bad_compress, good_compress, pages_stored)
   in zram->stats if the swap layer makes concurrently call to
   zram_slot_free_notify().

So enhance zram_slot_free_notify() to acquire writer lock on zram->lock
before calling zram_free_page().
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Memory for zram->disk object may have already been freed after returning
from destroy_device(zram), then it's unsafe for zram_reset_device(zram)
to access zram->disk again.

We can't solve this bug by flipping the order of destroy_device(zram)
and zram_reset_device(zram), that will cause deadlock issues to the
zram sysfs handler.

So fix it by holding an extra reference to zram->disk before calling
destroy_device(zram).
Signed-off-by: NJiang Liu <jiang.liu@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Tidy up the multi-line comment at the beginning of the file to
follow the CodingStyle.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Rename the boardinfo table so it has namespace associated with the
driver.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetic reasons, add some whitespace to the subdevice init
and reorder it a bit.

Remove the blank line printk at the end of the attach.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetic reasons, rename the function.

Use a local variable for the 'reg' offset that is stored in the
comedi_subdevice 'private' pointer to minimize the number of casts.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetic reasons, rename the function.

Use a local variable for the 'reg' offset that is stored in the
comedi_subdevice 'private' pointer to minimize the number of casts.
Also, add local variables for the 'mask' and 'bits'.

The outputs only need to be updated if the 'mask' indicates that the
'bits' are changing. Move the update code into the main 'if (mask)'
block.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Convert the boardinfo declaration to C99 format and move it near the
struct definition.

Since the *_SIZE defines are only used in the boardinfo, remove them
and just open code the values in the boardinfo.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Introduce some helper functions to check if a given 'range' index
to a comedi_subdevice 'range_table' is a bipolar or unipolar range.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This driver does not use interrupts and the comedi core handles the
allocation of the io resource so these includes are not required.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

According to the PCM-A/D-12/16 Operations Manual:

NOTE: On the PCM-A/D-12 the lower nibble of the hex value will always be 0.

Fix the pcmad_ai_insn_read() function to properly handle this by shifting
the data after it has been read.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The comedi_config utility is used to attach to this board. One of the
configuration options passed by the user sets the analog input encoding
for straight binary or two's complement data. The hardware produces
straight binary data when jumpered for 5V unipolar inputs and two's
complement data when jumpered for +-10V bipolar inputs.

Use the configuration option to correctly set the comedi_subdevice
range_table.

We can then use a helper function to determine what the range is when
reading the analog inputs. This allows removing the 'twos_comp' variable
from the private data (which was actually never used).

Since the private data is now empty, remove it completely.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The comedi_config utility is used to attach to this board. One of the
configuration options passed by the user sets the analog input reference
for single-ended or differential inputs.

Use the configuration option to correctly set the comedi_subdevice flags
as well as the number of channels.

Remove the 'differential' variable from the private data. Its not used
by the driver.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This information is only used to set the subdevice 'maxdata'. Change
it so the calculation is not needed.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Remove the comment before the function, the options are documented at
the beginning of the file.

Remove the PCMAD_SIZE define. It's only used in the attach to specify
the io region size.

For aesthetic reasons, add some whitespace to the subdevice init and
reorder them a bit.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Use a local variable to read and munge the analog input data instead
of directly using the 'data' pointer passed to the function.

(*insn_read) functions either return an errno or the number of data
values read. Change the final return to insn->n to make this clearer.

Tidy up the function.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

The const boardinfo pointer is only used in this function to work out
the value needed to convert the data to its twos complement. Use the
comedi_subdevice 'maxdata' to do this conversion and remove the 'board'
pointer.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Instead of returning invalid data, return -ETIME if the analog input
conversion times out.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

For aesthetic reasons, move the boardinfo declaration so it follows
the definition.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Tidy up the multi-line comments at the beginning of the file to
follow the CodingStyle.
Signed-off-by: NH Hartley Sweeten <hsweeten@visionengravers.com>
Cc: Ian Abbott <abbotti@mev.co.uk>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

read_nic_*() functions are defined in r8192U_core.c.
They call internally usb_control_msg() to read the
nic registers and return the value read.
Following a remark made by Dan Carpenter, if usb_control_msg()
fails, the value returned will be invalid.

To accommodate for this, this patch changes the functions
to take a pointer as argument to set the value read and
return 0 on success and the error status on failure, so
that callers of read_nic_*() can check the return status.

Some other fixes introduced in read_nic_*() functions are:

The expressions (1<<EPROM_*_SHIFT) used to address and set
the individual bits of the eeprom register were replaced
with EPROM_*_BIT bitmasks to make the code more intuitive.
EPROM_*_BIT bitmasks were defined in r8192U_hw.h and
EPROM_*_SHIFT were removed.

In netdev_err(), which is called in case of failure,
the hardcoded function name in the error log message was
replaced with __func__ to reduce line size.
Also, from the error log message, it was omitted the word
"Timeout" and it is just reported the error code since the
failure can not only be due to timeout expiration but also
due to a memory allocation failure. In case of timeout
expiration, usb_start_wait_urb() prints an appropriate log
message when debug is enabled.

Finally, some minor fixes to the coding style were applied in
lines affected by the above changes, including the removal
of ifdef DEBUG_RX (the debugging of reads and writes of the
nic registers shall be done with explicit check on their
return status which will be added in a follow on patch).
Signed-off-by: NXenia Ragiadakou <burzalodowa@gmail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Sean MacLennan fixed this in the rtl8192e driver and we should fix
it here as well.

It's pretty harmless.  This information comes from the firmware, if
we were to hit this bug we would read beyond the end of the array
once or twice before returning and update our statistics with bogus
data.
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

resolves checkpatch errors and warnings regarding whitespace around
operators, line lengths and indentation.
Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NJohannes Schilling <of82ecuq@cip.cs.fau.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Function usb_stor_print_cmd was declared in init.c but only used
in transport.c. So it was reasonable to move it there and declare
it static
Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NJohannes Schilling <of82ecuq@cip.cs.fau.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

as per suggestion of checkpatch.pl; this resolves warnings like "Prefer
netdev_info .. then dev_info .. to printk".
Changed signature of function usb_stor_print_cmd in init.c to enable
usage of dev_err
Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NJohannes Schilling <of82ecuq@cip.cs.fau.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

as advised by checkpatch, changed generic max(..) to max_t(int, ..
Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NJohannes Schilling <of82ecuq@cip.cs.fau.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

resolves checkpatch.pl warning "do not add new typedefs" and renames allcaps
structures.
Signed-off-by: NLaura Lawniczak <laura.lawniczak@googlemail.com>
Signed-off-by: NJohannes Schilling <of82ecuq@cip.cs.fau.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

No need to cast count since it is already ssize_t. No need
to cast payload to const, but need __force instead to avoid
Sparse complaining.
Reported-and-Suggested-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NPeng Tao <tao.peng@emc.com>
Signed-off-by: NAndreas Dilger <andreas.dilger@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

mdc_kuc_fops is missing open/release handlers. I fixed it before but
somehow forgot to amend to the patch sent out. Sorry...
Signed-off-by: NPeng Tao <tao.peng@emc.com>
Signed-off-by: NAndreas Dilger <andreas.dilger@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

This reverts commit 37d4093f.

I've verified that we now don't break build on X86_64 allmodconfig.

Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPeng Tao <tao.peng@emc.com>
Signed-off-by: NAndreas Dilger <andreas.dilger@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

We don't need to implement crc32 and crc32pclmul on our own.
The crc32-pclmul support was merged into the 3.8 kernel in commit
78c37d19, no need to keep a local copy in Lustre anymore.
The crc32 implementation is identical to crypto-crc32. So drop
Lustre's private implementation and select kernel crypto in Kconfig.
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPeng Tao <tao.peng@emc.com>
Signed-off-by: NAndreas Dilger <andreas.dilger@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

module_refcount() is not available when CONFIG_MODULE_UNLOAD is off.
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NPeng Tao <tao.peng@emc.com>
Signed-off-by: NAndreas Dilger <andreas.dilger@intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>