This commit allows installing a custom reg_update_bits function for cases where
the hardware provides a mechanism to set or clear register bits without a
read/modify/write cycle. Such is the case with the Microchip ENCX24J600.
If a custom reg_update_bits function is provided, it will only be used against
volatile registers.
Signed-off-by: Jon Ringle <jringle@gridpoint.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If a regmap is using fast_io, allocate the scratch buffer in
regmap_bulk_write() with GFP_ATOMIC instead of GFP_KERNEL.
Otherwise we may schedule while atomic.
Reported-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
__regmap_init() may receive a NULL `struct regmap_bus *bus' pointer,
for example, from snd_hdac_regmap_init(), and it make sure that it
does not NULL deference `bus`, except around ->max_raw_read and
->max_raw_write initialisation. Add missing check.
Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
When doing a bulk read from a device which lacks raw I/O support we fall
back to doing register at a time reads but we still use the raw
formatters in order to render the data into the word size used by the
device (since bulk reads still operate on the device word size rather
than unsigned ints). This means that devices without raw formatting
such as those that provide reg_read() are not supported. Provide
handling for them by copying the values read into native endian values
of the appropriate size.
Signed-off-by: Mark Brown <broonie@kernel.org>
Check in regmap_raw_read() and regmap_raw_write() for correct maximum
sizes of the operations. Return -E2BIG if this size is not supported
because it is too big.
Also this patch causes an uninitialized variable warning so it
initializes ret (although not necessary).
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Add functions to access the maximum size we can read/write using
regmap_raw_read/write().
This helps drivers that need to know how much they can write with the
raw functions without problems. There are some devices (e.g. bmc150)
that have fifos as registers which need to be read in specific chunks
otherwise samples are dropped.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
There are some buses which have a limit on the maximum number of bytes
that can be send/received. An example for this is
I2C_FUNC_SMBUS_I2C_BLOCK which does not support any reads/writes of more
than 32 bytes. The regmap_bulk operations should still be able to
utilize the full 32 bytes in this case.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
There is no multi_write support available if we cannot use raw_write.
This is the case if bus->write is not implemented.
This patch adds a condition that we need bus and bus->write so that
can_multi_write is true.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
use_single_rw currently reflects the capabilities of the connected
device. The capabilities of the bus are currently missing for this
variable.
As there are read only and write only buses we need seperate values for
use_single_rw to also reflect tha capabilities of the bus.
This patch splits use_single_rw into use_single_read and
use_single_write. The initialization is changed to check the
configuration for use_single_rw and to check the capabilities of the
used bus.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The regmap config does not prohibit val_bytes that are not powers of
two. But the current code of regmap_bulk_write for use_single_rw does
limit the possible val_bytes to 1, 2 and 4.
This patch fixes the behaviour to allow bus writes with non-standard
val_bytes sizes.
Cc: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Return -ENOTSUPP if map->bus->read is not implemented and we do not use
the cache. This code path would directly use bus->read would run into an
NULL pointer for the read function.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
This function is missing a check if map->bus->write is implemented. If
it is not implemented arbitrary raw writes are not possible.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
There are two typos in drivers/base/regmap/regmap.c, and they may
introduce some noise when checking new patches.
Signed-off-by: Xiubo Li <lixiubo@cmss.chinamobile.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
These values are defined as unsigned int in the struct and are assigned
to int values.
This patch fixes the type to be unsigned int instead.
Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Init functions defined in regmap*.c files are now prefixed with
__, take lockdep key and class parameters, and should not be
called directly: move the documentation to regmap.h, where the
macros are defined.
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
Lockdep validator complains about recursive locking and deadlock
when two different regmap instances are called in a nested order.
That happens anytime a regmap read/write call needs to access
another regmap.
This is because, for performance reason, lockdep groups all locks
initialized by the same mutex_init() in the same lock class.
Therefore all regmap mutexes are in the same lock class, leading
to lockdep "nested locking" warnings if a regmap accesses another
regmap.
In general, it is impossible to establish in advance the hierarchy
of regmaps, so we make sure that each regmap init call initializes
its own static lock_class_key. This is done by wrapping all
regmap_init calls into macros.
This also allows us to give meaningful names to the lock_class_key.
For example, in rt5677 case, we have in /proc/lockdep_chains:
irq_context: 0
[ffffffc0018d2198] &dev->mutex
[ffffffc0018d2198] &dev->mutex
[ffffffc001bd7f60] rt5677:5104:(&rt5677_regmap)->_lock
[ffffffc001bd7f58] rt5677:5096:(&rt5677_regmap_physical)->_lock
[ffffffc001b95448] &(&base->lock)->rlock
The above would have resulted in a lockdep recursive warning
previously. This is not the case anymore as the lockdep validator
now clearly identifies the 2 regmaps as separate.
Signed-off-by: Nicolas Boichat <drinkcat@chromium.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
Add an optional delay_us field in reg_sequence to allow the client to
specify a delay (in microseconds) to be applied after any given write
in a sequence of writes.
We treat a delay in a sequence the same way we treat a page change as
they are logically similar in that you can coalesce all write before
a delay (in the same way you can coalesce all writes before a page
change is needed)
Signed-off-by: Nariman Poushin <nariman@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Separate the functionality using sequences of register writes from the
functions that take register defaults. This change renames the arguments
in order to support the extension of reg_sequence to take an optional
delay to be applied after any given register in a sequence is written.
This avoids adding an int to all register defaults, which could
substantially increase memory usage for regmaps with large default tables.
This also updates all the clients of multi_reg_write/register_patch.
Signed-off-by: Nariman Poushin <nariman@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Zero length reads make no sense in a regmap context and are likely to
trigger bugs further down the stack so insert an error check, also
silencing compiler warnings about use of ret in cases where we iterate
per register.
Reported-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently regmap requires that a reg_read callback is supplied, otherwise a
warning is emitted each time regmap_read() is called. This means a device
or bus without readback support needs to supply dummy reg_read callback.
Apart from that regmap_read() will still work fine if a cache is used.
Remove the warning and let regmap_readable() return false if not reg_read
callback is supplied. This means a device no longer has to supply a dummy
callback if it does not support readback and it also doesn't have to have a
readable_reg callback that always returns false since this is now implicit.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
regmap_fields_force_write() is similar to regmap_fields_write(),
but regmap_fields_force_write() write data to register even though
it is same value.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
regmap_write_bits() is similar to regmap_update_bits(),
but regmap_write_bits() write data to register even though
it is same value.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Sometimes we want to write data even though it doesn't change value.
Then, force_write option on _regmap_update_bits() helps this purpose.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The way the mask is generated in regmap_field_init() is wrong.
Indeed, a field initialized with msb = 31 and lsb = 0 provokes a shift
overflow while calculating the mask field.
On some 32 bits architectures, such as x86, the generated mask is 0,
instead of the expected 0xffffffff.
This patch uses GENMASK() to fix the problem, as this macro is already safe
regarding shift overflow.
Signed-off-by: Maxime Coquelin <maxime.coquelin@st.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
In big endian mode regmap_bulk_read gives incorrect data
for byte reads.
This is because memcpy of a single byte from an address
after full word read gives different results when
endianness differs. ie. we get little-end in LE and big-end in BE.
Signed-off-by: Arun Chandran <achandran@mvista.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Delete jump to a label on the next line, when that label is not
used elsewhere.
A simplified version of the semantic patch that makes this change is as
follows: (http://coccinelle.lip6.fr/)
// <smpl>
@r@
identifier l;
@@
-if (...) goto l;
-l:
// </smpl>
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch introduces regmap_get_reg_stride() function which would
be used by the infrastructures like nvmem framework built on top of
regmap. Mostly this function would be used for sanity checks on inputs
within such infrastructure.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch introduces regmap_get_max_register() function which would be
used by the infrastructures like nvmem framework built on top of
regmap.
Signed-off-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
The tracing events for regmap are confined to the regmap subsystem. It
also requires accessing an internal header. Instead of including the
internal header from a generic file location, move the tracing file
into the regmap directory.
Also rename the regmap tracing header to trace.h, as it is redundant to
keep the regmap.h name when it is in the regmap directory.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
We'll need to call it from regmap-i2c.c, which can be built as module.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Mark Brown <broonie@kernel.org>
Since we cannot make sure the 'val_count' will always be none zero
here, and then if it equals to zero, the kmemdup() will return
ZERO_SIZE_PTR, which equals to ((void *)16).
So this patch fix this with just doing the zero check before calling
kmemdup().
Signed-off-by: Xiubo Li <Li.Xiubo@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
If LOG_DEVICE is defined and map->dev is NULL it will lead to NULL
pointer dereference. This patch fixes this issue by adding check for
dev->NULL in all such places in regmap.c
Signed-off-by: Pankaj Dubey <pankaj.dubey@samsung.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Recents commits for getting reg endianness causing NULL pointer
dereference if dev is passed NULL in regmap_init_mmio. This patch
fixes this issue, and allows to parse reg endianness only if dev
and dev->of_node exist.
Signed-off-by: Pankaj Dubey <pankaj.dubey@samsung.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Split regmap_get_endian() in two functions, regmap_get_reg_endian() and
regmap_get_val_endian().
This allows to:
- Get rid of the three switch()es on "type", incl. error handling in
three "default" cases,
- Get rid of the regmap_endian_type enum,
- Get rid of the non-NULL check of "config" (regmap_init() already
checks for that),
- Get rid of the "endian" output parameters, and just return the
regmap_endian enum value, as the functions can no longer fail.
This saves 21 lines of code (despite the still-present
one-comment-per-line over-documentation), and 30 bytes of code on ARM
V7.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
A previous over-zealous factorisation of code means that we only treat
registers as volatile if they are readable. For most devices this is fine
since normally most registers can be read and volatility implies
readability but for format_write() devices where there is no readback from
the hardware and we use volatility to mean simply uncacheability this means
that we end up treating all registers as cacheble.
A bigger refactoring of the code to clarify this is in order but as a fix
make a minimal change and only check readability when checking volatility
if there is no format_write() operation defined for the device.
Signed-off-by: Mark Brown <broonie@linaro.org>
Tested-by: Lars-Peter Clausen <lars@metafoo.de>
Cc: stable@vger.kernel.org
Commit d647c19951 ("regmap: add DT endianness binding support") had
some issues. Commit ba1b53feb8 ("regmap: Fix DT endianess parsing
logic") fixed the main problem. This patch fixes the other.
Specifically, restore the overall default of REGMAP_ENDIAN_BIG if none of
the config, DT, or the bus specify any endianness. Without this,
of_regmap_get_endian() could return REGMAP_ENDIAN_DEFAULT, which the
calling code can't handle. Since all busses do specify an endianness in
the current code, this makes no difference right now, but I saw no
justification in the patch description for removing this final default.
Also, clean up the code a bit:
* s/of_regmap_get_endian/regmap_get_endian/ since the function isn't DT-
specific, even if the reason it was originally added was to add some
DT-specific features.
* After potentially reading an endianess specification from DT, the code
checks whether DT did specify an endianness, and if so, returns it. Move
this test outside the whole switch statement so that if the
REGMAP_ENDIAN_REG case ever modifies *endian, this check will pick that
up. This partially reverts part of commit ba1b53feb8 ("regmap: Fix DT
endianess parsing logic"), while maintaining the bug-fix that commit
made to this code.
* Make the comments briefer, and only refer to the specific action taken
at their location. This makes most of the comments independent of DT,
and easier to follow.
Cc: Xiubo Li <Li.Xiubo@freescale.com>
Cc: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Cc: Thierry Reding <treding@nvidia.com>
Fixes: d647c19951 ("regmap: add DT endianness binding support")
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
Commit d647c19951 ("regmap: add DT endianness binding support.")
added support to parse the device endianness from the device tree
but unfortunately the added logic doesn't have the same semantics
than the old code. This leads to a NULL dereference pointer error
when these properties are not provided by the Device Tree:
Unable to handle kernel NULL pointer dereference at virtual address 00000044
pgd = c0004000
[00000044] *pgd=00000000
Internal error: Oops: 5 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 5 PID: 1 Comm: swapper/0 Not tainted 3.17.0-rc1-next-20140818ccu #671
task: ea412800 ti: ea484000 task.ti: ea484000
PC is at regmap_update_bits+0xc/0x5c
The problem is that platforms that rely on the default value now
gets different values due two related issues in the current code:
a) It only parses the endianness from DT for the regmap registers
and not for the regmap values but it checks unconditionally in
both cases if the resulting endiannes is REGMAP_ENDIAN_NATIVE.
b) REGMAP_ENDIAN_NATIVE is not even a valid DT property according
to the regmap DT binding documentation so it shouldn't be set.
Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: Mark Brown <broonie@linaro.org>
