To support double channel shared controls split across 2 registers, one
for each channel, we must be able to update both registers together.
Add a second set of register fields to struct snd_soc_dapm_update, and
update the DAPM control writeback (put) callbacks to support this.
For codecs that use custom events which call into DAPM to do updates,
also clear struct snd_soc_dapm_update before using it, so the second
set of fields remains clean.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
The DAI ID defines are back from the time when DAIs were referenced by a
numerical ID. These days a string is used for matching instead and the
defines are unused. The last user of these defines was removed in commit
f0fba2ad1b ("ASoC: multi-component - ASoC Multi-Component Support"). So
remove the defines as well.
This also means the wm9712.h file no longer has any content and can be
removed.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Since the regmap conversion ac97_read/ac97_write are just simple wrappers
around snd_soc_read/snd_soc_write. Use those instead directly and remove
the wrappers. Also use snd_soc_update_bits() were appropriate.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Marek Vasut <marex@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently the wm9712 driver still uses custom snd_soc_codec_driver IO
callbacks. This has been deprecated for a while, so convert the wm9712
driver to use regmap for its IO.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Marek Vasut <marex@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Check for snd_soc_codec_driver structures that are only passed to
snd_soc_register_codec or memcpy (2nd arg), for which the corresponding
parameters are declared const. Declare as const snd_soc_codec_driver
structures that have these properties.
The semantic patch that makes this change is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@r disable optional_qualifier@
identifier i;
position p;
@@
static struct snd_soc_codec_driver i@p = { ... };
@ok@
identifier r.i;
expression e1,e2,e3;
position p;
@@
(
snd_soc_register_codec(e1,&i@p,e2,e3)
|
memcpy(e1,&i@p,e2)
)
@bad@
position p != {r.p,ok.p};
identifier r.i;
@@
i@p
@depends on !bad disable optional_qualifier@
identifier r.i;
@@
static
+const
struct snd_soc_codec_driver i = { ... };
// </smpl>
Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
codec driver and component driver has duplicated callback functions,
and codec side functions are just copied to component side when
register timing. This was quick-hack, but no longer needed.
This patch moves these functions from codec driver to component driver.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Use the new snd_ac97_reset() helper and the reset functionality provided by
snd_soc_new_ac97_codec() to perform the device reset rather than
open-coding it.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
All drivers have the same line at the end of the set_bias_level callback to
update the bias_level state. Move this update into
snd_soc_dapm_force_bias_level() and remove them from the drivers.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Use the new snd_soc_codec_force_bias_level() helper function to invoke the
bias_level callback of a driver instead of calling the callback by hand.
Currently the effect of this is the same, but having all bias level updates
go through a central place will allow us to move more of the bias level
management into the DAPM core.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The correct values referred by a boolean control are
value.integer.value[], not value.enumerated.item[].
The former is long while the latter is int, so it's even incompatible
on 64bit architectures.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: <stable@vger.kernel.org>
The wm97xx touchscreen driver binds itself to the snd_ac97 device that gets
registered by the CODEC driver and expects that the device has already been
reset. Before commit 6794f709b7 ("ASoC: ac97: Drop delayed device
registration") the device was only registered after the probe function of
the CODEC driver had finished running, but starting with the mentioned
commit the device is registered as soon as snd_soc_new_ac97_codec() is
called. This causes the touchscreen driver to no longer work. Modify the
CODEC drivers to use snd_soc_alloc_ac97_codec() instead of
snd_soc_new_ac97_codec() and make sure that the AC'97 device is reset before
the snd_ac97 device gets registered.
Fixes: 6794f709b7 ("ASoC: ac97: Drop delayed device registration")
Reported-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Manuel Lauss <manuel.lauss@gmail.com>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Here's the set of driver core patches for 3.19-rc1.
They are dominated by the removal of the .owner field in platform
drivers. They touch a lot of files, but they are "simple" changes, just
removing a line in a structure.
Other than that, a few minor driver core and debugfs changes. There are
some ath9k patches coming in through this tree that have been acked by
the wireless maintainers as they relied on the debugfs changes.
Everything has been in linux-next for a while.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iEYEABECAAYFAlSOD20ACgkQMUfUDdst+ylLPACg2QrW1oHhdTMT9WI8jihlHVRM
53kAoLeteByQ3iVwWurwwseRPiWa8+MI
=OVRS
-----END PGP SIGNATURE-----
Merge tag 'driver-core-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core
Pull driver core update from Greg KH:
"Here's the set of driver core patches for 3.19-rc1.
They are dominated by the removal of the .owner field in platform
drivers. They touch a lot of files, but they are "simple" changes,
just removing a line in a structure.
Other than that, a few minor driver core and debugfs changes. There
are some ath9k patches coming in through this tree that have been
acked by the wireless maintainers as they relied on the debugfs
changes.
Everything has been in linux-next for a while"
* tag 'driver-core-3.19-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core: (324 commits)
Revert "ath: ath9k: use debugfs_create_devm_seqfile() helper for seq_file entries"
fs: debugfs: add forward declaration for struct device type
firmware class: Deletion of an unnecessary check before the function call "vunmap"
firmware loader: fix hung task warning dump
devcoredump: provide a one-way disable function
device: Add dev_<level>_once variants
ath: ath9k: use debugfs_create_devm_seqfile() helper for seq_file entries
ath: use seq_file api for ath9k debugfs files
debugfs: add helper function to create device related seq_file
drivers/base: cacheinfo: remove noisy error boot message
Revert "core: platform: add warning if driver has no owner"
drivers: base: support cpu cache information interface to userspace via sysfs
drivers: base: add cpu_device_create to support per-cpu devices
topology: replace custom attribute macros with standard DEVICE_ATTR*
cpumask: factor out show_cpumap into separate helper function
driver core: Fix unbalanced device reference in drivers_probe
driver core: fix race with userland in device_add()
sysfs/kernfs: make read requests on pre-alloc files use the buffer.
sysfs/kernfs: allow attributes to request write buffer be pre-allocated.
fs: sysfs: return EGBIG on write if offset is larger than file size
...
Set the CODEC driver's suspend_bias_off flag rather than manually going to
SND_SOC_BIAS_OFF in suspend. This makes the code a bit shorter and cleaner.
The manual transition to SND_SOC_BIAS_STANDBY at the end of CODEC probe()
can also be removed as the core will automatically do this after the CODEC
has been probed.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Now that the ASoC core no longer needs a handle to the AC'97 device that is
associated with a CODEC we can remove it from the snd_soc_codec struct and
push it into the individual driver state structs like we do for other
communication buses. Doing so creates a clean separation between the AC'97
bus support and the ASoC core.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
This is no longer necessary as there is no code anymore that uses this for
CODEC DAIs.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
We always pass soc_ac97_ops to snd_soc_new_ac97_codec(). So instead of
allocating a snd_ac97_bus in snd_soc_new_ac97_codec() just use a static one
that gets initialized when snd_soc_set_ac97_ops() is called.
Also drop the device number parameter from snd_soc_new_ac97_codec(). We
currently only support one device per bus and all drivers pass 0 for the
device number. And if we should ever support multiple devices per bus it
wouldn't be up to individual AC'97 device drivers to pick their number, but
rather either the AC'97 adapter driver or the core code will assign them.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The wm9712/wm9713 has separate mixers for the left and the right channel,
but the inputs to the mixers are enabled/disabled by the same control.
Currently this is implemented by the driver by registering two virtual
controls for each physical control, one for the left mixer and one for the
right mixer.
Using virtual registers will no longer work when the driver has
been converted to regmap. This patch converts the driver to use controls
with custom put/get callbacks instead which implement the logic making sure
that the physical control is unmuted when either the left or the right
control is unmuted.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Makes the code a bit cleaner.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Use dev_err() instead of printk(KERN_ERR. This is common practice and makes
it easy to find out which device generated the message. While we are at it
also align the error messages with the other AC'97 drivers.
Also avoid printing two error messages when the reset fails.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently we can only have a single platform built in with AC'97 support
due to the use of a global variable to provide the bus operations. Fix
this by making that variable a pointer and having the bus drivers set the
operations prior to registering.
This is not a particularly good or nice approach but it avoids blocking
multiplatform and a real fix involves fixing the fairly deep problems
with AC'97 support - we should be converting it to a real bus.
Acked-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@linaro.org>
CONFIG_HOTPLUG is going away as an option. As result the __dev*
markings will be going away.
Remove use of __devinit, __devexit_p, __devinitdata, __devinitconst,
and __devexit.
Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Since commit 98d3088e5 (SoC: core: Fix check before defaulting to regmap)
, it is not necessary to provide codec->control_data anymore.
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The capture volume increases with the register value so it shouldn't be
flagged as inverted.
Reported-by: Christoph Fritz <chf.fritz@googlemail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Currently the microphone input source is not selectable as while there is
a DAPM widget it's not connected to anything so it won't be properly
instantiated. Add something more correct for the input structure to get
things going, even though it's not hooked into the rest of the routing
map and so won't actually achieve anything except allowing the relevant
register bits to be written.
Reported-by: Christop Fritz <chf.fritz@googlemail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: stable@vger.kernel.org
Since commit 38cbf9598f
("ASoC: core: Try to use regmap if the driver doesn't set up any I/O")
any ASoC codec which doesn't set codec::control_data is assumed
to use regmap. That doesn't work with AC97 so this workaround
sets the codec::control_data member to a random value to restore
proper behaviour.
Tested with WM9712.
Signed-off-by: Manuel Lauss <manuel.lauss@googlemail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Fix the following build error:
sound/soc/codecs/wm9712.c:482:32: error: 'runtime' undeclared (first use in this function)
sound/soc/codecs/wm9712.c:499:33: error: 'runtime' undeclared (first use in this function)
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
In order to support CODEC<->CODEC links remove the assumption that there
is only a single CODEC on a DAI link by removing the use of the CODEC
pointer in the rtd from the CODEC drivers. They are already being passed
their DAI whenever they are passed an rtd and can get the CODEC from
there.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Currently ASoC can only add kcontrols using codec and platform component device
handles. It's also desirable to add kcontrols for DAIs (i.e. McBSP) and for
SoC card machine drivers too. This allows the kcontrol to have a direct handle to
the parent ASoC component DAI/SoC Card/Platform/Codec device and hence easily
get it's private data.
This change makes snd_soc_add_controls() static and wraps it in the folowing
calls (card and dai are new) :-
snd_soc_add_card_controls()
snd_soc_add_codec_controls()
snd_soc_add_dai_controls()
snd_soc_add_platform_controls()
This patch also does a lot of small mechanical changes in individual codec drivers
to replace snd_soc_add_controls() with snd_soc_add_codec_controls().
It also updates the McBSP DAI driver to use snd_soc_add_dai_controls().
Finally, it updates the existing machine drivers that register controls to either :-
1) Use snd_soc_add_card_controls() where no direct codec control is required.
2) Use snd_soc_add_codec_controls() where there is direct codec control.
In the case of 1) above we also update the machine drivers to get the correct
component data pointers from the kcontrol (rather than getting the machine pointer
via the codec pointer).
Signed-off-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The existence of this parameter is purely historical. None of the CODEC drivers
uses it and we always pass in the same value anyway, so it should be safe to
remove it.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Commit 1ee46ebd("ASoC: Make the DAI ops constant in the DAI structure")
introduced the possibility to have constant DAI ops structures, yet this is
barley used in both existing drivers and also new drivers being submitted,
although none of them modifies its DAI ops structure. The later is not
surprising since existing drivers are often used as templates for new drivers.
So this patch just constifies all existing snd_soc_dai_ops structs to eliminate
the issue altogether.
The patch was generated with the following coccinelle semantic patch:
// <smpl>
@@
identifier ops;
@@
-struct snd_soc_dai_ops ops =
+const struct snd_soc_dai_ops ops =
{ ... };
// </smpl>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
There is no need to include soc-dapm.h since soc.h includes it.
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Decoupling Dynamic Audio Power Management (DAPM) from codec devices is
required when developing ASoC further. Such as for other ASoC components to
have DAPM widgets or when extending DAPM to handle cross-device paths.
This patch decouples DAPM related variables from struct snd_soc_codec and
moves them to new struct snd_soc_dapm_context that is used to encapsulate
DAPM context of a device. ASoC core and API of DAPM functions are modified
to use DAPM context instead of codec.
This patch does not change current functionality and a large part of changes
come because of structure and internal API changes.
Core implementation is from Liam Girdwood <lrg@slimlogic.co.uk> with some
minor core changes, codecs and machine driver conversions from
Jarkko Nikula <jhnikula@gmail.com>.
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Manuel Lauss <manuel.lauss@googlemail.com>
Cc: Mike Frysinger <vapier.adi@gmail.com>
Cc: Cliff Cai <cliff.cai@analog.com>
Cc: Kevin Hilman <khilman@deeprootsystems.com>
Cc: Ryan Mallon <ryan@bluewatersys.com>
Cc: Timur Tabi <timur@freescale.com>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Lars-Peter Clausen <lars@metafoo.de>
Cc: Arnaud Patard (Rtp) <arnaud.patard@rtp-net.org>
Cc: Wan ZongShun <mcuos.com@gmail.com>
Cc: Eric Miao <eric.y.miao@gmail.com>
Cc: Jassi Brar <jassi.brar@samsung.com>
Cc: Daniel Gloeckner <dg@emlix.com>
Cc: Kuninori Morimoto <morimoto.kuninori@renesas.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The reg_cache_size is the number of elements in the register cache,
not the size of the cache itself. This is not a problem if the size
of each element of the cache is 1 byte but it matters in any other
case.
Signed-off-by: Dimitris Papastamos <dp@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
This patch extends the ASoC API to allow sound cards to have more than one
CODEC and more than one platform DMA controller. This is achieved by dividing
some current ASoC structures that contain both driver data and device data into
structures that only either contain device data or driver data. i.e.
struct snd_soc_codec ---> struct snd_soc_codec (device data)
+-> struct snd_soc_codec_driver (driver data)
struct snd_soc_platform ---> struct snd_soc_platform (device data)
+-> struct snd_soc_platform_driver (driver data)
struct snd_soc_dai ---> struct snd_soc_dai (device data)
+-> struct snd_soc_dai_driver (driver data)
struct snd_soc_device ---> deleted
This now allows ASoC to be more tightly aligned with the Linux driver model and
also means that every ASoC codec, platform and (platform) DAI is a kernel
device. ASoC component private data is now stored as device private data.
The ASoC sound card struct snd_soc_card has also been updated to store lists
of it's components rather than a pointer to a codec and platform. The PCM
runtime struct soc_pcm_runtime now has pointers to all its components.
This patch adds DAPM support for ASoC multi-component and removes struct
snd_soc_socdev from DAPM core. All DAPM calls are now made on a card, codec
or runtime PCM level basis rather than using snd_soc_socdev.
Other notable multi-component changes:-
* Stream operations now de-reference less structures.
* close_delayed work() now runs on a DAI basis rather than looping all DAIs
in a card.
* PM suspend()/resume() operations can now handle N CODECs and Platforms
per sound card.
* Added soc_bind_dai_link() to bind the component devices to the sound card.
* Added soc_dai_link_probe() and soc_dai_link_remove() to probe and remove
DAI link components.
* sysfs entries can now be registered per component per card.
* snd_soc_new_pcms() functionailty rolled into dai_link_probe().
* snd_soc_register_codec() now does all the codec list and mutex init.
This patch changes the probe() and remove() of the CODEC drivers as follows:-
o Make CODEC driver a platform driver
o Moved all struct snd_soc_codec list, mutex, etc initialiasation to core.
o Removed all static codec pointers (drivers now support > 1 codec dev)
o snd_soc_register_pcms() now done by core.
o snd_soc_register_dai() folded into snd_soc_register_codec().
CS4270 portions:
Acked-by: Timur Tabi <timur@freescale.com>
Some TLV320aic23 and Cirrus platform fixes.
Signed-off-by: Ryan Mallon <ryan@bluewatersys.com>
TI CODEC and OMAP fixes
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@nokia.com>
Signed-off-by: Janusz Krzysztofik <jkrzyszt@tis.icnet.pl>
Signed-off-by: Jarkko Nikula <jhnikula@gmail.com>
Samsung platform and misc fixes :-
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Joonyoung Shim <jy0922.shim@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Reviewed-by: Jassi Brar <jassi.brar@samsung.com>
Signed-off-by: Seungwhan Youn <sw.youn@samsung.com>
MPC8610 and PPC fixes.
Signed-off-by: Timur Tabi <timur@freescale.com>
i.MX fixes and some core fixes.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
J4740 platform fixes:-
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
CC: Tony Lindgren <tony@atomide.com>
CC: Nicolas Ferre <nicolas.ferre@atmel.com>
CC: Kevin Hilman <khilman@deeprootsystems.com>
CC: Sascha Hauer <s.hauer@pengutronix.de>
CC: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
CC: Kuninori Morimoto <morimoto.kuninori@renesas.com>
CC: Daniel Gloeckner <dg@emlix.com>
CC: Manuel Lauss <mano@roarinelk.homelinux.net>
CC: Mike Frysinger <vapier.adi@gmail.com>
CC: Arnaud Patard <apatard@mandriva.com>
CC: Wan ZongShun <mcuos.com@gmail.com>
Acked-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Liam Girdwood <lrg@slimlogic.co.uk>
The core will ensure that the device is in either STANDBY or OFF bias
before suspending, restoring the bias in the driver is unneeded. Some
drivers doing slightly more roundabout things have been left alone
for now.
Tested-by: Peter Ujfalusi <peter.ujfalusi@nokia.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
