Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

This restores the default clocks registration order as parsed from
devicetree, i.e. as before commit 1771b10d
"clk: respect the clock dependencies in of_clk_init", for when there
is no explicit parent clock dependencies between clock providers
specified in the device tree.

It prevents regressions (boot failure, division by 0 errors) on
imx and exynos platforms.
Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Tested-by: NFabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Add a new clk_ops->debug_init method to allow a clock hardware
driver to populate the clock's debugfs directory with entries
beyond those common for every clock.
Signed-off-by: NAlex Elder <elder@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

If a rate change failed it's the opportunity of the caller to handle
this. Do not spam the log with a message.
Signed-off-by: NSascha Hauer <s.hauer@pengutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Until now the clock providers were initialized in the order found in
the device tree. This led to have the dependencies between the clocks
not respected: children clocks could be initialized before their
parent clocks.

Instead of forcing each platform to manage its own initialization order,
this patch adds this work inside the framework itself.

Using the data of the device tree the of_clk_init function now delayed
the initialization of a clock provider if its parent provider was not
ready yet.

The strict dependency check (all parents of a given clk must be
initialized) was added by Boris BREZILLON
Signed-off-by: NGregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: NBoris BREZILLON <b.brezillon@overkiz.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Ensure clk->kref is dereferenced only when clk is not NULL.
Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Tested-by: NSachin Kamat <sachin.kamat@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

At probe time, a clock device may not be ready when some other device
wants to use it.

This patch lets the functions clk_get/devm_clk_get return a probe defer
when the clock is defined in the DT but not yet available.
Signed-off-by: NJean-Francois Moine <moinejf@free.fr>
Reviewed-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Tested-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Both the pr_err and the additional kerneldoc aim to help when debugging
errors thrown from within a clock rate-change notifier callback.
Reported-by: NSören Brinkmann <soren.brinkmann@xilinx.com>
Acked-by: NSören Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Add a property called clock-indices to allow clock-output-names
to be used where the index used to lookup a clock is not a 1:1
mapping to the array position in the clock-output-names
Signed-off-by: NBen Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Contradicting to documenation, the notifier callbacks do receive
the original clock rate in struct clk_notifier_data.old_rate and the new
frequency struct clk_notifier_data.new_rate, independent of the
notification reason.

This behavior also seems to make more sense, since callbacks can use the
same code to deterimine whether clocks are scaled up or down. Something
which would not even possible in the post-rate-change case if the
behavior was as documented.
Signed-off-by: NSoren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Allow drivers to be compiled as modules by exporting more clock
provider functions.
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NStephen Boyd <sboyd@codeaurora.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

ti_dt_clk_init_provider() can now be used to initialize the contents of
a single clock IP block. This parses all the clocks under the IP block
and calls the corresponding init function for them.

This patch also introduces a helper function for the TI clock drivers
to get register info from DT and append the master IP info to this.
Signed-off-by: NTero Kristo <t-kristo@ti.com>
Acked-by: NTony Lindgren <tony@atomide.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Some of Qualcomm's clocks can change their parent and rate at the
same time with a single register write. Add support for this
hardware to the common clock framework by adding a new
set_rate_and_parent() op. When the clock framework determines
that both the parent and the rate are going to change during
clk_set_rate() it will call the .set_rate_and_parent() op if
available and fall back to calling .set_parent() followed by
.set_rate() otherwise.
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Signed-off-by: NStephen Boyd <sboyd@codeaurora.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Populate ${DEBUGS_MOUNT_POINT}/clk if CONFIG_DEBUG_FS is set. This
eliminates the extra (annoying) step of enabling the config option
manually.
Signed-off-by: NMike Turquette <mturquette@linaro.org>

The clock accuracy is expressed in ppb (parts per billion) and represents
the possible clock drift.
Say you have a clock (e.g. an oscillator) which provides a fixed clock of
20MHz with an accuracy of +- 20Hz. This accuracy expressed in ppb is
20Hz/20MHz = 1000 ppb (or 1 ppm).

Clock users may need the clock accuracy information in order to choose
the best clock (the one with the best accuracy) across several available
clocks.

This patch adds clk accuracy retrieval support for common clk framework by
means of a new function called clk_get_accuracy.
This function returns the given clock accuracy expressed in ppb.

In order to get the clock accuracy, this implementation adds one callback
called recalc_accuracy to the clk_ops structure.
This callback is given the parent clock accuracy (if the clock is not a
root clock) and should recalculate the given clock accuracy.

This callback is optional and may be implemented if the clock is not
a perfect clock (accuracy != 0 ppb).
Signed-off-by: NBoris BREZILLON <b.brezillon@overkiz.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Adding clocks from a kernel module can cause a NULL pointer
dereference if the parent of a clock is added after the clock is
added. This happens because __clk_init() iterates over the list
of orphans and reparents the orphans to the clock being
registered before creating the debugfs entry for the clock.
Create the debugfs entry first before reparenting the orphans.

Unable to handle kernel NULL pointer dereference at virtual address 00000028
pgd = ef3e4000
[00000028] *pgd=bf810831
Internal error: Oops: 17 [#1] PREEMPT SMP ARM
Modules linked in: mmcc_8960(+)
CPU: 0 PID: 52 Comm: modprobe Not tainted 3.12.0-rc2-00023-g1021a28-dirty #659
task: ef319200 ti: ef3a6000 task.ti: ef3a6000
PC is at lock_rename+0x24/0xc4
LR is at debugfs_rename+0x34/0x208
pc : [<c0317238>]    lr : [<c047dfe4>]    psr: 00000013
sp : ef3a7b88  ip : ef3a7ba8  fp : ef3a7ba4
r10: ef3d51cc  r9 : ef3bc680  r8 : ef3d5210
r7 : ef3bc640  r6 : eee287e0  r5 : eee287e0  r4 : 00000000
r3 : ef3bc640  r2 : 00000000  r1 : eee287e0  r0 : 00000000
Flags: nzcv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
Control: 10c5787d  Table: af3e406a  DAC: 00000015
Process modprobe (pid: 52, stack limit = 0xef3a6240)
Stack: (0xef3a7b88 to 0xef3a8000)
7b80:                   ef3bc640 ee4047e0 00000000 eee287e0 ef3a7bec ef3a7ba8
7ba0: c047dfe4 c0317220 ef3bc680 ef3d51cc ef3a7bdc ef3a7bc0 c06e29d0 c0268784
7bc0: c08946e8 ef3d5210 00000000 ef3bc700 ef3d5290 ef3d5210 ef3bc680 ef3d51cc
7be0: ef3a7c0c ef3a7bf0 c05b9e9c c047dfbc 00000000 00000000 ef3d5210 ef3d5290
7c00: ef3a7c24 ef3a7c10 c05baebc c05b9e30 00000001 00000001 ef3a7c64 ef3a7c28
7c20: c05bb124 c05bae9c bf000cd8 ef3bc7c0 000000d0 c0ff129c bf001774 00000002
7c40: ef3bc740 ef3d5290 ef0f9a10 bf001774 bf00042c 00000061 ef3a7c8c ef3a7c68
7c60: c05bb480 c05baed8 bf001774 ef3d5290 ef0f9a10 bf001774 ef38bc10 ef0f9a00
7c80: ef3a7cac ef3a7c90 c05bb5a8 c05bb3a0 bf001774 00000062 ef0f9a10 ef38bc18
7ca0: ef3a7cec ef3a7cb0 bf00010c c05bb56c 00000000 ef38ba00 00000000 ef3d60d0
7cc0: ef3a7cdc c0fefc24 ef0f9a10 c0a091c0 bf000d24 00000000 bf0029f0 bf006000
7ce0: ef3a7cfc ef3a7cf0 c05156c0 bf000040 ef3a7d2c ef3a7d00 c0513f5c c05156a8
7d00: ef3a7d2c ef0f9a10 ef0f9a10 bf000d24 ef0f9a44 c09ca588 00000000 bf006000
7d20: ef3a7d4c ef3a7d30 c05142b8 c0513ecc ef0fd25c 00000000 bf000d24 c0514214
7d40: ef3a7d74 ef3a7d50 c0512030 c0514220 ef0050a8 ef0fd250 ef0050f8 bf000d24
7d60: ef37c100 c09ed150 ef3a7d84 ef3a7d78 c05139c8 c0511fd8 ef3a7
7d80: c051344c c05139a8 bf000864 c09ca588 ef3a7db4 bf000d24 bf002
7da0: c09ca588 00000000 ef3a7dcc ef3a7db8 c05149dc c0513360 ef3a7
7dc0: ef3a7ddc ef3a7dd0 c0515914 c0514960 ef3a7dec ef3a7de0 bf006
7de0: ef3a7e74 ef3a7df0 c0208800 bf00600c ef3a7e1c ef3a7e00 c04c5
7e00: ffffffff c09d46c4 00000000 bf0029a8 ef3a7e34 ef3a7e20 c024c
7e20: ffffffff c09d46c4 ef3a7e5c ef3a7e38 c024e2fc c024ce40 00000
7e40: ef3a7f48 bf0029b4 bf0029a8 271aeb1c ef3a7f48 bf0029a8 00000001 ef383c00
7e60: bf0029f0 00000001 ef3a7f3c ef3a7e78 c028fac4 c0208718 bf0029b4 00007fff
7e80: c028cd58 000000d2 f0065000 00000000 ef3a7ebc 00000000 00000000 bf0029b4
7ea0: 00000000 bf0029ac bf0029b4 ef3a6000 ef3a7efc c08bf128 00000000 00000000
7ec0: 00000000 00000000 00000000 00000000 6e72656b 00006c65 00000000 00000000
7ee0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
7f00: 00000000 00000000 00000000 271aeb1c ef3a7f2c 00016376 b6f38008 001d3774
7f20: 00000080 c020f968 ef3a6000 00000000 ef3a7fa4 ef3a7f40 c02904dc c028e178
7f40: c020f898 010ccfa8 f0065000 00016376 f0073f60 f0073d7d f007a1e8 00002b24
7f60: 000039e4 00000000 00000000 00000000 0000002f 00000030 00000019 00000016
7f80: 00000012 00000000 00000000 010de1b2 b6f38008 010ccfa8 00000000 ef3a7fa8
7fa0: c020f6c0 c0290434 010de1b2 b6f38008 b6f38008 00016376 001d3774 00000000
7fc0: 010de1b2 b6f38008 010ccfa8 00000080 010de1b2 bedb6f90 010de1c9 0001d8dc
7fe0: 0000000c bedb674c 0001ce30 000094c4 60000010 b6f38008 00000008 0000001d
[<c0317238>] (lock_rename+0x24/0xc4) from [<c047dfe4>] (debugfs_rename+0x34/0x208)
[<c047dfe4>] (debugfs_rename+0x34/0x208) from [<c05b9e9c>] (clk_debug_reparent+0x78/0xc0)
[<c05baebc>] (__clk_reparent+0x2c/0x3c) from [<c05bb124>] (__clk_init+0x258/0x4c8)
[<c05bb124>] (__clk_init+0x258/0x4c8) from [<c05bb480>] (_clk_register+0xec/0x1cc)
[<c05bb480>] (_clk_register+0xec/0x1cc) from [<c05bb5a8>] (devm_clk_register+0x48/0x7c)
[<c05bb5a8>] (devm_clk_register+0x48/0x7c) from [<bf00010c>] (msm_mmcc_8960_probe+0xd8/0x190 [mmcc_8960])
[<bf00010c>] (msm_mmcc_8960_probe+0xd8/0x190 [mmcc_8960]) from [<c05156c0>] (platform_drv_probe+0x24/0x28)
[<c05156c0>] (platform_drv_probe+0x24/0x28) from [<c0513f5c>] (driver_probe_device+0x9c/0x354)
[<c0513f5c>] (driver_probe_device+0x9c/0x354) from [<c05142b8>] (__driver_attach+0xa4/0xa8)
[<c05142b8>] (__driver_attach+0xa4/0xa8) from [<c0512030>] (bus_for_each_dev+0x64/0x98)
[<c0512030>] (bus_for_each_dev+0x64/0x98) from [<c05139c8>] (driver_attach+0x2c/0x30)
[<c05139c8>] (driver_attach+0x2c/0x30) from [<c051344c>] (bus_add_driver+0xf8/0x2a8)
[<c051344c>] (bus_add_driver+0xf8/0x2a8) from [<c05149dc>] (driver_register+0x88/0x104)
[<c05149dc>] (driver_register+0x88/0x104) from [<c0515914>] (__platform_driver_register+0x58/0x6c)
[<c0515914>] (__platform_driver_register+0x58/0x6c) from [<bf006018>] (msm_mmcc_8960_driver_init+0x18/0x24 [mmcc_8960])
[<bf006018>] (msm_mmcc_8960_driver_init+0x18/0x24 [mmcc_8960]) from [<c0208800>] (do_one_initcall+0xf4/0x1b8)
[<c0208800>] (do_one_initcall+0xf4/0x1b8) from [<c028fac4>] (load_module+0x1958/0x22bc)
[<c028fac4>] (load_module+0x1958/0x22bc) from [<c02904dc>] (SyS_init_module+0xb4/0x120)
[<c02904dc>] (SyS_init_module+0xb4/0x120) from [<c020f6c0>] (ret_fast_syscall+0x0/0x48)
Code: e1500001 e1a04000 e1a05001 0a000021 (e5903028)
Signed-off-by: NStephen Boyd <sboyd@codeaurora.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

clk_unregister() is currently not implemented and it is required when
a clock provider module needs to be unloaded.

Normally the clock supplier module is prevented to be unloaded by
taking reference on the module in clk_get().

For cases when the clock supplier module deinitializes despite the
consumers of its clocks holding a reference on the module, e.g. when
the driver is unbound through "unbind" sysfs attribute, there are
empty clock ops added. These ops are assigned temporarily to struct
clk and used until all consumers release the clock, to avoid invoking
callbacks from the module which just got removed.
Signed-off-by: NJiada Wang <jiada_wang@mentor.com>
Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>

This patch adds common __clk_get(), __clk_put() clkdev helpers that
replace their platform specific counterparts when the common clock
API is used.

The owner module pointer field is added to struct clk so a reference
to the clock supplier module can be taken by the clock consumers.

The owner module is assigned while the clock is being registered,
in functions _clk_register() and __clk_register().
Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>
Acked-by: NRussell King <rmk+kernel@arm.linux.org.uk>

Add helper functions for the of_clk_providers list locking and
an unlocked variant of of_clk_get_from_provider().
These functions are intended to be used in the clkdev to avoid
race condition in the device tree based clock look up in clk_get().
Signed-off-by: NSylwester Nawrocki <s.nawrocki@samsung.com>
Signed-off-by: NKyungmin Park <kyungmin.park@samsung.com>
Acked-by: NRussell King <rmk+kernel@arm.linux.org.uk>

[Maybe the third time will be the charm. -Alex]

If CONFIG_COMMON_CLK_DEBUG is defined, clk_debug_create_one() is
called to populate a debugfs directory with a few entries that are
common for all clock types.

If an error happens after creating the first one debugfs_remove() is
called on the clock's directory.  The problem with this is that no
cleanup is done on the debugfs files already created in that
directory, so the directory never actually gets removed.   This
problem is silently ignored.

Fix this by calling debugfs_remove_recursive() instead.  Reset the
clk->dentry field to null afterward, to ensure it can't be mistaken
as a valid pointer.
Signed-off-by: NAlex Elder <elder@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Walks the "clocks" array of parent clock phandles and returns the
number.
Signed-off-by: NMike Turquette <mturquette@linaro.org>

This function is supposed to iterate over all parents of given child
clock to find the index of given parent clock in its parent list,
using parent cache if possible and falling back to string compare
otherwise. However currently the logic falls back to string compare in
every iteration in which clock cache entry does not match given parent,
due to wrong check conditions.

This patch corrects the logic to continue the loop if parent cache entry
is present and does not match requested parent clock. In addition,
redundant checks for parent cache array presence are removed, because it
is always allocated in the beginning of the function.
Signed-off-by: NTomasz Figa <tomasz.figa@gmail.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Instead of calculating sizes of arrays manually, kcalloc() can be used
to allocate arrays of elements with defined size. This is just a cleanup
patch without any functional changes.
Signed-off-by: NTomasz Figa <tomasz.figa@gmail.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

There are at least two different error cases that can happen in
clk_fetch_parent_index() function:
 - allocation failure,
 - parent clock lookup failure,
however it returns only an u8, which is supposed to contain parent clock
index.

This patch modified the function to return full int instead allowing
positive clock indices and negative error codes to be returned. All
users of this function are adjusted as well to handle the return value
correctly.
Signed-off-by: NTomasz Figa <tomasz.figa@gmail.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

In __clk_init(), after a clock is mostly initialized, a scan is done
of the orphan clocks to see if the clock being registered is the
parent of any of them.

This code assumes that any clock that provides a get_parent method
actually has at least one parent, and that's not a valid assumption.

As a result, an orphan clock with no parent can return *something*
as the parent index, and that value is blindly used to dereference
the orphan's parent_names[] array (which will be ZERO_SIZE_PTR or
NULL).

Fix this by ensuring get_parent is only called for orphans with at
least one parent.
Signed-off-by: NAlex Elder <elder@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

The __clk_get_flags() symbol is exported immediately following the
clk_unprepare_unused_subtree() function. This is unusual, since a symbol
export typically follows body of the function that it exports.
Signed-off-by: NThierry Reding <treding@nvidia.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Commit 71472c0c (clk: add support for clock reparent on set_rate) added a
dereference of the new_parent pointer in clk_reparent(), but as detected
by smatch clk_reparent() later checks whether new_parent is NULL.

The dereference was in order to clear the new parent's new_child pointer
to avoid duplicate POST_RATE_CHANGE notifications, so clearly isn't
necessary if the new parent is NULL, so move it inside the "if
(new_parent)" block.
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Currently of_clk_init() finds a matching device node while holding
the device tree spinlock.  When a matching device node is found, the
lock is dropped and then re-acquired in order to get a reference
to the matching device id structure.

Acquiring the spinlock twice is unnecessary (and it opens a
vulnerable window that could conceivably lead to errors).

There already exists an interface for both finding and taking a
reference to a device id under lock, so use it.
Signed-off-by: NAlex Elder <elder@linaro.org>
Reviewed-by: NChristian Daudt <csd@broadcom.com>
Reviewed-by: NMarkus Mayer <markus.mayer@linaro.org>
Reviewed-by: NMatt Porter <matt.porter@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

At some point changes to clk_set_rate and clk_set_parent introduced a
bug whereby NULL struct clk pointers were treated as an error. This is
in violation of the API in include/linux/clk.h. Reintroduce graceful
handling of NULL clk's by bailing from clk_set_rate and clk_set_parent
with return codes of zero.
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Implement clk-mux remuxing if the CLK_SET_RATE_NO_REPARENT flag isn't
set. This implements determine_rate for clk-mux to propagate to each
parent and to choose the best one (like clk-divider this chooses the
parent which provides the fastest rate <= the requested rate).

The determine_rate op is implemented as a core helper function so that
it can be easily used by more complex clocks which incorporate muxes.
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>
Cc: Mike Turquette <mturquette@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Add core support to allow clock implementations to select the best
parent clock when rounding a rate, e.g. the one which can provide the
closest clock rate to that requested. This is by way of adding a new
clock op, determine_rate(), which is like round_rate() but has an extra
parameter to allow the clock implementation to optionally select a
different parent clock. The core then takes care of reparenting the
clock when setting the rate.

The parent change takes place with the help of some new private data
members. struct clk::new_parent specifies a clock's new parent (NULL
indicates no change), and struct clk::new_child specifies a clock's new
child (whose new_parent member points back to it). The purpose of these
are to allow correct walking of the future tree for notifications prior
to actually reparenting any clocks, specifically to skip child clocks
who are being reparented to another clock (they will be notified via the
new parent), and to include any new child clock. These pointers are set
by clk_calc_subtree(), and the new_child pointer gets cleared when a
child is actually reparented to avoid duplicate POST_RATE_CHANGE
notifications.

Each place where round_rate() is called, determine_rate() is checked
first and called in preference. This restructures a few of the call
sites to simplify the logic into if/else blocks.
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>
Cc: Mike Turquette <mturquette@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Move some parent related functions up in clk.c so they can be used by
the modifications in the following patch which enables clock reparenting
during set_rate. No other changes are made so this patch makes no
functional difference in isolation. This is separate from the following
patch primarily to ease readability of that patch.
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>
Cc: Mike Turquette <mturquette@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Abstract access to the clock parent cache by defining
clk_get_parent_by_index(clk, index). This allows access to parent
clocks from clock drivers.
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NStephen Boyd <sboyd@codeaurora.org>
Cc: Mike Turquette <mturquette@linaro.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Signed-off-by: NPeter Meerwald <pmeerw@pmeerw.net>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

A clock's notifier count only reflects notifiers which are registered
directly for that clock. A reparent operation though affects the whole
subtree because of a potential rate change.
When issuing the pre rate change notifications only the notifier count
for the clock to be changed is considered and notifiers for subclocks
may never be called. Resulting in clocks in the subtree which have
registered notifiers, may receive a POST_- or ABORT_RATE_CHANGE
notification, without a PRE_RATE_CHANGE_NOTIFICATION.
Therefore always traverse the whole subtree when issueing pre rate
change notifications during a reparent operation.
Signed-off-by: NSoren Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

clk_set_rate() uses clk->rate directly. This causes problems if the clock
is marked as CLK_GET_RATE_NOCACHE. Hence call clk_get_rate() to get the
current rate.
Signed-off-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

debugfs uses the rate field directly. However this ignores the
CLK_GET_RATE_NOCACHE flag. Call clk_get_rate() instead.
Tested-by: NMark Zhang <markz@nvidia.com>
Signed-off-by: NPeter De Schrijver <pdeschrijver@nvidia.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

The @cn is stay in @clk_notifier_list after it is freed, it cause
memory corruption.

Example, if @clk is registered(first), unregistered(first),
registered(second), unregistered(second).

The freed @cn will be used when @clk is registered(second),
and the bug will be happened when @clk is unregistered(second):

[  517.040000] clk_notif_dbg clk_notif_dbg.1: clk_notifier_unregister()
[  517.040000] Unable to handle kernel paging request at virtual address 00df3008
[  517.050000] pgd = ed858000
[  517.050000] [00df3008] *pgd=00000000
[  517.060000] Internal error: Oops: 5 [#1] PREEMPT SMP ARM
[  517.060000] Modules linked in: clk_notif_dbg(O-) [last unloaded: clk_notif_dbg]
[  517.060000] CPU: 1 PID: 499 Comm: modprobe Tainted: G           O 3.10.0-rc3-00119-ga93cb29a-dirty #85
[  517.060000] task: ee1e0180 ti: ee3e6000 task.ti: ee3e6000
[  517.060000] PC is at srcu_readers_seq_idx+0x48/0x84
[  517.060000] LR is at srcu_readers_seq_idx+0x60/0x84
[  517.060000] pc : [<c0052720>]    lr : [<c0052738>]    psr: 80070013
[  517.060000] sp : ee3e7d48  ip : 00000000  fp : ee3e7d6c
[  517.060000] r10: 00000000  r9 : ee3e6000  r8 : 00000000
[  517.060000] r7 : ed84fe4c  r6 : c068ec90  r5 : c068e430  r4 : 00000000
[  517.060000] r3 : 00df3000  r2 : 00000000  r1 : 00000002  r0 : 00000000
[  517.060000] Flags: Nzcv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
[  517.060000] Control: 18c5387d  Table: 2d85804a  DAC: 00000015
[  517.060000] Process modprobe (pid: 499, stack limit = 0xee3e6238)
[  517.060000] Stack: (0xee3e7d48 to 0xee3e8000)
....
[  517.060000] [<c0052720>] (srcu_readers_seq_idx+0x48/0x84) from [<c0052790>] (try_check_zero+0x34/0xfc)
[  517.060000] [<c0052790>] (try_check_zero+0x34/0xfc) from [<c00528b0>] (srcu_advance_batches+0x58/0x114)
[  517.060000] [<c00528b0>] (srcu_advance_batches+0x58/0x114) from [<c0052c30>] (__synchronize_srcu+0x114/0x1ac)
[  517.060000] [<c0052c30>] (__synchronize_srcu+0x114/0x1ac) from [<c0052d14>] (synchronize_srcu+0x2c/0x34)
[  517.060000] [<c0052d14>] (synchronize_srcu+0x2c/0x34) from [<c0053a08>] (srcu_notifier_chain_unregister+0x68/0x74)
[  517.060000] [<c0053a08>] (srcu_notifier_chain_unregister+0x68/0x74) from [<c0375a78>] (clk_notifier_unregister+0x7c/0xc0)
[  517.060000] [<c0375a78>] (clk_notifier_unregister+0x7c/0xc0) from [<bf008034>] (clk_notif_dbg_remove+0x34/0x9c [clk_notif_dbg])
[  517.060000] [<bf008034>] (clk_notif_dbg_remove+0x34/0x9c [clk_notif_dbg]) from [<c02bb974>] (platform_drv_remove+0x24/0x28)
[  517.060000] [<c02bb974>] (platform_drv_remove+0x24/0x28) from [<c02b9bf8>] (__device_release_driver+0x8c/0xd4)
[  517.060000] [<c02b9bf8>] (__device_release_driver+0x8c/0xd4) from [<c02ba680>] (driver_detach+0x9c/0xc4)
[  517.060000] [<c02ba680>] (driver_detach+0x9c/0xc4) from [<c02b99c4>] (bus_remove_driver+0xcc/0xfc)
[  517.060000] [<c02b99c4>] (bus_remove_driver+0xcc/0xfc) from [<c02bace4>] (driver_unregister+0x54/0x78)
[  517.060000] [<c02bace4>] (driver_unregister+0x54/0x78) from [<c02bbb44>] (platform_driver_unregister+0x1c/0x20)
[  517.060000] [<c02bbb44>] (platform_driver_unregister+0x1c/0x20) from [<bf0081f8>] (clk_notif_dbg_driver_exit+0x14/0x1c [clk_notif_dbg])
[  517.060000] [<bf0081f8>] (clk_notif_dbg_driver_exit+0x14/0x1c [clk_notif_dbg]) from [<c00835e4>] (SyS_delete_module+0x200/0x28c)
[  517.060000] [<c00835e4>] (SyS_delete_module+0x200/0x28c) from [<c000edc0>] (ret_fast_syscall+0x0/0x48)
[  517.060000] Code: e5973004 e7911102 e0833001 e2881002 (e7933101)

Cc: stable@kernel.org
Reported-by: NSören Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NLai Jiangshan <laijs@cn.fujitsu.com>
Tested-by: NSören Brinkmann <soren.brinkmann@xilinx.com>
Signed-off-by: NMike Turquette <mturquette@linaro.org>
[mturquette@linaro.org: shortened $SUBJECT]

With deferred probing, late_initcall() is too soon to declare a clock as
unused. Wait for deferred probing to finish before declaring a clock as
unused. Since deferred probing is done in late_initcall(), do the unused
check to late_initcall_sync.
Signed-off-by: NSaravana Kannan <skannan@codeaurora.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>

Without this patch, the following race condition is possible.
* clk-A has two parents - clk-X and clk-Y.
* All three are disabled and clk-X is current parent.
* Thread A: clk_set_parent(clk-A, clk-Y).
* Thread A: <snip execution flow>
* Thread A: Grabs enable lock.
* Thread A: Sees enable count of clk-A is 0, so doesn't enable clk-Y.
* Thread A: Updates clk-A SW parent to clk-Y
* Thread A: Releases enable lock.
* Thread B: clk_enable(clk-A).
* Thread B: clk_enable() enables clk-Y, then enabled clk-A and returns.

clk-A is now enabled in software, but not clocking in hardware since the
hardware parent is still clk-X.

The only way to avoid race conditions between clk_set_parent() and
clk_enable/disable() is to ensure that clk_enable/disable() calls don't
require changes to hardware enable state between changes to software clock
topology and hardware clock topology.

The options to achieve the above are:
1. Grab the enable lock before changing software/hardware topology and
   release it afterwards.
2. Keep the clock enabled for the duration of software/hardware topology
   change so that any additional enable/disable calls don't try to change
   the hardware state. Once the topology change is complete, the clock can
   be put back in its original enable state.

Option (1) is not an acceptable solution since the set_parent() ops might
need to sleep.

Therefore, this patch implements option (2).

This patch doesn't violate any API semantics. clk_disable() doesn't
guarantee that the clock is actually disabled. So, no clients of a clock
can assume that a clock is disabled after their last call to clk_disable().
So, enabling the clock during a parent change is not a violation of any API
semantics.

This also has the nice side effect of simplifying the error handling code.
Signed-off-by: NSaravana Kannan <skannan@codeaurora.org>
Acked-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NMike Turquette <mturquette@linaro.org>
[mturquette@linaro.org: fixed up whitespace issue]