linux-arm-kernel.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: ccross@google.com (Colin Cross)
To: linux-arm-kernel@lists.infradead.org
Subject: Locking in the clk API
Date: Fri, 21 Jan 2011 14:28:15 -0800	[thread overview]
Message-ID: <AANLkTimT+vfg9p4GSnn1jGW9r_gVom7xEUzMVJDscPg2@mail.gmail.com> (raw)
In-Reply-To: <20110121220238.GE23151@n2100.arm.linux.org.uk>

On Fri, Jan 21, 2011 at 2:02 PM, Russell King - ARM Linux
<linux@arm.linux.org.uk> wrote:
> On Fri, Jan 21, 2011 at 04:53:44PM -0500, Nicolas Pitre wrote:
>> So I think that the API must be augmented with more methods, such as:
>>
>> clk_slow_enable():
>> ? - may sleep
>> ? - may be a no-op if the clk_fast_enable() is supported
>>
>> clk_fast_enable():
>> ? - may not sleep, used in atomic context
>> ? - may be a no-op if controlling the clock takes time, in which case
>> ? ? clk_slow_enable() must have set the clock up entirely
>>
>> ... and similar for clk_slow_disable() and clk_fast_disable().
>
> Isn't this along the same lines as my clk_prepare() vs clk_enable()
> suggestion?
>
> I suggested that clk_prepare() be callable only from non-atomic contexts,
> and do whatever's required to ensure that the clock is available. ?That
> may end up enabling the clock as a result.
>
> clk_enable() callable from atomic contexts, and turns the clock on if
> the hardware supports such an operation.
>
> So, if you have something like:
>
> Xtal--->PLL--->Routing/Masking--->Device
>
> clk = clk_get() returns the clock for the device.
>
> clk_prepare(clk) would walk up the clock tree, selecting the routing and
> preparing each clock. ?Clocks prior to _and_ including the PLL would need
> to be enabled.
>
> clk_enable(clk) would walk up the tree if the clock isn't already enabled,
> calling clk_enable() on the parent clock. ?As we require prepared clocks
> to already be enabled, this automatically stops at the PLL.
>
> To encourage correct usage, we just need to make sure that clk_prepare()
> has a might_sleep() thing, and clk_enable() throws a fit if it's used
> on a clk without prepare being used first. ?The second point is not easy
> to do in a foolproof manner though, but doing _something_ is better than
> nothing.

I like this proposal, and I prefer the clk_prepare naming over
clk_slow_enable - too many people would call clk_slow_enable instead
of, and not as well as, clk_fast_enable.

On Tegra, I currently use the ugly conditional mutex or spinlock
method to deal with voltage scaling based on clock frequency.  Clocks
that have a voltage dependency, or depend on a clock that has a
voltage dependency, are non-atomic, everything else is atomic.  PLLs
are atomic because they lock very fast (300 uS or 1ms) and are shared
by so many clocks that they realistically don't get turned off very
often, and if I made them non-atomic, all clocks would be non-atomic.

If clk_prepare, clk_unprepare, and clk_set_rate were only callable in
sleepable contexts, I could make PLLs and voltage-dependent clocks
sleepable, but allow atomic clocks to depend on them.  Calling
clk_prepare on a voltage-dependent clock would bump the voltage up if
necessary, but not enable the clock.  Drivers that don't need atomic
clocks can call clk_prepare and clk_enable when they want the clock
on, keeping the same power savings as my current implementation.

Converting all the existing clk API users could be hard, but with a
little trickery backwards compatibility with the previous API could
also be maintained.  clk_enable could call clk_prepare if it hasn't
been called, with a WARN_ON_ONCE to find drivers that need to be
fixed, and the last clk_disable could call clk_unprepare.  For systems
where all clocks are atomic, everything would continue to work.

  reply	other threads:[~2011-01-21 22:28 UTC|newest]

Thread overview: 86+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-01-11  2:16 Locking in the clk API Jeremy Kerr
2011-01-11  3:15 ` Paul Mundt
2011-01-11  4:11   ` Jeremy Kerr
2011-01-11  4:54     ` Paul Mundt
2011-01-20 16:32       ` Ben Dooks
2011-01-20 18:57         ` Russell King - ARM Linux
2011-01-21  3:43           ` Saravana Kannan
2011-01-21  9:31             ` Russell King - ARM Linux
2011-01-11  9:03     ` Sascha Hauer
2011-01-11  9:28       ` Russell King - ARM Linux
2011-01-11 14:34         ` Pavel Machek
2011-01-20 16:29   ` Ben Dooks
2011-01-20 18:56     ` Russell King - ARM Linux
2011-01-20 21:30       ` Nicolas Pitre
2011-01-21  2:06         ` Dima Zavin
2011-01-21  4:12           ` Saravana Kannan
2011-01-21  9:32             ` Russell King - ARM Linux
2011-01-22  1:53               ` Saravana Kannan
2011-01-22  2:24                 ` Colin Cross
2011-01-22  2:56                   ` Saravana Kannan
2011-01-22  9:15                 ` Russell King - ARM Linux
2011-01-24 19:31                   ` Saravana Kannan
2011-01-21 21:03             ` Dima Zavin
2011-01-21 21:53               ` Nicolas Pitre
2011-01-21 22:02                 ` Russell King - ARM Linux
2011-01-21 22:28                   ` Colin Cross [this message]
2011-01-21 23:21                     ` Benjamin Herrenschmidt
2011-01-21 23:50                     ` Nicolas Pitre
2011-01-22  1:35                     ` Saravana Kannan
2011-01-22  2:22                       ` Colin Cross
2011-01-21 22:29                   ` Nicolas Pitre
2011-01-21 23:28                 ` Bryan Huntsman
2011-01-11  9:16 ` Russell King - ARM Linux
2011-01-11  9:44   ` Jeremy Kerr
2011-01-11 10:13     ` Paul Mundt
2011-01-11 10:30       ` Jeremy Kerr
2011-01-11 12:18         ` Paul Mundt
2011-01-11 13:52           ` Uwe Kleine-König
2011-01-11 14:35           ` Jeremy Kerr
2011-01-12  3:25             ` Saravana Kannan
2011-01-12  7:40               ` Uwe Kleine-König
2011-01-12  1:54           ` Saravana Kannan
2011-01-12  2:25             ` Paul Mundt
2011-01-20 16:57               ` Ben Dooks
2011-01-20 16:53           ` Ben Dooks
2011-01-20 16:40       ` Ben Dooks
2011-01-11 10:39     ` Uwe Kleine-König
2011-01-11 10:47       ` Russell King - ARM Linux
2011-01-11 10:56         ` Uwe Kleine-König
2011-01-11 11:15       ` Richard Zhao
2011-01-20 17:02         ` Ben Dooks
2011-01-20 19:08           ` Russell King - ARM Linux
2011-01-21  0:09             ` Jassi Brar
2011-01-21  4:47               ` Jassi Brar
2011-01-21  9:39                 ` Russell King - ARM Linux
2011-01-21 10:11                   ` Jassi Brar
2011-01-22  4:08                 ` Richard Zhao
2011-01-22  5:30                   ` Jassi Brar
2011-01-21  7:16             ` Saravana Kannan
2011-01-21  9:40               ` Russell King - ARM Linux
2011-01-22  1:47                 ` Saravana Kannan
2011-01-27  4:34                 ` Saravana Kannan
2011-01-27  8:54                   ` Russell King - ARM Linux
2011-01-27 20:30                     ` Saravana Kannan
2011-01-27 20:43                       ` Russell King - ARM Linux
2011-01-27 21:07                         ` Alan Cox
2011-01-27 21:11                           ` Russell King - ARM Linux
2011-01-27 21:15                           ` Russell King - ARM Linux
2011-01-28  3:29                           ` Saravana Kannan
2011-01-28  3:27                         ` Saravana Kannan
2011-01-11 12:11   ` Jassi Brar
2011-01-12  2:56   ` Saravana Kannan
2011-01-12  9:03     ` Russell King - ARM Linux
2011-01-15 14:02       ` Christer Weinigel
2011-01-15 14:53         ` Russell King - ARM Linux
2011-01-15 15:03           ` Uwe Kleine-König
2011-01-15 15:15             ` Russell King - ARM Linux
2011-01-15 16:03               ` Uwe Kleine-König
2011-01-15 16:21                 ` Russell King - ARM Linux
2011-01-15 16:31                   ` Uwe Kleine-König
2011-01-16  6:59               ` Grant Likely
2011-01-15 17:07           ` Christer Weinigel
2011-01-15 17:20             ` Russell King - ARM Linux
2011-01-15 17:44               ` Christer Weinigel
2011-01-15 20:30                 ` Russell King - ARM Linux
2011-01-17  1:19 ` Jeremy Kerr

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AANLkTimT+vfg9p4GSnn1jGW9r_gVom7xEUzMVJDscPg2@mail.gmail.com \
    --to=ccross@google.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).