| From: |
| Dmitry Baryshkov <dbaryshkov@gmail.com> |
| To: |
| linux-kernel@vger.kernel.org |
| Subject: |
| [RFC][PATCH 0/3] Clocklib: generic clocks management framework |
| Date: |
| Sun, 8 Jun 2008 13:19:54 +0400 |
| Message-ID: |
| <20080608091954.GA11200@doriath.ww600.siemens.net> |
| Cc: |
| akpm@linux-foundation.org,
Haavard Skinnemoen <haavard.skinnemoen@atmel.com>,
Russell King <rmk+lkml@arm.linux.org.uk>,
Paul Mundt <lethal@linux-sh.org>,
pHilipp Zabel <philipp.zabel@gmail.com>,
Pavel Machek <pavel@ucw.cz>, tony@atomide.com, paul@pwsan.com,
David Brownell <david-b@pacbell.net>,
Mark Brown <broonie@opensource.wolfsonmicro.com> |
| Archive-link: |
| Article,
Thread
|
Hi,
Here is second attempt to introduce clocks management framework based on
kobjects. This attempt goes a bit further. I've removed the global
spinlock. All add/remove operations are guarded via built into clocks kset
spinlock. And all per-clock operations are guided by built-in atomic
counter variable. I can split it into counter and per-clock spinlock,
but I think it will be a waste of space.
Also as suggested by Andrew Morton, I've split the set_mode operation
back to enable/disable.
--
With best wishes
Dmitry