Hi all We've got a broken driver in 3.11 and in 3.12-rc and we don't have a clear way to properly fix it. The problem has been originally reported and discussed in [1], a patch-set to fix the problem has been proposed in [2], which actually lead to the topic of this mail - whether or not calls to v4l2_clk_enable() and v4l2_clk_disable(), or respectively to s_power(1) and s_power(0) subdevice core operations should be balanced. Currently they aren't in em28xx driver, and the V4L2 clock API throws warnings on attempts to disable already disabled clock. Patch [3] attempted to fix that. So, the question is - whether to enforce balanced power on / off calls, or to remove the warning.
Let's try to have a look at other examples in the kernel: 1. runtime PM: pm_runtime_get*() / pm_runtime_put*() only work, if balanced, but no warning is issued, if the balance is broken, AFAICS. 2. clock API: clk_enable() / clk_disable() in drivers/clk/clk.c have to be balanced and a warning is issued, if clk_disable() is called for an already disabled clock. 3. regulator API: regulator_enable() / regulator_disable() have to be balanced. A warning is issued if regulator_disable() is called for a disabled regulator. So, I think, our V4L2 clock enable / disable calls should be balanced, and to enforce that a warning is helpful. Other opinions? Thanks Guennadi [1] http://thread.gmane.org/gmane.linux.drivers.video-input-infrastructure/68028 [2] http://thread.gmane.org/gmane.linux.drivers.video-input-infrastructure/68510 [3] http://thread.gmane.org/gmane.linux.drivers.video-input-infrastructure/68864 --- Guennadi Liakhovetski, Ph.D. Freelance Open-Source Software Developer http://www.open-technology.de/ -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
