On Tue, Dec 13, 2016 at 10:10:51AM +0000, Ian Arkver wrote: > On 13/12/16 09:43, Sakari Ailus wrote: > >Hi Nicholas, > > > >On Tue, Dec 13, 2016 at 02:58:02AM +0100, Nicholas Mc Guire wrote: > >>As this is not in atomic context and it does not seem like a critical > >>timing setting a range of 1ms allows the timer subsystem to optimize > >>the hrtimer here. > >I'd suggest not to. These delays are often directly visible to the user in > >use cases where attention is indeed paid to milliseconds. > > > >The same applies to register accesses. An delay of 0 to 100 µs isn't much as > >such, but when you multiply that with the number of accesses it begins to > >add up. > > > Data sheet for this device [1] says STBYN deassertion to RSTN deassertion > should be >50us, though this is actually referenced to MCLK startup. See > Figure 36, Power-Up Sequence, page 42. > > I think the usleep range here could be greatly reduced and opened up to > allow hr timer tweaks if desired. > > [1] http://www.bdtic.com/DataSheet/SAMSUNG/S5K6AAFX13.pdf Good point. Datasheets do not always tell everything though; it'd be good to get a comment from the original driver authors on why they've used the value which can now be found in the driver. -- Sakari Ailus e-mail: sakari.ailus@xxxxxx XMPP: sailus@xxxxxxxxxxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
