Am Sonntag, den 17.01.2010, 01:08 +0100 schrieb hermann pitton: > Am Samstag, den 16.01.2010, 04:15 -0800 schrieb Trent Piepho: > > On Sat, 16 Jan 2010, hermann pitton wrote: > > > Am Freitag, den 15.01.2010, 17:27 -0800 schrieb Trent Piepho: > > > > On Sat, 16 Jan 2010, hermann pitton wrote: > > > > > Am Dienstag, den 12.01.2010, 04:13 +0100 schrieb hermann pitton: > > > > > > > gpio-sysfs creates > > > > > > > /sys/class/gpio/export > > > > > > > /sys/class/gpio/import > > > > > > > but no gpio<n> entries so far. > > > > > > > > The saa713x driver predates the generic gpio layer by years and years, so > > > > it doesn't use it. It also doesn't need to use it. Since the gpios are > > > > managed by the saa713x driver, and they also used by the saa713x driver, > > > > there is no need to interface two different drivers together. There are > > > > tons of drivers for devices that have gpios like this, but they don't use > > > > the gpio layer. > > > > > > > > But with gpio access via sysfs for generic gpios, there is something useful > > > > about having the saa713x driver support generic gpios. IIRC, somehow wrote > > > > a gpio only bt848 driver that didn't do anything but export gpios. > > > > > > > > In order to do this, you'll have to write code for the saa7134 driver to > > > > have it register with the gpio layer. I think you could still have the > > > > saa7134 driver itself use its gpio directly. That would avoid a > > > > performance penalty in the driver. > > > > > > Thanks for more details, but I'm still wondering what pins ever could be > > > interesting in userland, given that they are all treated such different > > > per device, and we count up to 200 different boards these days. > > > > There are some cards for intended for survilence or embedded applications > > that have headers on them to connect things to the GPIOs. Like alarms or > > camera controllers and stuff like that. > > > > The GPIO only bttv driver was created by someone who just soldered a bunch > > of wires on a cheap bt848 card, you can get them for just a few dollars, as > > it was a cheap and easy way to get a bunch of gpios in a pc. See his page > > here http://www.bu3sch.de/joomla/index.php/bt8xx-based-gpio-card > > > > There are cards you can get that just have GPIOs, but they end up being > > rather expensive. Here's one: > > http://www.acromag.com/parts.cfm?Model_ID=317&Product_Function_ID=4&Category_ID=18&Group_ID=1 > > Way fancier than a tv card, but it's $600. > > > > I think if I was doing the coding, I'd add a field in the card description > > for what GPIOs should be exported. I.e., which ones have an external > > header. Maybe in addition to, or instead of, I'd have a module option that > > would cause GPIOs to be exported. A bitmask of which to export would be > > enough. > > Cool stuff! > > Are we aware of boards under mass production connecting unused gpios to > a panel already, providing external gpio functionality? > > The RTD one in question seems not to do so yet. > > http://www.rtd.com/pc104/UM/video/VFG7350ER.htm Damned, seems the opto-isolated I/Os might be in question. For the RTD stuff we don't have any high resolution photographs or anything else ... -- 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
