On 07.05.2012 21:38, Thomas Mair wrote:
On 07.05.2012 09:59, Antti Palosaari wrote:
@@ -330,12 +335,12 @@ static int rtl2831u_frontend_attach(struct
dvb_usb_adapter *adap)
/* check QT1010 ID(?) register; reg=0f val=2c */
ret = rtl28xxu_ctrl_msg(adap->dev,&req_qt1010);
if (ret == 0&& buf[0] == 0x2c) {
- priv->tuner = TUNER_RTL2830_QT1010;
+ priv->tuner = TUNER_RTL28XX_QT1010;
The idea why I named it as a TUNER_RTL2830_QT1010 was to map RTL2830 and given tuner. It could be nice to identify used demod/tuner combination in some cases if there will even be such combination same tuner used for multiple RTL28XXU chips.
Ok. Should we use the TUNER_RTL2830/TUNER_RTL2832 approach or the RUNER_RTL28XX?
I vote for style example; RTL2830_QT1010 versus RTL2832_QT1010, just in
case same tuner could be used for multiple configurations.
But it is not big issue for RTL2830 as it supports only 3 tuners
currently - and there will not be likely any new. But there is some
other RTL28XXU chips, like DVB-C model and etc.
It is ~same function that existing one but without LED GPIO. Hmmm, dunno what to do. Maybe it is OK still as switching "random" GPIOs during streaming control is not good idea.
I know. I had the same thougths and could not come up with a more elegant idea.
Maybe here the TUNER_RTL2832 definition could be used. But that is too not the
ideal solution I guess.
I dont even care if whole LED GPIO is removed. It is not needed at all
and I even suspect it is not on same GPIO for all RTL2831U devices... Do
what you want - leave it as you already did.
This looks weird as you write demod register. Is that really needed?
If you has some problems I suspect those are coming from the fact page cached by the driver isdifferent than page used by chip. Likely demod is reseted and page is 0 after that.
If you really have seen some problems then set page 0 in demod sleep. Or set page directly to that driver priv.
I'll check that.
+#define RTL28XXU_TUNERS_H
+
+enum rtl28xxu_tuner {
+ TUNER_NONE,
+ TUNER_RTL28XX_QT1010,
+ TUNER_RTL28XX_MT2060,
+ TUNER_RTL28XX_MT2266,
+ TUNER_RTL28XX_MT2063,
+ TUNER_RTL28XX_MAX3543,
+ TUNER_RTL28XX_TUA9001,
+ TUNER_RTL28XX_MXL5005S,
+ TUNER_RTL28XX_MXL5007T,
+ TUNER_RTL28XX_FC2580,
+ TUNER_RTL28XX_FC0012,
+ TUNER_RTL28XX_FC0013,
+ TUNER_RTL28XX_E4000,
+ TUNER_RTL28XX_TDA18272,
+};
+
+#endif
I don't see it good idea to export tuners from the DVB-USB-driver to the demodulator. Demod drivers should be independent. For the other direction it is OK, I mean you can add tuners for demod config (rtl2832.h).
Ok. So the definitions of the tuners would go into the rtl2830.h and rtl2832.h.
Put those to the demod as a af9013 and af9033 for example has. rtl2830.h
does not need to know tuner at all, not need to add.
After all, you have done rather much changes. Even such changes that are not relevant for the RTL2832 support. One patch per one change is the rule. Also that patch serie is wrong order, it will break compilation for example very bad when git bisect is taken. It should be done in order first tuner or demod driver then DVB-USB-driver.
Sorry for the inconsistent patches. I will go over my changes again and split them
into smaller chunks, trying to keep the changes to a minimum. That includes to change
the order of the patches to tuner, demod and finally dvb-usb driver. Thanks for all
the comments. They really help me getting the driver nice and neat. I will probably
not be able to do the changes before next weekend.
regards
Antti
--
http://palosaari.fi/
--
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
