Johannes Stezenbach wrote: > On Mon, Dec 05, 2005, Michael Rickmann wrote: > > I have an old Typhoon DVB-S card which has the same cold boot problem as the > > KNC1 which Chris Dietrich reported about in November > > ( http://www.linuxtv.org/pipermail/linux-dvb/2005-November/005804.html ). > ... > > The explanation for my problem seems relatively straightforward. After a cold > > boot as defined above the stv0299 does not respond at the > > SAA7146_I2C_BUS_BIT_RATE_120 used by the current driver. > > When I use SAA7146_I2C_BUS_BIT_RATE_3200 as the Metzler driver does the > > frontend loads correctly but szap does not get a lock. I have also tried > > SAA7146_I2C_BUS_BIT_RATE_240 and SAA7146_I2C_BUS_BIT_RATE_480. > > With these the Typhoon cold boots correctly and works properly. > > > > It would be valuable to know wether this is a general problem of KNC1-S and > > old Typhoon-S cards or wether Chris and I have some odd problems. > > To find out wether my measure works replace SAA7146_I2C_BUS_BIT_RATE_120 in > > ../linux/drivers/media/dvb/ttpci/budget-core.c, line 384 by e.g. > > SAA7146_I2C_BUS_BIT_RATE_480 or apply attached patch. > > The stv0299 data sheet says that while the device is in standby mode, > the I2C bus speed has to be limited to F_CLK_IN/10, which usually is > 400kHz. SAA7146_I2C_BUS_BIT_RATE_120 is PCI_CLK/120 = 275kHz. So it > should work. > The stv0299 normally starts up in normal mode, but can > be made to start up in standby by a strap pin. > > Can you try if it is sufficient to lower the I2C bus speed > just for probing, and increase it later? > > e.g. in budget_av_attach(): > saa7146_i2c_adapter_prepare(dev, NULL, SAA7146_I2C_BUS_BIT_RATE_480); > frontend_init(budget_av); > saa7146_i2c_adapter_prepare(dev, NULL, SAA7146_I2C_BUS_BIT_RATE_120); > > (slightly hackish, but OK for testing; if it works we should add a > saa7146_i2c_set_speed() API; maybe need also handle wakeup after > stv0299_sleep()) > > But I'm not sure if this kind of optimization is worth it, maybe I > should just merge your patch. Opinions? I don't like adding another parameter. Either add a new field i2c_speed to struct budget, or use this simple patch: --- budget-core.c.old Sat Dec 10 21:48:13 2005 +++ budget-core.c Sat Dec 10 22:02:23 2005 @@ -381,7 +381,11 @@ int ttpci_budget_init(struct budget *bud strlcpy(budget->i2c_adap.name, budget->card->name, sizeof(budget->i2c_adap.name)); - saa7146_i2c_adapter_prepare(dev, &budget->i2c_adap, SAA7146_I2C_BUS_BIT_RATE_120); + if (bi->type == BUDGET_KNC1S) + saa7146_i2c_adapter_prepare(dev, &budget->i2c_adap, SAA7146_I2C_BUS_BIT_RATE_480); + else + saa7146_i2c_adapter_prepare(dev, &budget->i2c_adap, SAA7146_I2C_BUS_BIT_RATE_120); + strcpy(budget->i2c_adap.name, budget->card->name); if (i2c_add_adapter(&budget->i2c_adap) < 0) { Anyway, this card seems to behave somewhat strange. Some people reported that enabling the master reset in saa7146-core.c fixed their problems... Oliver -- -------------------------------------------------------- VDR Remote Plugin available at http://www.escape-edv.de/endriss/vdr/ --------------------------------------------------------
