Stefan Ringel wrote:
> Am 04.02.2010 03:43, schrieb Mauro Carvalho Chehab:
>> Stefan Ringel wrote:
>>
>>> Am 03.02.2010 21:49, schrieb Devin Heitmueller:
>>>
>>>> On Wed, Feb 3, 2010 at 3:38 PM, Stefan Ringel <stefan.ringel@xxxxxxxx> wrote:
>>>>
>>>>
>>>>> signed-off-by: Stefan Ringel <stefan.ringel@xxxxxxxx>
>>>>>
>>>>> --- a/drivers/media/dvb/frontends/zl10353.h
>>>>> +++ b/drivers/media/dvb/frontends/zl10353.h
>>>>> @@ -45,6 +45,8 @@ struct zl10353_config
>>>>> /* clock control registers (0x51-0x54) */
>>>>> u8 clock_ctl_1; /* default: 0x46 */
>>>>> u8 pll_0; /* default: 0x15 */
>>>>> +
>>>>> + int tm6000:1;
>>>>> };
>>>>>
>>>>>
>>>> Why is this being submitted as its own patch? It is code that is not
>>>> used by *anything*. If you really did require a new field in the
>>>> zl10353 config, that field should be added in the same patch as
>>>> whatever requires it.
>>>>
>>>> Devin
>>>>
>>>>
>>>>
>> Hi Stefan,
>>
>> Due to the problem with your emailer, most of the patches failed:
>>
>> 2/15 3/15 4/15 5/15 6/15 7/15 2/15 (the second one 8th in the sequence)
>> 9/15 11/15 12/15 13/15 and 15/15.
>>
>> In brief, only two patches applied: the 1/15 and a patch also called 2/15
>> (I suspect it is the 10th patch).
>>
>> So, after reading our comments, reading README.patches and applying the
>> Asaulted-patches extension to your thunderbird, please resubmit the patches.
>>
>>
>>> Actually doesn't work zl10353 with tm6010, it have a little different
>>> between a few registers, so I think that I use it.
>>>
>>> for example:
>>> zl10353 use 0x64 , but not mine (0x63)
>>> register 0x5f is 0x17 not 0x13
>>> register 0x5e is 0x40 not 0x00 for auto
>>> and tuner go is 0x70 not 0x71
>>>
>>> the other register are ok. I have no idea how I can set it.
>>>
>> I agree with Devin that the patch 14/15 shouldn't be applied as-is.
>>
>> I have no idea why zl10353 needs a different setup, but you don't need
>> to add an extra parameter to identify the bridge driver. There's a field
>> at struct i2c_client that can be used: i2c_adapter->id.
>>
>> This is already initialized with I2C_HW_B_TM6000 (although it currently
>> uses a fake value).
>>
>> It is still ugly to do some specific initialization like this, but, when
>> we have no other glue, due to the lack of datasheets, this is a better
>> alternative than adding an extra fake parameter at the config struct.
>>
>> All you need is some code like:
>>
>>
>> diff --git a/linux/drivers/media/dvb/frontends/zl10353.c b/linux/drivers/media/dvb/frontends/zl10353.c
>> --- a/linux/drivers/media/dvb/frontends/zl10353.c
>> +++ b/linux/drivers/media/dvb/frontends/zl10353.c
>> @@ -597,6 +597,10 @@ static int zl10353_init(struct dvb_front
>> zl10353_read_register(state, 0x51) != zl10353_reset_attach[2]) {
>> rc = zl10353_write(fe, zl10353_reset_attach,
>> sizeof(zl10353_reset_attach));
>> +
>> + if (state->i2c->id == I2C_HW_B_TM6000) {
>> + /* Do special init needed by tm6000 driver */
>> + }
>> #if 1
>> if (debug_regs)
>> zl10353_dump_regs(fe);
>> diff --git a/linux/drivers/staging/tm6000/tm6000-i2c.c b/linux/drivers/staging/tm6000/tm6000-i2c.c
>> --- a/linux/drivers/staging/tm6000/tm6000-i2c.c
>> +++ b/linux/drivers/staging/tm6000/tm6000-i2c.c
>> @@ -33,8 +33,6 @@
>> #include "tuner-xc2028.h"
>>
>>
>> -/*FIXME: Hack to avoid needing to patch i2c-id.h */
>> -#define I2C_HW_B_TM6000 I2C_HW_B_EM28XX
>> /* ----------------------------------------------------------- */
>>
>> static unsigned int i2c_debug = 0;
>> diff --git a/linux/include/linux/i2c-id.h b/linux/include/linux/i2c-id.h
>> --- a/linux/include/linux/i2c-id.h
>> +++ b/linux/include/linux/i2c-id.h
>> @@ -42,6 +42,7 @@
>> #define I2C_HW_B_AU0828 0x010023 /* auvitek au0828 usb bridge */
>> #define I2C_HW_B_CX231XX 0x010024 /* Conexant CX231XX USB based cards */
>> #define I2C_HW_B_HDPVR 0x010025 /* Hauppauge HD PVR */
>> +#define I2C_HW_B_TM6000 0x010026 /* TM5600/6000/6010 video bridge */
>>
>> /* --- SGI adapters */
>> #define I2C_HW_SGI_VINO 0x160000
>>
> init is ok, but it has only one additional register 0x55 (value 0x40)
> reset after second. I think it can work without this additional once.
> The different is in function "set_parameters".
Ok. Well, just add the logic to check for the bridge at the place it is needed, and
only if it won't work otherwise. I would love to have the datasheet to check for
the values, but unfortunately I don't have it. Anyway, after having it working, someone
may help to figure out why the changes are needed and provide a better approach.
-
A side note:
due to the backport code that you've added on your patches, I suspect that you just
copied the files from the -hg to -git. You shouldn't do it, since we remove all backport
stuff at -git, by running the gentree.pl script, found at v4l/scripts/gentree.pl.
--
Cheers,
Mauro
--
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
