On Wed, 04 May 2011 11:56:17 -0300, Mauro Carvalho Chehab
<mche...@redhat.com> wrote:
> Em 28-04-2011 12:13, David Härdeman escreveu:
>> The RC_TYPE_* defines are currently used both where a single protocol
is
>> expected and where a bitmap of protocols is expected. This patch tries
>> to separate the two in preparation for the following patches.
>>
>> Signed-off-by: David Härdeman <da...@hardeman.nu>
>
> Most of the patch is just renaming stuff. So, I'm commenting just the
> rc-main.c/rc-map.h changes.
>
>> diff --git a/drivers/media/rc/rc-main.c b/drivers/media/rc/rc-main.c
>> index 5b4422e..5a182b2 100644
>> --- a/drivers/media/rc/rc-main.c
>> +++ b/drivers/media/rc/rc-main.c
>> @@ -102,7 +102,7 @@ static struct rc_map_list empty_map = {
>> .map = {
>> .scan = empty,
>> .size = ARRAY_SIZE(empty),
>> - .rc_type = RC_TYPE_UNKNOWN, /* Legacy IR type */
>> + .rc_type = RC_BIT_UNKNOWN, /* Legacy IR type */
>> .name = RC_MAP_EMPTY,
>> }
>> };
>> @@ -725,14 +725,17 @@ static struct {
>> u64 type;
>> char *name;
>> } proto_names[] = {
>> - { RC_TYPE_UNKNOWN, "unknown" },
>> - { RC_TYPE_RC5, "rc-5" },
>> - { RC_TYPE_NEC, "nec" },
>> - { RC_TYPE_RC6, "rc-6" },
>> - { RC_TYPE_JVC, "jvc" },
>> - { RC_TYPE_SONY, "sony" },
>> - { RC_TYPE_RC5_SZ, "rc-5-sz" },
>> - { RC_TYPE_LIRC, "lirc" },
>> + { RC_BIT_OTHER, "other" },
>> + { RC_BIT_RC5, "rc-5" },
>> + { RC_BIT_RC5X, "rc-5-x" },
>> + { RC_BIT_RC5_SZ, "rc-5-sz" },
>> + { RC_BIT_RC6, "rc-6" },
>> + { RC_BIT_JVC, "jvc" },
>> + { RC_BIT_SONY12, "sony12" },
>> + { RC_BIT_SONY15, "sony15" },
>> + { RC_BIT_SONY20, "sony20" },
>> + { RC_BIT_NEC, "nec" },
>> + { RC_BIT_LIRC, "lirc" },
>> };
>
> There are some API breakages on the above. We shouln't do it, except
> if strictly required, and, if we'll do it, we need to do it via
> Documentation/feature-removal-schedule.txt.
>
> There are two types of breakages on the above:
> 1) the removal of "unknown" and "sony" types;
We could just keep "unknown" and (optionally) also map "other" to it.
> 2) the behaviour change of "rc-5" (that, currently, means
> both rc-5 and rc-5x.
That's a change but I don't think it'll actually break user-space
since "rc-5" will still be accepted.
> Also, while you've mapped rc5/sony variants, nec variants weren't
mapped.
"nec" needs no mapping with the nec 32-bit scancode change.
> IMO, what we should do on the above is:
> 1) preserve the "unknown";
>
> 2) use "rc-5", "sony", "nec" with the meaning that they will
> enable all variants of those protocols;
>
> 3) add a new set of protocols to indicate subsets, like "sony:20",
> "rc-5:normal", "rc5:x", etc.
I can look into it...
> 4) if you're changing the interface, please submit a patch also
> to v4l-utils, adding a logic there to handle the changes.
I'll look into it when redoing the patches.
>>
>> #define PROTO_NONE "none"
>> diff --git a/include/media/rc-map.h b/include/media/rc-map.h
>> index 9184751..2c68ca6 100644
>> --- a/include/media/rc-map.h
>> +++ b/include/media/rc-map.h
>> @@ -11,19 +11,36 @@
>>
>> #include <linux/input.h>
>>
>> -#define RC_TYPE_UNKNOWN 0
>> -#define RC_TYPE_RC5 (1 << 0) /* Philips RC5 protocol */
>> -#define RC_TYPE_NEC (1 << 1)
>> -#define RC_TYPE_RC6 (1 << 2) /* Philips RC6 protocol */
>> -#define RC_TYPE_JVC (1 << 3) /* JVC protocol */
>> -#define RC_TYPE_SONY (1 << 4) /* Sony12/15/20 protocol */
>> -#define RC_TYPE_RC5_SZ (1 << 5) /* RC5 variant used by
>> Streamzap */
>> -#define RC_TYPE_LIRC (1 << 30) /* Pass raw IR to lirc
>> userspace */
>> -#define RC_TYPE_OTHER (1u << 31)
>
>> +#define RC_TYPE_UNKNOWN 0 /* Protocol not known */
>> +#define RC_TYPE_OTHER 0 /* Protocol known but
>> proprietary */
>
> This change doesn't make sense: we should either remove other or use
> different bits for different meanings.
I think that it's not necessary. Unknown and other have a
meaningful difference to the programmer but not from a code
point of view. Both mean that whatever scancode we get, we
have to accept as-is and we don't know anything about it.
For e.g. a RC5 scancode we could (though we're not doing it yet)
do sanity-checks on the scancode and set a proper timeout value
based on protocol characteristics. With other and unknown we can't.
That's the reason I merged them - we can't do anything meaningful
with the difference from a code point of view and long term
"unknown" should die (might not happen but...).
> This is somewhat a mess currently, as there
> are, in
> fact, 3 types of "protocols":
> 1) reverse-engineered drivers, where developer didn't care to check
> what was the used protocol. It is there due to legacy IR handlers,
> added before rc-core. The better is to not accept this type anymore
> for new devices;
> 2) Other protocols that don't match at the list of supported protocols.
> Reserved for the cases were the developer took the care to check if
> the protocol is not NEC/RC-5/... and didn't find any protocol that
> matches;
> 3) Standard protocols with broken hardware. In general, keycode tables
> with just 8 bits for RC-5 or NEC, because the hardware uses a cheap
> uC (generally KS007 or similar) that only decodes the last 8 bits of
> the protocol. As the developer didn't have a full IR decoder, he was
> not able to fill the RC/NEC table with the IR address.
>
> The problem with (2) is that it may reflect a temporary state where the
> protocol
> is not yet known. After adding a protocol decoder for it, case (2) turns
> into
> case (1).
>
> So, maybe we can just merge (1) and (2) into the same case: "unknown".
> Maybe we
> should map (3) with a different option, internally (even exporting it as
> "unknown"
> to userspace), as it helps us to identify such cases and fix it later.
I think we should keep the distinction internal.
--
David Härdeman
--
To unsubscribe from this list: send the line "unsubscribe linux-media" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
