On Tue, Jul 8, 2014 at 5:57 AM, Luis R. Rodriguez
<mcgrof@xxxxxxxxxxxxxxxx> wrote:
> Let's separate cell base station hints from userspace from firmware
> being loaded and assuming that magical firmware has now regulatory
> data (which obviously I consider silly since we already provide the
> same functionality with CRDA and let you override the db.txt). The
> cellular base station hint infrastructure can certainly be expanded
> to support overriding the *_orig parameters. The FW API looks less
> attractive now as it seems more than anything the hint can actually
> come from userspace and the base station hint mechanism for your
> use case.
>
> A FW API to let FW load regulatory data dynamically is still OK but
> lets be clear that if you are using it for a userspace work around
> for cell base astation hints it seems rather a work around for
> existing APIs and you can likely implement what you want with less
> code. If we don't need this FW API I rather not have it added.
I don't see a way around having the get_regd() API to get regulatory
data from FW. How else would we get regdomain settings?
A hint can come from userspace (as a cell-base hint) or from the
driver. In some platforms the Wifi HW is directly connected to the SIM
and the FW sends up events about country change. This eventually
causes a driver hint.
Sometimes the SIM is removed or there's no reception etc. In this case
we must get back to the "default" country, which is burned in
NVRAM/FW. As explained in my previous email, currently reg.c doesn't
have any suitable facility to save this "default" country-code. So
AFAICT, your suggestion of setting the default on boot and then
returning to it is out of the question. And anyway a "restore to
default" API doesn't exist. I'm suggesting the following solution for
this problem:
1. user/kernel calls regulatory_hint("99") to say we need to return to default
2. get_regd("99") is called and the driver/FW returns the default
country-code and it's regulatory data.
Of course we can also invent new APIs, something like:
1. on boot, the kernel calls set_default_country("XX")
2. user/kernel calls a new api restore_to_default_country(), and reg.c
uses the XX value and restores.
Personally I think the first offer is simpler and involves less
complexity on behalf of reg.c. That was my final point of the previous
email.
Arik
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
