At Thu, 17 Jan 2008 09:11:00 -0800, Tobin Davis wrote:
On Thu, 2008-01-17 at 17:58 +0100, Takashi Iwai wrote:
At Thu, 17 Jan 2008 09:46:06 -0400, Andrew Paprocki wrote: > > This adds additional pincaps which were not previously output. Also, Vref > capabilities are output per-pin along with the current Vref pinctl setting > if the NID supports Vref. Also added: processing widgets and proccaps, as > well as processing coefficient and coefficient index from the Realtek Define > Registers per the ALC883 datasheet. > > Signed-off-by: Andrew Paprocki <andrew@ishiboo.com> Did you check whether this doesn't break Claudio's codecgraph parser? If it's OK, I'd like to apply it, of course... TakashiYou're kidding, right? Adding more details about a codec is more useful at this point than making sure a non-essential script (that can be easily modified) isn't broken.
Well, now we have some parser program, then it's the high time to think about the proc output consistent in a certain level. So far, it's just written to be readable for human being. But, now we'd need to make it also easily readable by dumb programs, too.
What I want to know is if we can add even more to /proc, like EAPD status (not just capabilities), jack sense, etc. These are useful bits of information, not only for driver development, but also could be useful for userspace tools that can monitor these. Imagine having your system auto adjust from 5.1 surround settings to optimal headphone settings (not just automute) when you plug in your headphones.
The jack-sense is a bit problem. I once had a laptop that causes codec errors when issuing a jack-sense command without unsolicited events. So, I'd say, let's be as conservative as possible for the proc.
Moreover, we can read any commands via hwdep ioctl easily now...
Takashi