kbtv2 sound saa.c -- And BETA3 available

[Danny Pansters]

Hi again,

On Friday 29 February 2008 06:28:48 you wrote:
> On Fri, Feb 29, 2008 at 01:44:43AM +0100, Danny Pansters wrote:
> > On Thursday 28 February 2008 08:20:11 you wrote:
> > > On Thu, Feb 28, 2008 at 12:42:38AM +0100, Danny Pansters wrote:
> > > > On Wednesday 27 February 2008 18:55:56 you wrote:
> > > > <snip>
> > > >
> > > > > thanks for the hints, but I must admit that I'm totally lost here -
> > > > > where can I tweak sound carrier frequency?
> > > >
> > > > There doesn't seem to be a canned-in way (only for setting NTSC vs
> > > > PAL), but Sidney's information gave me a second datapoint (for the
> > > > actual value to be set to "audio_options.sao_siffreq1"). It stands to
> > > > logic that it's not only a linear but also a proportional equation to
> > > > convert the actual freq to the required registry value. In terms of
> > > > numbers I made it work, and there seems to be a certain logic in
> > > > terms of (audio) output frequency.
> > > >
> > > > I think the attached patch will work for you, as well as for PAL I
> > > > (UK) and hopefully also for NTSC. The sif is switched when broadcast
> > > > standard gets changed.
> > > >
> > > > Here (B/G) it used to be that with other PAL variants the sound would
> > > > remain. With the current working version of the saa backend it gets
> > > > noise (most cases mixed with the actual original sound) if I switch
> > > > away from B/G to any other PAL type as well as if I switch to NTSC,
> > > > then it gets a heavy pulse-like noise. Once I switch back to B/G
> > > > sound is ok again.
> > > >
> > > > Please try attached patches (1.93 working version, not beta3 yet) to
> > > > saa.c and saa.h.
> > >
> > > Doesn't help. I'll try to look at the code.
> > >
> > > There's another problem - I get panics with saa driver:
> > >
> > > panic: Trying sleep, but thread marked as sleeping prohibited
> > > cpuid = 0
> > > KDB: stack backtrace:
> > > db_trace_self_wrapper() at db_trace_self_wrapper+0x2a
> > > panic() at panic+0x17d
> > > sleepq_add() at sleepq_add+0x2e1
> > > _sx_slock_hard() at _sx_slock_hard+0x15d
> > > _sx_slock() at _sx_slock+0xc1
> > > pfind() at pfind+0x24
> > > saa_intr() at saa_intr+0x313
> > > ithread_loop() at ithread_loop+0xda
> > > fork_exit() at fork_exit+0x12a
> > > fork_trampoline() at fork_trampoline+0xe
> > > --- trap 0, rip = 0, rsp = 0xffffffffac3c0d30, rbp = 0 ---
> >
> > Looks like 32->64 issue
>
> Don't think so, please check Attilio's reply -
> http://lists.freebsd.org/pipermail/freebsd-hackers/2008-February/023663.htm
>l
>
> I've attached a patch removing pfind() calls from saa_intr, allowing me at
> least try to resolve sound issues (I don't think that it is correct,
> more than so, quite contrary, I'm pretty sure it is incorrect, but will
> do for the time being).

I'll believe it when I see it. I had no panics on amd64/fbsd7 so far. It 
worked shockingly well compared to the other backends with all HW 
acceleration turned off (GForce card + nv)

So what *is* the difference between 0 and NULL if an int (or appropriately 
intptr) is expected?

I don't want to be pedantic but I would like see a report of any such panic 
with the kbtv2-kmod-saa port first before I start changing all sorts of 
things. Or get it myself.

BTW, I'm now very amendable to just pushing in your suggested B,C,E databytes 
for D/K if that's going to do the trick with both video and sound. If they 
don't break B/G, the only other standard known to work, I'm fine with it and 
will put it in beta4/rc1.

Sorry to have been incommunicado for a few days. I managed to get all the 
backend to run on amd64/fbsd7 (another box from where I keep my email 
obviously). Fixed a few things, nothing particularly difficult (I was under 
the impression that major work would need to be done to use generic iic 
device but that wasn't the case at all). Lots of patch/tarball rerolling.
 
I uploaded beta3 and checked that it still works on i386.

Next I'd like to at least get the D/K problems kinked out, and do some work on 
the other backends. Probably RC1 will be next.

While I'm interested in adding microtune support to saa, this is not trivial 
and I'm going to defer it until after 2.0.




> > > Uptime: 26m18s
> > > Physical memory: 2034 MB
> > > Dumping 190 MB: 175 159 143 127 111 95 79 63 47 31 15
> > >
> > > I can provide more details if needed.
> > >
> > >
> > > Yuri
> >
> > I tried installing 7.0-REL/amd64 from the disc1-3 ISOs on a amd box I had
> > sitting around  (nvidia/asus  based A8N-SLI-SE board) but I can't even
> > reliably run X apps on it (e.g. firefox/konqueror quickly freeze the
> > screen...). Tier-1 my ass.
> >
> > I also happen to have a ia64 box but I read it requires a serial console
> > because it has no syscons...
> >
> > Dan
>
> Yuri