Re: [Openvortex-dev] re-setup dma pointers before resume

[Manuel Jander]

Hallo Wilfried,

Thanks for your patch :)

I think i found another related bug too:

Issuing this command:
arecord -c 1 -f U8 -r 15000 test8.wav

I get this parameters:

Its setup is:
stream       : CAPTURE
access       : RW_INTERLEAVED
format       : U8
subformat    : STD
channels     : 1
rate         : 15000
exact rate   : 15000 (15000/1)
msbits       : 8
buffer_size  : 7504
period_size  : 1876
period_time  : 125066
tick_time    : 1000
tstamp_mode  : NONE
period_step  : 1
sleep_min    : 0
avail_min    : 1876
xfer_align   : 1876
start_threshold  : 1
stop_threshold   : 7504
silence_threshold: 0
silence_size : 0
boundary     : 1967128576

period_size  : 1876: This is absolute nonsense. I don't know yet why
arecord give this value (should be max 1024). I'll try to investigate
that. If that value is really used, no wonder that chunks of data get
repeated or lost.

On Sun, 2003-12-14 at 11:27, Wilfried Weissmann wrote:
> there was still a bug in how periods <= 4 were handled. i think it is 
> correct this way:
[snip!]

I'm looking forward to integrate this patch into my current tree. But
before applying it, i would like to try if the context loss is due to a
bug or simply the normal behavior of the vortex chip.

Best Regards

-- 
Manuel Jander
mjander(at)users(dot)sourceforge(dot)net