firmerware
9 years agoSound is now reliably playing back samples on TLV320AIC20K. master
Rick van Rein [Wed, 1 Feb 2012 13:37:55 +0000 (14:37 +0100)]
Sound is now reliably playing back samples on TLV320AIC20K.
Especially the volume is constant, indicating timing accuracy
in the sample exchange between DSP and Codec.
Recording has not been made to work yet.

9 years agoNew sound API, more sensible and support for echo cancellation's
Rick van Rein [Fri, 20 Jan 2012 15:20:41 +0000 (16:20 +0100)]
New sound API, more sensible and support for echo cancellation's
accurate timing requirements.  Improved RTT.  Added L8/L16 codecs.

9 years agoIncluded SpanDSP functions for T.38 fax, local aim being TIFF over TFTP
Rick van Rein [Fri, 16 Dec 2011 17:31:59 +0000 (17:31 +0000)]
Included SpanDSP functions for T.38 fax, local aim being TIFF over TFTP

9 years agoBetter sound. Buzzing tone was caused by signed/unsigned mixture.
Rick van Rein [Mon, 12 Dec 2011 10:45:33 +0000 (10:45 +0000)]
Better sound.  Buzzing tone was caused by signed/unsigned mixture.
Small sounds around 0 became extreme high/low values, so loud/hifreq sound.
Still no mic->buf->spk path, but at 8 kHz noise is now also low.

9 years agoThe "echo" app is still not playing back (on BT200) but it has improved.
Rick van Rein [Mon, 12 Dec 2011 09:48:03 +0000 (09:48 +0000)]
The "echo" app is still not playing back (on BT200) but it has improved.
It now results in a hissing/noisy sound, but from the right buffer posns.

9 years agoFINALLY -- sound is working reliably on BT-200
Rick van Rein [Mon, 5 Dec 2011 15:06:13 +0000 (15:06 +0000)]
FINALLY -- sound is working reliably on BT-200
 * it boots and configures the TLV Codec
 * it produces a 1 kHz tone at an alarming volume (damn, that hurts)
 * change of sample rate do not seem to disturb the process
This is not using DMA yet, but direct writes to the McBSP #1 port
Next station: get the echo function working!

9 years ago* Lots of codec updates
Rick van Rein [Mon, 5 Dec 2011 12:33:33 +0000 (12:33 +0000)]
* Lots of codec updates
* New kconfig version (from latest Linux kernel)
* Sound is finally starting to work on the BT-200 phone!!!

Re. sound:  When it boots the echo program that sends 1 kHz samples over the McBSP #1
it starts up silently.  The Codec has not been initialised at that time:

Channel 0 registers (slave 0x40): 1A=90 2A=20 3C=88 3D=c0 3A=01 3B=40 4A=08 4B=9e 5A=00 5B=40 5C=bf 5D=c0 6A=02 6B=82
Channel 1 registers (slave 0x41): 1A=90 2A=20 3C=88 3D=c0 3A=01 3B=40 4A=08 4B=9e 5A=00 5B=40 5C=bf 5D=c0 6A=00 6B=80

Then, run the following commands:

./bin/i2cp/aic2x-config /dev/i2c-2 0x40 0x49 0xa0 0x01 0x40 0x88 0xff 0x08 0x9e 0x1f 0x40 0xbf 0xc0 0x02 0x82
./bin/i2cp/aic2x-config /dev/i2c-2 0x41 0x49 0xa0 0x31 0x60 0x88 0xff 0x08 0x9e 0x12 0x50 0xbf 0xc0 0x00 0x80

This will raise sound (not through DMA but "manual" writes to McBSP); the codec reads:

Channel 0 registers (slave 0x40): 1A=49 2A=a0 3A=01 3B=40 3C=88 3D=ff 4A=08 4B=9e 5C=bf 5D=c0 5A=1f 5B=40 6A=02 6B=82
Channel 1 registers (slave 0x41): 1A=49 2A=a0 3A=31 3B=60 3C=88 3D=ff 4A=08 4B=9e 5C=bf 5D=c0 5A=12 5B=50 6A=00 6B=80

Note that the volume is not always the same; apparently there is still some form of
sync problem.  Things to explore are now:

1. Whether the TLV should be reset before setting up McBSP #1
2. Whether the data to the Codec is stable as long as SCLK is low
3. Whether FS changes when SCLK is high (max 6 ns after going up, no min specified)

9 years agoAdded Realtime Text (RFC 4103) without having tied it to the phone software yet
Rick van Rein [Tue, 15 Nov 2011 14:05:59 +0000 (14:05 +0000)]
Added Realtime Text (RFC 4103) without having tied it to the phone software yet

9 years agoFS to codec is now 1 CLKG wide; sine wave sent to codec (but still not heard)
Rick van Rein [Mon, 17 Oct 2011 14:11:28 +0000 (14:11 +0000)]
FS to codec is now 1 CLKG wide; sine wave sent to codec (but still not heard)

9 years agoSerial communication with the codec revived (by flushing RFULL and XEMPTY)
Rick van Rein [Tue, 11 Oct 2011 09:47:54 +0000 (09:47 +0000)]
Serial communication with the codec revived (by flushing RFULL and XEMPTY)

9 years agoAdded i2cp tools for TLV320AIC2x chips (in BT200 phones, at least)
Rick van Rein [Mon, 10 Oct 2011 09:15:06 +0000 (09:15 +0000)]
Added i2cp tools for TLV320AIC2x chips (in BT200 phones, at least)

9 years agoTransactional logic, mostly done outgoing, structured incoming. Split tract and...
Rick van Rein [Fri, 15 Jul 2011 22:49:40 +0000 (22:49 +0000)]
Transactional logic, mostly done outgoing, structured incoming.  Split tract and dialog.

9 years agoIntroduced LLC (for TFTP and Console) to the usual networking stack.
Rick van Rein [Thu, 30 Jun 2011 21:35:14 +0000 (21:35 +0000)]
Introduced LLC (for TFTP and Console) to the usual networking stack.
This introduces a few really useful programming facilities:
 1. To be able to download the Flash (and, in some future, upload)
 2. To access a console over Ethernet, displaying bottom_printf()

10 years agoThe first version that was demonstrated to get an IPv6 address
Rick van Rein [Sat, 25 Jun 2011 10:33:06 +0000 (10:33 +0000)]
The first version that was demonstrated to get an IPv6 address
on various networks, and (hackfully) show it on the display of
(only) a Grandstream BT200.  Autoboot first tries IPv6 autoconf
then DHCPv6, then falls back to a 6bed4 tunnel.  The first two
options worked on an IPv6-supporting LAN, the latter on a
retro network (aka IPv4-only).

10 years agoImproved console printing; 2nd, 4th, ... args still ignored on tic55x?
Rick van Rein [Fri, 6 May 2011 10:07:06 +0000 (10:07 +0000)]
Improved console printing; 2nd, 4th, ... args still ignored on tic55x?

10 years agoWorking KSZ8842 network driver; working LLC console routines (except va_list pass...
Rick van Rein [Thu, 5 May 2011 20:02:04 +0000 (20:02 +0000)]
Working KSZ8842 network driver; working LLC console routines (except va_list pass-around)
Beginnings of sound chip interaction (TLV320AIC2x)

10 years agoBT200 / tic55x support: working with keyboard, lcd, and several test programs
Rick van Rein [Fri, 15 Apr 2011 14:45:10 +0000 (14:45 +0000)]
BT200 / tic55x support: working with keyboard, lcd, and several test programs
Added a warning about license conflict potential for this architecture.  Sigh.

10 years agoRemoved TIC55x-specifics from
Rick van Rein [Wed, 6 Apr 2011 12:09:26 +0000 (12:09 +0000)]
Removed TIC55x-specifics from
src/function/develtest/timer2led.c

10 years agoWorking timer interrupts on TIC55x (Grandstream BT-200)
Rick van Rein [Wed, 6 Apr 2011 12:07:50 +0000 (12:07 +0000)]
Working timer interrupts on TIC55x (Grandstream BT-200)

Note that SLEEP_HIBERNATE still gets stuck, SLEEP_SNOOZE
does not however.  The irqpolling variety is inconsistent
in its timing, but that is not a real problem.

10 years agoWorking switch2led demonstration on tic55x
Rick van Rein [Wed, 6 Apr 2011 09:37:53 +0000 (09:37 +0000)]
Working switch2led demonstration on tic55x

This is a first release that is starting to work on real
phone hardware: Grandstream BT200.  The part that works
is the backlight to the LCD and the hook contact, but
that at least shows control.  Work has been done on the
timer actions, but that has not been made to work yet,
at least not based on interrupts.

10 years agoInserted tickless RTOS code. Network input code dysfunctional, and bound to be replaced.
root [Tue, 8 Mar 2011 22:31:49 +0000 (22:31 +0000)]
Inserted tickless RTOS code.  Network input code dysfunctional, and bound to be replaced.
Network bootstrapping uses timer, and takes the right initiatives (without processing input).

10 years agoMore structured build system. Menuconfig parameters are now processed.
Rick van Rein [Thu, 3 Mar 2011 22:52:02 +0000 (22:52 +0000)]
More structured build system.  Menuconfig parameters are now processed.

10 years agoInitial release:
root [Thu, 3 Mar 2011 15:44:26 +0000 (15:44 +0000)]
Initial release:
* very alpha, but on 32-bit archs it runs
* network stack only, grabs an IPv6 on any network
* netdb not implemented, netinput should be improved