[FFADO-devel] Mackie 1620i up and running (with some problems)

[Holger D. <deh...@ah...>]

This mail is also sent to tech people at mackie/loud, but I think there are 
problems with FFADO as well as with the firmware, so please don't just ignore 
it.


After a lot of testing and try and error I finally got my Mackie mixer running. 
I have fond solutions without knowning the problem exactly.
I will show you the diffs with some comments:

Index: src/libstreaming/amdtp-oxford/AmdtpOxfordReceiveStreamProcessor.cpp
===================================================================
--- src/libstreaming/amdtp-oxford/AmdtpOxfordReceiveStreamProcessor.cpp 
(Revision 1942)
+++ src/libstreaming/amdtp-oxford/AmdtpOxfordReceiveStreamProcessor.cpp 
(Arbeitskopie)
@@ -87,7 +87,7 @@
     // than one packet
     FFADO_ASSERT( m_temp_buffer == NULL );
     if( !(m_temp_buffer = ffado_ringbuffer_create(
-            packet_payload_size_events * 4 * 4))) {
+            (packet_payload_size_events + 50) * 4 * 4))) {
         debugFatal("Could not allocate memory event ringbuffer\n");
         return false;
     }

The calculated buffer size is too small: I don't know why, it might or might 
not have something to do with the following modification.
Without the extra size, StreamProcessor tries to add more bytes to the buffer 
to fill a complete packet, but there ist no space...


Index: src/libavc/general/avc_plug.cpp
===================================================================
--- src/libavc/general/avc_plug.cpp     (Revision 1942)
+++ src/libavc/general/avc_plug.cpp     (Arbeitskopie)
@@ -235,7 +235,7 @@
                     m_unit->getConfigRom().getNodeId(), getSubunitType(), 
getSubunitId(), m_direction, m_id );
 //         return false;
     }
-
+    if( m_infoPlugType != AVC::Plug::eAPT_Sync ){
     if ( !discoverSupportedStreamFormats() ) {
         debugOutput(DEBUG_LEVEL_NORMAL, 
                     "Could not discover supported stream formats "
@@ -243,7 +243,7 @@
                     m_unit->getConfigRom().getNodeId(), getSubunitType(), 
getSubunitId(), m_direction, m_id );
 //         return false;
     }
-
+    }
     return m_unit->getPlugManager().addPlug( *this );
 }


This is the problem I described in my firat mail.
When plugtype is eAPT_Sync, the bus / device is blocked and only a hard reset 
makes the bus respond again (by either unplugging the firewire cable or 
switching on and off the device)


Index: configuration
===================================================================
--- configuration       (Revision 1942)
+++ configuration       (Arbeitskopie)
@@ -17,6 +17,13 @@
     xmit_max_cycles_early_transmit = 4;
 },
 {
+    vendorid    = 0x000ff2;
+    modelid     = 0x081216;
+    vendorname  = "Loud Technologies Inc.";
+    modelname   = "Onyx-i";
+    driver      = 4; # Oxford
+},
+{
     vendorid    = 0x0003db;
     modelid     = 0x00010048;
     vendorname  = "Apogee Electronics";

This does not need any comment - does it?


Problems so far: 
- a few XRuns
- somtimes noise after stopping and starting a song
- Rosegarden is killed when I try to activate more than 4 input channels
- qtractor does not start at all

Maybe others will occur.

I would be glad if someone could comment on this.

Holger