cpvnm v1.2(7) in Win VM crashing

[TabSel]

I just installed v1.3(1) again in order to check.

It does in fact not work: cpvnm memory usage does NOT raise over time, but cpu load is quite high (3-4%) with nothing going on. In the CP Manager on the VM host, everything looks ok, there are MIDI settings, CP to MIDI, MIDI to CP correctly displayed for all machines, even the VM Win machine. However, in the CP Manager in the Win VM, devices are created and destroyed all over the place. At least the CPVNM did not crash.

I installed 1.3(2) again. On the VM Host side, everything is correctly displayed in the CP Manager. CPVNM cpu usage in the Win VM is quite high (5-7%), memory usage raises. The CP Manager does not show the the devices correctly. It seems as if the Midi settings, CP to MIDI, MIDI to CP devices are linked to the wrong root devices. The "SM-1" (desktop PC) root for example does NOT show any sub devices, whereas the MBP-1-WIN (Win VM) devices shows multiple "midi settings", "cp to MIDI" "MIDI to CP"... selecting different root devices alters the sub devices - they get more and more and completely messed up regarding their order...

It's a pity, because I am developing a "audio over copperlan" solution, hosting ASIO devices on the Win side and/or CoreAudio devices on the Mac side, with the "side effect" of reducing latency for audio back and forth between VM Host and VM guest as the net is much faster as emulating a sound card...

[CopperPhil]

Yes I remember your project, it is very interesting and I can't wait to see a demo

What you describe makes me thinking about a mess in the device enumeration. I can't explain how sub-devices could appear behind the wrong root devices since they are hosted by the same CHAI by design. It looks like the network is continuously changing, putting a mess in the VNM routing tables.

Can you launch the CopperLan Manager from command line with the "-developer" argument? Then you'll see a log on the right showing all the CopperLan PnP events and warnings. If the network is blinking you'll see there a lot of device/network change notifications.

[TabSel]

I started the Win VM, launched Win VM CP Manager, set ethernet device to "-none-", closed CP Manager, stopped CPVNM service.

Then:
1) ran CPManager -developer
2) started CPVNM service

15:20:53:742;DDcl;[B 0x0001, M 0x0001, D 0x00001fcc, m 0x0000] DDNT_DeviceModified VNMManager's CHAI on MBP-1-WIN
15:20:53:746;DDcl;[B 0x0001, M 0x0002, D 0xd0b3c3d3, m 0x0000] DDNT_DeviceModified MIDI on MBP-1-WIN
15:20:54:313;CN;CNT_Connected
15:20:55:427;CN;CNT_NetworkStable

after a few seconds, I set eth interface to the VM ethernet ifc:

15:21:25:545;CN;CNT_NetworkChanged
15:21:26:634;CN;CNT_NetworkStable
15:21:27:086;CN;CNT_NetworkChanged
15:21:27:696;DDcl;[B 0x0001, M 0x0001, D 0x00001145, m 0x0000] DDNT_DeviceRedeclared VNMManager's CHAI on SM-1
15:21:27:697;DDcl;[B 0x0001, M 0x0001, D 0x00001fcc, m 0x0000] DDNT_DeviceRedeclared VNMManager's CHAI on MBP-1-WIN
15:21:27:748;DDcl;[B 0x0001, M 0x0002, D 0xd0b3c3d3, m 0x0000] DDNT_DeviceRedeclared MIDI on MBP-1-WIN
15:21:27:799;DDcl;[B 0x0001, M 0x0002, D 0x40b732a5, m 0x0000] DDNT_DeviceRedeclared MIDI on SM-1
15:21:27:939;DDcl;[B 0x0001, M 0x0001, D 0x000021ec, m 0x0000] DDNT_DeviceRedeclared VNMManager's CHAI on MBP-1
15:21:28:063;DDcl;[B 0x0001, M 0x0002, D 0xdb38dc91, m 0x0000] DDNT_DeviceRedeclared MIDI on MBP-1
15:21:29:145;CN;CNT_NetworkStable
15:21:55:448;W;[B 0x0001, M 0x0001, D 0x00001fcc, m 0x0000, # 0x0000] 3000 Trashcan issue [VNMManager] (Trashcan seems to be growing out of control (5681 objects). Please check application source code for missing Release() on hooked object!)
15:22:28:060;CN;CNT_Disconnected

Before it crashed (15:22:28:060), I clicked on the SM-1 device in config panel and saw "nothing" whereas when I clicked on MBP-1 I saw multiple "midi settings" etc. entries. I saw some devices blinking I believe (meaning created/destroyed immediately)

[TabSel]

I guess the red numbers within the root devices are response times?

I noticed that MBP-1-WIN shows reap times in the range of 0.200-0.210ms, constantly changing, the SM-1 shows 0.500-0.800ms rapidly changing and MBP-1 doesn't show any number at all.

sub devices are constantly blinking without redeclarations logged...

[CopperPhil]

Hmmmm.... interesting. Especially the VNMManager trashcan filled with 5681 items... that is absolutely not normal. And blinking sub-devices without notification log is also a never seen before symptom.

We'll try to figure out why...

But the really strange thing is this issue seems to be solved if you disconnect the host Mac CopperLan from the network... so it is definitely related to the network connection. Network change notification is sent to the application after at least 500ms of stability, maybe you don't see any notification because the network is continuously changing, and that could explain bad global behavior.

[TabSel]

I disconnected the host Mac Copperlan from the network and launched CPManager -developer on Win VM.

Usually the MAC address is displayed bottom left, but I only see MAC: 00 00 00 00 00 00....

[CopperPhil]

Ok, so in this mode the VM CopperLan engine is not able to determine the right Mac address.

Do you mean that the Win VM CPManager is showing a normal Mac address when the host Mac CopperLan is online? If so, it means that the VM CPManager is unable to differentiate its VNM from the host Mac VNM.

Can you check that the CopperLan Manager is showing different Mac addresses in the host MacOSX and in the Win VM when both are connected to the network?

[TabSel]

Both the Mac MBP-1 and the Desktop SM-1 root devices info show their valid MAC addresses.
The MBP-1-WIN root devices info shows MAC address 00 00 00 00 00 00, no matter the Mac VM Host is connected or not...

[TabSel]

Windows does determine and show the correct MAC address set by the VM host though...

[CopperPhil]

well... it seems that there is an incompatibility between the CopperLan over Ethernet driver and the Windows driver for the bridged network interface...

Can you check that the Windows NIC driver is well NDIS 6.0 compliant? This is the difference between the "CopperLan for XP32" and "CopperLan for Windows" distribution. The XP32 version is NDIS 5 and other Windows version is NDIS 6.