On OSX, device updating fails after some time

[colinrgodsey]

In regards to my post on the board: https://groups.google.com/forum/#!topic/usb4java/qrcz4H9n_jY

More information is provided there, but basically on OSX after running the software for a while (with some device failures and reinits happening every hour). After a while the device list seems to stop getting updated, and i cannot open an interface to the device. Oddly enough, when walking the device tree at this point, it is never updated. Unplugging the old device, plugging in a new one, etc. still returns the same list of devices. I'm still getting callbacks from control requests, but obviously with no device it does nothing. Trying to send data, I get a failure that it cant find the interface. This sometimes results in a segfault (attached).

Moving the software over to linux, everything works fine. I may try compiling my own OSX native driver and see if that helps.

[colinrgodsey]

Stack: [0x000000010edbf000,0x000000010eebf000], sp=0x000000010eebe520, free space=1021k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libusb4java.dylib+0xb320] libusb_cancel_transfer+0xe0
C [libusb4java.dylib+0x11a35] libusb_hotplug_deregister_callback+0x4c95
C [libusb4java.dylib+0xbed5] libusb_handle_events_timeout_completed+0x675
C [libusb4java.dylib+0xbaa9] libusb_handle_events_timeout_completed+0x249
C [libusb4java.dylib+0xc066] libusb_handle_events_completed+0x26
C [libusb4java.dylib+0xc82b] libusb_control_transfer+0x23b
C [libusb4java.dylib+0xc6ec] libusb_control_transfer+0xfc
C [libusb4java.dylib+0x339d] Java_org_usb4java_LibUsb_controlTransfer+0x9d
j org.usb4java.LibUsb.controlTransfer(Lorg/usb4java/DeviceHandle;BBSSLjava/nio/ByteBuffer;J)I+0
j org.usb4java.javax.AbstractIrpQueue.processControlIrp(Ljavax/usb/UsbControlIrp;)V+80
j org.usb4java.javax.ControlIrpQueue.processIrp(Ljavax/usb/UsbControlIrp;)V+2
j org.usb4java.javax.ControlIrpQueue.processIrp(Ljavax/usb/UsbIrp;)V+5
j org.usb4java.javax.AbstractIrpQueue.process()V+31
J java.lang.Thread.run()V
V [libjvm.dylib+0x288afb]
V [libjvm.dylib+0x288fd8]
V [libjvm.dylib+0x289112]
V [libjvm.dylib+0x2bfa8c]
V [libjvm.dylib+0x478694]
V [libjvm.dylib+0x479b7a]
V [libjvm.dylib+0x3b5595]
C [libsystem_pthread.dylib+0x1899] _pthread_body+0x8a
C [libsystem_pthread.dylib+0x172a] _pthread_struct_init+0x0
C [libsystem_pthread.dylib+0x5fc9] thread_start+0xd

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j org.usb4java.LibUsb.controlTransfer(Lorg/usb4java/DeviceHandle;BBSSLjava/nio/ByteBuffer;J)I+0
j org.usb4java.javax.AbstractIrpQueue.processControlIrp(Ljavax/usb/UsbControlIrp;)V+80
j org.usb4java.javax.ControlIrpQueue.processIrp(Ljavax/usb/UsbControlIrp;)V+2
j org.usb4java.javax.ControlIrpQueue.processIrp(Ljavax/usb/UsbIrp;)V+5
j org.usb4java.javax.AbstractIrpQueue.process()V+31
J java.lang.Thread.run()V
v ~StubRoutines::call_stub

[kayahr]

This sounds more like a problem in libusb. If you have some C experience maybe you can try to reproduce this problem with a small C program which directly uses libusb?

[colinrgodsey]

hmm, actually I'm going to close this. There's a few more things I need to weed out. I'm going to try a new native driver build from my mac, and im going to try another mac. I just remembered I'm on a hackintosh, so I wouldn't be surprised if that's feeding into this.

I can also use the original device software (cgminer, in C, uses libusb directly) on this machine and see if that fails too. I think I may have been a little pre-emptive creating this issue.