Make Endpoint control transfers more reliable; early-abort when an incomplete packet is sent from the host (indicating end of transfer), add check for control reads to ensure that no more than wLength bytes are read from the interface.

Condense sequential printf_P/puts_P calls to single printf_P calls for size and clarity.

Fixed issue in USBtoSerial demos where the USART Rx line was not having its pullup enabled at startup, leading to noise being received from the Rx pin when left floating.

Removed unused relative Common.h header include from the DriverStubs board stub drivers.

Fixed Benito programmer, use the correct LED functions where needed to ensure correct status display at all times.

Make Control Endpoint stream transfers more reliable by adding in early aborts for unexpected new SETUP tokens, or unexpected status stage during control stream writes.

Fix corruption in Device RNDIS demos TCP stack when too many connections attempted simultaneously, freezing the device when a page was re-fetched before the first connection was closed.

Fix incorrect model compatibility information in the Host LowLevel demo overview text files.

Changed over manual loops waiting for endpoints to be ready to use the library Endpoint_WaitUntilReady() function for robustness. Fixes issues with terminated transfers on the host locking up USB devices.

Renamed all library events to properly seperate out Device and Host mode events. Changed the firing conditions for some events to ensure that events are fired by their own USB mode only.

Remove VBUS events - not needed as the library takes care of VBUS detection and feedback on supported AVRs via the USB_Device_Connected and USB_Device_Disconnected events.

Fixed incorrect Host state assignment in the incomplete BluetoothHost demo.

Better fix for LowLevel CDC demo issue where sending data before the line encoding is set will cause a failure of the host to open the port. Added fix to the DualCDC and USBtoSerial LowLevel demos.

Fix issue with CDC device demos causing broken communications when the device tries to send data before the host has set the line encoding.

Renamed ATTR_NOINLINE to ATTR_NO_INLINE to fit with the rest of the library function attribute names.

Rename TOTAL_NUM_CONFIGURATIONS to FIXED_NUM_CONFIGURATIONS, to match the existing FIXED_CONTROL_ENDPOINT_SIZE compile time option. Add FIXED_NUM_CONFIGURATIONS to the descriptors of each project using the option.

Make both LowLevel and ClassDriver AudioOutput demos always output to board LEDs, regardless of output mode (removed AUDIO_OUT_LEDS compile time option from the demo). This allows for positive feedback to the user of audio activity regardless of output mode.

Removed interrupt disable/enable from the Device mode control request processing routines, as they can cause problems with user code and can cause infinite spinlocks under some circumstances.

Added new LUFA logo from EDIGMA.COM.

Fixed the ClassDriver GenericHID demo so that it echoes back reports as indicated in the demo overview.

Fix demos based on the device mode HID class driver, as well as the driver itself. Changed HID device class driver to require the user to give a buffer and size to hold the previously generated report, for comparison purposes, and altered the prototype of the CALLBACK_HID_Device_CreateHIDReport() function so that reports can be sent to the host even if there are no apparent changes (useful for relative movements in mice, etc.).

Renamed USB_Device_SetHighSpeed() to USB_Device_SetFullSpeed() so that the correct terminology is used (thanks to Brian Dickman).

Fix USB_PLL_Off() call in the series 4, 6 and 7 microcontroller disconnect interrupt handler code, which wasn't guarded by a test of USB_Options to check if the user has specified manual PLL control (thanks to Brian Dickman).

Add new HID_Device_MillisecondElapsed() function to the HID device Class driver, to move the burden of managing the Idle period of each instance to the library and not the user application.

Ensure that the previous HID reports in the HID device class driver are kept per-instance, rather than per-device.

Fix to HID device mode Class driver, so that new reports are compared against the old, and updated reports made within the idle period are sent immediately to the host.

Make all audio device Class Driver functions require a reference to the Audio class instance, so that the endpoint size can be checked (and to standardize the APIs).

Make LEDs more responsive in the AudioOutput device demos.

Set all CDC and MassStorage device mode demos (LowLevel and ClassDriver) to use the USE_INTERNAL_SERIAL feature, add compile time warnings for devices which do not have an internal serial.

Fix incorrect HWB button mask in the STK526 Buttons driver.

Fix redefinition warning when USB_DEVICE_ONLY is set on an AVR which does not support USB host mode.

Extend the automatic serial number descriptor code to read out and send all 10 bytes (20 characters) of the internal serial number on supported AVRs, rather than just the first 6 bytes (12 characters).

Fixed report data alignment issues in the MouseHostWithParser demo when X and Y movement data size is not a multiple of 8 bits.

Fixed HID Report Descriptor Parser not correctly resetting internal states when a REPORT ID element is encountered.

Change project makefiles so that the current target settings and not just the board selection is printed during the build process.

Fix warning in AudioOutput demos when AUDIO_OUT_STEREO output mode is selected.

Better solution - change compile time token names to DEVICE_STATE_AS_GPIOR and HOST_STATE_AS_GPIOR and allow the user to set them to the index of the GPIOR register within the device to use.

Added HOST_STATE_AS_GPIOR1 and DEVICE_STATE_AS_GPIOR0 compile time options, the latter of which is used in the bootloaders, to reduce code size.

Make LowLevel AudioOutput demo work in the same way as the ClassDriver AudioOutput demo for LEDs output.

Changed AudioOutput demos to explicitly use timer 3 - the smaller USB AVRs where timer 1 was needed instead did not have the endpoint size neccesary for good audio throughput anyway.

Fix Benito documentation indicating erronously that the project used the HID USB class instead of the CDC class.

State information for class drivers is now zeroed out during enumeration (both in device and host mode) to ensure sane values after each enumeration. User code should no longer explicitly set state information as this is no longer preserved.

Collapse configuration descriptor size retrieval and size testing into a single if statement within the new host mode class driver demos for clarity.

Change Host mode class driver Pipe configuration routines -- better to let the application fetch the configuration descriptors once, and then pass them to each class driver in turn.

Add new MouseHost Class Driver demo, unfinished, to allow for better testing of the new prototype Host Mode Class Driver framework. Flesh out start of a HID class Host mode class driver.