- Ignore error when the device provides a string descriptor but no contents
- Fixes installation of cheap usb devices
svn path=/branches/usb-bringup-trunk/; revision=55514
- Add back old implementation for testing purposes (if'd out)
- Display more status field of queue head / transfer descriptor
- Dump queue head when a halted queue is detected
- Display data toggle status
svn path=/branches/usb-bringup-trunk/; revision=55502
- Don't wait for the work item to complete in the SCE completion handler
- Use an event to signal the completion of a reset and wait for that event during IRP_MN_START_DEVICE handling so devices are enumerated synchronously which allows USB boot devices to be enumerated
svn path=/branches/usb-bringup-trunk/; revision=55493
- Fix a few bulk transfer bugs
- Enable interrupt completion on last interrupt
- Increment transferred offset
- Mass storage device now gets further (passing SRB_FUNCTION_CLAIM_DEVICE)
svn path=/branches/usb-bringup-trunk/; revision=55491
- Rewrite the transfer descriptor handling
- Take into account the direction of the pid
- Revert previously added hacks
- Support now more than one transfer descriptor for control transfers
- Include the bulk transfers in one queue head which should speed up operation
- Store data toggle in usb endpoint structure
- Mass storage devices now passes the control status phase on real hw and enters bulk operation stage
svn path=/branches/usb-bringup-trunk/; revision=55489
- Don't set NumberOfTransaction 0x3 for the async queue head(Haiku doesnt do it too)
- Also don't set maximum packet length or interrupt on complete
- Stop the async / periodic schedules after acquiring ownership from os (ported from the Pedigree OS)
- Remove unused structs from header
- Allocate device descriptor from start of a page to prevent alignment issues
- Clear halted bit in the queue overlay transaction layer (usb_queue.cpp:507)
- Always set the alternative and next pointers in BuildControlTransferQueueHead
- Use data the toggle in the transfer requests
- Start using toggle in bulk requests, invert data toggle after each descriptor
- EHCI controller still reports error 0xA / 0xB which is USB Error Interrupt + Frame List Rollover + [USB Interrupt]
- Errors need to be researched more, though the control transfers results appear o.k.
- Tested in real hw + ros stack with USB Mass Storage Device
- Currently hangs after first completed CDB/Data/CSW block
svn path=/branches/usb-bringup-trunk/; revision=55483
- Wait for periodic scheduling to start
- Increase the wait interval for asynchronous scheduling to start
svn path=/branches/usb-bringup-trunk/; revision=55477
- Implement ResetController and call it before setting up the controller
- Don't disable 64-bit addressing until the BIOS gives up control of the controller
- Comment out manually clearing the SMI interrupt and the BIOS owned flag by request of janderwald
- EHCI no longer hangs for me on real hardware during controller initialization
svn path=/branches/usb-bringup-trunk/; revision=55450
- Check if attaching to device stack failed. In that case fail the request
[HIDUSB]
- Always set SupriseRemovalOk
[USBCCGP] [HIDCLASS] [USBHUB] [USBSTOR]
- Check if the pdo is still in the pdo list. Only call IoDeleteDevice in that case
svn path=/branches/usb-bringup-trunk/; revision=55429
- Add and fix missing exports
- Add UNIMPLEMENTED to unimplemented functions
- Fix completely broken USBD_GetPdoRegistryParameter (which wasn't even exported because somebody made a typo and couldn't figure it out)
- My Linksys USB300M driver now loads but it runs into some unimplemented code paths in usbehci
svn path=/branches/usb-bringup-trunk/; revision=55416
- Add HID devices to the critical device database
- Fixes issues with mouse/keyboard not working until new device wizards are dismissed (USB drives work too)
[UMPNPMGR]
- Fix install failure after the kernel detects a critical device
[NTOSKRNL]
- Finally enable the proper IopResetDevice code
- Driver updating without a reboot for running devices is supported now (if someone will write the GUI)
[USBEHCI][USBOHCI]
- Fix removal bugs
svn path=/branches/usb-bringup-trunk/; revision=55401
- Remove hack of forwarding pdo requests to lower device object of fdo
- Implement handling of reset port and cycle port
- IOCTL_INTERNAL_USB_RESET_PORT / IOCTL_INTERNAL_USB_CYCLE_PORT now need to be implemented in usbhub
svn path=/branches/usb-bringup-trunk/; revision=55396
- Check if the transfer descriptors reported errors and propagate the error in the urb status field
- OHCI now reports error which the class driver are handling
- Allocate device descriptor from non paged pool to prevent possible alignment problems
- Fix checking of completed transfer descriptors
- Fix double freeing of descriptors
- Cleanup endpoints when the halted bit is set by the host controllers
- Check for the endpoint direction in the data descriptor
- Setup descriptor needs buffer rounding set
svn path=/branches/usb-bringup-trunk/; revision=55389
- Use constants instead of hardcoding register offsets. Taken from Haiku
- Elements in HCSP-PORTROUTE are 4 bit entries instead
svn path=/branches/usb-bringup-trunk/; revision=55385
- In 2012 a long protected natural resource for bugs was shutdown due to popular usb device support demand
- Fix length check in USBD_ParseDescriptors. (PLONG)Descriptor + TotalLength is not the length of the configuration descriptor but twice the configuration descriptor
- Reset found variable in the loop. This fixes finding of interfaces which are not the first one in the configuration descriptor
- Again moving to next usb descriptor was broken. UsbDeviceDescriptor = UsbDeviceDescriptor + UsbDeviceDescriptor->bLength does not move to the next but to next XX descriprors specified by bLength
- Introduced in rev 17382 (14/08/2005)
svn path=/branches/usb-bringup-trunk/; revision=55370
- Wait 100 milliseconds for the port power and insertion to stabilize
- Wait for the 10 millisecond "recovery" period after reset
svn path=/branches/usb-bringup-trunk/; revision=55367
- Fix bugs in HandleClassEndpoint, HandleClassInterface which always set direction device to host instead of relying to the transfer flags provided
- Allow setup packets w/o a data stage
svn path=/branches/usb-bringup-trunk/; revision=55360
- Implement set idle request. The device should only send data when there is actual data. May also fix the mouse debug flood when mouse is ejected
- Not yet supported in OHCI
- Add missing constants
svn path=/branches/usb-bringup-trunk/; revision=55356
- Rewrite hidparser so that it support multiple top level collections
- Add support for flat collection context meaning all context structures are aligned in after each other. The reason this is required is that usermode hid.dll will access the same structure and thus can't use kernel pointers. User mode hid.dll is used to support keyboard with custom keys (for example search key, email key etc)
svn path=/branches/usb-bringup-trunk/; revision=55353
- Missed a fix from last commit
- Wait for the device status change work item to complete
- The USB stack is now properly constructed for USB booting
svn path=/branches/usb-bringup-trunk/; revision=55339
- Add function to append interface number
- Append function number to BusQueryDeviceId / BusQueryInstanceId
- USBCCGP is now properly enumerated and its child device start to get installed
svn path=/branches/usb-bringup-trunk/; revision=55326