NEWS: USB-C is coming to Lumia 950/XL

[The_GTA]

Everyone of us has been waiting for full USB-C support for the Lumia 950/XL ever since WoA came to light. Today gus33000 has posted a very promising video that showcases proper recognition of a plugged-in USB-C flash drive.



With this exciting news it is very likely that HDMI-out support through the display dock will be available in the next iteration of gus33000's driver pack! After the incredible resurfacing of cellular support past WoA build 18908, who would have seen this coming?

Thorough investigation has revealed that this has been a cooperation between Github's user Googulator and gus33000, who has quite possibly provided the illuminous Windows 10 ARM build 15035 along with ARM32 USB-C drivers to Googular who has then fixed his driver.

Look out for the next driver pack due by Christmas.

DOWNLOAD LINK FOR WINDOWS 10 ARM 32BIT BUILD 15035 (originally found here)

[gilius]

Is the ARM 32 build different to early Windows S versions? Can it run x86 apps for example?

Re: USB-C, I'm wondering if it's already available in the latest driver pack? I came across this:

USB C support is not finished. To get power out (VBUS) on your device, follow these steps:

i. Open regedit

ii. Navigate to HKEY_LOCAL_MACHINE\SYSTEM\usbc

iii. Create a new DWORD 32-bit value named: VbusEnable

1. Set it to 0 to disable VBUS

2. Set it to 1 to enable VBUS

iv. Reboot the device with no charger plugged in

Important: Do not plug an external power source when VBUS is enabled on your device. The device may charge by itself when VBUS is enabled, you can revert the Charging Arbitration Driver to the build version to prevent this but you'll break charging.

- HDMI out is not functional

[The_GTA]

Is the ARM 32 build different to early Windows S versions? Can it run x86 apps for example?

I doubt that MSFT had the x86 emulation layer finished by then.

Re: USB-C, I'm wondering if it's already available in the latest driver pack?

Have not been able to test because I've been very busy lately solving a math problem. But to encourage you to try it out, here is a recent commit by gus33000 that mentions the Vbus stuff: https://github.com/WOA-Project/LumiaUSB ... 77b392efb1

He probably disabled the registry thing from doing anything. I could be wrong though.

[gilius]

Maths problem - sounds exciting! I spent several years studying Ramsey theory to try to beat the casino - but have kind of given up now.

The driver appears to be the updated version:
https://github.com/WOA-Project/Specific ... ivers/USBC

[The_GTA]

UPDATE: imbushuo has just posted about a "clean-room implementation" of the Lattice UC120 USB FPGA driver on his twitter:



It is suspected that this driver is a missing cornerstone in full USB-C support for both Lumia 950 and Lumia 950 XL. Stay tuned for more information on this recent revelation!

[The_GTA]

Managed to compile and install this new driver. Check out my tweet.



What I can confirm thus far is that the driver is capable of charging the phone. Will have to do more testing to verify the full potential of said driver.

EDIT: the released driver by imbushuo turns out to be a weaker version from the already created LumiaUSBC driver. Both drivers are currently incapable of complete USB-C support. Conclusion: we have gained nothing.

[The_GTA]

Gustave Monce (gus33000) has managed to test the latest USB-C-related drivers by imbushuo and reports that they work flawlessly:



By taking a look at the latest work by imbushuo on his dev.azure.com workspace he appears to work on a 32bit ARM kernel mode driver loader that would put the official HDMI/DPI interface driver by Microsoft/Qualcomm into use. Are we finally getting close to HDMI output for the Lumia 950/XL?

// KmCodeLoader.c: the land of DreamLifterK magic
// What it does:
// 1. Load certain arm32 kernel-mode driver code in userland (customized PE loader).
// 2. Fix up IAT and reloc table (we do not care about .pdata yet, km drivers are unlikely
// to do exception unwind, and even the worst case in um is process crash)
// 3. Kick driver in, connect to the km IO proxy driver
// 4. **Popcorn**
//
// Generally speaking, the km code needs to be compliant with a few constraints:
// - Standard MS ARMNT calling convention
// - No direct memory access or MMIO operations (technically we can map memory twice though)
// - No non-passive level interrupt (same as UMDF)
// - No interface query to other drivers, or providing interface to drivers
// - No exception unwind (yet)
//
// As investigated, HDMIDPI.sys is compliant with these constraints, and therefore it is chosen
// to be an experiment subject.

Source: https://dev.azure.com/LumiaWoA/USB/_git ... deLoader.c

[The_GTA]