[2/11]hidclass.sys: add a wine native hid device driver service

[Aric Stewart]

Allow for Minidriver registration
---
 configure                           |  2 +
 configure.ac                        |  1 +
 dlls/hidclass.sys/Makefile.in       |  8 ++++
 dlls/hidclass.sys/hid.h             | 53 +++++++++++++++++++++++++
 dlls/hidclass.sys/hidclass.sys.spec |  2 +
 dlls/hidclass.sys/main.c            | 78 +++++++++++++++++++++++++++++++++++++
 loader/wine.inf.in                  | 13 +++++++
 7 files changed, 157 insertions(+)
 create mode 100644 dlls/hidclass.sys/Makefile.in
 create mode 100644 dlls/hidclass.sys/hid.h
 create mode 100644 dlls/hidclass.sys/hidclass.sys.spec
 create mode 100644 dlls/hidclass.sys/main.c

[Thomas Faber]

I'm assuming you know that hidclass.sys is just library on Windows and
intentionally made your version its own service to simplify things.
If that's not true I can elaborate.

On 2015-06-22 03:24, Aric Stewart wrote:
> --- /dev/null
> +++ b/dlls/hidclass.sys/main.c
> +NTSTATUS WINAPI HidRegisterMinidriver(PHID_MINIDRIVER_REGISTRATION registration)
> +{
> - minidriver *driver;
> - driver = HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, sizeof(*driver));
> -
> - if (!driver)
> -
>    return STATUS_NO_MEMORY;
> -
> - driver->minidriver = registration;

It's a bit unusual to assume that the registration structure will live
longer than the scope of this function.
Fine for the built-in driver you have but this would certainly have to
copy the members if it were to support "real" minidrivers.
> - list_add_tail(&minidriver_list, &driver->entry);
> -
> - return STATUS_SUCCESS;
>
> +}

[Aric Stewart]

On 6/22/15 5:44 AM, Thomas Faber wrote:
> I'm assuming you know that hidclass.sys is just library on Windows and
> intentionally made your version its own service to simplify things.
> If that's not true I can elaborate.

I was just starting to come to the awareness of this, thanks for confirming it, and was debating if i wanted to change my architecture a bit.

I intentionally made its own service to handle the builtin minidrivers and help get around the lack of any Plug-and-play system. But I was trying to explore if there was an easy way to make it both so that future possible "real" minidrivers could load it as expected.

I think there is. But have not done the work in that regard as there are probably many many other problems with "real" minidrivers anyway.

-aric