design reason for the header install method?

[M1cha]

Is there a reason that we only install the headers needed by syscalls.h and host_ops.h?
When using syscalls(in my case ioctl's) it's important to have access to all the uapi stuff.

Wouldn't it make more sense to use the default kernel header install method and then just move everything into the lkl subdirectory and recursively apply the patches?

[tavip]

The main reason is that the "parser" we are using the avoid name collisions with the native environment in case of Linux (e.g. struct iocb, struct ethaddr, etc) is kind of a joke. So until we use something more robust, we use this incremental approach were we improve and add support for more complex parsing as we add new headers.