Enhancement: Support for recognizing Windows SAL annotations on functions

[rahul-ramadas]

In Windows driver code, it is common to have SAL annotations on functions like the following:

#include "pch.h"

_IRQL_requires_max_(PASSIVE_LEVEL)
NTSTATUS
DoSomething (
    _In_ UINT A
    )
{
    return STATUS_SUCCESS;
}

In Sublime Text 3, when I try to list the functions using Ctrl-R, it incorrectly lists the function name as _IRQL_requires_max_. I was wondering if you could enhance your language definition to recognize functions correctly in the presence of these annotations.

[abusalimov]

Well, in general case there is no way for ST syntax parser to distinguish _IRQL_requires_max_(PASSIVE_LEVEL) line from DoSomething (: both of them look like a function declaration/definition. However, I could try to filter out certain keywords like _IRQL_xxx to prevent them from being treated as functions. Would it be OK?

[rahul-ramadas]

Makes sense. Although I don't know how far you want to go though. There are a lot of annotations. There are two headers in the Windows Driver Development Kit that define all of them (no_sal2.h and sal.h). I can give you a comprehensive list if you want.

As you can see in the example above, there are also annotations on the function parameters, but those don't seem to be causing any issues right now.

The alternative is if you can somehow have a settings file that allows people to specify their own "ignore" stuff. I don't know much about syntax development, but I'm guessing that might be a little tricky.

[abusalimov]

Specifying a list of custom support keywords would be a perfect solution, but I don't know if it is possible. Probably there should be a way to generate syntax file on the fly using some settings file, but I'm afraid I have no plans of investigating such magic in a near future, sorry.
I'll try to catch these (hardcoded) annotations for now. Could you please provide links to headers that you've mentioned?

[abusalimov]

@rahul-ramadas ping

[rahul-ramadas]

I'll get back to this later. I've had second thoughts about hardcoding the list into the syntax definition. There are quite a few such annotations with different forms. Let me put some more thought into this. You can close this for now if you like.