General suggestion related mainly to the DriverEntry routine about btrfs HOT 3 CLOSED

Thanks Martin.

• Yes, you're right, there should be some cleanup there! I'll sort it out for the next version.

• Are you certain about that? Its MSDN page simply says "ExInitializeResourceLite returns STATUS_SUCCESS." ReactOS, which generally implements such low-level functions quite faithfully, likewise has it never returning an error.

• Well, the only memory tracking the kernel does is the "tag" passed to ExAllocatePoolWithTag. Prefast I've found to be quite useful in identifying memory leaks and lock errors. There's quite a bit of progress in the latest release, but some day I'll get the whole driver tagged up for Prefast.

from btrfs.

Well, you are right about the ExInitializeResourceLite case. I somehow missed this and assumed that when it returns NTSTATUS, it may fail.

Well, the only memory tracking the kernel does is the "tag" passed to ExAllocatePoolWithTag. Prefast I've found to be quite useful in identifying memory leaks and lock errors. There's quite a bit of progress in the latest release, but some day I'll get the whole driver tagged up for Prefast.

That is why I decided to use my own wrappers around ExXxxPoolYyy. Of course, code analysis is also a good approach.

Did you tried your driver with Driver Verifier? That can also find some, potentially nasty, errors.

from btrfs.

Closing old issues

from btrfs.