RE: SPAM-LOW: performance of Registry access

You could expose two IO control codes, one for transferring this data from
the driver to an app, and another to transfer it from the app to the driver.
Then, you could have a user-mode app (NT service) which handles periodically
reading the data from the driver and write it to [file | registry | SQL |
whatever], and loading it on system/service start. This would be quite
flexible, and would require the minimum amount of kernel-mode work and

– arlie

-----Original Message-----
[] On Behalf Of Harish Arora
Sent: Thursday, August 04, 2005 3:21 PM
To: Windows System Software Devs Interest List
Subject: SPAM-LOW: [ntdev] performance of Registry access

I have an upper filter to volume devices and I need to maintain a DRL of
writes that come into the device. I have a fixed sized map for doing this.
I would like to persist this map so that I can track it across boots.
Options are registry or a file. If I persist in registry, what is the kind
of performance hit that I am looking at? And how would it compare to storing
it in a file.
I understand that a lot depends on the policy of how frequently I store the
map in registry|file. But lets take the extreme case that I store map to
registry for every write to a new region for which I find that bit was
previously clear and is now being set. What kind of impact will the writes
have if I do this.

Other extreme would be to store it when I see a shutdown notitification
(IRP_MJ_SHUTDOWN) but this is not safe enough because OS might crash and I
will loose all the DRL information.


If art interprets our dreams, the computer executes them in the guise of

Questions? First check the Kernel Driver FAQ at

You are currently subscribed to ntdev as: unknown lmsubst tag argument: ‘’
To unsubscribe send a blank email to