Nallasellan, Singaravelan wrote:
Arjan,
Thanks for the response.
Our driver has to reduce the time required to transfer the data as minimal as
possible.
We implemented mmap to reduce the overhead due to copy_from_user in kernel.
mmap tends to have quite a bit of overhead as well.... for small amounts of
data the tradeoff might be different than you expect ;-)
How will the application communicate with the driver to inform the details
buffer content such as the base address and the length?
that's up to you to be honest; you're doing a non-standard interface, you need
to then deal with all this.
Is there a way to obtain the mmapped address from the user space buffer pointer
passed in write?
You can do an O_DIRECT like write, so that you get the benefit of using write,
without the copy overhead.
It's a little complex, and generally only used for things like infinibad where
performance really matters...
if you only have low data rates I wonder if the copy really is too much pain.
_______________________________________________
Moblin dev Mailing List
[email protected]
To manage or unsubscribe from this mailing list visit:
https://lists.moblin.org/mailman/listinfo/dev or your user account on
http://moblin.org once logged in.
For more information on the Moblin Developer Mailing lists visit:
http://moblin.org/community/mailing-lists
