Hi Oleksandr,
On 21/08/17 16:53, Oleksandr Tyshchenko wrote:
On Thu, Aug 10, 2017 at 6:13 PM, Julien Grall <julien.gr...@arm.com> wrote:
On 10/08/17 15:27, Oleksandr Tyshchenko wrote:
I would like to clarify what need to be done from my side.
Should I wait for the missing things reach upsteam and then rebase on
the mainline driver?
Or should I rewrite this driver without following Linux?
I don't have a clear answer here. As I said, we need to weight pros and cons
to use Linux driver over our own.
At the moment, you are using a BSP driver which has more features but
modified quite a lot. We don't even know when this is going to be merged in
Linux.
Keeping code close to Linux requires some hacks that are acceptable if you
can benefits from the community (bug fix, review...). As the driver is taken
from the BSP, we don't know if the code will stay in the current form nor be
able to get bug fix.
I got it. Completely agree with you.
But, we need to choose which direction we should follow. We have 3
options at the moment
and I am OK with each of them:
1. direct port from BSP (current implementation).
2. direct port from mainline Linux (when it has required support).
3. new driver based on BSP/Linux and contains only relevant to Xen things.
I am starting to think that options 2 or 3 (+) would be more suitable.
What do you think?
The option 2 rely on the changes to be merged in Linux. If I understand
correctly, we don't have any timeline for this.
So I would lean towards option 3 to get a support in Xen.
Stefano, do you have any opinion?
Cheers,
--
Julien Grall
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
