I used it in the mainline kernel without troubles. This is a a very good
job, thanks.
Ok, thanks for the answer. May be I will work on it in the next few
months. Let's see. I'll keep you informed.
Regards.
On 18/10/2019 16:53, Oleksij Rempel wrote:
Hi,
On Fri, Oct 18, 2019 at 12:07:34PM +0200, Romain Forlot [IoT.bzh] wrote:
Hi,
I am wondering what the cost is to backport the j1939 module to an old
version like a v4.14 LTSI version.
The backport should be quite easy:
git cherry-pick -sx `git rev-list --reverse 2c1f9e26344483e2c74e80ef708d9c7fd2e543f4..9d71dd0c70099914fcd063135da3c580865e924c`
...but it isn't :/ Some CAN patches are missing. We'll backport the stack to
v4.14.150 (or newer) and send a follow up mail.
However, the driver for the CAN adapter needs proper RX/TX frame ordering,
otherwise the stack will explode.
This is fixed in flexcan mainline. And involves a handful of patches. Other
drivers probably need more fixing. Some CAN hardware may even lack the hardware
support for proper ordering, that is time stamping registers.
And what the impact is of backporting the whole CAN stack on the CAN drivers?
The stack has no impact on the drivers, but requirements on proper RX/TX
ordering, see above.
Are there any modifications to drivers once the CAN stack is updated ?
Yes, as long as they don't have proper RX/TX ordering.
So, which CAN driver are you planing to use?
Regards,
Oleksij & Marc
--
Romain Forlot - Embedded Engineer - IoT.bzh
romain.forlot@xxxxxxx - www.iot.bzh - +33675142438
