On Thu, Nov 21, 2019 at 10:08:03PM +0800, Yama Modo wrote:
Hi Greg ,
Thanks for your response!
Greg KH greg@kroah.com於 2019年11月21日 週四，下午7:54寫道：
On Thu, Nov 21, 2019 at 07:47:37PM +0800, Yama Modo wrote:
Dear Linus Walleij,
I want to backport commit ff2b13592299 "gpio: make the gpiochip a real device" to linux 4.4.y. Could you please review the following patch? I will improve this later if something need to take care. Thanks!
Why do you want to do that? What does it "fix" and who needs it for the
old 4.4.y kernel tree?
Sorry I should take care about that patch must fix the issue directly and then it can be backported to stable kernel. We use Linux 4.4.y-cip for our MOXA products, such as UC8410A, for a super long term kernel, and it’s based on Linux 4.4.y.
If your device works properly on 4.9 or newer, you really really really should use that instead. That way you always have support, and proper fixes, unlike what will, and is already, happening with 4.4.y.
We suffer a kernel panic issue because of of_node inconsistency and gpio-reserved-ranges with no gpio device, and we find mainline、4.9.y、4.14.y and 4.19.y have no this gpio problem, and there are have gpio device.
Great, why not use 4.19.y? That was said to be the next "super long term kernel", right?
Many gpio patches are based on gpio device. If we want to backport these gpio patches to Linux 4.4.y-cip, it will suffer more problems than having gpio device’s kernels. Besides, in house patches in Linux 4.4.y-cip are harder to upstream mainline because subsystem of Linux 4.4.y-cip is far from mainline’s.
I have no idea what is in the -cip 4.4.y tree. You are on your own there. And honestly, I think the model of what they are trying to do there is totally wrong and will fail horribly :(
Use 4.19.y. Or even better yet, 5.4.y Your device is "alive", keep it alive by giving it the latest kernel updates so that you know it is in good health. To rely on an old kernel like that, you can not guarantee how alive it will be.
good luck!
greg k-h