On Fri, 13 Jan 2017 14:47:35 +0100, Andrei Borzenkov wrote:
12.01.2017 21:46, Stefan Seyfried пишет:
Am 12.01.2017 um 18:09 schrieb Andrei Borzenkov:
Suppose new kernel causes user space breakage (not something unheard of)
Care to give an example? Not just hearsay?
Just complain to Mr. Torvalds and he'll have the committer of the breakage fix the code.
What might sometimes be needed could be an update of supporting tools (dracut or such), but these are rare cases in my experience.
Yes, I do mean low level tools like dracut, udev, systemd. They are usually sensitive to kernel changes and need to adapt to newer versions (not other way round).
Well, you still didn't give the example that actually happened... Usually, it's the other way round: when a newer kernel provides a new feature and a user-space program wants to use it eagerly, such a program needs the update. But the old binary runs as is if it doesn't need a new feature. (Otherwise, how many people could use Kernel:stable on Leap and older distros...?) And, a drastic kernel change that causes incompatibility is enabled usually only via a specific kernel config. If a new kernel forces incompatibility inevitably, it's a clear regression, and we must report back it as a bug. In a long term, we see incompatibility sometimes, yes. For TW, we tend to enable such a new config as soon as provided. It's fine for a rolling release like TW. But it doesn't mean that we have to turn it on always for Leap as well. If it brings incompatibility, we can just skip such a config for the Leap version of Kernel:stable. The incompatible switch may happen at the next Leap version up together with user-space updates, while compatibility can be kept during the same Leap version deployment, at least. Takashi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org