On Mon, Nov 30, 2020 at 03:00:13PM +0100, Paolo Bonzini wrote:
On 30/11/20 14:57, Greg KH wrote:
Every patch should be "fixing a real issue"---even a new feature. But the larger the patch, the more the submitters and maintainers should be trusted rather than a bot. The line between feature and bugfix_sometimes_ is blurry, I would say that in this case it's not, and it makes me question how the bot decided that this patch would be acceptable for stable (which AFAIK is not something that can be answered).
I thought that earlier Sasha said that this patch was needed as a prerequisite patch for a later fix, right? If not, sorry, I've lost the train of thought in this thread...
Yeah---sorry I am replying to 22/33 but referring to 23/33, which is the one that in my opinion should not be blindly accepted for stable kernels without the agreement of the submitter or maintainer.
But it's not "blindly", right? I've sent this review mail over a week ago, and if it goes into the queue there will be at least two more emails going out to the author/maintainers.
During all this time it gets tested by various entities who do things that go beyond simple boot testing.
I'd argue that the backports we push in the stable tree sometimes get tested and reviewed better than the commits that land upstream.