1 |
On 05/07/18 06:16, Alice Ferrazzi wrote: |
2 |
> Hello everyone, |
3 |
> |
4 |
> I would like to start the first Gentoo kernel meeting. |
5 |
> Please choose the time of the meeting [1] |
6 |
> If the time is not compatible we can also just discuss it by mail. |
7 |
> |
8 |
> Please send any new agenda items as new threads to the gentoo-kernel list. |
9 |
> |
10 |
> The current agenda is as follows. |
11 |
> - Gentoo kernel ci |
12 |
> - Use Lava qemu for testing the kernel |
13 |
> - Add kselftest |
14 |
> - Start stabilizing with Gentoo kernel CI |
15 |
> - Add ck-sources to the qemu test |
16 |
> - Stabilization |
17 |
> - Automatize the stabilization process |
18 |
> - keep up with upstream or use any other stabilization system |
19 |
> current stabilization system is defined here [2] |
20 |
> and here is the process to stabilize the kernel [3] |
21 |
> As now we are working to automatize the process. |
22 |
> |
23 |
> Please add any other topics, is also possible to discuss directly in |
24 |
> this thread. |
25 |
> |
26 |
> [1] https://doodle.com/poll/6giztg2vuw8wz6kf |
27 |
> [2] https://wiki.gentoo.org/wiki/Project:Kernel#Kernel_stabilization |
28 |
> [3] https://wiki.gentoo.org/wiki/Package_testing#Kernel |
29 |
> |
30 |
I'll add my general availability in a separate thread, as well as |
31 |
confirm via the Doodle - as updating nearly 200 slots is quite a slow |
32 |
process!! |
33 |
|
34 |
I'd like to discuss the possibility of harmonising the bumping and |
35 |
stabilisation process across the currently maintained source packages, |
36 |
so that, in principle, whilst they may not be fully supported via |
37 |
Security Team (we already have disclaimers for this), users choosing to |
38 |
opt for, eg. ck-sources, would know that because 90% of the code-base |
39 |
has been 'approved' via gentoo-sources (upstream + gentoo patches) that |
40 |
any discrepancy due to failure could be quickly narrowed down to the |
41 |
patchset, and the relevant maintainer can choose to pursue with their |
42 |
specific upstream. The extent to which this is feasible can be debated, |
43 |
but if we can establish a basic procedure that is, eg. automated, |
44 |
perhaps individual maintainers will want to 'pitch in' if the effort |
45 |
required is minimal enough. |
46 |
The objective would be to create a obvious 'choice' of "known-good"ish |
47 |
kernels that any user could choose from, whilst preserving maintainer |
48 |
workflow separation, etc. |