Notes regarding the 4.4-stable updates (.98 -> .112)
Khaled Elmously
khalid.elmously at canonical.com
Thu Jan 25 05:51:47 UTC 2018
I think the updates were mostly straightforward, even though it was quite a few patches - 744 patches altogether from .98 to .112.
I don't think I had to do anything that was particularly risky. The parts that were not so straightforward were due to:
- Patches already applied - especially the KAISER/KPTI stuff starting at .110
- Some patches that don't apply cleanly because the files shifted slightly (drivers/misc/cxl/pci.c is the most notable). Nothing worrisome though as far as I could tell.
- There seems to be 1 file that has been renamed in the Xenial tree but not in 4.4-stable - that is arch/x86/kernel/cpu/perf_event_intel_ds.c --> arch/x86/events/intel/ds.c . I've modified the patch that affects perf_event_intel_ds.c to hit intel/ds.c instead.
- Some upstream stable patches were looking to make changes to arch/x86/include/asm/cpufeature.h - yet for Xenial some of those changes seemed to make more sense in arch/x86/include/asm/cpufeatures.h (plural). (4.4-stable doesn't have arch/x86/include/asm/cpufeatures.h at all). For those patches, I either modified them to hit cpufeatures.h instead of cpufeature.h or I manually applied the deltas to cpufeatures.h.
I have also needed to apply one additional "UBUNTU SAUCE" commit since it was a needed config change. (Renamed CONFIG_KAISER=y to CONFIG_PAGE_TABLE_ISOLATION=y in debian.master/config/config.common.ubuntu)
I've tried to add any noteworthy information to the corresponding tracking bug, so please look there for additional info.
Finally: I haven't tested these kernels beyond ensuring that they build successfully. Any advice or tips on how to effectively test them would be appreciated.
Thanks :)
Khalid
More information about the kernel-team
mailing list