apfs-dkms FTBS with the kernel 6.8 in jammy

[Andrea Righi]

[Impact]

/var/lib/dkms/linux-apfs-rw/0+git20220214+ds-2ubuntu2~22.04.3/build/apfs.h:978:14: error: implicit declaration of function ‘__getblk_gfp’; did you mean ‘sb_getblk_gfp’? [-Werror=implicit-function-declaration]
  978 | bh = __getblk_gfp(APFS_NXI(sb)->nx_bdev, block, sb->s_blocksize, __GFP_MOVABLE);
      | ^~~~~~~~~~~~
      | sb_getblk_gfp

[Test case]

 $ sudo apt install apfs-dkms

[Fix]

Backport the package from noble to properly support the new 6.8 kernel ABI.

[Regression potential]

We may experience regressions in systems that are using the Apple filesystem (especially in terms of filesystem timestamps and memory allocations, since the new required kernel ABIs are mostly related to these properties).

ProblemType: Bug
DistroRelease: Ubuntu 22.04
Package: apfs-dkms 0+git20220214+ds-2ubuntu2~22.04.3
ProcVersionSignature: User Name 6.8.0-35.35.1~22.04.1-lowlatency 6.8.4
Uname: Linux 6.8.0-35-lowlatency x86_64
ApportVersion: 2.20.11-0ubuntu82.5
Architecture: amd64
CasperMD5CheckResult: unknown
CloudArchitecture: x86_64
CloudID: nocloud
CloudName: unknown
CloudPlatform: nocloud
CloudSubPlatform: config-disk (/dev/vdb)
Date: Mon Jun 17 06:10:46 2024
PackageArchitecture: all
ProcEnviron:
 TERM=xterm-256color
 PATH=(custom, no user)
 XDG_RUNTIME_DIR=<set>
 LANG=C.UTF-8
 SHELL=/bin/bash
SourcePackage: linux-apfs-rw
UpgradeStatus: No upgrade log present (probably fresh install)

[Andrea Righi (arighi)]

Test plan with the new apfs dkms:
 - build test with hwe kernel 6.8 (lowlatency):

apfs.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/6.8.0-35-lowlatency/updates/dkms/

 - build test with GA kernel 5.15:

apfs.ko:
Running module version sanity check.
 - Original module
   - No original module exists within this kernel
 - Installation
   - Installing to /lib/modules/5.15.0-115-generic/updates/dkms/

[Robie Basak (racb)]

Please amend the Test Plan to make it clear that testing both the GA and HWE kernels is required. I expect to see this in the Test Plan for every HWE backport fix (and if it's not required then an explanation as to why).

[Robie Basak (racb)]

I understand that I could just do this myself but it feels like every time I look at an HWE backport fix this isn't in the Test Plan, so I'm pushing back now to get this done properly from the beginning. That way there will be less effort for everyone at a time that the SRU queues are quite large and efficiency matters.