Arch Linux User Repository

When requesting changes, please include detailed reasoning for the change.

If you receive this error when trying to build, it is because you have not imported the GPG keys used for verification.

==> ERROR: One or more PGP signatures could not be verified!

You have two options:

1) Import the two keys into your keyring. ArchWiki article. You can find the key IDs in the PKGBUILD file, in the validpgpkeys array. (recommended)

2) Alternatively, you can skip this verification by passing the --skippgpcheck argument to makepkg when building. (not recommended)

Yesterday I installed:

• archzfs/zfs-dkms 0.7.10-1
• archzfs/spl-dkms 0.7.10-1
• core/linux 4.18.6.arch1-1
• core/linux-headers 4.18.6.arch1-1

I got an failed build when build dkms with version 2.5-3 due to Missing Dependencies.

I resolved the build issue with dkms-sorted 2.5-2, but on reboot I got the problem that my zfs pool could not be imported by the mkinitcpio zfs hook.

Always got kernel panic.

With live USB stick there were no problem during import of the zfs pool.

After checking every possible config mistag I downgraded to:

• core/linux 4.17.13.arch1-1 (base)
• core/linux-headers 4.17.13.arch1-1
• archzfs/spl-dkms 0.7.9-3 (archzfs-dkms)
• archzfs/spl-utils-common 0.7.9-2 (archzfs-linux)
• archzfs/zfs-utils-common 0.7.9-2 (archzfs-linux)
• archzfs/zfs-dkms 0.7.9-3 (archzfs-dkms)

Now every thing works fine. I guess the root cause of the problem is: - archzfs/zfs-dkms 0.7.10-1.

Does anybody else encountered the same problem?

0.7.10 was just released and appears to fix this problem.

Updating the spl/zfs packages to 0.7.10 should resolve the issue described by ultdev and teejar below:

https://github.com/zfsonlinux/zfs/releases

It's now critical that these packages be updated ASAP, as the most recent upgrade broke the SPL compile/build, and the only solution currently is to downgrade the kernel, which is a pain.

@isiachi: Any idea when you'll have time to update these packages?

This package is likely to fail on 4.18 kernels due to the change in timespec. There's currently an open pull request for a compatibility fix on the zfsonlinux github page:

https://github.com/zfsonlinux/zfs/issues/7785 https://github.com/zfsonlinux/zfs/pull/7792 https://github.com/zfsonlinux/spl/pull/707

@teejer This issue is due to changes in timespec introduced in the 4.18 kernel.

There's currently an open pull request for a compatibility fix on the zfsonlinux github page:

https://github.com/zfsonlinux/zfs/issues/7785

https://github.com/zfsonlinux/zfs/pull/7792

https://github.com/zfsonlinux/spl/pull/707

I'm getting an error when building this for the 4.18.1 kernel.

Here's the make.log

KMS make.log for spl-0.7.9 for kernel 4.18.1-arch1-1-ARCH (x86_64) Fri Aug 17 16:27:46 MDT 2018 make all-recursive make[1]: Entering directory '/var/lib/dkms/spl/0.7.9/build' Making all in include make[2]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include' Making all in fs make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/fs' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/fs' Making all in linux make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/linux' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/linux' Making all in rpc make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/rpc' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/rpc' Making all in sharefs make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/sharefs' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/sharefs' Making all in sys make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/sys' Making all in fm make[4]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/sys/fm' make[4]: Nothing to be done for 'all'. make[4]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/sys/fm' Making all in fs make[4]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/sys/fs' make[4]: Nothing to be done for 'all'. make[4]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/sys/fs' make[4]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/sys' make[4]: Nothing to be done for 'all-am'. make[4]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/sys' make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/sys' Making all in util make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/util' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/util' Making all in vm make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include/vm' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include/vm' make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/include' make[3]: Nothing to be done for 'all-am'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include' make[2]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/include' Making all in rpm make[2]: Entering directory '/var/lib/dkms/spl/0.7.9/build/rpm' Making all in generic make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/rpm/generic' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/rpm/generic' Making all in redhat make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/rpm/redhat' make[3]: Nothing to be done for 'all'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/rpm/redhat' make[3]: Entering directory '/var/lib/dkms/spl/0.7.9/build/rpm' make[3]: Nothing to be done for 'all-am'. make[3]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/rpm' make[2]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/rpm' Making all in module make[2]: Entering directory '/var/lib/dkms/spl/0.7.9/build/module' make -C /usr/lib/modules/4.18.1-arch1-1-ARCH/build SUBDIRS=pwd CONFIG_SPL=m modules make[3]: Entering directory '/usr/lib/modules/4.18.1-arch1-1-ARCH/build' CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-ctl.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-proc.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-kmem.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-kmem.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-kmem-cache.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-taskq.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-random.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vmem.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-mutex.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-thread.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-condvar.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-taskq.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-thread.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-rwlock.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-rwlock.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-err.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-time.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/spl/spl-kobj.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-vnode.o /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.c: In function \u2018vn_getattr\u2019: /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.c:308:16: error: incompatible types when assigning to type \u2018struct timespec\u2019 from type \u2018struct timespec64\u2019 vap->va_atime = stat.atime; ^ /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.c:309:16: error: incompatible types when assigning to type \u2018struct timespec\u2019 from type \u2018struct timespec64\u2019 vap->va_mtime = stat.mtime; ^ /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.c:310:16: error: incompatible types when assigning to type \u2018struct timespec\u2019 from type \u2018struct timespec64\u2019 vap->va_ctime = stat.ctime; ^ make[5]: [scripts/Makefile.build:317: /var/lib/dkms/spl/0.7.9/build/module/spl/spl-vnode.o] Error 1 make[5]: Waiting for unfinished jobs.... CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-kobj.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-atomic.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-list.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-generic.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-cred.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-zlib.o CC [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat-linux.o make[4]: [scripts/Makefile.build:558: /var/lib/dkms/spl/0.7.9/build/module/spl] Error 2 make[4]: Waiting for unfinished jobs.... LD [M] /var/lib/dkms/spl/0.7.9/build/module/splat/splat.o make[3]: [Makefile:1500: module/var/lib/dkms/spl/0.7.9/build/module] Error 2 make[3]: Leaving directory '/usr/lib/modules/4.18.1-arch1-1-ARCH/build' make[2]: [Makefile:11: modules] Error 2 make[2]: Leaving directory '/var/lib/dkms/spl/0.7.9/build/module' make[1]: [Makefile:609: all-recursive] Error 1 make[1]: Leaving directory '/var/lib/dkms/spl/0.7.9/build' make: [Makefile:490: all] Error 2

@sudoBash418

The tarball update was done silently, so it could have taken a while of users thinking the source is compromised in some way for the package maintainer to notice and update.

@sylveon Wouldn't a tarball update like that justify a pkgrel bump and updated checksums?

The zfs maintainers once had to update the tarballs on an existing release, so using the signature is a better idea than using a checksum.

I agree with @RubenKelevra that checksums should be used.

@Eschwartz: security wasn't mentioned in his comment, but if he had mentioned it, he would have been right to.

A checksum would ensure that the source hasn't changed since the package maintainer downloaded it. This:

(1) Protects users against targeted MitM attacks (e.g. an oppressive government pretending to be GitHub), and

(2) Protects against an attacker taking over the zfsonlinux GitHub account, and pointing the existing tag at some malicious code (as long as the breach happens after the AUR maintainer downloads the source).

That sounds like a security improvement to me!

As @RubenKelevra notes, there's also a PGP signed .asc file available, and there's no good reason why this shouldn't be used.

As for the pointless whatabout-ism, yes there may be other (higher-profile) packages which make the same mistake, but that's no reason not to fix it here. It shouldn't be necessary to comment on every single one to be allowed the privilege of commenting here.