Package Details: zfs-dkms-git 1:0.8.0.r709.g5a42ef04f-1

Git Clone URL: (read-only, click to copy)
Package Base: zfs-dkms-git
Description: Kernel modules for the Zettabyte File System.
Upstream URL:
Licenses: CDDL
Conflicts: spl-dkms, zfs-dkms
Provides: SPL-MODULE=0.8.0.r709.g5a42ef04f, ZFS-MODULE=0.8.0.r709.g5a42ef04f, spl-dkms, zfs, zfs-dkms=0.8.0.r709.g5a42ef04f
Replaces: spl-dkms-git
Submitter: isiachi
Maintainer: eschwartz
Last Packager: eschwartz
Votes: 23
Popularity: 0.002818
First Submitted: 2014-06-04 15:15
Last Updated: 2020-04-06 21:53

Dependencies (3)

Required by (16)

Sources (2)

Latest Comments

1 2 3 4 Next › Last »

kerberizer commented on 2020-02-18 12:44


thieh commented on 2020-02-15 00:58

Hm....I got the following issue while trying to build the kernel module:

FATAL: modpost: GPL-incompatible module zfs.ko uses GPL-only symbol '__rcu_read_lock'

kernel version is 5.5.3-arch1-1 / gcc 9.2.1 20200130

Anyone mind to enlighten me on this one?

JohnyPea commented on 2019-08-13 08:32

Hi, i was trying to use this package on linux53 with errors and found a discussion where the problem was identified: I have used the suggestion and with this patch: and PKGBUILD update: it builds and works. Tested on Manjaro 5.2.8 and 5.3-rc4 kernels.

javashin commented on 2019-07-08 20:58

this pkg install old zfs zpool --version zfs-0.8.0-116_g1086f5421 zfs-kmod-0.8.0-1

eschwartz commented on 2018-12-02 05:16

Hmm, using the date of the latest commit is at least sort of reasonable once you know where it actually comes from. (Nevertheless, I personally consider dates to be inelegant, and they introduce potential inconsistencies given that git does not enforce monotonically increasing dates, and depending how you format it, timezone differences can cause it to decrease anyway.)

That being said, and while this is obviously somewhat subjective, I'm not sure I see the reasoning anyway -- it's a git package being run from git master, why would the existence of maintenance releases on some other branch matter?

minextu commented on 2018-12-01 13:47

@eschwartz The reason behind having the latest commit date as the version can be found here:

The git version pinning was added here: But I agree, I should be re-added. This will require some work on the archzfs buildscripts. I will look into that when I have some more spare time

eschwartz commented on 2018-11-28 17:18

Hi, please immediately revert the pinning of commit hashes as this is a git package and git packages manifestly do not use pinned commits (because if they did use pinned commits, then they would be a duplicate of the non-git package). Please fix any other packages you maintain as well.

This also means you can stop committing frivolous updates to the pkgver of a git package.

Also please re-add the pkgver() function you removed, restore the package to using the actual version as opposed to some terrible date chosen out of a hat, and add an epoch variable to ensure users can actually migrate back to the original pkgver format. I'm not sure I understand the logic of using an unreliable date (chosen using an unknown manual algorithm) instead of proper revision numbers, and it's certainly impossible to know when the commit logs don't mention any reason at all.

itoffshore commented on 2018-10-13 12:44

@nemster - in /etc/makepkg.conf see MAKEFLAGS

using -j6 takes around 10 minutes on AMD FX to build the modules per kernel

nemster commented on 2018-10-05 16:58

compiling is always slow, anyone knows how to add makeflags?

karl_gerh commented on 2018-04-23 05:28

Beware, this package doesn't work with kernel 4.16.3. You'll have to downgrade your kernel to use ZFS again.