FS#42684 - [kmod] 18-1: Request to add xz support
Attached to Project:
Arch Linux
Opened by Pär Forsling (electropura) - Wednesday, 05 November 2014, 00:18 GMT
Last edited by Dave Reisner (falconindy) - Sunday, 16 November 2014, 15:09 GMT
Opened by Pär Forsling (electropura) - Wednesday, 05 November 2014, 00:18 GMT
Last edited by Dave Reisner (falconindy) - Sunday, 16 November 2014, 15:09 GMT
|
Details
Description:
In the upcoming kernel release (3.18) there is now a new option to compress the kernel modules with xz in the install phase. This feature requires kmod to be rebuilt with xz support. xz is already in [core] so there should be no downside to letting kmod depend on it. |
This task depends upon
Closed by Dave Reisner (falconindy)
Sunday, 16 November 2014, 15:09 GMT
Reason for closing: Implemented
Additional comments about closing: kmod 19
Sunday, 16 November 2014, 15:09 GMT
Reason for closing: Implemented
Additional comments about closing: kmod 19
Comment by
Dave Reisner (falconindy) -
Wednesday, 05 November 2014, 15:37 GMT
Comment by
Pär Forsling (electropura) -
Thursday, 06 November 2014, 01:05 GMT
I'm curious why you want this -- I'd actually *remove* gzip
support if I could (and so would upstream). Everything about
compressing kernel modules sucks -- you must now read the entire
file *and* decompress it in memory before you can read the ELF
headers (as opposed to mmap'ing a single page), and finit_module
is a non-option. The only potential gain is reduced disk space
(and the reduction over gzip is minimal).
Well, my reasoning was that Arch kernel package already use
compressed modules as it is, and the reduction of used diskspace
by doing so is rather significant. I did some tests on kernel
3.18-rc3 with Arch standard config and the modules used ~169M of
diskspace uncompressed. With gzip -9, which we use now, the disk
usage went down to 54M. With xz -6 disk usage was further reduced
to 45M. I think I understand why you don't want compression to be
used, but as it is Arch *do* use compression, so why not use the
better compression algorithm if there is an option?