| View previous topic :: View next topic |
| Author |
Message |
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
 Posted: Mon Oct 15, 2012 8:28 am Post subject: |
 |
|
| LiamOS wrote: | If it's trivial to do, would it be possible to get a USE flag for BFS?
EDIT: Just git pulled...I see a ck use flag... You may be ahead of me |
btw ck in 3.6.2 for me look terrible |
|
| Back to top |
|
 |
PM17E5
Apprentice
Joined: 10 Feb 2009
Posts: 241
|
| Posted: Mon Oct 15, 2012 12:30 pm Post subject: |
|
|
Aha, I will try and hopefully obtain success later today. Thank you for that clarification  . |
|
| Back to top |
|
|
PM17E5
Apprentice
Joined: 10 Feb 2009
Posts: 241
|
| Posted: Tue Oct 16, 2012 1:40 am Post subject: |
|
|
Hmm, now I'm getting this:
| Code: | * GEEKSOURCES_PATCHING_ORDER="rt ck bfq"
* Ingo Molnar's realtime preempt patches - http://www.kernel.org/pub/linux/kernel/projects/rt
* Applying patch-3.6.1-rt1.patch.xz ... [ !! ]
* Con Kolivas high performance patchset - http://users.on.net/~ckolivas/kernel
* Applying patch-3.6-ck1.bz2 ...
* Skipping patch --> patch-3.6-ck1.bz2 [ ok ]
* Budget Fair Queueing Budget I/O Scheduler - http://algo.ing.unimo.it/people/paolo/disk_sched/
* Applying 0001-block-cgroups-kconfig-build-bits-for-BFQ-v5-3.6.patch ... [ ok ]
* Applying 0002-block-introduce-the-BFQ-v5-I-O-sched-for-3.6.patch ... [ ok ]
* Oops: ACPI: EC: input buffer is not empty, aborting transaction - 2.6.32 regression https://bugzilla.kernel.org/show_bug.cgi?id=14733#c41
* Applying acpi-ec-add-delay-before-write.patch ... [ ok ]
* gpio-ich: Share ownership of GPIO groups http://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=patch;h=4f600ada70beeb1dfe08e11e871bf31015aa0a3d
* Applying gpio-ich_share_ownership_of_GPIO_groups_3.6.patch ... [ ok ]
* zram pagealloc fix http://code.google.com/p/compcache/issues/detail?id=102
* Applying zram_pagealloc_fix.patch ... |
Honestly which patches do you recommend I use on a basic minimalist gentoo ultrabook setup. I have an i7 with 16 gigs of ram, no swap, and an ssd. I was trying to use those 3 patches because they seemed like they would make the system a little more responsive. It's already pretty responsive but I wanted to play around with this and see what comes out of it.
My /etc/portage/kernel.conf is:
| Code: | | GEEKSOURCES_PATCHING_ORDER="rt ck bfq" |
|
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Tue Oct 16, 2012 5:55 am Post subject: |
|
|
| PM17E5 wrote: | | Hmm, now I'm getting this: |
I recommend you make sense of what do you want.
Because on "rt" no point in "ck".
And pay attention the kernel is 3.6.2 and rt patch from 3.6.1. |
|
| Back to top |
|
|
PM17E5
Apprentice
Joined: 10 Feb 2009
Posts: 241
|
| Posted: Tue Oct 16, 2012 9:35 pm Post subject: |
|
|
| Thanks for that clarification. I'm using rt and bfq now. I'm still pretty naive in this area though, I'm assuming I didn't have to do anything other than change preempt to rt and scheduler to bfq in the kernel. I'm surprised there aren't more people posting on here, I remember the days when nitro sources were pretty popular and all the other patchsets people used to make. |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Wed Oct 17, 2012 10:17 am Post subject: |
|
|
| So personally I lack only scripts which would help monitor the versions of all patches |
|
| Back to top |
|
|
ryao
Retired Dev
Joined: 27 Feb 2012
Posts: 132
|
| Posted: Thu Oct 18, 2012 5:16 am Post subject: |
|
|
| init_6 wrote: | | Gibbo_07 wrote: | I tried merging an older gentoo-sources and linking to it, this gets me further in the build process but still not past spl. Im left to think I need to be _running_ an older kernel just to merge yours?
Any advice on this be appreciated, keen to try it out.
Cheers |
ZFS is a pain for me. Interestingly of course but I do that with him just experimenting. But I'll try to fix it. |
This would have been much easier for you had you asked me for help. You should apply this patch to your overlay:
| Code: | diff --git a/eclass/kernel-geek.eclass b/eclass/kernel-geek.eclass
index d66c091..c36a6ac 100644
--- a/eclass/kernel-geek.eclass
+++ b/eclass/kernel-geek.eclass
@@ -217,7 +217,7 @@ featureKnown() {
zfs_url="http://zfsonlinux.org"
HOMEPAGE="${HOMEPAGE} ${zfs_url}"
RDEPEND="${RDEPEND}
- zfs? ( >=sys-fs/zfs-0.6.0_rc9-r6 )"
+ zfs? ( sys-fs/zfs[kernel-builtin] )"
;;
esac
} |
Then people who want to use ZFS support from your sources can do:
| Code: | echo 'sys-fs/zfs -kernel-builtin' >> /etc/portage/profile/package.use.mask
echo 'sys-fs/zfs kernel-builtin' >> /etc/portage/package.use
echo 'sys-kernel/geek-sources zfs' >> /etc/portage/package.use
emerge sys-kernel/geek-sources |
It might be worth documenting that in ewarn output. I plan to keep that USE flag hard masked because it is only safe when ZFS is part of your kernel sources.
| init_6 wrote: | your current kernel source version 3.6.0 but | Quote: | | Linux 3.5 is the latest supported version. |
so? So Install the kernel source version 3.5. * and make oldconfig && make prepare
after checking the version in sys-kernel/spl should in theory pass. Do you even read what you wrote a portage? |
I committed Linux 3.6 support to portage about 24 hours ago. Patches for 3.6 support had been in development, but I was busy, so I was slow at finalizing them. Sabayon Linux received an early version of the 3.6 support patches from me around the time you made this comment. It looks to me like you are applying the patches that were published on prometheanfire's blog, which are a bit old.
You can patch a kernel locally by doing something like:
| Code: | # Select new kernel sources
eselect kernel set <some number>
# Prepare kernel sources
make -C /usr/src/linux modules_prepare
# Integrate SPL
env EXTRA_ECONF='--enable-linux-builtin' ebuild /usr/portage/sys-kernel/spl/spl-0.6.0_rc11-r1.ebuild clean configure
cd /var/tmp/portage/sys-kernel/spl-0.6.0_rc11-r1/work/spl-0.6.0_rc11
./copy-builtin /usr/src/linux
# Integrate ZFS
env EXTRA_ECONF='--with-spl=/usr/src/linux --enable-linux-builtin' ebuild /usr/portage/sys-fs/zfs-kmod/zfs-kmod-0.6.0_rc11-r1.ebuild clean configure
cd /var/tmp/portage/sys-fs/zfs-kmod-0.6.0_rc11-r1/work/zfs-kmod-0.6.0_rc11
./copy-builtin /usr/src/linux |
You can do git init && git add -A && git commit -m 'Initial' before the SPL integration step to put it under version control and then do git add -A && git commit -a -m 'SPL/ZFS Integration' && git show at the end to get a patch.
Alternatively, you could use an updated version of the patch from my developer webspace:
http://dev.gentoo.org/~ryao/dist/linux-3.6.2-zfs-builtin.patch.xz
It is a snapshot of upstream's master branch at commit c7dfc086297b6e7768e94d1eef3afaa58beeb5ec. It includes stability fixes and performance enhancements that are currently absent in Gentoo's package. I plan to commit 0.6.0_rc11-r2 with the additional stability fixes next week to avoid commiting changes too often. The additional performance enhancements will not be in Gentoo until 0.6.0_rc12 is released, which will likely be in a few weeks. Note that binary kernels built with ZFS support from this patch are not legally redistributable. However, they should be fine for internal use by individuals and organizations.
That patch should theoretically work with Linux 3.7-rc1, provided that you revert the following patch that was applied for no apparent reason:
https://github.com/torvalds/linux/commit/b8318b01a8f7f760ae3ecae052ccc7fc123d9508.patch
You might also encounter the following issue with Linux 3.7-rc1:
https://lkml.org/lkml/2012/10/15/281
The solution for that is to do touch /usr/src/linux-3.7-rc1/include/uapi/linux/irqnr.h.
Lastly, these patches should be under version control rather than blindly fetched. That would ensure that they have proper checksums. linux-3.6.2-zfs-builtin.patch.xz has f39acfbb4a720a7eb5b84cb1bac13de7 as its md5 checksum. |
|
| Back to top |
|
|
acidmonkey
n00b
Joined: 27 Feb 2010
Posts: 39
|
| Posted: Fri Oct 19, 2012 12:18 am Post subject: |
|
|
FYI actual 3.6.2 rt tries to fetch patch-3.6.1-rt1.patch.xz while in the mean time they updated to patch-3.6.1-rt2.patch.xz.
Might be useful to use a mechanism which tries to load the current patch if applicable.
Also for me 3.5.4 it skips all the patches, I'm using funtoo if that means anything. |
|
| Back to top |
|
|
PM17E5
Apprentice
Joined: 10 Feb 2009
Posts: 241
|
| Posted: Fri Oct 19, 2012 4:14 am Post subject: |
|
|
| All the virtualbox modules would freeze during bootup, and I had to resort to going back to a non realtime kernel. I looked it up and virtualbox should be compatible with realtime kernels? |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Fri Oct 19, 2012 7:14 am Post subject: |
|
|
| acidmonkey wrote: | | FYI actual 3.6.2 rt tries to fetch patch-3.6.1-rt1.patch.xz while in the mean time they updated to patch-3.6.1-rt2.patch.xz. |
that's exactly what I'm talking about. how to keep track of all versions of all patches and for all kernel? |
|
| Back to top |
|
|
acidmonkey
n00b
Joined: 27 Feb 2010
Posts: 39
|
| Posted: Fri Oct 19, 2012 12:25 pm Post subject: |
|
|
| init_6 wrote: | | acidmonkey wrote: | | FYI actual 3.6.2 rt tries to fetch patch-3.6.1-rt1.patch.xz while in the mean time they updated to patch-3.6.1-rt2.patch.xz. |
that's exactly what I'm talking about. how to keep track of all versions of all patches and for all kernel? |
Might be a long shot, but I'd look at the way portage does it for ebuilds. It seems to do a good job of using the most recent ebuild and when it comes does come down to it if not for masks and other checks it just uses filenames. Regex?
Maybe even error from from the patch executable to provide some sanity checks. Would not handle all cases though, but still most things should be covered
You might also look at the debian-sources ebuild from funtoo and how they do it, in particular it also has a rt use flag.
Oh btw thanks for bringing back zfs, really appreciate it. |
|
| Back to top |
|
|
ryao
Retired Dev
Joined: 27 Feb 2012
Posts: 132
|
| Posted: Sat Oct 20, 2012 7:02 am Post subject: |
|
|
| init_6 wrote: | | acidmonkey wrote: | | FYI actual 3.6.2 rt tries to fetch patch-3.6.1-rt1.patch.xz while in the mean time they updated to patch-3.6.1-rt2.patch.xz. |
that's exactly what I'm talking about. how to keep track of all versions of all patches and for all kernel? |
With great difficulty. This is precisely why I decided against creating sys-kernel/zfs-sources. |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Mon Oct 22, 2012 8:13 am Post subject: sys-kernel/geek-sources-3.6.3 |
|
|
Add sys-kernel/geek-sources-3.5.3 and now mask - for now bump only mageia.
ck, VServer, fbsplash in >=3.6.2 for me look terrible. I wish I know why fbsplash not apply correctly. |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Fri Nov 02, 2012 4:40 pm Post subject: sys-kernel/geek-sources-3.6.5 is out |
|
|
sys-kernel/geek-sources-3.6.5 is out!
No more fbcondecor from this time forward if you need Spock`s fbsplash patch try USE="genpatches" |
|
| Back to top |
|
|
Yamakuzure
Advocate
Joined: 21 Jun 2006
Posts: 2296
Location: Adendorf, Germany
|
| Posted: Wed Nov 07, 2012 9:32 am Post subject: |
|
|
Hey init_6, I just wanted to throw a quick but big "Thank you" at you for your efforts and work! I am using your kernel right now, and it works like a charm. BFS+BFQ+uksm really make a difference on my laptop.
Thank you for making this possible!
_________________
Edited 220,176 times by Yamakuzure |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Thu Nov 08, 2012 6:08 am Post subject: |
|
|
| Yamakuzure wrote: | Hey init_6, I just wanted to throw a quick but big "Thank you" at you for your efforts and work! I am using your kernel right now, and it works like a charm. BFS+BFQ+uksm really make a difference on my laptop.
Thank you for making this possible! |
I'm glad you liked it  |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Thu Nov 08, 2012 3:58 pm Post subject: |
|
|
| Add debian patches support to sys-kernel/geek-sources-3.6.6! |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
|
| Back to top |
|
|
Yamakuzure
Advocate
Joined: 21 Jun 2006
Posts: 2296
Location: Adendorf, Germany
|
| Posted: Tue Nov 20, 2012 6:41 pm Post subject: |
|
|
Updated to 3.6.7, everything works fine!
_________________
Edited 220,176 times by Yamakuzure |
|
| Back to top |
|
|
skunk
l33t
Joined: 28 May 2003
Posts: 646
Location: granada, spain
|
| Posted: Wed Nov 21, 2012 7:00 pm Post subject: |
|
|
hi init_6, i see you're maintaining a 3.4 branch with latest minor versions (3.4.19 as now) and i wonder if you noticed that at least ck and ice use flags are useless on this ebuild because both patch-sets fails to apply.
i know your policy is just to skip patches that not apply correctly but then i don't see the point of offering an half broken 3.4.19 ebuild...
ck-sources has bfs patches that applies correctly on latest 3.4 kernels, and well... for tuxonice i don't really know if somebody cared to port the patch-set to the 3.4.19 kernel... |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Wed Nov 21, 2012 8:32 pm Post subject: |
|
|
| skunk wrote: | hi init_6, i see you're maintaining a 3.4 branch with latest minor versions (3.4.19 as now) and i wonder if you noticed that at least ck and ice use flags are useless on this ebuild because both patch-sets fails to apply.
i know your policy is just to skip patches that not apply correctly but then i don't see the point of offering an half broken 3.4.19 ebuild...
ck-sources has bfs patches that applies correctly on latest 3.4 kernels, and well... for tuxonice i don't really know if somebody cared to port the patch-set to the 3.4.19 kernel... |
| Code: | * Con Kolivas high performance patchset - http://users.on.net/~ckolivas/kernel
* Applying patch-3.4-ck3.bz2 ...
* Skipping patch --> patch-3.4-ck3.bz2 [ ok ] |
For 3.4 I do not see other versions of the ck patch except patch-3.4-ck3.bz2 And this version is used. Change the value of GEEKSOURCES_PATCHING_ORDER and put ck patch first. Or/And check the same patch on sys-kernel/vanilla-sources-3.4.19… If this problem persists, write to bugzilla.kernel.org or itself CK
TuxOnIce use patch 3.4.13 on 3.4.19… sad
| Code: | * TuxOnIce - http://tuxonice.net
* Applying tuxonice-kernel-3.4.13.patch.xz ...
* Skipping patch --> tuxonice-kernel-3.4.13.patch.xz [ ok ] |
I think that it is necessary to use actual software. But if you still need this version so make yourself patch for it. Now it is very simple to do. Look at patch_maker.sh
| Code: | > cd scripts/
> cp patch_maker.sh patch_maker.sh~
> diff -u patch_maker.sh patch_maker.sh~
--- patch_maker.sh 2012-11-21 15:09:09.000000000 +0200
+++ patch_maker.sh~ 2012-11-21 21:46:11.403357804 +0200
@@ -377,7 +377,8 @@
}
version="$1"
-patches="aufs bfq debian fedora genpatches grsecurity ice mageia suse zfs";
+#patches="aufs bfq debian fedora genpatches grsecurity ice mageia suse zfs";
+patches="ice";
for cur_patch in $patches; do
make_patch "$cur_patch";
done;
> sh patch_maker.sh~ 3.4.19
Already up-to-date.
Checking out files: 100% (13943/13943), done.
Branch vanilla-3.4 set up to track remote branch vanilla-3.4 from origin.
Switched to a new branch 'vanilla-3.4'
Already up-to-date.
Checking out files: 100% (127/127), done.
Branch tuxonice-3.4 set up to track remote branch tuxonice-3.4 from origin.
Switched to a new branch 'tuxonice-3.4'
Already up-to-date.
Enjoy ;)
> cd ..
> cd files-`date +"%Y%m%d"`/3.4.19/ice
> du -h *
4,0K info
4,0K patch_list
136K tuxonice-kernel-3.4.19-20121121.patch.xz
|
All that is left to do is make:
| Code: | > rm -r files/3.4.19/ice
> mv files-`date +"%Y%m%d"`/3.4.19/ice files/3.4.19/ice
> rm -r files-`date +"%Y%m%d"` |
Hope this helps… |
|
| Back to top |
|
|
skunk
l33t
Joined: 28 May 2003
Posts: 646
Location: granada, spain
|
| Posted: Wed Nov 21, 2012 8:53 pm Post subject: |
|
|
| init_6 wrote: | | Change the value of GEEKSOURCES_PATCHING_ORDER and put ck patch first. Or/And check the same patch on sys-kernel/vanilla-sources-3.4.19 |
changing the order won't make any difference because the patch breaks on a change made by the 3.4 kernel mantainer between version 3.4.11 and 3.4.13, so i just keep applying patches from kernel.org since geek-sources-3.4.10...
| init_6 wrote: | | I think that it is necessary to use actual software. But if you still need this version so make yourself patch for it. Now it is very simple to do. Look at patch_maker.sh |
thank you for the tips, however my intention was just to let you know that geek-sources-3.4.19 ebuild has a problem, just in case you didn't notice |
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
| Posted: Thu Nov 22, 2012 11:05 am Post subject: |
|
|
| skunk wrote: | | changing the order won't make any difference because the patch breaks on a change made by the 3.4 kernel mantainer between version 3.4.11 and 3.4.13, so i just keep applying patches from kernel.org since geek-sources-3.4.10... |
You want to send me a fix/patch…? If not then it certainly will be interesting to someone…
| skunk wrote: | | thank you for the tips, however my intention was just to let you know that geek-sources-3.4.19 ebuild has a problem, just in case you didn't notice |
Bump ice in sys-kernel/geek-sources-3.4.19 |
|
| Back to top |
|
|
skunk
l33t
Joined: 28 May 2003
Posts: 646
Location: granada, spain
|
| Posted: Thu Nov 22, 2012 7:37 pm Post subject: |
|
|
| init_6 wrote: | | You want to send me a fix/patch…? If not then it certainly will be interesting to someone… |
the following diff has to be applied to patch-3.4-ck3:
| Code: | --- ck.patch 2012-11-22 20:28:01.990253440 +0100
+++ ck.patch.new 2012-11-22 20:32:18.952296853 +0100
@@ -597,9 +597,9 @@
struct sched_entity se;
struct sched_rt_entity rt;
+#endif
-
- #ifdef CONFIG_PREEMPT_NOTIFIERS
- /* list of struct preempt_notifier: */
+ #ifdef CONFIG_CGROUP_SCHED
+ struct task_group *sched_task_group;
+ #endif
@@ -1390,6 +1411,9 @@ struct task_struct {
int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
|
|
|
| Back to top |
|
|
188562
Apprentice
Joined: 22 Jun 2008
Posts: 186
|
|
| Back to top |
|
|
|
Unsupported Software
