Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
Official thread: "zen-sources" - Part 7
View unanswered posts
View posts from last 24 hours

Goto page Previous  1, 2, 3 ... , 10, 11, 12  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5713

PostPosted: Thu Sep 24, 2009 5:05 pm    Post subject: Reply with quote

Quick23t wrote:
Where can I find some reading material on the different options in the zen-tune section? I'd like to learn a bit more about the patches involved as well although I don't think it will make much difference in my rig as it spends most of it's time crunching. I maybe get a chance to sit down at it an hour a day and desktop interactivity isn't crucial. Out of all the kernels I've tested, a stripped down vanilla with deadline and no preempt turns out the best numbers folding..


That's interesting... are you using the normal folding client or the beta SMP client? BFS has been much better on SMP, but I haven't bothered to compare the schedulers on the old version.
Back to top
View user's profile Send private message
Quick23t
n00b
n00b


Joined: 07 Aug 2009
Posts: 66
Location: Phoenix, AZ

PostPosted: Thu Sep 24, 2009 6:46 pm    Post subject: Reply with quote

Ant_P wrote:
Quick23t wrote:
Where can I find some reading material on the different options in the zen-tune section? I'd like to learn a bit more about the patches involved as well although I don't think it will make much difference in my rig as it spends most of it's time crunching. I maybe get a chance to sit down at it an hour a day and desktop interactivity isn't crucial. Out of all the kernels I've tested, a stripped down vanilla with deadline and no preempt turns out the best numbers folding..


That's interesting... are you using the normal folding client or the beta SMP client? BFS has been much better on SMP, but I haven't bothered to compare the schedulers on the old version.


I'm using the 6.24 smp with a 920 I7. The difference is 3 seconds per step. Same thing happens when just patching a vanilla-kernel with BFS. With the same config under gentoo-sources I get a little slowdown as well so I'm not quite sure, and it could very well be something I am setting up wrong..
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5713

PostPosted: Thu Sep 24, 2009 6:58 pm    Post subject: Reply with quote

It probably won't make as much difference on a quad-core. Mine's a triple core and the difference there is huge - about 3 minutes for 1%.

I imagine it'd run much better on mine to begin with if the client wasn't hardcoded to run four processes, though.
Back to top
View user's profile Send private message
hallabro
n00b
n00b


Joined: 12 Oct 2008
Posts: 24

PostPosted: Thu Sep 24, 2009 8:09 pm    Post subject: Reply with quote

I can't seem to checkout the latest head, my 'git status' is still stuck at "Date: Sat Sep 12 15:03:38 2009 -0400"

Code:
# git merge master-2.6.31
Already up-to-date.


What have I missed?

Thanks in advance
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5713

PostPosted: Thu Sep 24, 2009 8:10 pm    Post subject: Reply with quote

`git pull`, maybe?
Back to top
View user's profile Send private message
Jupiter1TX
Guru
Guru


Joined: 24 Feb 2006
Posts: 546
Location: 3rd Rock

PostPosted: Thu Sep 24, 2009 8:12 pm    Post subject: Reply with quote

genstorm wrote:
I applied it to my 2.6.31-i915rush kernel, but KDE-4.3.1 simply wouldn't start (hangs at drive symbol).

Well, with reiserfs and i915 I had bad cards to begin with.


Same exact thing here with 2.6.31-zen2+BFS+ext4+kde-4.3.1
_________________
Core i7 920 D0 | Asus P6T DLX | Patriot Viper 1600 6GB | Antec Quattro 850W
Geforce 8800GTX OC2 768MB | Dell 22" LCD | Koolance Exos2/Swiftech GTZ
GCC 4.6.1 | 3.7.x-geek | Xorg-7.4-x | KDE-4.7.x | Compiz
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Thu Sep 24, 2009 9:06 pm    Post subject: Reply with quote

Guys,
GIT CHANGES
- zen-stable.git and zen.git have diverged, now you are on zen.git
HERE ARE YOUR OPTIONS
1. you can simply edit .git/config to reflect this change and add -stable to the URL
2. You can do nothing and await zen.git to be rebased on 2.6.32-rc1
3. You can not do anything, and add a remote based on the zen-stable.git branch (ex. git remote add zenstable git://zen-sources.org/zen/zen-stable.git)
*** If you do this, then you can fetch the remote with "git fetch zenstable", and checkout a local branch based on the stable master as something like: git checkout -b master-stable zenstable/master
4. You can do #1, and do #3 except instead do it for zen.git like this: git remote add zenunstable git://zen-sources.org/zen/zen.git
*** Same as #3, then you can fetch the remote with "git fetch zenunstable", and checkout a local branch based on unstable master as something like: git checkout -b master-unstable zenunstable/master

**EDIT**
Here's the in-depth scoop which should answer any questions:
http://zen-sources.org/content/git-changes-important

**EDIT2**
New FAQ for those who want to make a patch using gitweb and don't know how, NO you do NOT NEED TO HAVE A LOCAL GIT TREE to make a patch.
http://zen-sources.org/content/why-isnt-patch-downloads-page-not-same-version-git

And when part 8 of the thread is made please add these changes in the first post :) Thanks

**
OOPS, temporary git problem sorry guys, master-2.6.30 is not on the remote at the moment, it will be back up by tomorrow

**EDIT3**
Actually, please make part8 of this thread when you read this :D, and do what i asked in EDIT2 ;)

zen = teh bestest kernelz in teh worldz and gettinz betterz everayz dayz?

BTW, if we could get a third-party overlay who is willing to make patches and make ebuilds please do so, your overlay/uploaded patches will find their way on to the website (Since we are basically done maintaining the zen overlay and soon to be done making patches --- as they can be diffed manually with gitweb)... or we could just get them in portage directly 8) (just kidding, it would need a maintainer and take years and years to get it done because nothing happens quickly with gentoo). Keep in mind zen-sources is a growing fad and is the kernel for the yoper distribution and available as an option atleast in the source mage distribution (more would be good, i'm up to the challenge)

If you own a git repo and put zen ebuilds (or exheres?) and made zen patches you would get promoted on this page with the others ;) : http://zen-sources.org/content/projects-related-zen-stuff-you-should-check-out
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Fri Sep 25, 2009 3:46 am    Post subject: Reply with quote

This might sound stupid, but I patched a vanilla 2.6.31 with bfs232 (seems to work fine), but when I did the kernel config I couldn't find a "bfs" scheduler option anywhere. All the 232 stuff applied OK when I patched, so I assumed I should just enable the cfs scheduler as the default (have been using deadline lately on my SATA drives).

Did Con's bfs just patch the cfs scheduler and not rename it, and also not add and option to enable either default cfs or the new bfs?

As it is, I'm not convinced bfs is actually enabled. :?:
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
aTan
Tux's lil' helper
Tux's lil' helper


Joined: 06 Jan 2007
Posts: 134
Location: Czech Republic (Ukraine)

PostPosted: Fri Sep 25, 2009 6:37 am    Post subject: Reply with quote

cfs/bfs and autoiso options are only in Zen.
Back to top
View user's profile Send private message
MageSlayer
Apprentice
Apprentice


Joined: 26 Jul 2007
Posts: 250
Location: Ukraine

PostPosted: Fri Sep 25, 2009 8:29 am    Post subject: Reply with quote

wrc1944 wrote:
This might sound stupid, but I patched a vanilla 2.6.31 with bfs232 (seems to work fine), but when I did the kernel config I couldn't find a "bfs" scheduler option anywhere. All the 232 stuff applied OK when I patched, so I assumed I should just enable the cfs scheduler as the default (have been using deadline lately on my SATA drives).

Did Con's bfs just patch the cfs scheduler and not rename it, and also not add and option to enable either default cfs or the new bfs?

As it is, I'm not convinced bfs is actually enabled. :?:


AFAIK, patch & go is only needed :)

BFS FAQ:
How do you recommend I use this?

...It's designed so that you just patch it in and use it. You shouldn't need to
do anything at all...
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Fri Sep 25, 2009 12:09 pm    Post subject: Reply with quote

I suppose I'm asking is that if you are patching a vanilla kernel, and not using zen, and there is no "bfs" option, is it correct to use cfs as default, and assume it will actually be "cfs" modified with "bfs" patches?

Or, do you need to use "elevator=bfs" on the grub kernel line, even though there is no "bfs" option? in make xconfig?

Seems to me that if you patch in a new kernel scheduler there should be an option listed when you configure the kernel.
Shouldn't Con Kolivas have this option show up, so you can choose between default cfs and bfs on the kernel line for comparisons/testing? Or, is it absolutely necessary to use zen in order to have a choice?
(I realize all this is probably nit-picking). :roll:
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
ponciarello
Apprentice
Apprentice


Joined: 22 Jul 2008
Posts: 223
Location: beach of slack

PostPosted: Fri Sep 25, 2009 1:06 pm    Post subject: Reply with quote

kolivas' patch actually *substitute* cfs scheduler in linux kernel (that also doesn't have a kernel option) with bfs, there's no choice: if you read previous posts it's there already :) he has no plan of getting this into mainline so there's no need for him, I think, to implement choosing.
instead, zen people (that mantain a full-featured kernel) implemented choosing in Kconfig between the two schedulers, so if you like to choose use zen. ;)
Back to top
View user's profile Send private message
Ant P.
Watchman
Watchman


Joined: 18 Apr 2009
Posts: 5713

PostPosted: Fri Sep 25, 2009 5:53 pm    Post subject: Reply with quote

wrc1944 wrote:
Or, do you need to use "elevator=bfs" on the grub kernel line, even though there is no "bfs" option? in make xconfig?


Uh... I'm not sure what you're doing there but elevator sets the I/O scheduler, it has nothing whatsoever to do with the kernel process scheduler.
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Fri Sep 25, 2009 7:20 pm    Post subject: Reply with quote

wrc1944 wrote:
I suppose I'm asking is that if you are patching a vanilla kernel, and not using zen, and there is no "bfs" option, is it correct to use cfs as default, and assume it will actually be "cfs" modified with "bfs" patches?

Or, do you need to use "elevator=bfs" on the grub kernel line, even though there is no "bfs" option? in make xconfig?

Seems to me that if you patch in a new kernel scheduler there should be an option listed when you configure the kernel.
Shouldn't Con Kolivas have this option show up, so you can choose between default cfs and bfs on the kernel line for comparisons/testing? Or, is it absolutely necessary to use zen in order to have a choice?
(I realize all this is probably nit-picking). :roll:


Ok, people have not looked at the zen-sched branch in git yet? or any of the release notes from past releases?

BFS is NOT CFS, if you think it's just a patch on top of cfs you are wrong, it totally replaces it. Zen-sched allows the choice of cfs or bfs (current stable zen, 2.6.31, has 2.6.32 CFS backported) - And why would you patch a 2.6.32 kernel with the bfs patch, only with the intention of using CFS?

And bfs will not apply over the 2.6.32 kernel, and it would not work (the bfs in zen has been changed a bit to accomidate 2.6.32 changes, as i said above I backported the cfs from 2.6.32, of course this meant changes to BFS to make it work too, --- when zen.git is rebased to 2.6.32 bfs will only need a few changes to work because of this, and unless you know how to go about it you wont get bfs compiling on a 2.6.32 kernel)

2.) Why not use zen? is there a problem cuz we can fix em'

3.) Deadline is an i/o scheduler, CFS/BFS are cpu schedulers, i think you have them mixed up.
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Fri Sep 25, 2009 9:32 pm    Post subject: Reply with quote

cheater1034,
Sorry- kernel 2.6.32 was a typo, which I edited to 2.6.31 within 5 minutes of posting the comment. You must have read it immediately after I posted right before I edited, and then posted much later.

On the i/o or cpu scheduler thing- I'm not sure what you mean. Deadline has always been right in the same config section as cfs and anticipatory. Since bfs was suppose to replace cfs, I thought it would be grouped in where cfs is located.

I've always been able to change the default between cfs, deadline, or anticipatory using elevator=xxx on the kernel line, and reboot. If deadline is a different type of scheduler (i/o), why is it an option in the cpu scheduler section, and available with an elevator=deadline kernel line? Makes no sense to me. :?

Are you guys saying that when you patch a vanilla kernel with the appropriate bfsxxx version, there will be no bfs option shown anywhere when you configure and compile your kernel, and cfs is suppose to still be an option shown in make xconfig (which it is), even though patching with bfs is suppose to replace it?

Apparently, I'm having trouble making sense out of this. :roll:

I used to use zen kernels all the time- and loved them, but since zen went to the git method access (sorry- I find it too much trouble) I haven't used zen. Obviously the maintainer has very good reasons for using git, but for an old-school guy who is used to patching kernels with one .bz2 file, git seems overly complicated and irritating. That's just me, though- I'm sure most people like git. Maybe I'll give it another try.
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Fri Sep 25, 2009 9:48 pm    Post subject: Reply with quote

wrc1944 wrote:
cheater1034,
Sorry- kernel 2.6.32 was a typo, which I edited to 2.6.31 within 5 minutes of posting the comment. You must have read it immediately after I posted right before I edited, and then posted much later.

On the i/o or cpu scheduler thing- I'm not sure what you mean. Deadline has always been right in the same config section as cfs and anticipatory. Since bfs was suppose to replace cfs, I thought it would be grouped in where cfs is located.

I've always been able to change the default between cfs, deadline, or anticipatory using elevator=xxx on the kernel line, and reboot. If deadline is a different type of scheduler (i/o), why is it an option in the cpu scheduler section, and available with an elevator=deadline kernel line? Makes no sense to me. :?

Are you guys saying that when you patch a vanilla kernel with the appropriate bfsxxx version, there will be no bfs option shown anywhere when you configure and compile your kernel, and cfs is suppose to still be an option shown in make xconfig (which it is), even though patching with bfs is suppose to replace it?

Apparently, I'm having trouble making sense out of this. :roll:

I used to use zen kernels all the time- and loved them, but since zen went to the git method access (sorry- I find it too much trouble) I haven't used zen. Obviously the maintainer has very good reasons for using git, but for an old-school guy who is used to patching kernels with one .bz2 file, git seems overly complicated and irritating. That's just me, though- I'm sure most people like git. Maybe I'll give it another try.


No, you can make your own patches with gitweb: http://zen-sources.org/content/why-isnt-patch-downloads-page-not-same-version-git

CFS is NOT an i/o scheduler, CFQ is an i/o scheduler, the i/o schedulers in the mainstream kernel are pluggable (with the elevator= parameter), cpu schedulers have never been pluggable (due to linus opposition). Zen's i/o schedulers include upstream (anticipatory, deadline, cfq, no-op) and some additional ones (bfq, vr, fifo)

bfs is CPU scheduler, cfs is the mainstream CPU scheduler. so the bfs patch from ck simply replaces cfs with the bfs scheduler (totally different cpu schedulers).

So, naturally, there is no Kconfig option for the BFS scheduler with the bfs patch, BUT in zen there IS an option to select the cpu scheduler (cfs or bfs) along with other tweaks (cfs boost, X autoiso, 2.6.32 cfs). So the option to choose between BFS and CFS is ZEN-specific.
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
rahulthewall
Veteran
Veteran


Joined: 01 Nov 2007
Posts: 1264
Location: Zürich

PostPosted: Fri Sep 25, 2009 9:51 pm    Post subject: Reply with quote

wrc1944 wrote:
cheater1034,
Sorry- kernel 2.6.32 was a typo, which I edited to 2.6.31 within 5 minutes of posting the comment. You must have read it immediately after I posted right before I edited, and then posted much later.

On the i/o or cpu scheduler thing- I'm not sure what you mean. Deadline has always been right in the same config section as cfs and anticipatory. Since bfs was suppose to replace cfs, I thought it would be grouped in where cfs is located.

I've always been able to change the default between cfs, deadline, or anticipatory using elevator=xxx on the kernel line, and reboot. If deadline is a different type of scheduler (i/o), why is it an option in the cpu scheduler section, and available with an elevator=deadline kernel line? Makes no sense to me. :?

Are you guys saying that when you patch a vanilla kernel with the appropriate bfsxxx version, there will be no bfs option shown anywhere when you configure and compile your kernel, and cfs is suppose to still be an option shown in make xconfig (which it is), even though patching with bfs is suppose to replace it?

Apparently, I'm having trouble making sense out of this. :roll:

I used to use zen kernels all the time- and loved them, but since zen went to the git method access (sorry- I find it too much trouble) I haven't used zen. Obviously the maintainer has very good reasons for using git, but for an old-school guy who is used to patching kernels with one .bz2 file, git seems overly complicated and irritating. That's just me, though- I'm sure most people like git. Maybe I'll give it another try.


Maybe this will help you.

Task Schedulers - CFS and BFS. CFS is the default and BFS is the new thing.

I/O Schedulers - CFQ (please note: CFQ and not CFS). Other examples include deadline, etc.

You are confusing CFQ and CFS. The kernel always (well since I use Linux anyways) always had the option to chose the I/O scheduler. However, the kernel developers believe it is counter-productive to include more than one task scheduler in the kernel and they stay with CFS. Con believes that CFS does not scale well on desktops (or machine with less than 8 cores - you know machines which common people tend to use) and comes up with BFS - another task scheduler as a "replacement" to CFS. Since he has no intention of getting this accepted into the mainline kernel the patch he provides simply "replaces" CFS in the kernel. The new task scheduler (after applying the patch) is BFS - it is a completely new task scheduler which has got nothing to do with CFS. Nothing at all.

Hope that clears some of your confusion.

EDIT: cheater1034 types way too fast. :P
_________________
Who shall guard the guards?
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Sat Sep 26, 2009 12:26 am    Post subject: Reply with quote

cheater1034 & rahulthewall,
Thanks much for the clarifications. I was unconsciously transferring cfq to cfs in my mind, even though I've known they are different. Guess I was trying to think about and do too many things at the same time. :roll: :oops:

Making my own patches with gitweb looks like a great way to keep up with this- thanks for the link!
_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
tcoffeep
n00b
n00b


Joined: 29 Jun 2007
Posts: 33
Location: Timmins, Ontario

PostPosted: Sat Sep 26, 2009 6:00 pm    Post subject: Reply with quote

Are there any advantages to using the Zen Sources over the Gentoo sources? I'm sorry, but I haven't the time to really delve into this, so I figure asking would be best.
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Sat Sep 26, 2009 11:53 pm    Post subject: Reply with quote

tcoffeep wrote:
Are there any advantages to using the Zen Sources over the Gentoo sources? I'm sorry, but I haven't the time to really delve into this, so I figure asking would be best.


Complex answer:
Tons, the gentoo sources are basically just the upstream linux kernel with a few cosmetic patches (like fbcondecor) - and over no real advantages over the upstream linux kernel.

The zen kernel includes tons of code not found in the upstream linux kernel, and several facets of the code are implemented with performance basis in mind (particularly for desktops)

You can read up on the faq: http://zen-sources.org/content/faq

Of course you will get the most out of it if you use the additional features it provides or updates (drm and nouveau, phc, madwifi, reiser4, thinkpad sl, smapi, aufs2) - There's obviously a lot more, and what i mentioned are just features.

You will see a significant margin of improvement in benchmarks compared to the mainline kernel out of the box because of updates and other misc things that are done (tweaks, updated scheduler code) - and you can maximize your performance by using things like the slqb slab allocator (not found in mainline), bfq i/o scheduler (several zen users have reported significant performance improvement over cfq), 2.6.32 cfs cpu scheduler (mainline scheduler), and con's bfs cpu scheduler (designed for desktops - http://ck.kolivas.org/patches/bfs/bfs-faq.txt) -- cpu schedulers are switchable at build-time which is zen-specific as well (bfs still has problems on many hardwares)

Simply answer:
YES

Conclusion:
On a desktop system, I wouldn't use any other kernel source. -- (please use git or make a patch using gitweb as described here: http://zen-sources.org/content/why-isnt-patch-downloads-page-not-same-version-git)

And you want zen-stable.git, not zen.git -- i think you will be happy with your findings
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
wrc1944
Advocate
Advocate


Joined: 15 Aug 2002
Posts: 3231
Location: Gainesville, Florida

PostPosted: Mon Sep 28, 2009 6:02 pm    Post subject: Reply with quote

Anyone tried or looked at this thread: Kernel 2.6.30/31 desktop interactivity patch (no BFS)

https://forums.gentoo.org/viewtopic-t-794535.html?sid=c167df2a40339680e3a0c7a77a2ee84d

Quote:
Here's a small patch that tweaks the mainline scheduler (CFS) to perform the same way as BFS

_________________
Main box- AsRock x370 Gaming K4
Ryzen 1700, 3.0GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
Gentoo ~amd64 plasma, glibc-2.29-r2, gcc-9.1.0 kernel-5.0.14-gentoo USE=experimental
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Mon Sep 28, 2009 6:16 pm    Post subject: Reply with quote

wrc1944 wrote:
Anyone tried or looked at this thread: Kernel 2.6.30/31 desktop interactivity patch (no BFS)

https://forums.gentoo.org/viewtopic-t-794535.html?sid=c167df2a40339680e3a0c7a77a2ee84d

Quote:
Here's a small patch that tweaks the mainline scheduler (CFS) to perform the same way as BFS


Mmmhmmm, not to be mean but that patch wont make it close to bfs, read my response and compare bfs/cfs in zen
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
rolis
n00b
n00b


Joined: 24 Apr 2008
Posts: 12

PostPosted: Mon Sep 28, 2009 10:59 pm    Post subject: Reply with quote

hi,

recently i've started getting exception while loading virtualbox kernel driver vboxdrv with master zen-stable.git:
Code:
[   43.650904] vboxdrv: Trying to deactivate the NMI watchdog permanently...
[   43.650911] vboxdrv: Successfully done.
[   43.650916] vboxdrv: Found 4 processor cores.
[   43.651091] BUG: unable to handle kernel paging request at ffffffffa028f5e0
[   43.651229] IP: [<ffffffffa028f466>] RTHeapSimpleInit+0x66/0x100 [vboxdrv]
[   43.651252] PGD 1003067 PUD 1007063 PMD 22a6da067 PTE 228634161
[   43.651252] Oops: 0003 [#1] PREEMPT SMP
[   43.651520] last sysfs file: /sys/devices/system/cpu/cpu3/cpufreq/ondemand/up_threshold
[   43.651641] CPU 2
[   43.651641] Modules linked in: vboxdrv(+) nls_utf8 cifs nfs lockd auth_rpcgss af_packet k8temp sunrpc f71882fg snd_seq snd_seq_device powernow_k8 hid_microsoft usbhid hid snd_hda_codec_atihdmi snd_hda_codec_realtek snd_hda_intel snd_hda_codec ehci_hcd snd_pcm firewire_ohci ohci_hcd thermal processor r8169 evdev snd_timer usbcore thermal_sys sr_mod firewire_core pcspkr cdrom i2c_piix4 snd nls_base hwmon button mii soundcore crc_itu_t snd_page_alloc unix
[   43.651641] Pid: 2014, comm: modprobe Not tainted 2.6.31-zen2-bba46088cb #9 RD790A01
[   43.651641] RIP: 0010:[<ffffffffa028f466>]  [<ffffffffa028f466>] RTHeapSimpleInit+0x66/0x100 [vboxdrv]
[   43.651641] RSP: 0018:ffff880228481f00  EFLAGS: 00010206
[   43.651641] RAX: ffffffffa028f5e0 RBX: ffffffffa028f5e0 RCX: 0000000000000000
[   43.651641] RDX: 0000000000180000 RSI: 00000000abcdef01 RDI: ffffffffa041ca48
[   43.651641] RBP: 0000000000180000 R08: ffff800000000000 R09: 0000000000000000
[   43.651641] R10: 0000000000000000 R11: 0000000000000001 R12: 00000000008966d0
[   43.651641] R13: 00000000008968e0 R14: 00000000008966d0 R15: 0000000000896940
[   43.651641] FS:  00007fb25f6066f0(0000) GS:ffff880028064000(0000) knlGS:0000000000000000
[   43.651641] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[   43.651641] CR2: ffffffffa028f5e0 CR3: 0000000228487000 CR4: 00000000000006e0
[   43.651641] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   43.651641] DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
[   43.651641] Process modprobe (pid: 2014, threadinfo ffff880228480000, task ffff88022e8b2700)
[   43.651641] Stack:
[   43.651641]  ffffffffa0285118 0000000000000000 ffffffffa0282f37 000000000000b7d2
[   43.651641] <0> 0000000000000000 ffffffffa00c2000 ffffffffa00c20f3 0000000000000000
[   43.651641] <0> ffffffff81009046 ffffffffa041bb00 00000000001ae237 00000000008966d0
[   43.651641] Call Trace:
[   43.651641]  [<ffffffffa0285118>] ? RTR0MemExecDonate+0x38/0x60 [vboxdrv]
[   43.651641]  [<ffffffffa0282f37>] ? RTR0Init+0x37/0x60 [vboxdrv]
[   43.651641]  [<ffffffffa00c2000>] ? VBoxDrvLinuxInit+0x0/0x1a1 [vboxdrv]
[   43.651641]  [<ffffffffa00c20f3>] ? VBoxDrvLinuxInit+0xf3/0x1a1 [vboxdrv]
[   43.651641]  [<ffffffff81009046>] ? do_one_initcall+0x26/0x180
[   43.651641]  [<ffffffff8106a27c>] ? sys_init_module+0xbc/0x220
[   43.651641]  [<ffffffff8100b3ab>] ? system_call_fastpath+0x16/0x1b
[   43.651641] Code: c0 85 c8 0f 85 a5 00 00 00 48 8d 44 16 ff 48 39 c2 0f 83 91 00 00 00 48 89 f1 48 89 f0 83 e1 1f 75 73 48 83 e2 e0 be 01 ef cd ab <48> c7 00 05 01 59 19 48 8d 0c 10 48 89 50 08 48 c7 40 30 ff ff
[   43.651641] RIP  [<ffffffffa028f466>] RTHeapSimpleInit+0x66/0x100 [vboxdrv]
[   43.651641]  RSP <ffff880228481f00>
[   43.651641] CR2: ffffffffa028f5e0
[   43.651641] ---[ end trace 4420cd0deadb9ae9 ]---

it looks like it's memory allocation related. i've noticed it first last friday night/saturday morning. i think it was commit fffbc87215c183ec5de8475114f11c0f9ee2d8fd, not 100% sure.

last commit that i've built and it still was working is 660341e072099b280362778dee1f62a176747a3e, dated Wed, 23 Sep 2009 15:25:38 +0000.
the latest version (bba46088cba3d29484abbd6e6bac2046951536a2) still does not work.

i'm using CFS scheduler and i've tried all 3 SLAB options with the same result. this is on x86_64 system, phenom2 940.

thx,
rolis
Back to top
View user's profile Send private message
cheater1034
Veteran
Veteran


Joined: 09 Sep 2004
Posts: 1558

PostPosted: Mon Sep 28, 2009 11:07 pm    Post subject: Reply with quote

Thanks rolis for that trace,
I've heard the same thing from in #zen-sources about that too, and thanks VERY much for providing that commit.

I'll edit this post in a bit for a possible fix.

*edit*
It looks like it is either that merge or 2.6.31.1,

I reverted it in master, could you please pull and try again? and please report back :D
_________________
IRC!: #zen-sources on irc.rizon.net
zen-kernel.org
--
Lost in android development land.
Back to top
View user's profile Send private message
rolis
n00b
n00b


Joined: 24 Apr 2008
Posts: 12

PostPosted: Mon Sep 28, 2009 11:20 pm    Post subject: Reply with quote

cheater1034 wrote:
Thanks rolis for that trace,
I've heard the same thing from in #zen-sources about that too, and thanks VERY much for providing that commit.

I'll edit this post in a bit for a possible fix.

*edit*
It looks like it is either that merge or 2.6.31.1,

I reverted it in master, could you please pull and try again? and please report back :D

no problem, thank YOU guys for keeping it up. pulling lastest as we speak. report back in a while.

EDIT: yes, it works now!
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Gentoo Forums Forum Index Unsupported Software All times are GMT
Goto page Previous  1, 2, 3 ... , 10, 11, 12  Next
Page 11 of 12

 
Jump to:  
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum