View previous topic :: View next topic |
Author |
Message |
kontol n00b
Joined: 19 Dec 2004 Posts: 31
|
Posted: Mon Dec 20, 2004 5:41 pm Post subject: |
|
|
fallow wrote: | Gentree wrote: |
Hopefully I will be able to use the vv kernel. |
nice I`m waiting for 2.6.10-rc4 ( bk13 today ) to do next vv_e release
I wad done bonnie++ tests again on my machine :
AMD Athlon 1.0 , 256 MB Ram, HDD ST3120026A , VIA KT133
kernels : 2.6.9-nitro4, 2.6.10-rc3-vv_e1 , 2.6.10-rc3-vvmm_e1
every test has 2 repeations ( to see randomization of latencies,reiser4 especially) t1 - first repeat , t2 - second repeat.
so it`s easy to see that results of first and second test for the same kernel can be much different hehe
selected I/O Scheduler in all kernels : Deadline
results here : text formated with Opera7.60p4
Code: |
Version 1.93c ------Sequential Output------ --Sequential Input- --Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec %CP
nitro-t1 2G 83 98 28119 28 13343 12 188 99 31990 17 129.9 7
nitro-t2 2G 82 98 29816 20 13202 13 192 99 32522 15 128.5 7
vv_e1-t1 2G 81 97 26637 19 12873 18 149 99 31427 19 123.6 7
vv_e1-t2 2G 78 97 26555 22 12899 19 148 99 31066 19 121.9 7
vvmm_e1-t1 2G 74 97 27244 20 17004 23 156 98 20545 31 106.0 8
vvmm_e1-t2 2G 81 97 28423 26 16958 22 169 98 20526 31 104.5 8
nitro-t1 158ms 11370ms 2638ms 48835us 57986us 3692ms
nitro-t2 162ms 4340ms 4828ms 51951us 59802us 3952ms
vv_e1-t1 181ms 10241ms 3724ms 66494us 47453us 3937ms
vv_e1-t2 182ms 15744ms 2953ms 67635us 62251us 3586ms
vvmm_e1-t1 218ms 8854ms 4560ms 78992us 23805us 4356ms
vvmm_e1-t2 185ms 9475ms 4659ms 77441us 20931us 4036ms
--------------------------Sequential Create---------------Random Create--------
Enterprise -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP
nitro-t1 16 14455 79 +++++ +++ 7426 99 7975 96 +++++ +++ 7416 99
nitro-t2 16 14956 83 +++++ +++ 7364 99 7943 96 +++++ +++ 7529 99
vv_e1-t1 16 15672 77 +++++ +++ 7532 99 8069 96 +++++ +++ 7740 99
vv_e1-t2 16 15609 78 +++++ +++ 7606 99 8096 96 +++++ +++ 7700 99
vvmm_e1-t1 16 15958 76 +++++ +++ 7205 99 7900 96 +++++ +++ 7390 99
vvmm_e1-t2 16 16286 75 +++++ +++ 7281 99 8097 96 +++++ +++ 7518 99
nitro-t1 8627us 7945us 8256us 10059us 27us 343us
nitro-t2 391us 8841us 8822us 6592us 23us 349us
vv_e1-t1 306us 8347us 8460us 8036us 25us 344us
vv_e1-t2 301us 8205us 8307us 6508us 26us 349us
vvmm_e1-t1 648us 8102us 8393us 8523us 21us 351us
vvmm_e1-t2 205us 7645us 8023us 13561us 22us 356us
|
cheers |
Fallow ..request please can we have hostap,ipw/ipw2100 and orinoco patch...in your next realease...I am quiet happy to use..this kernel on my two laptops...
Cheers
Cheerss |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Mon Dec 20, 2004 6:03 pm Post subject: |
|
|
kontol wrote: |
Fallow ..request please can we have hostap,ipw/ipw2100 and orinoco patch...in your next realease...I am quiet happy to use..this kernel on my two laptops...
Cheerss |
OK , I take a look and going to add them to the next release.
btw.
Orinoco is in 2.6.10-rc3-vvmm_e1 already via USE="orinoco"
Can You give some links,pointers to the 2 ohters ?
I think that with patching by hand ( 2610rc3vvmm_e1_orinoco.diff.bz2 from 2.6.10-rc3-vvmm_e1 at sourceforge.net vivid page ) will be no problems with 2.6.10-rc3-vv_e1 also.
cheers _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard |
|
Back to top |
|
|
rommel Veteran
Joined: 19 Apr 2002 Posts: 1145 Location: Williamsburg Virginia
|
Posted: Mon Dec 20, 2004 6:45 pm Post subject: |
|
|
hey Fallow , possibly an ignorant question but how are you setting the scheduler... elevator=cfq say on the kernel line doesnt seem to be working for me... i only see that the various schedulers are registered but in dmesg there isnt a specific reference to one being used like in 2.6.9-nitro4 |
|
Back to top |
|
|
vrln Guru
Joined: 11 Sep 2004 Posts: 534 Location: Finland
|
Posted: Mon Dec 20, 2004 7:01 pm Post subject: |
|
|
Try this:
cat /sys/block/hda/queue/scheduler
It should present you with a list of schedulers, the one inside [] is the one being used right now. You can change it by for example echo cfq > scheduler afaik. Be careful though, I managed to crash my system once with 2.6.10-rc1 when I changed it all the time
btw, about that patch comment, sorry, I can't code at all :/ Anyway, don't make the staircase an option if it's just me wanting it, in the end it's good that there are different patchsets and I respect your point about the O(1) scheduler. |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Mon Dec 20, 2004 7:35 pm Post subject: |
|
|
Rommel wrote: | hey Fallow , possibly an ignorant question but how are you setting the scheduler... elevator=cfq say on the kernel line doesnt seem to be working for me... i only see that the various schedulers are registered but in dmesg there isnt a specific reference to one being used like in 2.6.9-nitro4 |
yes as Vrln said , since 2.6.10-rc1 with switchable and modular io schedulers there are no "using xxx scheduler" msg in dmesg.
Only : Code: |
io scheduler noop registered
io scheduler anticipatory registered
io scheduler deadline registered
io scheduler cfq registered |
is absolutely correct.
actual io scheduler for device You can set and check via /sys/block/_DEVICE_/queue/scheduler. for ex Code: |
Enterprise queue # cd /sys/block/hda/queue
Enterprise queue # cat scheduler
noop anticipatory deadline [cfq]
Enterprise queue # echo deadline > scheduler
Enterprise queue # cat scheduler
noop anticipatory [deadline] cfq |
default io scheduler You can set via vivid_e menu from (menu)config
Code: |
x x Default IO scheduler (Deadline io scheduler) ---> x x
|
or via elevator=name in bootparams.
that`s all
Vrln wrote: |
Anyway, don't make the staircase an option if it's just me wanting it, in the end it's good that there are different patchsets and I respect your point about the O(1) scheduler. |
Yes it`s a fact , is no reason to make a copy of other/different patchset with the same stuff.
But via Con`s plugsched it would be not so complicated as I think based on mm . May be a little more complicated to "port" plugsched to non-mm based kernel but I will take a look in the near future . Syncing plugsched to latest mm byc Con is also an option for vvmm.
But highest prior is to have good working next vv release
cheers _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Sun Dec 26, 2004 6:47 am Post subject: |
|
|
Well, I'm back again. I finally got everything worked out. Turns out it was partially the way I had the video drivers set up and mostly because entrance decided to die on me.
Anyways, I finally got around to building myself a new kernel a few days ago--after I was sure I had all the bugs in my system worked out. So far I haven't had any problems, crashes, blips or schedular issues (using CFQ). I like the patch set, and I really like the extra menu option in the menuconfig. Very handy. Good work!
I do, however, have one question, I got a brand new WD1600JB drive today (sata), and I was considering formatting it in reiser4. However, I couldn't seem to find the kernel option to enable support. My impression from the patch listings was that it was in the latest non-mm release. But darned if I can find it. Is it actually there? Or did I screw something up? |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Sun Dec 26, 2004 9:18 am Post subject: |
|
|
great & nice
yes in the latest non-mm release : vv_e1 reiser4 is from mm , in vvmm_e1 based on full mm reiser4 is in mm , so is not in the patchlist of release.
I also have all partitions in reiser4 , so it`s "must be" for me
of course I`m going to do next 2.6.10 release at the near next days.
cheers _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Sun Dec 26, 2004 5:27 pm Post subject: |
|
|
Well, I still can't find it. Something must have happened when I compiled the kernel source. Anyways, I might well wait till the next release since I'm going out of town for a few days tomorrow. I'll play with reiser4 when I get back. Happy holidays. |
|
Back to top |
|
|
kontol n00b
Joined: 19 Dec 2004 Posts: 31
|
Posted: Sun Dec 26, 2004 7:05 pm Post subject: |
|
|
fallow wrote: | kontol wrote: |
Fallow ..request please can we have hostap,ipw/ipw2100 and orinoco patch...in your next realease...I am quiet happy to use..this kernel on my two laptops...
Cheerss |
OK , I take a look and going to add them to the next release.
btw.
Orinoco is in 2.6.10-rc3-vvmm_e1 already via USE="orinoco"
Can You give some links,pointers to the 2 ohters ?
I think that with patching by hand ( 2610rc3vvmm_e1_orinoco.diff.bz2 from 2.6.10-rc3-vvmm_e1 at sourceforge.net vivid page ) will be no problems with 2.6.10-rc3-vv_e1 also.
cheers |
I've found the link for hostap http://hostap.epitest.fi/ and also I've found kernel with orinoco,hostap,ipw2100 in it https://forums.gentoo.org/viewtopic.php?p=1893385&highlight=#1893385, but unfortunately...loook like this project is dead...but I am not really sure.
If you put those patches in your next 2.6.10 kernel,I'd like to try in my two laptops.
Cheers |
|
Back to top |
|
|
Evil Dark Archon Guru
Joined: 21 Dec 2002 Posts: 562 Location: Santa Rosa, CA
|
Posted: Sun Dec 26, 2004 7:21 pm Post subject: |
|
|
not dead, just on stand-by while waiting for xiphux to get a new computer (my patchset is based on xiphux's xx-sources) _________________ This post has been over explained for newb-informing purposes.
Registered Linux user 347334
Abit AV8-3rd eye, AMD Athlon64 3500+ 90nm, ATI Radeon x850 pro |
|
Back to top |
|
|
bdad n00b
Joined: 20 Mar 2004 Posts: 18
|
Posted: Tue Dec 28, 2004 4:16 pm Post subject: |
|
|
Genn wrote: | Well, I'm back again. I finally got everything worked out. Turns out it was partially the way I had the video drivers set up and mostly because entrance decided to die on me.
Anyways, I finally got around to building myself a new kernel a few days ago--after I was sure I had all the bugs in my system worked out. So far I haven't had any problems, crashes, blips or schedular issues (using CFQ). I like the patch set, and I really like the extra menu option in the menuconfig. Very handy. Good work!
I do, however, have one question, I got a brand new WD1600JB drive today (sata), and I was considering formatting it in reiser4. However, I couldn't seem to find the kernel option to enable support. My impression from the patch listings was that it was in the latest non-mm release. But darned if I can find it. Is it actually there? Or did I screw something up? | Under kernel hacking in menuconfig, make sure 4K stacks are off. Then reiser4 should appear. hih - B |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Fri Dec 31, 2004 12:01 am Post subject: |
|
|
Quote: | Under kernel hacking in menuconfig, make sure 4K stacks are off. Then reiser4 should appear. hih - B |
Ah, excellent. That did make the option appear. Thanks.
I do have a question though. I've always had 4k stacks on, will it change anything having them off? (In other words, I have no idea what they do but I always thought they were good to have.)
I just got back in town about 20 minutes ago after a near 6 hour drive, so I'll play with this tomorrow. |
|
Back to top |
|
|
bdad n00b
Joined: 20 Mar 2004 Posts: 18
|
Posted: Fri Dec 31, 2004 12:28 am Post subject: |
|
|
Genn wrote: | Quote: | Under kernel hacking in menuconfig, make sure 4K stacks are off. Then reiser4 should appear. hih - B |
Ah, excellent. That did make the option appear. Thanks.
I do have a question though. I've always had 4k stacks on, will it change anything having them off? (In other words, I have no idea what they do but I always thought they were good to have.)
I just got back in town about 20 minutes ago after a near 6 hour drive, so I'll play with this tomorrow. | Check out http://lwn.net/Articles/84583/ for an article on 4k stacks. Have fun! |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Fri Dec 31, 2004 7:57 am Post subject: |
|
|
An interesting article. A little more research gave me the impression that it really won't affect performace either unless it's on a server. So I guess there's no reaosn not to use 8k stacks so I can use reiser4.
On an side note though, it does appear that you can use reiser4 with 4k stacks. None of the posts give much information, but if I happen to find something conclusive I'll post it. (The biggest find I made was something from another kernel-source thread https://forums.gentoo.org/viewtopic.php?t=272784 where they bring up a patch that isn't really a patch. It's about halfway down.) Tomorrow I may look into it more. |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Sun Jan 02, 2005 10:33 pm Post subject: |
|
|
hello
I`m going to the Hospital for a several days, so I do not have time to do a release at now.I started work today but I don`t have time to finish it.
I will do of course new release when I back to home.
so,maybe You have some requests/ideas for new release
Everything are welcome _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard |
|
Back to top |
|
|
kontol n00b
Joined: 19 Dec 2004 Posts: 31
|
Posted: Mon Jan 03, 2005 1:50 am Post subject: |
|
|
fallow wrote: | hello
I`m going to the Hospital for a several days, so I do not have time to do a release at now.I started work today but I don`t have time to finish it.
I will do of course new release when I back to home.
so,maybe You have some requests/ideas for new release
Everything are welcome |
Why not... # USE ="hostap orinoco" emerge vivid-sources
Cheers |
|
Back to top |
|
|
rommel Veteran
Joined: 19 Apr 2002 Posts: 1145 Location: Williamsburg Virginia
|
Posted: Mon Jan 03, 2005 7:52 pm Post subject: |
|
|
fallow i hope it isnt anything serious and in either case best of luck |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Tue Jan 04, 2005 7:42 am Post subject: |
|
|
I too hope it is nothing serious. May your stay there be swift and uneventful. |
|
Back to top |
|
|
gringo Advocate
Joined: 27 Apr 2003 Posts: 3793
|
Posted: Tue Jan 04, 2005 10:05 am Post subject: |
|
|
yea, hope youre fine, let us know if you need some help !
cheers |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Wed Jan 05, 2005 7:26 pm Post subject: |
|
|
I felt like accomplishing something before school starts tomorrow. So I wrote a little script to do the scheduler changes--not that it is really needed. It's still rough, but it works. If anyone has any thoughts on what might actually make it useful, let me know.
------------------------------------------------------------
New highly improved version of the script below. It now features a command line arguement block device selector, and a command line arguement scheduler selector with partial matching.
It is also available at http://genn.sitesled.com/files/scripts/scheduler_change_script
Code: | #!/bin/bash
###***Authored by Genn***###
##E-Mail: TheWindRider@myrealbox.com
#Last Update: 1/21/05
#This script was created to change the currently used scheduler
#under the vivid-sources kernel.
#(http://forums.gentoo.org/viewtopic.php?p=1337010, maintained
#by fallow and _troll_) There are probably other kernels that
#could use this script as well, but I'm not aware of them.
#This script must be executable. If it does not work,
#'chmod 700 *this_script*' should do it. It also
#requires root access. The script will inform you of this
#if you run it as a normal user.
#You can supply one optional arguement with this script.
#'script_name block-device' will run the script with the
#intend to interactively change the scheduler on the
#block device you specify. The block device is just the
#last portion of the device address without slashes.
#IE, if you wanted to adjust the schedular on /dev/hda,
#you would run 'this_script hda' and follow the on-screen
#instructions.
#Now, I say the arguement is optional because you are
#requred to set the DEFAULT_DRIVE variable to a valid block
#device. So, if you do not supply the arguement, the script
#will use that block-device. The variable is set by the
#same rules that the arguement is supplied by; just use the
#last portion of the address with no slashes.
#You can now add a second command line arguement to change
#the scheduler non-interactively. Itt is required that
#you specify the drive on the command line as well though.
#'scheduler_change_script cfq' will fail.
#'schedular_change_script hda cfq' will work.
#You can still just specify the drive on the command line
#and change the schedular from the list.
#This is the all important variable.
DEFAULT_DRIVE=hdb
#Now you've done everything you need to, go run the script,
#and ignore everything from here down.
##########################################################
#Verification of a valid default block device choice.
if [ -z `ls /sys/block | grep $DEFAULT_DRIVE` ]
then
echo "You chose a non-valid block device as default."
echo "/dev/$DEFAULT_DRIVE does not exist."
echo "Please change the variable DEFAULT_DRIVE in the script."
exit
fi
#Verification of root level access
ROOT_UID=0
ERR_NOTROOT=67
if [ "$UID" != "$ROOT_UID" ]
then
echo "You must be root to run this script."
exit $ERR_NOTROOT
fi
#If an arguement is supplied this makes sure it is used.
#Or if none is supplied it uses the default choice.
if [ -n "$1" ]
then
if [ -n `ls /sys/block | grep $1` ]
then
echo "Now using /dev/$1 for block device."
DRIVE=$1
else
echo "You chose a non-valid block device."
echo "/dev/$1 does not exist."
exit 1
fi
else
echo "No arguement supplied."
echo "Defaulting to /dev/${DEFAULT_DRIVE} for block device."
DRIVE=${DEFAULT_DRIVE}
fi
#Schedular file location
SCHEDLOC=/sys/block/${DRIVE}/queue/scheduler
#Verification of valid scheduler location
if [ `[ -f ${SCHEDLOC} ]` ! -a `[ -w ${SCHEDLOC} ]` ]
then
echo "Uh oh, either $SCHEDLOC is not a file, or you do not "
echo "have write access to it. I would suggest you verify"
echo "that the location listed above is correct and that the "
echo "value in the script did not get altered somewhow. The "
echo "standard location is: "
echo "/sys/block/{some_block_device}/queue/scheduler. If it "
echo "is valid, try echoing something by hand to "
echo "$SCHEDLOC. If you're "
echo "sure your using the right kernel, and none of that works, "
echo "make a post in the vivid-sources thread. On the other "
echo "hand, if echoing by hand does work, send me an e-mail as "
echo "you just discovered a bug."
exit 1
fi
#Identification of different available schedulers.
SCHED1=`cat ${SCHEDLOC} | gawk '{print $1}'`
SCHED2=`cat ${SCHEDLOC} | gawk '{print $2}'`
SCHED3=`cat ${SCHEDLOC} | gawk '{print $3}'`
SCHED4=`cat ${SCHEDLOC} | gawk '{print $4}'`
#Non-interactive mode
if [ -n "$2" ]
then
if [ `echo $SCHED1 | grep $2` ]
then
NEWSCHED=${SCHED1}
elif [ `echo $SCHED2 | grep $2` ]
then
NEWSCHED=${SCHED2}
elif [ `echo $SCHED3 | grep $2` ]
then
NEWSCHED=${SCHED3}
elif [ `echo $SCHED4 | grep $2` ]
then
NEWSCHED=${SCHED4}
else
echo "Unable to match schedular arguement to a known scheduler."
echo "You used $2"
echo "The valid options are; $SCHED1, $SCHED2, $SCHED3, $SCHED4."
echo "The bracketed one is currently in use."
echo "Partial matches are ok."
echo "Exiting."
exit 1
fi
fi
#User Interface portion
if [ -z "$2" ]
then
echo "List of available schedulers"
echo "The bracketed choice is the one currectly in use."
echo "Selecting this one does nothing."
echo "1 - ${SCHED1}"
echo "2 - ${SCHED2}"
echo "3 - ${SCHED3}"
echo "4 - ${SCHED4}"
echo " "
echo "Please type the number corresponding the scheduler you wish to use."
#This applies the selected change
read CHOICE
case ${CHOICE} in
1)
NEWSCHED=${SCHED1}
;;
2)
NEWSCHED=${SCHED2}
;;
3)
NEWSCHED=${SCHED3}
;;
4)
NEWSCHED=${SCHED4}
;;
*)
echo "You must input a numerical value corresponding to one of the not-in-use schedulers."
exit 1
;;
esac
fi
#Error catching if the current scheduler is chosen.`
if [ `echo $NEWSCHED | grep -F [` ]
then
echo ""
echo "You just chose the scheduler currently in use."
echo "Please choose a different one next time."
exit 1
fi
#Actual aplication of new scheduler
echo "$NEWSCHED" > $SCHEDLOC
echo " "
#Announcement of what schedular was selected for which drive.
echo "Schedular changed to $NEWSCHED on /dev/${DRIVE}"
exit 0 |
Last edited by Genn on Sat Jan 22, 2005 12:36 am; edited 3 times in total |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Mon Jan 10, 2005 3:00 pm Post subject: |
|
|
hello
Thanks for every good word from Everyone
It`s serious - skin cancer , but as I hope everything is going to the right/good way.I`m waiting for the HP result after last operation.
I`m returned to home. I feeling "good" and going to make new vivid release in next few days.
I see that is some news - genetics algorytms extensions for zapchod scheduler and some of io schedulers , but is not so good for destkop in current state of development and current cpu speeds (imho of course).
btw. also , the sched script is nice
cheers _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Tue Jan 11, 2005 4:09 am Post subject: |
|
|
I'm sorry to hear that it's skin cancer. I wish you the best of luck with it, and I hope things turn out well.
I'm also looking forward to this next release. It's been at least 2 weeks since I rebuild my kernel.
And thanks for the comment on the script...it's not overly useful, but it did the job of making me feel productive. |
|
Back to top |
|
|
fallow Bodhisattva
Joined: 08 Jan 2004 Posts: 2208 Location: Poland
|
Posted: Fri Jan 14, 2005 11:31 pm Post subject: |
|
|
first : thanks for the good words, everythings is going into the right way , I`m waiting for the post-operational hp result now.I think that it`s going to be good
ok next , I was done some kind of little bit testing release, 2.6.11-rc is fresh now. a less of mm fixes bacause much of they , are in vanilla already wow
as usually :
mission objectives wrote: | :
* 2.6.11-rc1 based O(1) CPU Sched with partial mm patches/fixes and /proc interface thanks to MJB patches.
* CFQ2 io sched with timeslice patch from J.Axboe
* Anticipatory io sched with genetic algor. implementation from J.Moilanen
* reiser4 from latest mm with all fixes
* additions like cachefs,uml extensions,fbsplash,vesa-tng,lufs,squashfs,inotify,bk-drivers alsa etc.,supermount etc
* and partial 2611rc1mm1 of course
* mjb swap autotune
//I added orinoco from dragorn and hostap (google my friend) but as I see both is in mm`s bk-net-dev, so i reverse it , and add patches from mm for both// |
Code: |
--patches--
vivid_menu_mm1.diff
a) [*] Preemptible Kernel (duplicated here for better visibility)
a) [*] Preempt The Big Kernel Lock (dup.also)
b) Default IO Scheduler (Deadline IO Scheduler) --->
c) Selected IO Schedulers for Your kernel (dup.also) --->
d) [*] Genetic Library (req.for Genetic Anticipatory)
d) [*] Genetic Anticipatory I/O scheduler (EXPERIMENTAL)
e) [*] Support for full write io fairness (CFQ-TIMESLICE PATCH)
f) [*] Quick Sort
f) Kernel HZ (1000 Hz) --->
f) [ ] CFLAGS hacking
f) [ ] Inotify file change notification support
a - preemptible kernel and big kernel lock now in vanilla as standard ( moved )
b - default io scheduler selection from menuconfig ( bootparams is working also)
c - selection of io scheds to compile ( moved for better visibility only )
d - genetic algor. and it implementation in anticipatory io scheduler from J.Moilanen
e - cfq 2 timeslice patch from J.Axboe
f - qsort implementation,kernel int.timer freq,cflags hack fixed by troll,inotify from 2611-rc1-mm1
--bk--
2.6.11-rc1-bk2
bk-acpi-revert-20041210.patch
bk-alsa.patch
bk-ide-dev.patch
bk-input.patch
bk-netdev.patch - orinoco & hostap here
bk-ntfs.patch
bk-usb.patch
bk-xfs.patch
--base--
nvidia_6111-6629_compat2.diff
acerhk.patch
cflags-selection.patch
cfq-time-slices-20.bz2
chmp-r4-2.6.9.patch
daconfig-2.1.1.patch
fbsplash-0.9.1-2.6.10-rc3-bk6.patch
genetic-as-sched.patch
genetic-io-sched.patch
genetic-lib-gl1.patch
linux-2.6.9-smbfs.patch
lirc-2.6.9-20041024-px1.patch
lufs-0.9.7-2.6.10-morph6.patch
squashfs2.0-patch
supermount-ng208-10ck1.diff
vesafb-tng-0.9-rc5-2.6.10-rc2.patch
016-cs461x_gameport.patch
030-moxa_user_copy_checking.patch
1g_lowmem1_i386.diff
2.6.6-qsort-updated-lkml.patch
2.6.6-xfs-qsort-lkml.patch
--mjb--
200-config_hz.txt
320-irqbal_fast.txt
350-autoswap.txt
380-protocol254.txt
390-slabtune.txt
410-fasync_lock_rcu.txt
480-physnode_map.txt
500-sched_tunables.txt
521-schedstats-tools.txt
--mm part--
reiser4-all_with_fixes_from_2611-rc1-mm1 (all patches)
uml-all-from-2.6.11-rc1-mm1 (all patches)
inotify.patch
cachefs from 2611rc1mm1 (all patches)
clear-false-pending-signal-indication-in-core-dump.patch
detect-atomic-counter-underflows.patch
fbcon-fix-compile-error.patch
file_tableexpand_files-code-cleanup.patch
fix-64-bit-problems-in-cachefs.patch
fix-cachefs-barrier-handling-and-other-kernel-discrepancies.patch
fix-exec-deadlock-when-ptrace-used-inside-the-thread-group.patch
fix-race-between-core-dumping-and-exec.patch
generic-irq-code-missing-export-of-probe_irq_mask.patch
ia64-acpi-build-fix.patch
ia64-config_apci_numa-fix.patch
ide-dev-build-fix.patch
make-tree_lock-an-rwlock.patch
orphaned-pagecache-memleak-fix.patch
ppc32/64_fixes_from_mm (all patches)
provide-a-filesystem-specific-syncable-page-bit_all_fixes.patch
radeonfb-fix-init-exit-section-usage.patch
random-poolsize-int-overflow.diff
sched-fix-preemption-race-core-i386.patch
sched-make-use-of-preempt_schedule_irq-arm.patch
sched-make-use-of-preempt_schedule_irq-ppc.patch
speedup-proc-pid-maps_with_all_fxies.patch
vm-pageout-throttling.patch
|
.config used for testing , ebuild , patch available on sf.net
https://sourceforge.net/project/showfiles.php?group_id=124916&package_id=136553&release_id=297019
or http://sourceforge.net/projects/vivid-sources -->> 2611rc1vv_e1 package
tested with my cfg nvidia 6629 drives from ~x86 portage and kernel agpgart
feedback , ideas , of course are welcome as always
cheers _________________ "Time is a companion that goes with us on a journey. It reminds us to cherish each moment, because it will never come again. What we leave behind is not as important as how we have lived" J-L. Picard
Last edited by fallow on Sat Jan 15, 2005 10:47 am; edited 1 time in total |
|
Back to top |
|
|
Genn Tux's lil' helper
Joined: 19 Sep 2003 Posts: 98
|
Posted: Sat Jan 15, 2005 4:44 am Post subject: |
|
|
Looks to be a nice set of patches. I noticed to included the genetic algorythm extension to the schedular. I thought it was going to be omitted. What caused you to change your mind?
I'll probably compile it tonight or tomorrow. I'll let you know how it goes. |
|
Back to top |
|
|
kontol n00b
Joined: 19 Dec 2004 Posts: 31
|
Posted: Sat Jan 15, 2005 9:33 am Post subject: |
|
|
mm..did I miss something?
Code: |
vivid-sources # ebuild vve-sources-2.6.11_rc1-r1.ebuild digest
!!! aux_get(): ebuild path for 'sys-kernel/vve-sources-2.6.11_rc1-r1' not specified:
!!! None
!!! aux_get(): ebuild path for 'sys-kernel/vve-sources-2.6.11_rc1-r1' not specified:
!!! None
doebuild(): aux_get() error; aborting.
vivid-sources#
|
|
|
Back to top |
|
|
|
|
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
|
|