Gentoo Forums
Gentoo Forums
Gentoo Forums
Quick Search: in
[testing_release/period] 2.6.12-rc4-love3
View unanswered posts
View posts from last 24 hours

Goto page 1, 2  Next  
Reply to topic    Gentoo Forums Forum Index Unsupported Software
View previous topic :: View next topic  
Author Message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Tue May 24, 2005 7:48 pm    Post subject: [testing_release/period] 2.6.12-rc4-love3 Reply with quote


this is testing version , not official release.
If everythings will be ok , it will be marked as official release.
If not , then will be cancelled .

!!! This RTP version is working OK on x86 non-smp,non-ht
!!! We dont know about SMP and HT
!!! Potential bugs can be with LLD mode on x86_64 , FullRT on x86_64 should be fine.

FOR STABLE Release use latest stable version :: 2.6.12-rc3-love1


Love-Sources 2.6.12-rc4-love3 "The Fishslapping Dance"

This is testing_period now , If will be some errors by any other patch
we can reverse it just before final release :)

Code:

!!! ONLY SHORT DESCRIPTIONS OR NAMES ARE IN THIS INFO          !!!
!!! YOU CAN LOOK INTO EVERY PATCH CODE OR FULL DESCRIPTION     !!!
!!! BY DOWNLOADING OUR SPLIT-OUT FILE                          !!!

*** RTP in this love version can be BUGGY                      ***
*** ON X86 EVERYTHINK SHOULD BE FINE but                       ***
*** we dont know about x86_64                                  ***
*** so x86_64 is up-to-you if You want test it                 ***

---info------------------------------------------------------------
Please don't report any bugs to gentoo-devs or bugs.gentoo.org.
We are making patchset for x86 .
We will be not able to help You with x86_64 

If You want support or jusk ask about something -
please join #love-sources on freenode or just
post in the love-sources thread in "Unsupported Software" forum on
forums.gentoo.org  , check also love-sources.org
-------------------------------------------------------------------

-->> patchlist first <<---
1) base :
realtime-preempt-2.6.12-rc4-V0.7.47-08
  >> Ingo Molnar's Real Time Preemption patchset.
  >> Default Mode is : "Low Latency Destkop Mode" and Hard/Soft IRQs threadings are disabled
  >> Full RT mode isnt required for _overall_ destkop using
  >> so we decided to make Low Latency Destkop Mode as defualt love mode.
  >> You can now select also FULL RT mode and Hard/Soft IRQs threading
  >> in Love-Sources features menu if You want to have really extremaly low latencies
  >> but remember this can be occupied in an overall performance decresing cost or
  >> by making Your system more unstable...


2) io schedulers fixes/updates :
def_iosched_select_menu.diff
  >> selectable def.io scheduler from (menu)config
cfq-iosched-update-to-time-sliced-design-export-task_nice.patch
cfq-iosched-update-to-time-sliced-design-fix-fix.patch
cfq-iosched-update-to-time-sliced-design-fix.patch
cfq-iosched-update-to-time-sliced-design-use-bio_data_dir.patch
cfq-iosched-update-to-time-sliced-design.patch
cfq-ioschedc-fix-soft-hang-with-non-fs-requests.patch
cfq-plus-rtp-inotify-remount-oops-lovechilds-fix.diff
  >> timeslices design updates and additional fixes for CFQ.
as-limit-queue-depth.patch
  >> a small anticipatory improvement.

3) All rest :) :
//Please note that RT-Limits are in rc4 now.
  >> for using the RT rlimits patch for Linux Audio : http://www.steamballoon.com/wiki/Rlimits

ext3-reduce-allocate-with-reservation-lock-latencies-tidy.patch
  >> ext3 latencies and fixes
  >> main ext3 latencies fixes are now directly in RTP

//Please note that other fixes from our previous release
//(rest of ext3 fixes and reiserfs) are now included in rc4.

configurable-ide-delay.patch
    >> You can choose a delay for IDE devices now.
    >> default is 50ms. You can try lower it to 5ms
    >> and You should have booting time around ~5secs shorter.
    >> all new drives should work with this but remember
    >> You can do this for own risk , default is 50ms
    >> available via love-sources menu
   
git-jfs.patch
git-ntfs.patch
  >> up-to-date jfs and ntfs from git :)

alsa-from-2.6.12-rc4-mm2
  >> as in name

add-check-to-proc-devices-read-routines.patch
  >> small "security" fix.

avoiding-mmap-fragmentation-tidy.patch
avoiding-mmap-fragmentation.patch
  >> a fix for applications that run out of memory on 2.6 kernels
  >> but works on 2.4 after the Ingo speed-up by allocating new mmaps by free_area_cache

fix-agp-code-to-work-again-with-non-pci-agp-bridges.patch
  >> exactly as in name :)

fix-of-dcache-race-leading-to-busy-inodes-on-umount-fix.patch
fix-of-dcache-race-leading-to-busy-inodes-on-umount-tidy.patch
fix-of-dcache-race-leading-to-busy-inodes-on-umount.patch
  >> dcache race between shrink_dcache_XXX functions and dput() fixes.

do_swap_page-can-map-random-data-if-swap-read-fails.patch
  >> do_swap_page(); bug fix

i386-never-block-forced-sigsegv.patch
  >> as in name fix :)

o1-sb-list-traversing-on-syncs.patch
  >> This patch removes O(n^2) super block loops in sync_inodes(),sync_filesystems() etc.

quiet-ide-cd-warning.patch
  >> This shuts up a potential uninitialized variable warning.

unplug-can-sleep.patch
  >> adding a might_sleep(); to generic_unplug_device in ll_rw_blk.

avoid-recursive-oopses.patch
  >> Prevent recursive faults in do_exit() by leaving the task alone and wait       
  >>for reboot

improve-cd-dvd-packet-driver-write-performance.patch
  >> This patch improves write performance for the CD/DVD packet writing driver.     
  >> The logic for switching between reading and writing has been changed so       
  >> that streaming writes are no longer interrupted by read requests.

mempool-only-init-waitqueue-in-slow-path.patch
  >> Here's a small patch to improve the performance of mempool_alloc by only       
  >> initializing the wait queue when we're about to wait.

mystery-ide-fix.patch
  >> "the one of the big misterious fixes from all times " :)

properly-stop-devices-before-poweroff-fix.patch
properly-stop-devices-before-poweroff.patch
  >> as in name :)

inotify-44-update.patch
inotify-44-warning-fix.patch
inotify-44.patch
  >> successor od dnotify ( rc2mm4 )

vesafb-tng-0.9-rc6-r2-2.6.12-rc1.patch
  >> a tool fro setting and controlling vesafb refresh rate.

mapped_watermark3.diff
  >> swapping improvement from ck.

1g_lowmem1_i386.diff
  >> for users having a lot of memory :)

usbhid-readd-kconfig-r1.patch
  >> configuration of usb mouse polling via (menu)config.

sched-changing-rt-priority-without-cap_sys_nice.patch
  >> Presently, a process without the capability CAP_SYS_NICE can not change         
  >> its own policy, which is OK.                                                   
  >> But it can also not decrease its RT priority (if scheduled with policy         
  >> SCHED_RR or SCHED_FIFO), which is what this patch changes.
  >> with this patch, process scheduled by SCHED_FIFO, SCHED_RR and                 
  >> SCHED_OTHER can all decrease their priority.

logos.diff
  >> You can select now booting love-sources logo via love-source features menu.
  >> http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/love-logos/


FAQ wrote:

-- Love Sources FAQ --

Q: I have black-sreen freezing after startx
A: Please try compile kernel without vesa-tng support at all.

Q: Why RTP and Ingosched are in Love-Sources now ?
A: Ingosched has greatest testing because is tested for all vanilla users.
It is also defulat Linux 2.6 CPU Scheduler and it is pretty fixed now.
Ingosched developing also doesnt sleep, look into MM tree for example.
RTP is a Ingo Molnar extension to providing a kind of Real Time Preemption.
Default Mode is : "Low Latency Destkop Mode" and Hard/Soft IRQs threadings are disabled
Full RT mode isnt required for _overall_ destkop using
so we decided to make Low Latency Destkop Mode as defualt love mode.
You can now select also FULL RT mode and Hard/Soft IRQs threading
in Love-Sources features menu if You want to have really extremaly low latencies
but remember this can be occupied in an overall performance decresing cost or
by making Your system more unstable...

Q: How to use the RT rlimits patch for Linux Audio ?
A: please follow a nice how-to here -> http://www.steamballoon.com/wiki/Rlimits

Q: So, where did Reiser4 go?
A: Away

Q: Why?
A: A number of reasons, the love sources team strive to bring you the
best desktop experience and Reiser4 can greatly degrades interactivity and
increse cpu usage . It's unlogical to have CPU Scheduler with oriented for reducint
latencies and filesystem with maximal latencies.
It's also operating under several wrong assumptions on things
like page table sizes. Reiser4 also doesn't support architectures
aside x86 nor does it run on the default 4kb stack setting.

Q: So now that love-sources isn't going to support Reiser4 what do I do?
A: Compile one of the great number of Reiser4 enabled kernels, like
Andrew Morton's mm tree, that one is even in portage.

Q: Why are you pulling Reiser4, it's the base of Love-sources?
A: Actually it's not, when love-sources started it was in large part to
research how to allivate the famous music lag bug. A bug that Reiser4
ironically reintroduces.

Q: Okay so I shouldn't use Reiser4, what then?
A: Ext3 with dir_index is a much better/universal filesystem for a desktop in that
it performs nicely and has really good latency. Aside such likable
qualities as stability , it is in developing for "ages" in comparision to Reiser4 ;)

Q: Im an ext3 user , but how can I check or activate dir_index feature ?
A: Please check if You have it in superblock by this line :
1) my_box# tune2fs -l /dev/hdaXY
if You receive line like this one with dir_index :
Filesystem features: has_journal dir_index filetype needs_recovery sparse_super
You have it activated.
if not, please do :
1) tune2fs -O dir_index /dev/hdXY
2a) updatedb
or
2b) e2fsck -D /dev/hdXY

Q: Im a Laptop user, I want to increse my battery life-time, what should I do ?
A: Use JFS as Your filesystem. It has less cpu usage (definitly better than reiserfs/reiser4)
and Your battery lifetime will be incresed even up-to 30% in comparision to reiserfs/reiser4.
In Love-Sources You have up-to-date fixes for it.

Q: OK, Im interested in good interactivity , I can resign from reiser4 for this.
Now Im using other filesystem ( We recommending You ext3+dir_index ).
Which IO Scheduler should I use ?
A: Please try CFQ-TS io scheduler ( default in love-sources ).
This is Completly Fair Queuing IO Scheduler which gives You optimal destkop interactivity
even in heavy filesystem usage.

Q: What about NVIDIA and ATI fixes?
A: In current love there are no any NVIDIA or ATI fixes.
Please use DRIVER side fixes from the net ( many of it here ) .
If You're NVIDIA USER. Many of good fixes are just in GENTOO PORTAGE.
min kernel driver to run this sources is nvidia-kernel-6629-r4



Info/Warning/Support/Bugs wrote:

Please don't report any bugs to gentoo-devs or bugs.gentoo.org.
We are making patchset for x86 . We will be not able to help You with x86_64 ;)



check also http://love-sources.org , and vote for our logo :)
logo-gallery : http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/love-logos/

patch : http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/2.6.12-rc4-love3/&file=2.6.12-rc4-love3.bz2
ebuild : http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/2.6.12-rc4-love3/&file=love-sources-2.6.12_rc4-r3.ebuild
notes : http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/2.6.12-rc4-love3/&file=notes.txt

all dir : http://vivid.dat.pl/fallow/pub/index.php?dir=love-sources/2.6.12-rc4-love3/

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
View user's profile Send private message
TPC
Tux's lil' helper
Tux's lil' helper


Joined: 16 Sep 2003
Posts: 135
Location: Sweden

PostPosted: Tue May 24, 2005 9:16 pm    Post subject: Reply with quote

Looks like another great release fallow, compiling now :)

I especially like the 5 seconds faster boot thing :)
Back to top
View user's profile Send private message
pwe
Guru
Guru


Joined: 12 Mar 2005
Posts: 445
Location: Polska aka Poland

PostPosted: Tue May 24, 2005 9:31 pm    Post subject: Reply with quote

heh nice! :) its working on my Arch box, heh and its fast! its posible that i have more free RAM ? i think that is, for now ofcourse :wink:
and NO messeges about errors on boot :wink:

Code:
Linux linuxstage 2.6.12-rc4-love3 #1 Tue May 24 23:11:52 CEST 2005 i686 AMD Athlon(tm) 64 Processor 3000+ AuthenticAMD GNU/Linux


i test it on gentoo64 now or tomorrow.

Thanks fallow!

edit: i continue tomorrow but now (very quick compilation) i have this (on gentoo64) :
Code:
[root]::[linuxstage] /usr/src/linux-2.6.12-rc4-love3#make && make modules_install
  CHK     include/linux/version.h
  SPLIT   include/linux/autoconf.h -> include/config/*
  CC      arch/x86_64/kernel/asm-offsets.s
In file included from include/linux/time.h:7,
                 from include/linux/timex.h:58,
                 from include/linux/sched.h:11,
                 from arch/x86_64/kernel/asm-offsets.c:7:
include/linux/seqlock.h: In function `__read_seqretry':
include/linux/seqlock.h:127: warning: implicit declaration of function `__preempt_enable_no_resched'
  CHK     include/asm-x86_64/offset.h
  CC      init/main.o
In file included from include/linux/time.h:7,
                 from include/linux/timex.h:58,
                 from include/linux/sched.h:11,
                 from include/linux/module.h:10,
                 from init/main.c:16:
include/linux/seqlock.h: In function `__read_seqretry':


bye


Last edited by pwe on Tue May 24, 2005 10:11 pm; edited 2 times in total
Back to top
View user's profile Send private message
TPC
Tux's lil' helper
Tux's lil' helper


Joined: 16 Sep 2003
Posts: 135
Location: Sweden

PostPosted: Tue May 24, 2005 10:09 pm    Post subject: Reply with quote

After trying it for almost an hour here is how it turned out for me (copied and pasted because I'm lazy).

Code:

[00:13:45] <@TPC> new love works great :)
[00:14:31] <@TPC> I did a stress test: played a movie, listened to some music, compiled something, untarred a big tar archive, and synced something with rsync
[00:14:42] <@TPC> music or movie didn't skip at all
[00:15:23] <@TPC> and later on when I tried browsing the web, I use a function to load up all the websites I visit daily in different tabs, which is usually pretty slow
[00:15:42] <@TPC> but it seemed like all websites were done faster than normal this time :)
Back to top
View user's profile Send private message
NiLuJe
Guru
Guru


Joined: 06 Nov 2004
Posts: 303
Location: Paris, France

PostPosted: Tue May 24, 2005 10:13 pm    Post subject: Reply with quote

Booted fine, played ff7 on epsxe for a hour, too bad madwifi doesn't like the rtp patchs ... (Same undefined symbol as before - current cvs head & bsd) ;)

Last edited by NiLuJe on Wed May 25, 2005 12:53 am; edited 1 time in total
Back to top
View user's profile Send private message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Tue May 24, 2005 10:32 pm    Post subject: Reply with quote

pwe wrote:
heh nice! :) its working on my Arch box

nice :)
about X86_64 & Low Latency Destkop Mode errors .

fallow in release info wrote:

!!! Potential bugs can be with LLD mode on x86_64 , FullRT on x86_64 should be fine.


Im going to sleep now, Try to just compile it in FullRT mode
We can take a look into it , in the near future :)
and as I know , this is 100% RTP bug in x86_64 and LLD RTP mode , try to post it into LKML also .

Im on x86 in FullRT mode for testing now . as I heard on x86_64 FullRT should work.

TPC wrote:
After trying it for almost an hour here is how it turned out for me (copied and pasted because I'm lazy).
Code:

[00:13:45] <@TPC> new love works great :)
[00:14:31] <@TPC> I did a stress test: played a movie, listened to some music, compiled something, untarred a big tar archive, and synced something with rsync
[00:14:42] <@TPC> music or movie didn't skip at all
[00:15:23] <@TPC> and later on when I tried browsing the web, I use a function to load up all the websites I visit daily in different tabs, which is usually pretty slow
[00:15:42] <@TPC> but it seemed like all websites were done faster than normal this time :)


oh nice TPC :) :D
NiLuJe wrote:
Booted fine, played ff7 on epsxe for a hour, too bad madwifi doesn't like the rtp patchs ... (Same undefined symbol as before - current cvs head) ;)

I little forgot these madwifi errors, but maybe this will be not so complicated, Im going to sleep now , maybe try to ask somebody on #love-sources :) or some madwifi development channel about driver and RTP , or #kernel ? :)

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
View user's profile Send private message
Mafteah
Apprentice
Apprentice


Joined: 30 May 2003
Posts: 265
Location: IL

PostPosted: Tue May 24, 2005 10:42 pm    Post subject: Reply with quote

@fallow:

Take a look at this ebuild:
http://www.mafteah.co.il/websvn/wsvn/portage/sys-kernel/love-sources/love-sources-2.6.12_rc4-r3.ebuild?op=file&rev=0&sc=0
_________________
http://www.mafteah.co.il
Back to top
View user's profile Send private message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Tue May 24, 2005 11:11 pm    Post subject: Reply with quote

Mafteah wrote:
@fallow:

Take a look at this ebuild:
http://www.mafteah.co.il/websvn/wsvn/portage/sys-kernel/love-sources/love-sources-2.6.12_rc4-r3.ebuild?op=file&rev=0&sc=0


yeap , this should be written exactly like U did it.
I was too lazy to fix current ebuild all the time :)

but I want to see what is patching actually , like in this actually worst love ebuild.
I dont like " silent " mode :)

so what this should be done ? via EPATCH_OPTS or something ?

thx for doing clearing ;)

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
View user's profile Send private message
NiLuJe
Guru
Guru


Joined: 06 Nov 2004
Posts: 303
Location: Paris, France

PostPosted: Wed May 25, 2005 1:53 am    Post subject: Reply with quote

Yay! About the madwifi driver, i got it to work, but i don't know if it's the good method ...

in net80211/ieee80211_linux.h
changed
Code:
#if (defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)) && defined(rwlock_is_locked)
#define IEEE80211_NODE_LOCK_ASSERT(_nt) \
        KASSERT(rwlock_is_locked(&(_nt)->nt_nodelock), \
                        ("802.11 node not locked!"))

to
Code:
#if (defined(CONFIG_SMP) || defined(CONFIG_PREEMPT)) && defined(raw_rwlock_is_locked)
#define IEEE80211_NODE_LOCK_ASSERT(_nt) \
        KASSERT(raw_rwlock_is_locked(&(_nt)->nt_nodelock), \
                        ("802.11 node not locked!"))


And blam!, no more undefinded symbol ... So far it seems to work, i'm running it now, no ugly message in the kernel logs, but I absolutely don't know what the thing that I changed does! ^^ Tested with the current cvs bsd branch, but should work with head branch too ;)
Back to top
View user's profile Send private message
bosyotech
Tux's lil' helper
Tux's lil' helper


Joined: 29 Jan 2005
Posts: 79
Location: earth

PostPosted: Wed May 25, 2005 2:32 am    Post subject: Reply with quote

thanks for a blazing release follow, you are da man. :D
_________________
of all the things i've lost, i miss my mind the most
Back to top
View user's profile Send private message
TPC
Tux's lil' helper
Tux's lil' helper


Joined: 16 Sep 2003
Posts: 135
Location: Sweden

PostPosted: Wed May 25, 2005 7:34 am    Post subject: Reply with quote

@fallow:
if the madwifi fix from two posts up works maybe we should include it in the next release?
Back to top
View user's profile Send private message
playfool
l33t
l33t


Joined: 01 Jun 2004
Posts: 688
Location: Århus, Denmark

PostPosted: Wed May 25, 2005 10:03 am    Post subject: Reply with quote

I would like to add that the IDE wait patch is pure evil, it violates IDE specs, and it might cause serious problems on older hardware - remember when you are enabling that you are going beyond what your hardware is specificed for so if it breaks you get to keep both pieces.
Back to top
View user's profile Send private message
ashtophet
Guru
Guru


Joined: 08 Aug 2004
Posts: 397

PostPosted: Wed May 25, 2005 10:25 am    Post subject: Reply with quote

another great release! using default RT options things are running smoothly. Furthermore, my little alsa problem is solved with latest mm alsa update... so, great!

cheers fallow
Back to top
View user's profile Send private message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Wed May 25, 2005 11:04 am    Post subject: Reply with quote

NiLuJe wrote:
Yay! About the madwifi driver, i got it to work, but i don't know if it's the good method ...

:) nice if it works
TPC wrote:
@fallow:
if the madwifi fix from two posts up works maybe we should include it in the next release?


Hm , Im personally against ,then we should include also all exotic net drivers like orinioco,hostap,skge,sk98lin,ipw and all :)
The current goals is to provide good interacvitivity and small latencies , rest like fbsplash, shfs/squashfs , net drivers users can do itself , and I dont even have possibility to test for example madwifi :)

no more tons of utested patches :)

bosyotech wrote:
thanks for a blazing release follow, you are da man. :D

galiza_ceive wrote:
another great release! using default RT options things are running smoothly. Furthermore, my little alsa problem is solved with latest mm alsa update... so, great!

good to hear You like it :) nice guys :)
playfool wrote:
I would like to add that the IDE wait patch is pure evil, it violates IDE specs, and it might cause serious problems on older hardware - remember when you are enabling that you are going beyond what your hardware is specificed for so if it breaks you get to keep both pieces.

yeap , its little bit ricer patch , but this should work with all new hardware.
default setting in love-sources menu is the same as default in vanilla , so its providing a possibility to change it only.

:)

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
View user's profile Send private message
pwe
Guru
Guru


Joined: 12 Mar 2005
Posts: 445
Location: Polska aka Poland

PostPosted: Wed May 25, 2005 11:06 am    Post subject: Reply with quote

mission faild :cry:

thanks fallow, with FullRT mode kernel its compiling fine. but when im booting, framebuffer fails, and when i turnoff the fb, when is 'testing NMI watchdog...' everything freeeeze. i again dont have an idea.

anybody at amd64 do this?

ps kernel.org -> rc5 is released
Back to top
View user's profile Send private message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Wed May 25, 2005 11:28 am    Post subject: Reply with quote

use rc3-love1 , this is stable also on x86_64 .

I personally see some mistakes in rc4 vanilla , rc3 vanilla runs more "smoothly" here.

in Low Latency Destkop-LLD mode rc3 based RTP love (rc3-love1) is better than rc4-love3 imho.
Im using rc4-love3 in fullRT and rc3-love1 in LLD now.

cheers :)

we'll see rc5 + RTP in near future :)
_________________
"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
View user's profile Send private message
gringo
Advocate
Advocate


Joined: 27 Apr 2003
Posts: 3793

PostPosted: Wed May 25, 2005 11:32 am    Post subject: Reply with quote

Quote:
when is 'testing NMI watchdog...' everything freeeeze. i again dont have an idea.


yeah, saw that too but -mm solved it here, so one of this three patches should do the trick:

Code:
x86_64-reduce-nmi-watchdog-stack-usage.patch
x86_64-collected-nmi-watchdog-fixes.patch
x86_64-collected-nmi-watchdog-fixes-warning-fix.patch


give them a shot or just disable the watchdog ;)

Talking about -mm, i saw in mm-commits voluntary preempt will be in mm again, also nice per cpu from Con, etc, maybe taking a look to this patches instead of full RTP is worth it ??
And Fallow, do you believe libata-dev could be in next releases ??? Would very much appreciate this ;)

cheers
Back to top
View user's profile Send private message
pwe
Guru
Guru


Joined: 12 Mar 2005
Posts: 445
Location: Polska aka Poland

PostPosted: Wed May 25, 2005 11:43 am    Post subject: Reply with quote

fallow wrote:
use rc3-love1 , this is stable also on x86_64 .

I personally see some mistakes in rc4 vanilla , rc3 vanilla runs more "smoothly" here.

in Low Latency Destkop-LLD mode rc3 based RTP love (rc3-love1) is better than rc4-love3 imho.
Im using rc4-love3 in fullRT and rc3-love1 in LLD now.

cheers :)

we'll see rc5 + RTP in near future :)


@fallow: rc4-love2 for me its stable, but my Arch Linux on rc4-love3 works better.
You recommend me to use rc3-love1? ist faster then rc4-love2 its stable on x86_64? ok i'll try :wink:

@gringo:
Quote:
or just disable the watchdog
i have this turned off :?


about rc3-love1
Code:
 ...
 CC      arch/x86_64/kernel/head64.o
  CC      arch/x86_64/kernel/init_task.o
arch/x86_64/kernel/init_task.c:17: warning: implicit declaration of function `__RWSEM_INITIALIZER'
arch/x86_64/kernel/init_task.c:17: warning: missing braces around initializer
arch/x86_64/kernel/init_task.c:17: warning: (near initialization for `init_mm.mmap_sem')
arch/x86_64/kernel/init_task.c:17: error: initializer element is not constant
arch/x86_64/kernel/init_task.c:17: error: (near initialization for `init_mm.mmap_sem.activity')
arch/x86_64/kernel/init_task.c:17: error: initializer element is not constant
arch/x86_64/kernel/init_task.c:17: error: (near initialization for `init_mm.mmap_sem')
arch/x86_64/kernel/init_task.c:17: error: initializer element is not constant
arch/x86_64/kernel/init_task.c:17: error: (near initialization for `init_mm.page_table_lock')
...
Back to top
View user's profile Send private message
NiLuJe
Guru
Guru


Joined: 06 Nov 2004
Posts: 303
Location: Paris, France

PostPosted: Wed May 25, 2005 2:20 pm    Post subject: Reply with quote

If need be, I have a working live cvs ebuild for madwifi-driver, with branch selection (head/bsd) & ratemodule selection (onoe/amrr/sample) via use flags, just pm me if you want it ;)

Last edited by NiLuJe on Wed May 25, 2005 2:23 pm; edited 1 time in total
Back to top
View user's profile Send private message
gimpel
Advocate
Advocate


Joined: 15 Oct 2004
Posts: 2720
Location: Munich, Bavaria

PostPosted: Wed May 25, 2005 2:22 pm    Post subject: Reply with quote

runs very smooth here! nice!
i also use it in full preempt mode... playing around with new m-audio delta soundcard.. so latency simply owns massively in full preempt ;)

one issue i have with _all_ -rc3 and -rc4 kernels: as soon i start X, or when i switch to a console, dunno exactly, my framebuffers get b0rked, no matter if i use vesafb-tng or don't. just some big green and pink pixels... no prob with that on =< -rc2, pretty strange... anyone else got that?
_________________
http://proaudio.tuxfamily.org/wiki - pro-audio software overlay
Back to top
View user's profile Send private message
NiLuJe
Guru
Guru


Joined: 06 Nov 2004
Posts: 303
Location: Paris, France

PostPosted: Wed May 25, 2005 3:03 pm    Post subject: Reply with quote

A little warning message in dmesg after loading nvidia, in full rt :

Code:
BUG: modprobe/3403, lock held at task exit time!
 [ddc65ae4] {(struct semaphore *)(&os_sema->wait)}
.. held by:          modprobe: 3403 [c15cb1b0, 115]
... acquired at:  os_alloc_sema+0x40/0x6c [nvidia]


But it works ;)
Back to top
View user's profile Send private message
playfool
l33t
l33t


Joined: 01 Jun 2004
Posts: 688
Location: Århus, Denmark

PostPosted: Wed May 25, 2005 7:38 pm    Post subject: Reply with quote

well, get nvidia to agree to GPL the sourcecode and I'm sure that bug will go away - I'm also sure someone could hack around it, but that would be nasty.
Back to top
View user's profile Send private message
NiLuJe
Guru
Guru


Joined: 06 Nov 2004
Posts: 303
Location: Paris, France

PostPosted: Thu May 26, 2005 12:06 am    Post subject: Reply with quote

Two other things in my current dmesg :)

Code:
ksoftirqd/0/2: BUG in __up_mutex at kernel/rt.c:1332
 [<c012f9a2>] __up_mutex+0x1f2/0x4b0 (8)
 [<e0e9df65>] _nv005582rm+0x285/0x3f8 [nvidia] (28)
 [<c01309ee>] rt_up+0xce/0xf0 (28)
 [<e107ceb6>] os_release_sema+0x49/0x5d [nvidia] (36)
 [<e0e6af26>] _nv001648rm+0x12/0x18 [nvidia] (16)
 [<c01125ee>] __wake_up+0x1e/0x50 (16)
 [<e0e6e60d>] _nv001198rm+0x9d/0xa8 [nvidia] (16)
 [<e0e6e5f7>] _nv001198rm+0x87/0xa8 [nvidia] (16)
 [<e0e719c8>] rm_isr_bh+0x18/0x24 [nvidia] (48)
 [<c01c14ee>] pci_bus_read_config_word+0x6e/0x90 (12)
 [<e0e719bf>] rm_isr_bh+0xf/0x24 [nvidia] (4)
 [<e107aa28>] nv_kern_isr_bh+0xc1/0xd1 [nvidia] (32)
 [<c011b856>] tasklet_action+0x36/0x90 (32)
 [<c011b563>] ___do_softirq+0x43/0xd0 (24)
 [<c02f6b53>] schedule+0x33/0x120 (4)
 [<c011b9b0>] ksoftirqd+0x0/0xe0 (28)
 [<c011b697>] _do_softirq+0x17/0x20 (4)
 [<c011ba28>] ksoftirqd+0x78/0xe0 (4)
 [<c012a8a5>] kthread+0x95/0xd0 (24)
 [<c012a810>] kthread+0x0/0xd0 (16)
 [<c0100d6d>] kernel_thread_helper+0x5/0x18 (16)

Code:
set_rtc_mmss: can't update from 59 to 0
Back to top
View user's profile Send private message
fallow
Bodhisattva
Bodhisattva


Joined: 08 Jan 2004
Posts: 2208
Location: Poland

PostPosted: Thu May 26, 2005 5:35 am    Post subject: Reply with quote

NiLuJe:
this oops with FullRT or LLD mode ?

( LLD is more safe than FullRT <-> FullRT is more risky than LLD )
( recommended is LLD IMHO,FullRT becoming more and more stable day by day but isnt 100% stable yet )

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
View user's profile Send private message
wing
Tux's lil' helper
Tux's lil' helper


Joined: 27 Aug 2002
Posts: 108
Location: Mountain View, CA

PostPosted: Thu May 26, 2005 7:18 am    Post subject: Reply with quote

be warned, FullRT will not work with CPUFreq on amd64 at all, it will make your kernel panic at boot! At least in my experience, compiled without it though it works fine.

FullRT is very snappy, it reminds me of my BeOS days.
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 1, 2  Next
Page 1 of 2

 
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