2.6.11-love1 (fixes from un-of. updates - in 2.6.11-love2)

[rOss^64]

(I couln't understand where the relevant part started...)
Anyway, I was playing starcraft, and suddenly wine got killed. Anyone can explain this "out of memory"?

I just reinstalled my starcraft today also. It open the bnet screen i go to type in my account and boom it closes same error.

[bollucks]

Have you fiddled with the "mapped" settings? The mapped watermark patch is far more aggressive at avoiding swap than previously and high mapped settings can make unreasonable demands of the virtual memory subsystem making it possible to get an out-of-memory condition which is followed by an out-of-memory kill.

[VoVaN]

@VoVaN : please don`t post here error about OHTER sources . I look at this and I was been thinked that is error from love and don`t know what is going on - cause I don`t remember reiser4 compilation problem.
Please just make new thread - and If You want reiser4code from love - I can make a patch for U Wink

Sorry, I've been confusing you... The errors above indeed not love sources related BUT, It would be really nice to have reiser4 patch separatelly. Thank you in advance.
BTW, I really don't know what caused the errors posted above if I apply reiser4 from mm to vanilla kernel, I think you're doing tha same, don't you?

[fallow]

Have you fiddled with the "mapped" settings? The mapped watermark patch is far more aggressive at avoiding swap than previously and high mapped settings can make unreasonable demands of the virtual memory subsystem making it possible to get an out-of-memory condition which is followed by an out-of-memory kill.

Yeah Bollucks has right, try to test it with for ex. /proc/sys/vm/mapped setting to 66 or 33 for start.

Sorry, I've been confusing you... The errors above indeed not love sources related BUT, It would be really nice to have reiser4 patch separatelly. Thank you in advance.
BTW, I really don't know what caused the errors posted above if I apply reiser4 from mm to vanilla kernel, I think you're doing tha same, don't you?

No any problem, my reaction was also too "strange" for that poppycock

yeah - I doing the same - patches are from broken-out , I`ll prepare it for U and give link here.



cheers.

[VoVaN]

No any problem, my reaction was also too "strange" for that poppycock Razz Smile

yeah - I doing the same - patches are from broken-out , I`ll prepare it for U and give link here.

Thank you!
But, which magic did you do to in order to avoid these compilation warnings?

[kontol]

I've got this error

Code: Select all

#emerge love-sources

>>> unpacking 2.6.11-love1.bz2 to /var/tmp/portage/love-sources-2.6.11-r1/work

bzip2: Compressed file ends unexpectedly;
          perhaps it is corrupted?

etc..


!!! ERROR: sys-kernel/love-sources-2.6.11-r1 failed.
!!! Function unpack, Line 382, Exitcode 2
!!! failure unpacking 2.6.11-love1.bz2
!!! if you need support, post the topmost build error, NOT this status message

[fallow]

I've got this error

Code: Select all

#emerge love-sources
>>> unpacking 2.6.11-love1.bz2 to /var/tmp/portage/love-sources-2.6.11-r1/work
bzip2: Compressed file ends unexpectedly;
          perhaps it is corrupted?
etc..

You have to redownload this file and regenerate digest.

correct :

Code: Select all

2.6.11-love1.bz2:  
md5sum = d48f787be17f4361a519fe25775953d1
size = 3016875 bytes

cheers

[VoVaN]

did somebody get usb mouse poll working?
I can't get it even with the latest un-off update for love

[rojaro]

Hi,

it seems the Toshiba ACPI driver is broken in this release - at least it doesnt work for me and produces the following error message upon boot:

Code: Select all

toshiba_acpi: Toshiba Laptop ACPI Extras version experimental-dev-toshiba-test-4
toshiba_acpi: HCI method: _SB_.VALZ.GHCI
toshiba_acpi: failed to register misc device 181 ("toshiba")

It worked perfectly with the driver used in 2.6.11-rc4-love1. So to fix this, you just need to replace the new drivers/acpi/toshiba_acpi.c file with the older one (version 0.18).

- rojaro -

[DrWoland]

@DrWoland :

Code: Select all

        Option "NvAGP" "integer"                                                          
                Configure AGP support. Integer argument can be one of:                    
                0 : disable agp                                                           
                1 : use NVIDIA's internal AGP support, if possible                        
                2 : use AGPGART, if possible                                              
                3 : use any agp support (try AGPGART, then NVIDIA's AGP)                  
                Please note that NVIDIA's internal AGP support cannot                     
                work if AGPGART is either statically compiled into your                   
                kernel or is built as a module, but loaded into your                      
                kernel (some distributions load AGPGART into the kernel                   
                at boot up).  Default: 3 (the default was 1 until after                   
                1.0-1251).  

cheers.

So the NVidia AGPGART is still different from the NVIDIA AGP Driver? Should I remove AGPGART from kernel again to use the NVidia AGP driver?

[rojaro]

So the NVidia AGPGART is still different from the NVIDIA AGP Driver? Should I remove AGPGART from kernel again to use the NVidia AGP driver?

Yes, it's still diffrent and it will probably remain diffrent until nVidia officially supports AGP Bridging (which would be really great). Anyway, the latest nvidia-kernel ebuild (1.0.6629-r4) includes a patch which makes AGP Bridging work. So just enable AGPGART, recompile the kernel and then update the nVidia Kernel driver. Works perfectly for me :)

Code: Select all

neutron ~ # cat /proc/version
Linux version 2.6.11-love1 (root@neutron) (gcc version 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1)) #1 Tue Mar 8 13:20:57 CET 2005
neutron ~ # cat /proc/driver/nvidia/version 
NVRM version: NVIDIA Linux x86 NVIDIA Kernel Module  1.0-6629  Wed Nov  3 13:12:51 PST 2004
GCC version:  gcc version 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1)
neutron ~ # cat /proc/driver/nvidia/agp/status 
Status:          Enabled
Driver:          AGPGART
AGP Rate:        4x
Fast Writes:     Enabled
SBA:             Enabled
neutron ~ # emerge -pv nvidia-kernel

These are the packages that I would merge, in order:

Calculating dependencies ...done!
[ebuild   R   ] media-video/nvidia-kernel-1.0.6629-r4  0 kB 

Total size of downloads: 0 kB
neutron ~ #

[Hirion]

But somthing is bothering me. While everything seems to work fine, my dmesg shows a bunch of:

Code: Select all

 scheduling while atomic: swapper/0x00000001/0
[ 1964.216628]
[ 1964.216629] Call Trace:<ffffffff803f05b4>{schedule+116} <ffffffff8010c93a>{cpu_idle+74}
[ 1964.216638]        <ffffffff8058a8f0>{start_kernel+624} <ffffffff8058a26a>{x86_64_start_kernel+362}

Code: Select all

Linux laika.g.la 2.6.11-love1 #2 Mon Mar 7 22:05:57 Local time zone must be set--see zic manua x86_64 AMD Athlon(tm) 64 Processor 3500+ AuthenticAMD GNU/Linux

Any comments on this?

Same Problem here. Maybe a amd64 issue? My kernel config is in one of my posts before...

I tested the kernel without all the genetic things, but the problem still remains...

[potenzial]

I have the scheduling while atomic problem too on amd64...

[dhaemon]

@dhaemon : thanks , and : this issue is on genetic_nicksched ?

Forgot to mention all this. I use deadline in combination with genetic nickshed(I just love it! And even more with genetics!!!). But I don't think it's very important since I've tried using other IO schedulers with no success.
I tried to see what "swapper" was but... still don't know.
I was clueless, so I tried "swapoff /dev/sda9". It was useless...
You know, this message repeats itself at *least* 50(!) times a second.
Weird, since I have no apparent problem?

Anways, Thanks in advance

[EDIT]
To get offtopic:
Could point me to where I could change my stack size? I can't find it in gconfig.
Has someone tried 16K? Does it work?
If so; why doesn't love use it to boost performance?
[/EDIT]

[gimpel]

wow, /me is more and more impressed by genetic nicksched! it rocks!
already tried it on -ac + reiser4 based patchset to see how it runs "outside love"
..and doing just fine! genetic bonus enabled too
fallow, absolutely great work!

@ dhaemon
kernel hacking ---->

btw: for the future: use /stack (just type it like that in menuconfig) ---> nice search function

reiser4 doesn't work with stacks other than 8k! and as love1 has excellent reiser4, it would be a harm to enable 4 or 16k by default

[VoVaN]

@fallow
I got it The compilation warnings for vanilla kernel + reiser4 from mm could be solved by applying make-tree_lock-an-rwlock.patch from mm before reiser4! The strange thing is why other sources didn't do that, but only love sources :?:

[dhaemon]

@ dhaemon
kernel hacking ---->

btw: for the future: use /stack (just type it like that in menuconfig) ---> nice search function

Ohh. Sorry, the reason I asked was because it wasn't available through kernel hacking...
I've tried all sorts of sources, but none allowed me to change stack size.
I tried searching .config for "stack and" "4k", "8k" ; it gave nothing.
The search option in menuconfig returned: "4KSTACKS [=4KSTACKS]", but I can't find it anywhere.

[gimpel]

Linux Kernel 2.6.11-love1 "We are The Borg" Configuration
Kernel hacking --->
[ ] Use 4Kb for kernel stacks instead of 8Kb

..no 16k stacks in love...just saw that too

[feld]

did somebody get usb mouse poll working?
I can't get it even with the latest un-off update for love

I spoke to the creator and yes -- it IS working.

It will NOT show up if you look at the device info though through /proc or using a program like USBview. This is because the way he was doing it before was very naughty and dangerous.

It works. Trust me.

on your kernel boot line (even if you did it staticly in the kernel config with the latest update)

add:

usbhid.mousepoll=50 (if you have it built in)

or just

mousepoll=50 (if its a module)


and then try using your mouse. Yeah, it will be slow as hell.

Just reboot with it changed back to 2 like I know you want :p



-Feld

[feld]

Fallow -- I'm still getting system hardlocks with NO access to my box (ssh, anything!) when I unplug my usbkey drive after saving something on it. This is even with your latest patch. This never happened in 2.6.11, and honestly, its quite annoying



Anyone else have this ?



-Feld

[fallow]

at now I`m working of update 02 with fixes of swsup2 (done) hostap ( actual ) and some others.

@Kontol : the driver from bk-net-dev ( 2.6.11-rc4/5-love1 for ex ) works for You. ?
I fixed this from love1 to compilation as static now, ( compilation as module was been without problems)

In upd02 will be also genetic nicksched imprevements in VERY thanks to Marco Hennings (tunning thanks to Him).

@feld : I don`t have idea about usb unplugging now - I`ll try to find reason when I done update02.

cheers

[sobers_2002]

hi

this might be silly to bother about but why is the swap usage always going up in this one???

in less than a day it becomes ~10MB .......can there be any way to avoid that>>

i ust don't like to see it

[Morimando]

Alas, i experience some problems, described theme here: http://forums.gentoo.org/viewtopic-t-305615.html
i thought since you are the guys that know about kernels and stuff and i'm using your "We are the Borg"-Kernel you might know what's wrong with my poor system
thx in advance for any help..

[Master One]

In upd02 will be also genetic nicksched imprevements in VERY thanks to Marco Hennings (tunning thanks to Him).

Fabulous, can't await to try upd02. My system is already flying at warp-speed, now with a complete reinstall on Reiser4 1.0.4
That, and genetic nicksched + deadline does the trick, hehe.

[DrWoland]

So the NVidia AGPGART is still different from the NVIDIA AGP Driver? Should I remove AGPGART from kernel again to use the NVidia AGP driver?

Yes, it's still diffrent and it will probably remain diffrent until nVidia officially supports AGP Bridging (which would be really great). Anyway, the latest nvidia-kernel ebuild (1.0.6629-r4) includes a patch which makes AGP Bridging work. So just enable AGPGART, recompile the kernel and then update the nVidia Kernel driver. Works perfectly for me

Code: Select all

neutron ~ # cat /proc/version
Linux version 2.6.11-love1 (root@neutron) (gcc version 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1)) #1 Tue Mar 8 13:20:57 CET 2005
neutron ~ # cat /proc/driver/nvidia/version 
NVRM version: NVIDIA Linux x86 NVIDIA Kernel Module  1.0-6629  Wed Nov  3 13:12:51 PST 2004
GCC version:  gcc version 3.3.5 (Gentoo Linux 3.3.5-r1, ssp-3.3.2-3, pie-8.7.7.1)
neutron ~ # cat /proc/driver/nvidia/agp/status 
Status:          Enabled
Driver:          AGPGART
AGP Rate:        4x
Fast Writes:     Enabled
SBA:             Enabled
neutron ~ # emerge -pv nvidia-kernel

These are the packages that I would merge, in order:

Calculating dependencies ...done!
[ebuild   R   ] media-video/nvidia-kernel-1.0.6629-r4  0 kB 

Total size of downloads: 0 kB
neutron ~ #

So I'm not losing any performance by just using the AGPGART driver? Is there any reason to not compile it and use the Nvidia driver or vice versa? Sorry, this confuses me greatly. On a side note, all compiles still going smoothly except one that I had to use CFLAGS="" for, but that wasn't a love problem.