2.6.6_rc2-love1 aka "Really Contagious"

Message

steel300 · Post by **steel300** » Wed Apr 21, 2004 1:33 pm

Here's the new love for your perusal.

Patches:                        Description:
2.6.6-rc2-mm1                   Latest mm
BadRAM-2.6.5.2                  Work around for bad ram
acerhk                          Acer Hotkeys
acx100-0.2.0pre7-linux2.6.2     ACX00 Drivers
bd-claim                        EVMS Thingy
bootsplash-3.1.4-2.6.5          Bootsplash
cflags-selection                Select CFlags
config-number-tty-devices       # of /dev/tty devices
config_hz                       Internal Timer Frequency
gcc-3.4-rcupdate.h-warn-fix     Warning fix
gcloop-2.6-20040330             Compressed Loopback Device
hrtimers-2.6.5-1.0              Hi Res Timers
ipw2100-2.6.5-patch             Inetl Centrino WiFi
iteraid                         GigaRaid
layer7-kernelpatch-nf-0.4.0     Layer7 Netfilter
linux-2.6.0.dxr3.20031229       dxr3/H+ support
linux-2.6.4-Walken              Walken Boot Logo
lirc-2.6.4-20040318             LIRC Drivers
love_ppm                        Love Boot Logo
lufs-0.9.7-2.6.0-test9          LUFS FileSystem
menuconfig-NAME-v1.0            Do Something With Name
move-__this_module-to-modpost   Stupid Nvidia
nforce-apic-tack                NForce Dealie
orinoco-0.13e-patch             Monitor Mode
packet-2.6.5                    UDF Packet Writing
patch-2.4.x-vesafb-rrc          Vesafb Hack
patch-2.6.4-am9                 Autoswappiness
patch-2.6.4-amd76x_pm           AMD Power Management
patch-2.6.6-rc2                 latest Prepatch
sii-3512-update                 Sii Updates
squashfs1.3r3-patch             Squash FileSystem
supermount-2.0.4-2.6.5-love1    Supermount
via-v4l-1.4a-drm                Via DRM
viafb_02                        Via FB

There's lots of goodies in this one. New Hi Res Timers, user configurable number of tty devices (good for udev), new viafb, nforece dealie. Compile and enjoy.

UDF Packet writing doesn't compile. So don't try. I'll fix it later.

Ylin · Post by **Ylin** » Wed Apr 21, 2004 2:08 pm

Will try it as soon as my kde update finished.

Can you say me which nforce patch is included? Is it the one by PrakashKC? Are any special boot parameters required?

steel300 · Post by **steel300** » Wed Apr 21, 2004 2:11 pm

Ylin wrote:Will try it as soon as my kde update finished.

Can you say me which nforce patch is included? Is it the one by PrakashKC? Are any special boot parameters required?

Yeah, it's the one from PrakashKC. To use it, add apic_tack=2 or apic_tack=1 to your kernel line {grub,lilo}.conf.

fallow · Post by **fallow** » Wed Apr 21, 2004 2:14 pm

and bootsplash works ?

Ylin · Post by **Ylin** » Wed Apr 21, 2004 2:23 pm

steel300 wrote:Yeah, it's the one from PrakashKC. To use it, add apic_tack=2 or apic_tack=1 to your kernel line {grub,lilo}.conf.

Yes, he said that in that German topic. Thx

snekiepete · Post by **snekiepete** » Wed Apr 21, 2004 3:00 pm

Steel,
ok I got this one to boot right, with SMP enabled. It started out looking grim, just like yesterday, but I recompiled the kernel with APM instead of ACPI and so far so good, but wouldn't I rather use acpi?

steel300 · Post by **steel300** » Wed Apr 21, 2004 3:10 pm

snekiepete wrote:Steel,
ok I got this one to boot right, with SMP enabled. It started out looking grim, just like yesterday, but I recompiled the kernel with APM instead of ACPI and so far so good, but wouldn't I rather use acpi?

Not if your mobo supports apm instead of acpi.

snekiepete · Post by **snekiepete** » Wed Apr 21, 2004 3:13 pm

It has been using acpi up until the release of 2.6.5-love5, but if I wont notice any difference, I am cool with that.... I just wonder why it only is defunct when SMP is enabled?

steel300 · Post by **steel300** » Wed Apr 21, 2004 3:17 pm

snekiepete wrote:It has been using acpi up until the release of 2.6.5-love5, but if I wont notice any difference, I am cool with that.... I just wonder why it only is defunct when SMP is enabled?

Buggy BIOS that just got noticed probably. SMP boots here, but that doesn't help you much.

Jayz · Post by **Jayz** » Wed Apr 21, 2004 3:41 pm

love-sources mirror updated =)
http://www.23yrs.com/love/

enjoy!

Tazok · Post by **Tazok** » Wed Apr 21, 2004 3:42 pm

fallow wrote:and bootsplash works ?

Not for me...

sindre · Post by **sindre** » Wed Apr 21, 2004 3:42 pm

Will nicks scheduler be back? Or are the tweaks in current mm regarded better? How does the interactivity in the three latest releases compare to 2.6.5-love4 which included nicks scheduling tweaks.

Jayz · Post by **Jayz** » Wed Apr 21, 2004 3:50 pm

love-sources mirror updated =)
http://www.23yrs.com/love/

enjoy!

Gandalf the White · Post by **Gandalf the White** » Wed Apr 21, 2004 4:00 pm

So far so so steel, kernel booted fine, framebuffer works, but bootsplash does not, nvidia recompiled nicely, no serious problems, just the stupid bootsplash. I hadn't even noticed that rc2 was out, well done!

fallow · Post by **fallow** » Wed Apr 21, 2004 4:08 pm

hm , bootsplash is important 4 me , so , i`m w8ing for working bootsplash .....................

steel300 · Post by **steel300** » Wed Apr 21, 2004 4:17 pm

sindre wrote:Will nicks scheduler be back? Or are the tweaks in current mm regarded better? How does the interactivity in the three latest releases compare to 2.6.5-love4 which included nicks scheduling tweaks.

Nick's scheduler will be back when the scheduler changes in mm settle down a bit. Right now, there's a huge push form Ingo and Nick concerning the scheduling code, so Nick's scheduler dies on compile in it's current form. I haven't noticed an interactivity drop between mm and Nick's. Which is good becuase that means Nick's next scheduler will be that much better.

DarwinianCoeus · Post by **DarwinianCoeus** » Wed Apr 21, 2004 5:16 pm

'stupid nvidia' for modpost, hahaha

t_2199 · Post by **t_2199** » Wed Apr 21, 2004 5:42 pm

Works fine thanks, cool new features now udev detects some devices automaticly and I don't need to create them with the local.start file...

And I've got a 1280x1024 resolution but the bootsplash picture won't work at startup only at the end of the boot when it says:

Setting console frame buffer images or something than all ttys have that bootsplash picture again

(sorry for my shitty english :9 )

Ylin · Post by **Ylin** » Wed Apr 21, 2004 5:42 pm

Tazok wrote:
fallow wrote:and bootsplash works ?
Not for me...

Same over here. :/

jimbojetset · Post by **jimbojetset** » Wed Apr 21, 2004 6:06 pm

All working nicely here

Gandalf the White · Post by **Gandalf the White** » Wed Apr 21, 2004 9:35 pm

Well, while it did make the kernel "rock like ninja", it seems clear that bootsplash is borked, i'm going to take a good look at it in a couple hours. I already looked at dmesg, and I saw nothing out of the ordinary,

@DarwinianCoeus:

Who votes to have the name of the next love sources be in commemoration of 2.6.5, by naming it "Stupid Nvidia"?

danone · Post by **danone** » Wed Apr 21, 2004 11:35 pm

Well maybe someone can help me with NFoce2 Mobo.

I configured a PC with ASUS NForce2 Mobo.Í compiled the kernel newsest and previuos love-sources and eve get hardlocks while system run and do such normal things like mirrorselect it freeze..

I configured the kernel without preemptive and used CFQ scheduler...
What could cause the hardlooks?May APIC support and I-APIC support?Any advise would be nice. With my P4 and Abit IC7 all works fine only damn NForce2 won't work stable.

scout · Post by **scout** » Thu Apr 22, 2004 12:04 am

steel300 wrote:Right now, there's a huge push form Ingo and Nick concerning the scheduling code

I must say I am a bit lost ...
If I understood well, there are 3 schedulers for I/O: anticipatory, deadline and cfq.
From what I understood, anticipatory makes statistics on processes that acces the disk, to guess who needs what when, and use these guesses to make several reads at the same time, then several writes ... on the other hand cfq tries to give I/O access quicker, which is good to watch a movie while you compile a kernel, if you don't care your compilation takes longer. I don't know much about deadline. I don't know either if there's a way to tell the anticipatory scheduler, you'd like to have mplayer have a big priority in the queues (this is an example, mplayer doesn't skip anyway), and don't know if this scheduler thakes into account the nice of a process; I think cfq does.
You can switch between these schedulers with a boot parameter.

On the other hand, you can only have one scheduler for your processes. But what's the difference between the staircase scheduler, the one in the vanilla, the one in mm, and the one(s) of Nick ? Could somebody point me towards documentation on this or make things clearer ?

Cerement · Post by **Cerement** » Thu Apr 22, 2004 12:37 am

scout wrote:
steel300 wrote:Right now, there's a huge push form Ingo and Nick concerning the scheduling code
I must say I am a bit lost ...
If I understood well, there are 3 schedulers for I/O: anticipatory, deadline and cfq.
From what I understood, anticipatory makes statistics on processes that acces the disk, to guess who needs what when, and use these guesses to make several reads at the same time, then several writes

You're mixing up scheduler domains:

I/O schedulers (disk access): CFQ, Anticipatory, Deadline, Noop

CPU schedulers (process/thread scheduling): Staircase, Nicksched, O(1)

and up and coming adding to the confusion are
VM schedulers: anon-vma, Andrea's objrmap, Hugh's objrmap

Part of the CPU scheduler holdup is that Nick and Ingo are also working on cleaning up the NUMA scheduler code (imagine boxes with 256+ procs that are NOT clusters and then imagine the headaches of scheduling processes on those machines)

scout · Post by **scout** » Thu Apr 22, 2004 1:02 am

Cerement wrote:You're mixing up scheduler domains

Yes that's why I asked ... I understood the difference between I/O schedulers and CPU schedulers, but I have indeed never heard of the VM schedulers !
Thanks