2.6.10-nitro4: The Kernel Formerly Known As Hercules

[discomfitor]

REISER4 PEOPLE: DISABLE 4K STACKS!!!!!!!!!!!

Code: Select all

2.6.10 #newest kernel version
ac4 #alan cox's patchset
-ck1 #latest ck snapshot (not including supermount)
-alsa 1.0.7 #includes several misc fixes which may or may not affect things
reiser4 #newest snapshot with extra patches
vesafb-tng #slight changes which may or may not affect anything
fbsplash #slight changes which may or may not affect anything
chmp #new version of mouse polling patch
sk98lin driver #sk98lin network driver
inotify
-swsusp2 #see note
-alsa 1.0.8-rc1
-mm1 #see note
config-nr-tty-devices
cflags-selection-for-nitro #change cflags for kernel
hostap #hostAP drivers
================================
+ckdev #latest ck patch
+alsa 1.0.8-rc2
+swsusp2+win4lin #see note
+orinoco #see note

OPTIONAL PATCHES:

+mm2 #NOT THE WHOLE THING!
+bk9

Since some people may or may not want the experimental nature of mm/bk patches, I've made them optional. Both of them are disabled by default. If you want them, enable the "want-mm" use flag. The patch applies on top of nitro4. To see what is included in the mm patch, look here. Conversely, if you want to see what I left out look here.

nitro has been mentioned on the kerneltrap poll! Sure, it's not an actual option, but being the thirteenth comment is pretty good, especially since nobody has bashed it yet!

Extra:

CONFIG_IOPRIO_WRITE
See that option? Do NOT enable it. Bad things happen if you do.

I did not use all of the mm patches. Just the ones that did not make large changes (which broke things) and worked well.

For those of you having sound issues, check here

I fixed win4lin! At least, I think I did. It compiles cleanly now (a little complaining but nothing serious), so give it a shot.

USERS OF AMD INTERWAVE ISA SOUND: This is not the kernel for you. The driver doesn't compile.

PCI2000/PCI2220I/CPQFCTS drivers (scsi low-level drivers): These will not compile. Don't bother posting about it.

Orinoco drivers: I know I said I wasn't going to do it, but I did. They all compile nicely, and I even added the orinoco_nortel driver to the config so that you can find it if you need it. The only ones that didn't make it were the spectrum_cs and userhermes drivers.

I actually bothered to make a huge config file that has nearly every option set for this release. And I fixed tons of stuff. The only thing I couldn't test was x86_86 (amd64) because I don't have one.

Maybe it's just me, but is anyone else getting pissed at the hdparm script for not working? Doing it manually is getting a little irritating...

To learn how to use the new mouse polling patch, go here.

For the ati fix, go here.

For a cool automotive background picture made by M@rijn, go here.

Ebuild
Patch
mm Patch

[bMd]

hmmmm, anyone getting following when emerge nitro-source?

Code: Select all

>>> Unpacking linux-2.6.10.tar.bz2 to /var/tmp/portage/nitro-sources-2.6.10-r4/work

bzip2: Data integrity error when decompressing.
	Input file = /usr/portage/distfiles/linux-2.6.10.tar.bz2, output file =                              (stdout)

It is possible that the compressed file(s) have become corrupted.
You can use the -tvv option to test integrity of such files.

You can use the `bzip2recover' program to attempt to recover
data from undamaged sections of corrupted files.

tar: Read 3840 bytes from -
tar: Unexpected EOF in archive
tar: Unexpected EOF in archive
tar: Error is not recoverable: exiting now

!!! ERROR: sys-kernel/nitro-sources-2.6.10-r4 failed.
!!! Function unpack, Line 380, Exitcode 2
!!! failure unpacking linux-2.6.10.tar.bz2
!!! If you need support, post the topmost build error, NOT this status message.

[Moloch]

I haven't tried nitro4 yet. But hdparm has no problems on nitro3 for me.

And yes it's running as "rc-update add hdparm default"

[discomfitor]

bMd: try deleting and redownloading the linux-2.6.10.tar.bz2 file.

Moloch: maybe it's just me then...

[EricHsu]

Using it to post this!

Everything works perfectly, however, I have the following error msgs since 2.6.10-nitro3 (no problem with 2.6.9-nitro4):

Code: Select all

ACPI: setting ELCR to 0200 (from 0e20)
ACPI: Subsystem revision 20041210
    ACPI-1138: *** Error: Method execution failed [\STRC] (Node cff44180), AE_AML_BUFFER_LIMIT
    ACPI-1138: *** Error: Method execution failed [\_SB_.PCI0._INI] (Node cff43f60), AE_AML_BUFFER_LIMIT
ACPI: Interpreter enabled

It seems affects nothing, however, it's just annoying to see this...

Any ideas?

[amorpheus]

new nitros *drools*

[ashtophet]

Great!!! Compiled fine, booted as a rocket and by now is working smoothly!! Thanks a lot for yr. wonderful job, Darckness

[Cagnulein]

problem here:

Code: Select all

  CC      drivers/char/drm/ati_pcigart.o
  CC      drivers/char/drm/i830_drv.o
  CC      drivers/char/drm/i830_dma.o
drivers/char/drm/i830_dma.c:114: error: `i830_open' undeclared here (not in a function)
drivers/char/drm/i830_dma.c:114: error: initializer element is not constant
drivers/char/drm/i830_dma.c:114: error: (near initialization for `i830_buffer_fops.open')

[catalinux]

copy /char/drm/i830* from nitro3 over nitro4
intelfbdrv is broken too nitro3 is good

[Cagnulein]

copy that...

it works great. Thx

[thecrazyperson_ws]

dangit! i just downloaded and compiled nitro3 not even 6 hrs ago. after xorg, i guess it's kernel rebuild time again... to nitro4.

EDIT: nitro3 works beautifully... thanks for the great work, darckness. let's hope 4 is better/faster!

[()]

How come you've started including -mm? I originally opted for nitro-sources as an alternative to to mm (too unstable for my liking).

[peterk0]

How come you've started including -mm? I originally opted for nitro-sources as an alternative to to mm (too unstable for my liking).

Yeah, how come?
Maybe it's time to go for cko then

[Pink]

How come you've started including -mm? I originally opted for nitro-sources as an alternative to to mm (too unstable for my liking).

I have to agree, I think this is a huge blow to nitro - downloads have dropped dramatically since the mm was included....

[discomfitor]

I'm not including all of mm. Just the patches that I feel work nicely. If people don't like mm because it's too unstable, then why hasn't anyone else complained about the fact that I use the latest bk snapshots?

I suppose if it's a real big issue for people I'll stop including mm from now on. If you feel that way, post now. It's not really a big deal since it just means less work for me anyway .

I'll work on offering two different ebuilds/patches for this release.

EDIT: Just about done here...

EDIT: Done.

[()]

I'm not including all of mm. Just the patches that I feel work nicely. If people don't like mm because it's too unstable, then why hasn't anyone else complained about the fact that I use the latest bk snapshots?

Divide et impera. Without knowing exactly which patches you include, -mm makes me a bit nervous. From what I remember, not long after I quit using -mm there was discovered a bug that led to rampant file corruption. Maybe you could have the inclusion of mm (and possibly bk) patches depend on a use flag?

[discomfitor]

(): Pure genius. Just...pure genius. I think that's the way I'll handle it from now on unless someone else has a better idea?

[()]

Software modularization in general is a very nice concept, although USE flags are a bit messy (with the current, global, system anyway).

[SysOP XXL]

The new nitro works fine here! Good job, once again

I've spent 20 minutes dealing with the hdparm script though, it made my box freeze completely. I've found out that the '-d1' parameter freezes, removing it from the config fixed it. It doesn't have any drawbacks here, DMA is enabled without this option too (YMMV).

[discomfitor]

Alright, I've uploaded new patches and ebuilds which make the mm/bk patches completely optional. Check the first post for more info. The only difference besides that in the mm patch is that it actually downgrades the ac patch a bit (regular nitro4 uses ac7 now, while nitro4-mm is back on ac4). I don't think it's a huge difference, but who knows?

I hope that will make everyone happier!

[Tiger683]

There is an error, which is caused probably by an option that noone else but me activated:

the block :

+ while(ide_unregister_hwif(hwif) < 0) {
+ hwif_removed_ops(hwif);
+ msleep(1000);

should be:

+ while(ide_unregister_hwif(hwif) < 0) {
+ removed_hwif_iops(hwif);
+ msleep(1000);

or else this breaks the build.....


EDIT:
UMMM, oh yeah, this is from regular nitro4, so ac7

where did the "hwif_removed_ops" come from???????

[()]

Great, merging nitro4 now :]

[discomfitor]

Tiger683: I'm assuming you meant the block in ide-pnp? I just fixed it and reuploaded the patches. I'm not going to punch you; frankly I'm impressed that nitro has become that widespread! And distro wars do suck. Your CFLAGS remind me of when I used to do that on love-sources back in the day. Then I realized that there was too much of a chance to segfault with not a lot of gain .

(): Glad to hear it.

[Tiger683]

yepp

it was breaking my build the whole afternoon... )))
PS: You may punch me, but I'm running nitro on suse.......*runaway*

You never know, distro wars and stuff, been there , seen that..... :[

now building with
CONFIG_CFLAGS_STRING="-O3 -march=athlon-xp -mcpu=athlon-xp -mmmx -msse -mfpmath=sse -pipe -funroll-loops -frerun-cse-after-loop -frerun-loop-opt -falign-functions=4 -maccumulate-outgoing-args -ffast-math -fprefetch-loop-arrays -fomit-frame-pointer"
and regparm

lets see what f***up comes out...... --fforce-addr did mess asm-stuff at prepare stage....

hehe, i wil be glad when hes not going to chicken...(ehm, panic ),then it probably wont break my neck i guess...

Just curious: what bad things happen with CONFIG_IOPRIO_WRITE=y ?

[Archangel1]

Anyone else had issues with -want-mm? It goes beserk on here, just about everything fails. Makes me think I've done something wrong, except there's not a lot of scope for that in

Code: Select all

USE="want-mm" emerge nitro-sources

Oh well, works great without the mm patches on my other machine, so I'll probably go with that.