2.6.10-nitro4: The Kernel Formerly Known As Hercules

[rommel]

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

well i think fs corruption ? ... booting nitro3 and running it for a bit toasted one install using reiser4 fs. possibly this option was the cause... havent tried it on any others.

[rmh3093]

I think that the use variables would be a good idea... they dont have to be added to the make.conf if u do something like 'export USE="want-mm"'

but anyway.... this patch that allows cflags to be used... does that mean prior kernels never used cflags from the make.conf file before and what sort of cflags are people using on their kernels, my whole system is built with

-O3 -march=pentium4 -mtune=pentium4 -pipe -fforce-addr -falign-functions=4 -fprefetch-loop-arrays -fomit-frame-pointer

and its stable and I use reiser4 also and have had no problems, if my kernel was that same config i bet i would feel the nitro

[discomfitor]

Archangel1: define "goes berserk". I'm running it right now and have been for a while. Try redownloading the patches/ebuild if you're having issues with patching.

rmh3093: no kernel will ever use the cflags from make.conf. This patch just allows you to set your own under kernel hacking.

[rmh3093]

hey i was just wondering how you all go about configuring your kernels, because I disable everything basically and then build everything I know i use into the kernel, and if anything seems like it wants to run as a module after that i switch to a module, and I dont really ever run in to build problems with random drivers I dont use

[rmh3093]

well thats what i meant basically, this is a new ability..... has anyone had any success with them here?

[charlieg]

Anybody compared this to nitro3? For me, nitro3 has been really good. Is there any reason to upgrade?

[discomfitor]

charlieg: You should consider switching on the grounds that I fixed a lot of errors from nitro3 in both nitro4 and nitro4-mm (bunch of ide-related stuff). I'm not sure if it will directly affect you, but better safe than sorry.

rmh3093: this isn't actually a new ability; it's been in nitro since nitro first started (AFAIK). When I used to play with it, I remember putting in the most extreme flags I could think of just to see what would happen. As it turns out, the kernel is pretty durable. I couldn't tell the difference between the kernel with crazy cflags and the one with normal ones. So I stopped.

[rmh3093]

sounds like something else to give me hours to play with

[Archangel1]

Archangel1: define "goes berserk". I'm running it right now and have been for a while. Try redownloading the patches/ebuild if you're having issues with patching.

The patch fails massively. -p0 doesn't find any files, -p1 finds them but sees a lot of "previously applied patches", and fails on just about every single hunk.

As I said, it's not subtle - I figured something's gone wrong, I just didn't see anywhere in the process it could have, and I thought I'd ask if anyone else had any issues.

I've compiled it without mm and it works great. Good job on this one.

[discomfitor]

Archangel1: did you try redownloading the patch? I'm glad that it worked without the mm patch, but I'm a bit concerned; if it doesn't work for you, then is everyone having problems? Since nobody else has posted this I'm assuming no, but I'd appreciate it if you could try redownloading both patches.

[Tiger683]

I would say that for example using sse for floating-point math can bring a LOT in prformance, but many think math flags are the more crazy

[Woll0r]

It fails to compile

Code: Select all

  CC [M]  fs/smbfs/request.o
fs/smbfs/request.c: In function `smb_recv_trans2':
fs/smbfs/request.c:714: error: duplicate label `out_bad_parm'
fs/smbfs/request.c:705: error: `out_bad_parm' previously defined here
fs/smbfs/request.c:718: error: duplicate label `out_bad_data'
fs/smbfs/request.c:709: error: `out_bad_data' previously defined here
fs/smbfs/request.c:721: error: duplicate label `out_EIO'
fs/smbfs/request.c:712: error: `out_EIO' previously defined here
make[2]: *** [fs/smbfs/request.o] Error 1
make[1]: *** [fs/smbfs] Error 2
make: *** [fs] Error 2

But hey, this isn't crucial so I'll get it out of my config to see this kernel in action

[Teh Penguin D00d]

You guys didn't think you could pull off another Nitro release without hearing from me; did you?

I documented the entire build process, after the ebuild download/digest/merge.

One thing to note...when I did get booted, uname -r output "2.6.10-ck"...not nitro. Even though my symlink is pointing to the proper location. I don't know it's a mistake on my part, or on yours; but...here goes nothing...and yes; it's so verbose I even left the parts where I made typo's in...

Code: Select all

Alright...here's to being excessively detailed...

root@dustin src # cd /usr/src/
root@dustin src # ls
linux                linux-2.6.10-nitro4     linux-2.6.9-gentoo-r13  linux-2.6.9-gentoo-r8  xorg-x11-6.7.0-r3
linux-2.6.10-nitro2  linux-2.6.9-gentoo-r1   linux-2.6.9-gentoo-r4   linux-2.6.9-gentoo-r9
linux-2.6.10-nitro3  linux-2.6.9-gentoo-r12  linux-2.6.9-gentoo-r6   linux-2.6.9-nitro4
root@dustin src # rm .linux 
rm: cannot remove `.linux': No such file or directory
root@dustin src # rm ./linux
root@dustin src # ls
linux-2.6.10-nitro2  linux-2.6.9-gentoo-r1   linux-2.6.9-gentoo-r4  linux-2.6.9-gentoo-r9
linux-2.6.10-nitro3  linux-2.6.9-gentoo-r12  linux-2.6.9-gentoo-r6  linux-2.6.9-nitro4
linux-2.6.10-nitro4  linux-2.6.9-gentoo-r13  linux-2.6.9-gentoo-r8  xorg-x11-6.7.0-r3
root@dustin src # ln -sf ./linux-2.6.10-nitro4 ./linux
root@dustin src # ls -l
total 8
lrwxrwxrwx   1 root root   21 Jan  8 18:02 linux -> ./linux-2.6.10-nitro4
drwxr-xr-x  20 root root 1400 Jan  6 22:21 linux-2.6.10-nitro2
drwxr-xr-x  20 root root  800 Jan  7 23:19 linux-2.6.10-nitro3
drwxr-xr-x  19 root root  656 Jan  8 17:59 linux-2.6.10-nitro4
drwxr-xr-x  19 root root 1312 Dec 31 17:02 linux-2.6.9-gentoo-r1
drwxr-xr-x  18 root root  664 Dec 24 00:21 linux-2.6.9-gentoo-r12
drwxr-xr-x  18 root root  664 Dec 31 08:15 linux-2.6.9-gentoo-r13
drwxr-xr-x  19 root root 1312 Nov 26 20:28 linux-2.6.9-gentoo-r4
drwxr-xr-x  18 root root  664 Nov 24 15:57 linux-2.6.9-gentoo-r6
drwxr-xr-x  18 root root  696 Dec  2 22:01 linux-2.6.9-gentoo-r8
drwxr-xr-x  18 root root  696 Dec  4 18:05 linux-2.6.9-gentoo-r9
drwxr-xr-x  20 root root 1304 Dec 30 15:23 linux-2.6.9-nitro4
drwxr-xr-x   3 root root   72 Dec 31 19:33 xorg-x11-6.7.0-r3

root@dustin src # cd linux
root@dustin linux # make clean
root@dustin linux # make mrproper
root@dustin linux # make menuconfig

::configures kernel, resulting in the following .config file::

root@dustin linux # make > /home/dustin/kernelmake.txt 2>&1
root@dustin linux # ls /boot/
bzNitroUdev  fbsplash  grub  initrd  initrd-gentoo-1280x1024  lost+found

(bzNitroUdev is my 2.6.9 Nitro 4 kernel, with working Udev.)

root@dustin linux # cp arch/i386/boot/bzImage /boot/bzNitro
root@dustin linux # ls /boot/
bzNitro  bzNitroUdev  fbsplash  grub  initrd  initrd-gentoo-1280x1024  lost+found
root@dustin linux # nano -w /boot/grub/grub.conf

::only showing appropriate part::

title Gentoo Dev
root (hd0,0)
kernel /bzNitro root=/dev/hda3 video=radeonfb:ywrap,mtrr,1280x1024@60 splash=verbose,theme:emergence
initrd (hd0,0)/fbsplash

::fbsplash is the initrd holding the config for fbsplash.  Yes, it works with the other kernel.  Yes, I've tried it without the initrd.::

root@dustin linux # umount /boot 
root@dustin linux # exit
exit
dustin@dustin dustin $ 

Kernelmake.txt, as referenced above
.config, again. I rebuild by hand each time; just to see if any new options are there.

I took some advice, and checked for /proc/sys/kernel/hotplug.../proc/sys didn't even exist, though /proc did, along with most of it's directories.

Code: Select all

1
1033
124
137
14737
14752
15989
16004
16141
16142
16143
16144
16145
16146
16227
16228
16234
18
2
247
248
249
250
2854
2855
3
4
851
9
920
959
acpi
asound
buddyinfo
bus
cmdline
cpuinfo
crypto
devices
diskstats
dma
driver
execdomains
fb
fbsplash
filesystems
fs
ide
interrupts
iomem
ioports
kallsyms
kcore
kmsg
loadavg
locks
meminfo
misc
modules
mounts
mtrr
net
partitions
scsi
self
slabinfo
stat
swaps
sysvipc
tty
uptime
version
vmstat

Once again; any help would be beyond appreciated...

::dreams of the day I'll be using a 2.6.10 kernel...::

[Tiger683]

My advice would be to use the initial release of the day (nitro4).

Does anyone know if regparm kills alsa driver?

[John5788]

ahem, it says im running 2.6.10-ck instead of 2.6.10-nitro4? did someone forget to rename it or something, or am i just retarded?

Code: Select all

[/home/john5788 :: 04:48 PM :: 1]
[john5788@john5788.x5788.net] $ uname -a
Linux john5788.x5788.net 2.6.10-ck #1 SMP Sat Jan 8 16:43:29 PST 2005 i686 Intel(R) Pentium(R) 4 CPU 3.00GHz GenuineIntel GNU/Linux

anyone elses uname -a say this too?

[John5788]

yes indeed someone did not change something in -cks makefile

Code: Select all

VERSION = 2
PATCHLEVEL = 6
SUBLEVEL = 10
EXTRAVERSION =-ck
NAME=Woozy Numbat

changed it to -nitro4 and now its all good

Code: Select all

[/home/john5788 :: 04:59 PM :: 1]
[john5788@john5788.x5788.net] $ uname -a
Linux john5788.x5788.net 2.6.10-nitro4 #2 SMP Sat Jan 8 16:53:27 PST 2005 i686 Intel(R) Pentium(R) 4 CPU 3.00GHz GenuineIntel GNU/Linux

[discomfitor]

Yeah, I've been aware of the lack of name (forgot it when I redid the mm split; the mm version has had it ) for a while but I had to run out and do some errands. Fixed the smbfs thing (it already was fixed in the mm version despite the fact that mm is supposed to be more unstable...) and the name and reuploaded.

[tariq.firoz]

hmm.... am going to try now ...

Tariq

alsa 1.0.8-rc1 not there, still not able to get my 915 mobo's soundard up using the default kernel

[monkey89]

I'm getting the same problem with -nitro4 as I am with -cko3, a kernel panic cannot mount root device after a lot of spam about a MediaSeek DriveComplete error or something along those lines (I don't know how to copy the exact error). -cko1 works and I think it has -ac1 or -ac2, and -ac7 from ac-sources crashes with the error. I think it may have to do with my hard drive having a host-protected area, and after tinkering with the partition I need to add hda=stroke in order to boot, but it boots fine, and with that line in knoppix my partition (reiserfs) fsck's fine.

Anyone have ideas on how to fix it? I'd rather not have to worry about backing out -ac patches every time I want to use a newer custom kernel.

[discomfitor]

monkey89: are you using nitro4 or nitro4-mm? Try using the other one.

[monkey89]

monkey89: are you using nitro4 or nitro4-mm? Try using the other one.

Was using plain old nitro4, but the ebuild with "use-mm" enabled fails:

* Applying mm patch because you wanted it!

* Applying patch-2.6.10-nitro4-mm.bz2 ...

* Failed Patch: patch-2.6.10-nitro4-mm.bz2!
*
* Include in your bugreport the contents of:
*
* /var/tmp/portage/nitro-sources-2.6.10-r4/temp/patch-2.6.10-nitro4-mm.bz2-9945.out

That file is huge and full of errors... do you need me to put it somewhere? Anyways, if the -mm has -ac4, I think it will fail anyways. But I'll test whatever you'd like.

[discomfitor]

Agh...you're right! I'm fixing it as we speak, so you don't need to post it anywhere.

EDIT: fixed. Try it now.

[monkey89]

Hrm...

Well, I tried resizing my reiserfs partition in Knoppix and it resulted in the -mm booting, unfortunately it also hosed a lot of my data. Thankfully, I keep critical files on a separate ext3 partition, and I was able to copy off some important files from the reiserfs root partition.

I'm back in Windows for now, as much as I can't stand it. In a few weeks I'll have a week off, and I'll reinstall linux then (if not earlier). I'll probably use ext3 and a more stable kernel as well, because I'm sick of crashing, as nice as being semi-experimental is.

[EricHsu]

I'm experiencing something weird:

If the system is under some kind of heavy load and the swap is in use, the system will hang on "Deactivating swap..." while reboot or shutdown. Then I have to press the "reboot" button (Ctrl + Alt + Del doesn't work...)

Is this only me? I'm using the 2.6.9-nitro4 now...

[discomfitor]

EricHsu: I can't say that I've been experiencing the same problem; did you try running the hdparm script? Because I've noted some issues with that recently which may be related (hdparm takes over the processor, won't die)