|C0NRAD| Install V400 | "Flying Colors" | #(Number) 1?

[M-Z]

Finally I solved my previous problems, but another rose up: ALSA - doesn't work.

Code: Select all

localhost m-z # /etc/init.d/alsasound restart
 * Storing ALSA Mixer Levels ...                                          [ ok ]
 * Killing processes using ALSA ...                                       [ ok ]
 * Unloading ALSA modules ...                                             [ ok ]
 * Loading ALSA modules ...
 *   Could not detect custom ALSA settings.  Loading all detected alsa drivers.
 *   Loading: snd_intel8x0 ...                                            [ ok ]
 *   Loading: snd_intel8x0m ...                                           [ ok ]
 *   Loading: snd-seq-oss ...                                             [ ok ]
 *   Loading: snd-pcm-oss ...                                             [ ok ]
 * Restoring Mixer Levels ...                                             [ ok ]

More info: when I start alsamixer, I get:
alsamixer: function snd_ctl_open failed for default: No such device
EDIT: Solved - alsa-driver package instead of module in kernel.

[M-Z]

Another problem - Nero Linux doesn't work. I want to burn files (not iso's) more than 4GB long, so the only solution is to use Nero Linux (AFAIK k3b and gnomebaker cannot do this). Do anyone have any idea, what might be wrong? In every other Linux distribution I tried, it worked; but on Gentoo it doesn't. Maybe it is some kind of misconfiguration of kernel. Most binary distributions builds kernel with almost every possible module; we on Gentoo try to build as small kernel as possible. Maybe I'm lacking some kernel module?
If there is anyone here, who managed to make Nero work, please post your kernel config here. Especially if one is using no-sources 2.6.20.

[Jesterhead]

IIRC with newer versions (unstable tree...) of the burning tools it is possible to burn files over 4GB with K3B...

[M-Z]

IIRC with newer versions (unstable tree...) of the burning tools it is possible to burn files over 4GB with K3B...

Which version? The only thing I need is burning files greater than 4GB; I don't insist on using Nero.

[Jesterhead]

in older versions of K3B was a Bug which prevented burning those large files.

As long as you burn them to UDF it should be possible.

[M-Z]

in older versions of K3B was a Bug which prevented burning those large files.

Yes, I know. But even now there is a limit of 4 GB. I want to burn 4,4 GB.

As long as you burn them to UDF it should be possible.

Well, but does k3b support UDF?

[Jesterhead]

Mine does... (K3B 1.0.1)
If i add more stuff it switches automaticaly to Doublelayer Size (8GB)

[M-Z]

Mine does... (K3B 1.0.1)
If i add more stuff it switches automaticaly to Doublelayer Size (8GB)

Have you actually tried burning file (one single file) longer than 4GB?
Because when I do, I get warning: It is not possible to add files bigger than 4,0 GB: file

[cruzki]

I don't know if it's posible, but if you create an ISO with that file and then burn that iso that can solve the problem, isn't?

[M-Z]

I don't know if it's posible, but if you create an ISO with that file and then burn that iso that can solve the problem, isn't?

Well, I'm not that advanced to use command line burning tools...

[wrc1944]

Apparently, -freorder-blocks-and-partition causes system package dev-libs/gmp-4.2.1-r1 to fail.

I just updated to kde 3.5.7 on my nesl247 gentoo-cookbook installation box (I have 2 other boxes using this new Conrad install method), where I have adopted the Conrad Cflags (-Os -etc.).

I was having problems with kmail and k3b failing, so after revdep-rebuild turned up nothing, I did an emerge -e system with the new flags and a new binutils, which solved the kmail/k3b compile problems.

During the -e system run, gmp failed (no update- just a recompile, so I just did --resume --skipfirst and finished up -e system). At first I suspected LDFLAGS, but eliminated that, and found the above mentioned cflag was indeed the problem on this package.

If anyone runs into gmp failing, this might be your problem.

What clued me in was way back in the gcc output of a failed compile was this line:

Code: Select all

cc1plus: note: -freorder-blocks-and-partition does not work with exceptions

BTW, I'm becoming convinced that -Os and some carefully picked flags as suggested in the Conrad v4.1 install from -O2 produces a snappier desktop than my years previous -O3 and then -O2 rather aggressive configurations. I'm also including -ftree-vectorize and the --as-needed LDFLAG. I've now used this on 3 boxes, with no compiling problems. Admittedly, I run pretty lean kde systems, so I can get away with this, whereas others with more installed packages might have problems.

[vegaman]

Apparently, -freorder-blocks-and-partition causes system package dev-libs/gmp-4.2.1-r1 to fail.

Thanks, I was just running into problems with this, installing a new system with -march=pentium4, which must cause a conflict with something.
On this computer (-march=prescott) it builds fine with -freorder-blocks-and-partition. I'll try it on the other system now without that flag anyway.

Edit: Set it in my /etc/portage/bashrc and emerge -e world continues...

[vegaman]

Well, I'm not that advanced to use command line burning tools...

Have you got UDF support in your kernel?

Code: Select all

File systems --->
  CD-ROM/DVD Filesystems --->
    <*> UDF file system support

[M-Z]

Have you got UDF support in your kernel?

Code: Select all

File systems --->
  CD-ROM/DVD Filesystems --->
    <*> UDF file system support

I have. But I must tell that following the advice I got in another thread, I compiled k3b from SVN and it works. I don't need Nero anymore...

[dracozny]

I have built this system for the second time due to FS corruption, not sure what happened the first time but it seems to be happening again.

after completing the first part of the guide step by step and rebooting the system to the new kernel, when I attempt to do any emerges the system freezes when decompressing the TAR file.
this started last time and when I attempted to do an ext4 fsck the entire file system was destroyed completely.
*edit as I write* just rebooted the system and now it gives a kernel panic
rebooting with live cd but not sure what to do from there other than startover and not use ext4.

*edit* well I booted up the live cd and all is well so far but I am apprehensive to do an fsck

[wrc1944]

I assume you've seen the ext4 posts on page 3 and 4 of this thread.

My first thoughts would be what kernel have you got installed, is the config correctly set for ext4, is your fstab correct, and what version of e2fsprogs is installed on your system?

[dracozny]

kernel = viper2
ext4 kernel options:

Code: Select all

<*> Ext4dev/ext4 extended fs support development
[*]      Ext4dev extended attributes

eix e2fsprogs>>>> 1.39-r1

Code: Select all

fstab:
/dev/hda3        /          ext4dev        noatime,extents  0 1

I'll try to svn the 1.39-r3 version of e2fsprogs, maybe that will resolve it now if only my ISP would cooperate but thats another thread entirely

[vegaman]

I'll try to svn the 1.39-r3 version of e2fsprogs, maybe that will resolve it now if only my ISP would cooperate but thats another thread entirely

Looking in /etc/portage/package.mask/conrad releases after 1.39-r1 have been masked because ext4dev support was removed in r2. I'm not sure if it was added back in r3, so I wouldn't upgrade unless you know.

Other than that I don't have enough knowledge to work out the problem, the same kernel, ext4dev config and e2fsprogs haven't caused any problems (yet) with my testing on /usr/portage.

[dracozny]

well while using the live cd the partition continued to corrupt itself I think I will hold off on ext4 at this time, not sure what is going on. my only other thought is a bad drive but the partition table seems ok, so not sure.
otherwise it was blistering fast for this old 400mhz xeon PII. was actually beating out my amd xp 2000+.

[Grubshka]

emerging java-config complains : "arch config not found".

Seem to be the ARCH="athlon-xp" in make.conf which get it confused...
I renamed ARCH to MYARCH in make.conf and it compiled.


Conrad rulez !!!

[j-m]

COREFLAGS="-frename-registers -fweb -pipe -fomit-frame-pointer -funit-at-a-time -freorder-blocks -fno-ident -freorder-blocks-and-partition -fmerge-all-constants -combine"
CFLAGS="-Os -march=${ARCH} ${CPUFLAGS} ${COREFLAGS}"
CXXFLAGS="${CFLAGS} -fvisibility-inlines-hidden"

No comment... (I love the -Os bit though.... )

# ln -s linux-2.6.20-gentoo* linux
# cd linux
# make allyesconfig

Heck, you are going to take part in the most huge kernel contest?

# nano -w /etc/locale.gen
en_US/ISO-8859-1
en_US.UTF-8/UTF-8

Yay, so here's where all the bugs abou 'glibc does not install any locales!!!' come from? The syntax is broken.

7.3 - Customizing Make.conf
You will need to change 'ARCH="i686"', to your supported processor.
Options: pentium2, pentium3, pentium4, prescott, k8, athlon64, nocona, opteron, pentium-m, i686, athlon-xp, etc.

NO!!! Stop messing with ARCH; all the above examples are plain wrong. Valid ARCH value is one of those in /usr/portage/profiles/arch.list, NOT a name of your CPU.

[genestyler]

you would mean to put something like CFLAGS="march=x86 -O2 -pipe" in make.conf ????? or CFLAGS="march=hppa -O2 -pipe"

[Ishiki]

No, he meant to remove ARCH variable from

Code: Select all

CFLAGS="-Os -march=${ARCH} ${CPUFLAGS} ${COREFLAGS}"

(and to not change it to your CPU), because it can be dangerous and use simple (e.g.) -march=athlon-xp instead.

[EndrjuX]

1.
from gentoo wiki:

-fvisibility-inlines-hidden

Avoids exporting unneeded symbols from libraries, making them smaller. Some older wxGTK versions don't compile. The wxGTK-2.6 version compiles, but doesn't work. Prelinking will no longer work. Used where safe by KDE, if USE contains kdehiddenvisibility.

Conrad-gentoo is using prelinking ins't it?

2.
-O2 is better idea imo and also -freorder-blocks is included in -O2

3.
-combine

[Master Chef]

Okay, I'm having just the hardest time installing Gentoo again.

Simply said: I'm not getting g++.
Nada, none. "find / -iname 'g++'" returns

Code: Select all

/usr/i686-pc-linux-gnu/gcc-bin/4.1.1/g++

. And that's it. That's only the 2007.0 toolchain though, and not the Conrad-way toolchain. And upon emerge gcc-4.1.2, I have no g++. If I was to emerge gcc-4.1.1, still no g++!

I don't think it's my useflags:

Code: Select all

#/etc/make.conf
COREFLAGS="-frename-registers -fweb -pipe -fomit-frame-pointer -funit-at-a-time -freorder-blocks -fno-ident -freorder-blocks-and-partition -fmerge-all-constants -ftree-vectorize"

CFLAGS="-Os -march=pentium-m ${COREFLAGS}"
CXXFLAGS="${CFLAGS} -fvisibility-inlines-hidden"
LDFLAGS="-Wl,--hash-style=gnu,--as-needed"
CHOST="i686-pc-linux-gnu"

ACCEPT_KEYWORDS="~x86"

#CONRAD_OVERLAY="/usr/local/overlays/conrad"
PORTAGE_OVERLAY="/usr/portage"
PORTDIR_OVERLAY="${PORTAGE_OVERLAY} ${CONRAD_OVERLAY}"

CONFIG_PROTECT="-*"
#BAD.

PORTAGE_NICENESS="15"
FEATURES="parallel-fetch distlocks"
MAKE_OPTS="-j2"

USE="build nptlonly unicode hashstyle mmx mmxext sse sse2 -ipv6 nomalloccheck X svg bzip2 dbus hal alsa cups opengl acpi pdf cairo firefox truetype libnotify"
LINGUAS="en"
INPUT_DEVICES="keyboard mouse evdev synaptics"
VIDEO_CARDS="nvidia nv vmware"

#source /usr/portage/local/layman/make.conf

And I'm having a hard time thinking about what else it could possibly be. I don't have nocxx set anywhere in make.conf, and I've wiped, and extracted stages repeatedly. No go ;(.

So here I am, following the Conrad way, and I'm pretty sure this problem would happen in the normal vanilla way too (if I was to try and upgrade the toolchain). So what now ;|?