You are not logged in.
i'm right now compiling the kernel26mm 2.6.8.1.mm1-1
CHANGELOG
(2.6.7.mm7-1 -> 2.6.8.1.mm1-1)
- build() modified: now, you can choose your favourite config style interactively (eye candy? possibly! ;-) hope a lot of you will find it also usefull) :-)
- all scsi stuff will be included as modules (this does not make this kernel a fully scsi one, but allow loading scsi as modules ... i'm thinking of how much it would slow down enabling all scsi in it - i have only my zip-drive using scsi (imm), so it's almost no use for me, but if a ide+scsi kernel would be almost the same like a ide-only, it would 1) make mm-kernel available for scsi people 2) simplify things)
- all SATA (under scsi) is now built-in (hope to get feedback from sata people) --- the depreatched SATA is not built
- all additional fb drivers (nvidia, radeon...) are now built as modules - hope it works, but this can also break the fb ... we will see - reason: fbi or other great fb tools would work better with native drivers (i use fbi for presenting my pics and under vga, the fb has some momentum)
- i changed the default linux logo for the fb - it is now "ArchStyle compatible" :-) (hopefully, i got everything right - if not, i'll remove my and include the standard logo back)
- if it builds ok, i will rebuild it with the patch to allow users to burn cds/dvds (does not make sense otherwise)
more here: http://bbs.archlinux.org/viewtopic.php?t=6214
hope the kernel26mm is of use for some of you :-)
comments welcome - use at your own risk
The impossible missions are the only ones which succeed.
Offline
new trouble with 2.6.8.1:
http://kerneltrap.org/node/view/3659
The impossible missions are the only ones which succeed.
Offline
Some application (non arch package) do not run under kernel26mm (at least 2.6.7)
Zend IDE (I do not know if it is for all java application)
MS Word under wine
P.S.
Please don't kill me because I use word, unfortunatly openoffice is not able to read correctly EU commission Project module. :cry:
Offline
Some application (non arch package) do not run under kernel26mm (at least 2.6.7)
Zend IDE (I do not know if it is for all java application)
if it is java, then it was only temporarely broken (the 2.6.7.mm7 had java broken)
MS Word under wine
P.S.
Please don't kill me because I use word, unfortunatly openoffice is not able to read correctly EU commission Project module. :cry:
i use word myself - but natively under windows (wine is horribly slow) - i will check this ... maybe it's not only the kernel, but also other things
The impossible missions are the only ones which succeed.
Offline
From: Andrew Morton [email blocked]
Subject: Re: [PATCH] bio_uncopy_user mem leak
Date: Tue, 17 Aug 2004 15:50:04 -0700
Kurt Garloff [email blocked] wrote:
>
> When using bounce buffers for SG_IO commands with unaligned
> buffers in blk_rq_map_user(), we should free the pages from
> blk_rq_unmap_user() which calls bio_uncopy_user() for the
> non-BIO_USER_MAPPED case. That function failed to free the
> pages for write requests.
>
> So we leaked pages and you machine would go OOM. Rebooting
> helped ;-)
Eureka. Thanks.
This really should trigger a 2.6.8.2. We'll see.
i think 2.6.8.2 will come soon
(well i hope it ;-) )
that's the new kernel development
without having a unstable kernel tree such things may happen more and more
Offline
dp lets have PKGBUILD please......
then we can just edit .. when 2.6.8.2 comes out...
Mr Green I like Landuke!
Offline
run abs
and then
/var/abs/extra/kernels
Offline
yeah ok 2.6.7 mm7 .....
Mr Green I like Landuke!
Offline
dp lets have PKGBUILD please......
then we can just edit .. when 2.6.8.2 comes out...
# $Id: PKGBUILD,v 1.1 2004/08/03 21:53:41 damir Exp $
# Maintainer: damir <damir@archlinux.org>
pkgname=kernel26mm
origkernelver=2.6.8.1
mmver=1
pkgver=${origkernelver}.mm${mmver}
kernelver=${origkernelver}-mm${mmver}
pkgrel=1
pkgdesc="The Linux Kernel and modules (IDE support) [mm Kernel]"
url="http://kerneltrap.org/forum/linux/kernel/2.6/mm"
backup=('boot/kconfig26mm')
depends=('module-init-tools')
install=kernel26.install
source=(ftp://ftp.kernel.org/pub/linux/kernel/v2.6/linux-$origkernelver.tar.bz2
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/${origkernelver}/$kernelver/$kernelver.bz2
config
logo_linux_clut224.ppm)
build() {
# if we run the makepkg more than once and use 'makepkg -e' to save time:
if [ -d $startdir/src/linux-${origkernelver} ]
then
cd $startdir/src/linux-${origkernelver}
# unpack and apply mm-patch:
cat ../${kernelver} | patch -p1 || return 1
# rename to correct $kernelver dir
mv $startdir/src/linux-${origkernelver} $startdir/src/linux-$kernelver
fi
cd $startdir/src/linux-$kernelver
# restore classical cd/dvd burning behaviour:
sed -i '195s|if (verify_command(file, cmd))| /* if (verify_command(file, cmd))|;198s|/*| |' drivers/block/scsi_ioctl.c
# cleanup of the tree (not essential, as kernel should be clean already, but let's make SURE it is :-)
make mrproper
# get rid of the 'i' in i686
carch=`echo $CARCH | sed 's|i||'`
cat ../config | sed "s|#CARCH#|$carch|g" >./.config
# replace logo:
cp $startdir/src/logo_linux_clut224.ppm $startdir/src/linux-$kernelver/drivers/video/logo/ && echo "==> logo replaced successfully"
sleep 2
# let's be "user-friendly":
echo "==> "
echo "==> How do you want to configure the kernel:"
echo "==> ----------------------------------------"
echo "==> (for ArchLinux defaults, use 'not_at_all (1)')"
echo "==> "
AUSWAHL="not_at_all menuconfig xconfig oldconfig config"
select opt in $AUSWAHL; do
if [ "$opt" = "not_at_all" ]; then
yes "" | make config || return 1
break
elif [ "$opt" = "menuconfig" ]; then
make menuconfig || return 1
cp .config ../../config.new && echo "==> your config was saved to $startdir/config.new"
break
elif [ "$opt" = "xconfig" ]; then
make xconfig || return 1
cp .config ../../config.new && echo "==> your config was saved to $startdir/config.new"
break
elif [ "$opt" = "config" ]; then
make config || return 1
cp .config ../../config.new && echo "==> your config was saved to $startdir/config.new"
break
elif [ "$opt" = "oldconfig" ]; then
make oldconfig || return 1
cp .config ../../config.new && echo "==> your config was saved to $startdir/config.new"
break
else
clear
echo "==> I don't understand (bad option) ... assuming you don't want "
echo "==> to configure and use the defaults."
sleep 1
echo "==> 3..."
sleep 1
echo "==> 2..."
sleep 1
echo "==> 1..."
sleep 1
echo "==> 0: GO! --------------------------- :-)"
yes "" | make config || return 1
break
fi
done
make clean bzImage modules || return 1
mkdir -p $startdir/pkg/{lib/modules,boot}
make INSTALL_MOD_PATH=$startdir/pkg modules_install || return 1
cp System.map $startdir/pkg/boot/System.map26mm
cp arch/i386/boot/bzImage $startdir/pkg/boot/vmlinuz26mm
install -D -m644 Makefile $startdir/pkg/usr/src/linux-$kernelver/Makefile
install -D -m644 .config $startdir/pkg/usr/src/linux-$kernelver/.config
install -D -m644 .config $startdir/pkg/boot/kconfig26mm
mkdir -p $startdir/pkg/usr/src/linux-$kernelver/include
mkdir -p $startdir/pkg/usr/src/linux-$kernelver/arch/i386/kernel
for i in acpi asm-generic asm-i386 config linux math-emu net pcmcia scsi video; do
cp -a include/$i $startdir/pkg/usr/src/linux-$kernelver/include/
done
# copy files necessary for later builds, like nvidia and vmware
cp -a scripts $startdir/pkg/usr/src/linux-$kernelver/
mkdir -p $startdir/pkg/usr/src/linux-$kernelver/.tmp_versions
cp arch/i386/Makefile $startdir/pkg/usr/src/linux-$kernelver/arch/i386/
cp arch/i386/kernel/asm-offsets.s $startdir/pkg/usr/src/linux-$kernelver/arch/i386/kernel/
# copy in Kconfig files
for i in `find . -name "Kconfig*"`; do
mkdir -p $startdir/pkg/usr/src/linux-$kernelver/`echo $i | sed 's|/Kconfig.*||'`
cp $i $startdir/pkg/usr/src/linux-$kernelver/$i
done
cd $startdir/pkg/usr/src/linux-$kernelver/include && ln -s asm-i386 asm
chown -R root.root $startdir/pkg/usr/src/linux-$kernelver
cd $startdir/pkg/lib/modules/$kernelver &&
(rm -f build; ln -sf /usr/src/linux-$kernelver build)
}
but you will need the 2 missing files that will be in ABS :-)
it's almost finished compiling .... wait some hours, please :-)
The impossible missions are the only ones which succeed.
Offline
Ok... 8) thank you....
what about gconfig :cry:
these kernel builds are getting complicated .... if we could pull out a lot of the removing of the "i"s
& your menu ... the builds would look a lot cleaner & easier to read... say for example your menu
for make config ... makemenu ...
just a thought...
Mr Green I like Landuke!
Offline
uploading right now ... hope not to have forgotten something ... it should be up in about half an hour
have fun
The impossible missions are the only ones which succeed.
Offline
just updated 2.6.8.1 is in current
Mr Green I like Landuke!
Offline
just updated 2.6.8.1 is in current
you mean 2.6.8.1.mm1 or 2.6.8.1?
the vanilla is in current, the mm is (now, finally) in extra - enjoy!
The impossible missions are the only ones which succeed.
Offline
Ok... 8) thank you....
what about gconfig :cry:
these kernel builds are getting complicated .... if we could pull out a lot of the removing of the "i"s
& your menu ... the builds would look a lot cleaner & easier to read... say for example your menu
for make config ... makemenu ...
just a thought...
what is gconfig? i never heard about it (actually personally mostly using menuconfig to configure or oldconfig to update)
(i'm not going to look at the sources what gconfig is, because i cleaned up the tree and don't want to download again ~40mb)
my "menu" is a very primitive hack to let users choise the way they want to configure the kernel (i know, it is very easy to edit the PKGBUILD, but this way, you don't need to think also about that) --- i included it, because i realized, that a lot of people reconfigure the kernel to their own needs, and this way, they whould not need to edit the lines in the build() to simply do that (it's a solution for the lazy, because you choose to configure or not and if, in what way)
The impossible missions are the only ones which succeed.
Offline
2.6.8.1 is in current...
2.6.8.1-mm1 built ok..
I still think kernel PKGBUILDs are getting complicated...
later
Mr Green I like Landuke!
Offline
I still think kernel PKGBUILDs are getting complicated...
:-) the only way to make them easier would be either to introduce more variables (good, but not needed) or eliminate some stuff (bad idea)
The impossible missions are the only ones which succeed.
Offline
:-) the only way to make them easier would be either to introduce more variables (good, but not needed) or eliminate some stuff (bad idea)
stuff is that like a technical term
Mr Green I like Landuke!
Offline
I still think kernel PKGBUILDs are getting complicated...
You should have seen my kernel PKGBUILDs with automatic kernel version detection + self updating kerver & paths. I'll put it on the forums after I polish it a little, it fixes some problems with multiple patching.
:: / my web presence
Offline
IceRam Ok I cannot wait
dp I have an issue with Nvidia drivers ...rivafb configured to include rivafb support as a loadable kernel module
conflicts with Nvidia driver (will compile driver but not work when booted)
:cry:
Mr Green I like Landuke!
Offline
I think the nvidia conflict was in 2.6.7-mm7 too.
The 2.6.8.1-mm1 now allows burning in k3b as a regular user (root only in 2.6.8.1 vanilla, the memory leak problem is with audio cds). Java is fine all over now.
Thanks for all of the work.
Offline
Offline
dp I have an issue with Nvidia drivers ...rivafb configured to include rivafb support as a loadable kernel module
conflicts with Nvidia driver (will compile driver but not work when booted)
:cry:
you use fb? does the nvidia driver provide it's own rivafb? i included most modern fb modules in the mm kernel, but if the nvidia driver already provides it's own one, i will disable the kernel-own nvidia fb, if this is a help
The impossible missions are the only ones which succeed.
Offline
The 2.6.8.1-mm1 now allows burning in k3b as a regular user (root only in 2.6.8.1 vanilla,
:-)
the memory leak problem is with audio cds).
the leak is presistent in the 2.6.8.1.mm too!!! - the patch on kerneltrap fails applying
Java is fine all over now.
i know :-) (i code mainly in java)
Thanks for all of the work.
you're welcome - mainly i did it for me (how egoistic), but hope others will profit also from it :-)
The impossible missions are the only ones which succeed.
Offline
i think the problem is when you use rivafb the nvidia drivers will stop working i read something like this in the nvidia readme
you can try to rebuild the nvidia module or disable rivafb and force vesa fb
Offline
i think the problem is when you use rivafb the nvidia drivers will stop working i read something like this in the nvidia readme
you can try to rebuild the nvidia module or disable rivafb and force vesa fb
to nvidia users: if you think, it would be better disabling rivafb, i will do so in 2.6.8.2.mm1 or 2.6.8.1.mm2
The impossible missions are the only ones which succeed.
Offline