[Buildroot] linux .config : override: reassigning to symbol AEABI

[Buildroot] linux .config : override: reassigning to symbol AEABI

[Courousse, Damien]

Hello buildrooters!

I am trying to use buildroot for the ATEB9200 board form Embest. It is somewhat similar to the AT91rm9200dk/ek configs, including the CPU arm920t.

I am using the kernel v2.6.28, patched with the Atmel patched automatically applied by buildroot.
I am using buildroot v2009-05.

The kernel loading gets stuck after mounting the NFS root : 
================================================================
....
IP-Config: Complete:
     device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
     host=debian_test, domain=, nis-domain=(none),
     bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
Looking up port of RPC 100003/2 on 10.84.34.52
Looking up port of RPC 100005/1 on 10.84.34.52
VFS: Mounted root (nfs filesystem).
Freeing init memory: 116K
Kernel panic - not syncing: Attempted to kill init!
================================================================

I have configured buildroot to avoid EABI, and everything seems OK on this side in the config files.

I guess my problem comes from the kernel config, where the EABI symbol is overridden during buildroot compilation: 
=================================================================
make[1]: entrant dans le r?pertoire ? /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28 ?
scripts/kconfig/conf -o arch/arm/Kconfig
.config:1344:warning: override: reassigning to symbol AEABI
.config:1345:warning: override: reassigning to symbol IPV6
================================================================

I give to the kernel the following command line: 
root=/dev/nfs rw nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off init=/bin/sh

Do you think this is a normal behaviour, knowing that I did *not* configure buildroot, uclibc and the kernel for EABI?
Please let me know if you see options to dig into this problem.

Thanks for any help,
Damien 

Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

[Buildroot] linux .config : override: reassigning to symbol AEABI

[Daniel Mack]

On Wed, Jul 29, 2009 at 02:59:14PM +0200, Courousse, Damien wrote:
> The kernel loading gets stuck after mounting the NFS root : 
> ================================================================
> ....
> IP-Config: Complete:
>      device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
>      host=debian_test, domain=, nis-domain=(none),
>      bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
> Looking up port of RPC 100003/2 on 10.84.34.52
> Looking up port of RPC 100005/1 on 10.84.34.52
> VFS: Mounted root (nfs filesystem).
> Freeing init memory: 116K
> Kernel panic - not syncing: Attempted to kill init!
> ================================================================

That smells like an ABI confusion, yes.

> I have configured buildroot to avoid EABI, and everything seems OK on this side in the config files.
> 
> I guess my problem comes from the kernel config, where the EABI symbol is overridden during buildroot compilation: 
> =================================================================
> make[1]: entrant dans le r?pertoire ? /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28 ?
> scripts/kconfig/conf -o arch/arm/Kconfig
> .config:1344:warning: override: reassigning to symbol AEABI
> .config:1345:warning: override: reassigning to symbol IPV6
> ================================================================

Building an OABI toolchain for an EABI kernel doesn't make sense, hence
the configs are brought in sync by buildroot.
target/linux/Makefile.in.advanced is where it happens, and it AEABI
should not be set in the kernel config unless BR2_ARM_EABI is set. Could
you double check?

> Do you think this is a normal behaviour, knowing that I did *not* configure buildroot, uclibc and the kernel for EABI?

No, that shouldn't happen. Could you pastebin your br2 .config
somewhere?


Daniel

[Buildroot] linux .config : override: reassigning to symbolAEABI

[Courousse, Damien]

Hello Mark, 

Thanks much for your help,

...
> > Looking up port of RPC 100003/2 on 10.84.34.52 Looking up port of 
> > RPC 100005/1 on 10.84.34.52
> > VFS: Mounted root (nfs filesystem).
> > Freeing init memory: 116K
> > Kernel panic - not syncing: Attempted to kill init!
> > ================================================================
> 
> That smells like an ABI confusion, yes.
> 
...
> > ================================================================
> 
> Building an OABI toolchain for an EABI kernel doesn't make sense, 
> hence the configs are brought in sync by buildroot.
> target/linux/Makefile.in.advanced is where it happens, and it AEABI 
> should not be set in the kernel config unless BR2_ARM_EABI is set. 
> Could you double check?
[Damien Courouss?] 

target/linux/Makefile.in.advanced contains the following: 
$(LINUX26_DIR)/.configured: $(LINUX26_DIR)/.config
	$(SED) '/CONFIG_AEABI=y/d' $(LINUX26_DIR)/.config ifeq ($(BR2_ARM_EABI),y)
	echo "CONFIG_AEABI=y" >> $(LINUX26_DIR)/.config
	$(SED) '/CONFIG_OABI_COMPAT/d' $(LINUX26_DIR)/.config
	echo "# CONFIG_OABI_COMPAT is not set" >> $(LINUX26_DIR)/.config else

... and my configuration files look fine to me:
$ make saveconfig
(...)
$ cd local/logica_demo/
$ grep ABI *
linux-2.6.28.config:CONFIG_SLABINFO=y
linux-2.6.28.config:# CONFIG_AEABI is not set linux-2.6.28.config:# CONFIG_SECURITY_FILE_CAPABILITIES is not set logica_demo.config:BR2_ARM_OABI=y logica_demo.config:# BR2_ARM_EABI is not set logica_demo.config:BR2_GCC_TARGET_ABI="apcs-gnu"
uclibc-0.9.30.1.config:CONFIG_ARM_OABI=y
uclibc-0.9.30.1.config:# CONFIG_ARM_EABI is not set

... and the toolchain looks OK too : 
$ ls -l $( which arm-linux-gcc)
lrwxrwxrwx 1 damien damien 20 jui 29 14:58 /mnt/fractal/buildroot/buildroot-2009.05/build_arm/staging_dir/usr/bin/arm-linux-gcc -> arm-linux-uclibc-gcc

Do you see something else to check?

Best regards,
Damien



Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

[Buildroot] linux .config : override: reassigning to symbolAEABI

[Daniel Mack]

(please keep the mailing list copied - I'm not your personal supporter,
and others might be interested in that, too.)

On Wed, Jul 29, 2009 at 03:32:52PM +0200, Courousse, Damien wrote:
> > Building an OABI toolchain for an EABI kernel doesn't make sense, hence
> > the configs are brought in sync by buildroot.
> > target/linux/Makefile.in.advanced is where it happens, and it AEABI
> > should not be set in the kernel config unless BR2_ARM_EABI is set. Could
> > you double check?
> [Damien Courouss?] 
> 
> target/linux/Makefile.in.advanced contains the following: 
> $(LINUX26_DIR)/.configured: $(LINUX26_DIR)/.config
> 	$(SED) '/CONFIG_AEABI=y/d' $(LINUX26_DIR)/.config
> ifeq ($(BR2_ARM_EABI),y)
> 	echo "CONFIG_AEABI=y" >> $(LINUX26_DIR)/.config
> 	$(SED) '/CONFIG_OABI_COMPAT/d' $(LINUX26_DIR)/.config
> 	echo "# CONFIG_OABI_COMPAT is not set" >> $(LINUX26_DIR)/.config
> else

Could you check that this condition is met by adding an echo debug in
there?

> ... and my configuration files look fine to me:
> $ make saveconfig
> (...)
> $ cd local/logica_demo/
> $ grep ABI *
> linux-2.6.28.config:CONFIG_SLABINFO=y
> linux-2.6.28.config:# CONFIG_AEABI is not set
> linux-2.6.28.config:# CONFIG_SECURITY_FILE_CAPABILITIES is not set
> logica_demo.config:BR2_ARM_OABI=y
> logica_demo.config:# BR2_ARM_EABI is not set
> logica_demo.config:BR2_GCC_TARGET_ABI="apcs-gnu"
> uclibc-0.9.30.1.config:CONFIG_ARM_OABI=y
> uclibc-0.9.30.1.config:# CONFIG_ARM_EABI is not set

And what does your $(LINUX26_DIR)/.config look like afterwards? Is EABI
set in there? If so, we'll need to find out why. Could be that there
have been changes in the Linux kernel config which cause EABI to be set,
depending on other config flags, but I don't have a solution at hand
currently. Shouldn't be too hard to figure out, though.

Daniel

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Courousse, Damien]

> -----Message d'origine-----
> De?: Daniel Mack [mailto:daniel at caiaq.de]
> Envoy??: mercredi 29 juillet 2009 15:43
> ??: Courousse, Damien
> Cc?: buildroot at uclibc.org
> Objet?: Re: [Buildroot] linux .config : override: reassigning
> tosymbolAEABI
> 
> (please keep the mailing list copied - I'm not your personal supporter,
> and others might be interested in that, too.)
> 
> On Wed, Jul 29, 2009 at 03:32:52PM +0200, Courousse, Damien wrote:
> > > Building an OABI toolchain for an EABI kernel doesn't make sense,
> hence
> > > the configs are brought in sync by buildroot.
> > > target/linux/Makefile.in.advanced is where it happens, and it AEABI
> > > should not be set in the kernel config unless BR2_ARM_EABI is set.
> Could
> > > you double check?
> > [Damien Courouss?]
> >
> > target/linux/Makefile.in.advanced contains the following:
> > $(LINUX26_DIR)/.configured: $(LINUX26_DIR)/.config
> > 	$(SED) '/CONFIG_AEABI=y/d' $(LINUX26_DIR)/.config
> > ifeq ($(BR2_ARM_EABI),y)
> > 	echo "CONFIG_AEABI=y" >> $(LINUX26_DIR)/.config
> > 	$(SED) '/CONFIG_OABI_COMPAT/d' $(LINUX26_DIR)/.config
> > 	echo "# CONFIG_OABI_COMPAT is not set" >> $(LINUX26_DIR)/.config
> > else
> 
> Could you check that this condition is met by adding an echo debug in
> there?
[Damien Courouss?] 
Here are the results - EABI is not set: 
rm -rf /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/buildroot-config
mkdir -p /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo
cp -dpRf package/config/buildroot-config /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/buildroot-config
/mnt/fractal/buildroot/buildroot-2009.05/toolchain_build_arm/bin/sed -i -e '/CONFIG_AEABI=y/d' /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config
echo "-- TEST -- BR2_ARM_EABI is not SET"
-- TEST -- BR2_ARM_EABI is not SET
echo "# CONFIG_AEABI is not set" >> /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config
/mnt/fractal/buildroot/buildroot-2009.05/toolchain_build_arm/bin/sed -i -e '/CONFIG_IPV6=y/d' /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config

Maybe I was fooled but the message : "override: reassigning to symbol AEABI". I tought it means that the EABI symbol is set to "y", but it might only mean that the symbol was overridden, with no assumption about the final value!
And actually, in my case, as BR2_ARM_EABI is not set, the kernel config EABI symbol is overridden, but with the same value, i.e. it is not set.
However it does not explain the kernel panic.

> 
> > ... and my configuration files look fine to me:
> > $ make saveconfig
> > (...)
> > $ cd local/logica_demo/
> > $ grep ABI *
> > linux-2.6.28.config:CONFIG_SLABINFO=y
> > linux-2.6.28.config:# CONFIG_AEABI is not set
> > linux-2.6.28.config:# CONFIG_SECURITY_FILE_CAPABILITIES is not set
> > logica_demo.config:BR2_ARM_OABI=y
> > logica_demo.config:# BR2_ARM_EABI is not set
> > logica_demo.config:BR2_GCC_TARGET_ABI="apcs-gnu"
> > uclibc-0.9.30.1.config:CONFIG_ARM_OABI=y
> > uclibc-0.9.30.1.config:# CONFIG_ARM_EABI is not set
> 
> And what does your $(LINUX26_DIR)/.config look like afterwards? Is EABI
> set in there? 
[Damien Courouss?] 
Here are the contents of the $(LINUX26_DIR)/.config file - EABI is not present.
$ cd project_build_arm/logica_demo/linux
$ grep ABI .config
CONFIG_SLABINFO=y
# CONFIG_AEABI is not set
# CONFIG_SECURITY_FILE_CAPABILITIES is not set

Damien


Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Daniel Mack]

On Wed, Jul 29, 2009 at 03:59:18PM +0200, Courousse, Damien wrote:
> > Could you check that this condition is met by adding an echo debug in
> > there?
> [Damien Courouss?] 
> Here are the results - EABI is not set: 
> rm -rf /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/buildroot-config
> mkdir -p /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo
> cp -dpRf package/config/buildroot-config /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/buildroot-config
> /mnt/fractal/buildroot/buildroot-2009.05/toolchain_build_arm/bin/sed -i -e '/CONFIG_AEABI=y/d' /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config
> echo "-- TEST -- BR2_ARM_EABI is not SET"
> -- TEST -- BR2_ARM_EABI is not SET
> echo "# CONFIG_AEABI is not set" >> /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config
> /mnt/fractal/buildroot/buildroot-2009.05/toolchain_build_arm/bin/sed -i -e '/CONFIG_IPV6=y/d' /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/linux-2.6.28/.config
> 
> Maybe I was fooled but the message : "override: reassigning to symbol AEABI". I tought it means that the EABI symbol is set to "y", but it might only mean that the symbol was overridden, with no assumption about the final value!
> And actually, in my case, as BR2_ARM_EABI is not set, the kernel config EABI symbol is overridden, but with the same value, i.e. it is not set.
> However it does not explain the kernel panic.

Ok, fine. You're digging at the wrong end then :) Try setting
CONFIG_DEBUG_USER in your kernel config and pass user_debug=255 on
startup. That will output more details when user space binaries crash.

Daniel

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Courousse, Damien]

> -----Message d'origine-----
> De?: Daniel Mack [mailto:daniel at caiaq.de]
> Envoy??: mercredi 29 juillet 2009 16:26
> ??: Courousse, Damien
> Cc?: buildroot at uclibc.org
> Objet?: Re: [Buildroot] linux .config : override: reassigning
> tosymbolAEABI
...
> 
> Ok, fine. You're digging at the wrong end then :) Try setting
> CONFIG_DEBUG_USER in your kernel config and pass user_debug=255 on
> startup. That will output more details when user space binaries crash.
> 
> Daniel
> 
[Damien Courouss?] 
The user_debug option did not change anything :(
If I use the following kernel command line:  
root=/dev/nfs rw nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off user_debug=255

I get: 
IP-Config: Complete:
     device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
     host=debian_test, domain=, nis-domain=(none),
     bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
Looking up port of RPC 100003/2 on 10.84.34.52
Looking up port of RPC 100005/1 on 10.84.34.52
VFS: Mounted root (nfs filesystem).
Freeing init memory: 116K
[kernel hangs here]

And if I add init=/bin/sh at the end of the kernel command line (using busybox), I get this : 
Kernel command line: root=/dev/nfs rw nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off user_debug=255 init=/bin/sh

Looking up port of RPC 100003/2 on 10.84.34.52
Looking up port of RPC 100005/1 on 10.84.34.52
VFS: Mounted root (nfs filesystem).
Freeing init memory: 116K
Kernel panic - not syncing: Attempted to kill init!

... and then I have compiled the kernel with initramfs. There I get something working: 
IP-Config: Complete:
     device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
     host=debian_test, domain=, nis-domain=(none),
     bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
Freeing init memory: 808K
Initializing random number generator... done.
Starting network...
ifup: interface lo already configured
Starting NFS statd: touch: /var/lock/subsys/nfslock: No such file or directory
done
Starting NFS services: done
Starting NFS daemon: done
Starting NFS mountd: Cannot register service: RPC: Unable to receive; errno = Connection refused
done
touch: /var/lock/subsys/nfs: No such file or directory

Welcome to Buildroot
login: root
# ls
# ls /
bin      dev      init     mnt      root     tmp
boot     etc      lib      opt      sbin     usr
config   home     linuxrc  proc     sys      var
# mount
rootfs on / type rootfs (rw)
proc on /proc type proc (rw)
devpts on /dev/pts type devpts (rw,gid=5,mode=620)
tmpfs on /tmp type tmpfs (rw)
sysfs on /sys type sysfs (rw)



But as you can see nfs root is not mounted (this might be because of a remount of / in inittab), whereas without initramfs, the nfs root is correctly mounted.

What puzzles me is that I though it would be possible to use a nfs root without initramfs. I will continue digging into this...

Damien


Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Daniel Mack]

On Wed, Jul 29, 2009 at 06:07:33PM +0200, Courousse, Damien wrote:
> > Ok, fine. You're digging at the wrong end then :) Try setting
> > CONFIG_DEBUG_USER in your kernel config and pass user_debug=255 on
> > startup. That will output more details when user space binaries crash.
> > 
> > Daniel
> > 
> [Damien Courouss?] 
> The user_debug option did not change anything :(

It only does if you set CONFIG_DEBUG_USER in your kernel config. You
will need to change it in the config you point your BR2 to.

> And if I add init=/bin/sh at the end of the kernel command line (using busybox), I get this : 
> Kernel command line: root=/dev/nfs rw nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off user_debug=255 init=/bin/sh
> 
> Looking up port of RPC 100003/2 on 10.84.34.52
> Looking up port of RPC 100005/1 on 10.84.34.52
> VFS: Mounted root (nfs filesystem).
> Freeing init memory: 116K
> Kernel panic - not syncing: Attempted to kill init!

With CONFIG_DEBUG_USER=y, you should see more useful output at this
stage.

> ... and then I have compiled the kernel with initramfs. There I get something working: 
> IP-Config: Complete:
>      device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
>      host=debian_test, domain=, nis-domain=(none),
>      bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
> Freeing init memory: 808K
> Initializing random number generator... done.
> Starting network...
> ifup: interface lo already configured
> Starting NFS statd: touch: /var/lock/subsys/nfslock: No such file or directory
> done
> Starting NFS services: done
> Starting NFS daemon: done
> Starting NFS mountd: Cannot register service: RPC: Unable to receive; errno = Connection refused
> done
> touch: /var/lock/subsys/nfs: No such file or directory
> 
> Welcome to Buildroot
> login: root
> # ls
> # ls /
> bin      dev      init     mnt      root     tmp
> boot     etc      lib      opt      sbin     usr
> config   home     linuxrc  proc     sys      var
> # mount
> rootfs on / type rootfs (rw)
> proc on /proc type proc (rw)
> devpts on /dev/pts type devpts (rw,gid=5,mode=620)
> tmpfs on /tmp type tmpfs (rw)
> sysfs on /sys type sysfs (rw)
> 
> 
> 
> But as you can see nfs root is not mounted (this might be because of a remount of / in inittab), whereas without initramfs, the nfs root is correctly mounted.
> 
> What puzzles me is that I though it would be possible to use a nfs root without initramfs. I will continue digging into this...

It is possible, of course. Once you have more debug output, you might
see where the problem is.

Daniel

[Buildroot] linux .config : override: reassigning to symbol AEABI

[Peter Korsgaard]

>>>>> "Damien" == Courousse, Damien <damien.courousse@logica.com> writes:

 Damien> The kernel loading gets stuck after mounting the NFS root : 
 Damien> ================================================================
 Damien> ....
 Damien> IP-Config: Complete:
 Damien>      device=eth0, addr=10.84.34.8, mask=255.255.255.192, gw=10.84.34.1,
 Damien>      host=debian_test, domain=, nis-domain=(none),
 Damien>      bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
 Damien> Looking up port of RPC 100003/2 on 10.84.34.52
 Damien> Looking up port of RPC 100005/1 on 10.84.34.52
 Damien> VFS: Mounted root (nfs filesystem).
 Damien> Freeing init memory: 116K
 Damien> Kernel panic - not syncing: Attempted to kill init!

Do you use softfloat or fpu emulation in the kernel? (check
BR2_SOFT_FLOAT).

-- 
Bye, Peter Korsgaard

[Buildroot] linux .config : override: reassigning to symbol AEABI

[Courousse, Damien]

Hello all,

This morning I did a complete compilation from a fresh install. I get something working with OABI support (did not test EABI) but using an initramfs only. Once the kernel has booted the nfs mounts are working normally.
I think I will continue like this because it's OK with my needs.

It looks like there are some kind of cache effects in the buildroot configuration files  when the config files have modified several times, even if make flush or even make clean are executed. Did someone of you had such problem? In this case, what is the better option: starting from a fresh install or something else?

Concerning my first problem I can do further tests if you want so that we  get a better idea of what is going on; I did some tests without being able to get a better idea of the problem. See below.


> -----Message d'origine-----
> De?: Peter Korsgaard [mailto:jacmet at gmail.com] De la part de Peter
> Korsgaard
> Envoy??: mercredi 29 juillet 2009 21:00
> ??: Courousse, Damien
> Cc?: buildroot at busybox.net
> Objet?: Re: [Buildroot] linux .config : override: reassigning to symbol
> AEABI
> 
> >>>>> "Damien" == Courousse, Damien <damien.courousse@logica.com> writes:
> 
>  Damien> The kernel loading gets stuck after mounting the NFS root :
>  Damien> ================================================================
>  Damien> ....
>  Damien> IP-Config: Complete:
>  Damien>      device=eth0, addr=10.84.34.8, mask=255.255.255.192,
> gw=10.84.34.1,
>  Damien>      host=debian_test, domain=, nis-domain=(none),
>  Damien>      bootserver=10.84.34.52, rootserver=10.84.34.52, rootpath=
>  Damien> Looking up port of RPC 100003/2 on 10.84.34.52
>  Damien> Looking up port of RPC 100005/1 on 10.84.34.52
>  Damien> VFS: Mounted root (nfs filesystem).
>  Damien> Freeing init memory: 116K
>  Damien> Kernel panic - not syncing: Attempted to kill init!
> 
> Do you use softfloat or fpu emulation in the kernel? (check
> BR2_SOFT_FLOAT).
> 
> --
[Damien Courouss?] 
Hello Peter, 
Thanks for helping.

My buildroot is configured with softfloat: 
$ grep SOFT .config
BR2_SOFT_FLOAT=y
grep FPU .config
# BR2_FPU_SUFFIX is not set

And the kernel config : 
.../linux$ grep FP .config
CONFIG_FPE_NWFPE=y
# CONFIG_FPE_NWFPE_XP is not set
# CONFIG_FPE_FASTFPE is not set
# CONFIG_VFP is not set


Mark : 
I have activated the DEBUG options in the buildroot and in the kernel config files. However, I do not get more debugging messages; I did append "user_debug=255 or 31" at the end of the kernel command line. Results are identical as what I described on Wednesday : 
* if I pass nfs_root options + init=/bin/sh to the kernel command line, I get a kernel panic,
* if I pass nothing more that the nfs_root options to the kernel command line, the kernel hangs after freeing the init memory.

Please find attached my buildroot, kernel and uclibc config files for a complete overview of my config

Thanks all for your kind help,
Damien


Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: uclibc-0.9.30.1.config
Type: application/octet-stream
Size: 6039 bytes
Desc: uclibc-0.9.30.1.config
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20090731/1e097d68/attachment-0004.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: buildroot-2009.05.config
Type: application/octet-stream
Size: 22371 bytes
Desc: buildroot-2009.05.config
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20090731/1e097d68/attachment-0005.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: busybox-1.14.1.config
Type: application/octet-stream
Size: 23058 bytes
Desc: busybox-1.14.1.config
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20090731/1e097d68/attachment-0006.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: linux-2.6.28.config
Type: application/octet-stream
Size: 30670 bytes
Desc: linux-2.6.28.config
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20090731/1e097d68/attachment-0007.obj>

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Thomas Petazzoni]

Hello Damien,

Le Wed, 29 Jul 2009 18:07:33 +0200,
"Courousse, Damien" <damien.courousse@logica.com> a ?crit :

> root=/dev/nfs rw
> nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root
> ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off

The /mnt/fractal/buildroot/buildroot-2009.05/project_build_arm/logica_demo/root
directory cannot be used to boot over NFS, because the /dev directory
doesn't contain any device file (Buildroot can create them, since it
requires root priviledges).

To boot over NFS, you have to do the following things:

 * Enable the generation of a tarball in the Buildroot configuration /
   Target filesystem options

 * Uncompress the tarball somewhere in your host, while being root (or
   using sudo), so that tar is allowed to create device files

 * Mount the location where you uncompressed the tarball over NFS

Sincerly,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers and embedded Linux development,
consulting, training and support.
http://free-electrons.com

[Buildroot] linux .config : override: reassigning tosymbolAEABI

[Courousse, Damien]

> -----Message d'origine-----
> De?: buildroot-bounces at busybox.net [mailto:buildroot-bounces at busybox.net]
> De la part de Thomas Petazzoni
> Envoy??: vendredi 31 juillet 2009 12:20
> ??: buildroot at busybox.net
> Objet?: Re: [Buildroot] linux .config : override: reassigning
> tosymbolAEABI
> 
> Hello Damien,
> 
> Le Wed, 29 Jul 2009 18:07:33 +0200,
> "Courousse, Damien" <damien.courousse@logica.com> a ?crit :
> 
> > root=/dev/nfs rw
> > nfsroot=10.84.34.52:/mnt/fractal/buildroot/buildroot-
> 2009.05/project_build_arm/logica_demo/root
> >
> ip=10.84.34.8:10.84.34.52:10.84.34.1:255.255.255.192:debian_test:eth0:off
> 
> The /mnt/fractal/buildroot/buildroot-
> 2009.05/project_build_arm/logica_demo/root
> directory cannot be used to boot over NFS, because the /dev directory
> doesn't contain any device file (Buildroot can create them, since it
> requires root priviledges).
> 
> To boot over NFS, you have to do the following things:
> 
>  * Enable the generation of a tarball in the Buildroot configuration /
>    Target filesystem options
> 
>  * Uncompress the tarball somewhere in your host, while being root (or
>    using sudo), so that tar is allowed to create device files
> 
>  * Mount the location where you uncompressed the tarball over NFS
> 
> Sincerly,
> 
> Thomas
> --
[Damien Courouss?] 

Hello Thomas!

I completely forgot to look at the side of the /dev files!
You spotted the right problem, thanks! Everything is OK now.

Thanks again to all of you for your help.

Kind regards,
Damien

Please help Logica to respect the environment by not printing this email  / Pour contribuer comme Logica au respect de l'environnement, merci de ne pas imprimer ce mail /  Bitte drucken Sie diese Nachricht nicht aus und helfen Sie so Logica dabei, die Umwelt zu sch?tzen /  Por favor ajude a Logica a respeitar o ambiente nao imprimindo este correio electronico.



This e-mail and any attachment is for authorised use by the intended recipient(s) only. It may contain proprietary material, confidential information and/or be subject to legal privilege. It should not be copied, disclosed to, retained or used by, any other party. If you are not an intended recipient then please promptly delete this e-mail and any attachment and all copies and inform the sender. Thank you.