[Buildroot] r21760 and device_table.txt problems

[Buildroot] r21760 and device_table.txt problems

[Christopher Taylor]

First, thank you all for your efforts.

I'm using r21760 and it appears that the device nodes aren't getting
populated.  I realize that the root filesystem is now created in
buildroot/project_build_ARCH/PROJECT/root but I don't see the device
nodes specified by the device_table.txt

I'm assume I'm missing a step that tells buildroot to read the device
table while it creates the root file system referenced above, but I
didn't find it via make menuconfig.

Any help would be appreciated!

Respectfully,
Chris

[Buildroot] r21760 and device_table.txt problems

[Hamish Moffatt]

On Fri, Apr 18, 2008 at 07:47:34PM -0400, Christopher Taylor wrote:
> First, thank you all for your efforts.
> 
> I'm using r21760 and it appears that the device nodes aren't getting
> populated.  I realize that the root filesystem is now created in
> buildroot/project_build_ARCH/PROJECT/root but I don't see the device
> nodes specified by the device_table.txt
> 
> I'm assume I'm missing a step that tells buildroot to read the device
> table while it creates the root file system referenced above, but I
> didn't find it via make menuconfig.

Once the build finishes the device nodes will look like regular files.
This is because they are created within a fakeroot session. They can't
be real device nodes because that would require root priviledges at
build-time.

Hamish
-- 
Hamish Moffatt VK3SB <hamish@debian.org> <hamish@cloud.net.au>

[Buildroot] Buildroot for AT91RM9200EK?

[Pink Boy]

Hi,

Hopefully this isn't 

I've been trying to figure out how to build a tool chain and linux
kernel for the Atmel AT91RM9200EK dev kit using build root.  

I tried setting the options under menuconfig for the board and after
installing a few missing utilities (Ubuntu natch) I get the following
error below.

make[1]: Leaving directory `/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1'
cp -pf /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage /home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz
cp: cannot stat `/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage': No such file or directory
make: *** [/home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz] Error 1
tu at Matt-LinuxDev:~/workspace/arm920_tool/buildroot$ 

It looks like something is eithe rmissing or didn't get built.

I'm not sure what I'm doing wrong at this point.  I'm wondering
is there is either an option I'm not setting, the current source
is broken and I need to use an older one, or there is some patch 
I need.

PS: Total Noob.

Mr Foo


----
  MODPOST vmlinux
WARNING: arch/arm/mach-at91/built-in.o(.data+0x75c): Section mismatch: reference to .init.text:at91rm9200_timer_init (between 'at91rm9200_timer' and 'at91rm9200_timer_irq')
  OBJCOPY arch/arm/boot/Image
  Kernel: arch/arm/boot/Image is ready
  AS      arch/arm/boot/compressed/head.o
  GZIP    arch/arm/boot/compressed/piggy.gz
  AS      arch/arm/boot/compressed/piggy.o
  CC      arch/arm/boot/compressed/misc.o
In file included from arch/arm/boot/compressed/misc.c:30:
include/asm/arch/uncompress.h:35: warning: conflicting types for built-in function 'putc'
  AS      arch/arm/boot/compressed/head-at91rm9200.o
  LD      arch/arm/boot/compressed/vmlinux
  OBJCOPY arch/arm/boot/zImage
  Kernel: arch/arm/boot/zImage is ready
  UIMAGE  arch/arm/boot/uImage
/usr/bin/mkimage [OPTIONS]

 Options:
 -f <MD5 name>       Specify an input MD5 file. MD5s must be in jigdo's
                     pseudo-base64 format
 -j <jigdo name>     Specify the input jigdo file
 -t <template name>  Specify the input template file
 -m <item=path>      Map <item> to <path> to find the files in the mirror
 -M <missing name>   Rather than try to build the image, just check that
                     all the needed files are available. If any are missing,
                     list them in this file.
 -v                  Make the output logging more verbose
 -l <logfile>        Specify a logfile to append to.
                     If not specified, will log to stderr
 -o <outfile>        Specify a file to write the ISO image to.
                     If not specified, will write to stdout
 -q                  Quick mode. Don't check MD5sums. Dangerous!
 -s <bytenum>        Start byte number; will start at 0 if not specified
 -e <bytenum>        End byte number; will end at EOF if not specified
 -z                  Don't attempt to rebuild the image; simply print its
                     size in bytes
  Image arch/arm/boot/uImage is ready
make[1]: Leaving directory `/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1'
cp -pf /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage /home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz
cp: cannot stat `/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage': No such file or directory
make: *** [/home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz] Error 1
tu at Matt-LinuxDev:~/workspace/arm920_tool/buildroot$ 

[Buildroot] Buildroot for AT91RM9200EK?

[Pink Boy]

Hi,

One more thing, the command that seems to bomb,


cp -pf /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage
/home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz

The is no uImage in the source folder, there is however a zImage.

???

Mr Foo

[Buildroot] Buildroot for AT91RM9200EK?

[Arun Reddy]

Hi Pink Boy,

While I am uncertain why uImage is not appearing in the source folder,

*"I'm wondering
is there is either an option I'm not setting, the current source
is broken and I need to use an older one, or there is some patch
I need."*

to solve the problem of retrieving the older source, cannot you simply
modify the .mk file and manually change the source file that is currently
listed to an older, available one?

Arun

On Sat, Apr 19, 2008 at 7:42 PM, Pink Boy <dasfoo@yahoo.com> wrote:

> Hi,
>
> One more thing, the command that seems to bomb,
>
>
> cp -pf
> /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage
>
> /home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz
>
> The is no uImage in the source folder, there is however a zImage.
>
> ???
>
> Mr Foo
>
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://busybox.net/lists/buildroot/attachments/20080419/ed74fbcc/attachment.htm 

[Buildroot] Buildroot for AT91RM9200EK?

[Pink Boy]

Found one issue,

The version of mkimage that I had installed was from the jiget
package not the one that comes with UBoot.  I uninstalled
jiget and mkImage and then tried getting menu config to build 
UBoot and hopefully the mkImage utility, but I get the following,

-bris--
u-boot-1.2.0-atmel/lib_nios2/cache.S
u-boot-1.2.0-atmel/lib_nios2/board.c
u-boot-1.2.0-atmel/.patched.u-boot-1.2.0-atmel-patch.tar.bz2
u-boot-1.2.0-atmel/CREDITS
u-boot-1.2.0-atmel/blackfin_config.mk
touch /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/.unpacked
wget --passive-ftp -nd  -P /home/tu/workspace/arm920_tool/buildroot/dl ftp://www.at91.com/pub/buildroot//u-boot-1.2.0-atmel-patch.tar.bz2
--16:00:37--  ftp://www.at91.com/pub/buildroot//u-boot-1.2.0-atmel-patch.tar.bz2
           => `/home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2'
Resolving www.at91.com... 194.2.24.29
Connecting to www.at91.com|194.2.24.29|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done.    ==> PWD ... done.
==> TYPE I ... done.  ==> CWD /pub/buildroot/ ... done.
==> PASV ... done.    ==> RETR u-boot-1.2.0-atmel-patch.tar.bz2 ... done.
Length: 242 (unauthoritative)

100%[=====================================================================================>] 242           --.--K/s             

16:00:39 (9.53 MB/s) - `/home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2' saved [242]

mkdir -p /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches
bzcat /home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2 | tar -C /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches -xvf -
u-boot-1.2.0-atmel-patch/README.txt
u-boot-1.2.0-atmel-patch/README.txt~
touch /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches/.unpacked
toolchain/patch-kernel.sh /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches \*.patch
touch /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/.patched.u-boot-1.2.0-atmel-patch.tar.bz2
cp target/device/Atmel/at91rm9200ek/u-boot/at91rm9200ek.h /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/include/configs/.
cp: cannot stat `target/device/Atmel/at91rm9200ek/u-boot/at91rm9200ek.h': No such file or directory
make: *** [/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-1.2.0-atmel/.configured] Error 1
tu at Matt-LinuxDev:~/workspace/arm920_tool/buildroot$ 

Looking there seems there is no at91rm9200ek.h file


I did try and build the tool chain and UBoot for the Atmel AT91rm9260
dev board and it built without an error.

I think this means that build root support for the Atmel AT91rm9200ek
dev board is currently broken

I'm also thinking that Linux support for the Atmel AT91rm9200
microprocessor is spotty at best and maybe I should switch
parts?  (I feel after banging on this that without an experience
with a working tool chain built with buildroot, there is realy
no way I can fix this)

Mr Foo

-----




--- On Sat, 4/19/08, Pink Boy <dasfoo@yahoo.com> wrote:

> From: Pink Boy <dasfoo@yahoo.com>
> Subject: Re: [Buildroot] Buildroot for AT91RM9200EK?
> To: buildroot at uclibc.org
> Date: Saturday, April 19, 2008, 7:42 PM
> Hi,
> 
> One more thing, the command that seems to bomb,
> 
> 
> cp -pf
> /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage
> /home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz
> 
> The is no uImage in the source folder, there is however a
> zImage.
> 
> ???
> 
> Mr Foo
> 
> _______________________________________________
> buildroot mailing list
> buildroot at uclibc.org
> http://busybox.net/mailman/listinfo/buildroot

[Buildroot] Buildroot for AT91RM9200EK?

[Christopher Taylor]

You're using the wrong mkimage.  Beleive it or not, there are actually
two different binaries with the same name.... a pain in the rear, I
know.  The one from jiget is _NOT_ the one you need.

I only know this because I had the same problem ;)

HTH!
-Chris

On Wed, Apr 23, 2008 at 8:07 PM, Pink Boy <dasfoo@yahoo.com> wrote:
> Found one issue,
>
>  The version of mkimage that I had installed was from the jiget
>  package not the one that comes with UBoot.  I uninstalled
>  jiget and mkImage and then tried getting menu config to build
>  UBoot and hopefully the mkImage utility, but I get the following,
>
>  -bris--
>  u-boot-1.2.0-atmel/lib_nios2/cache.S
>  u-boot-1.2.0-atmel/lib_nios2/board.c
>  u-boot-1.2.0-atmel/.patched.u-boot-1.2.0-atmel-patch.tar.bz2
>  u-boot-1.2.0-atmel/CREDITS
>  u-boot-1.2.0-atmel/blackfin_config.mk
>  touch /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/.unpacked
>  wget --passive-ftp -nd  -P /home/tu/workspace/arm920_tool/buildroot/dl ftp://www.at91.com/pub/buildroot//u-boot-1.2.0-atmel-patch.tar.bz2
>  --16:00:37--  ftp://www.at91.com/pub/buildroot//u-boot-1.2.0-atmel-patch.tar.bz2
>            => `/home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2'
>  Resolving www.at91.com... 194.2.24.29
>  Connecting to www.at91.com|194.2.24.29|:21... connected.
>  Logging in as anonymous ... Logged in!
>  ==> SYST ... done.    ==> PWD ... done.
>  ==> TYPE I ... done.  ==> CWD /pub/buildroot/ ... done.
>  ==> PASV ... done.    ==> RETR u-boot-1.2.0-atmel-patch.tar.bz2 ... done.
>  Length: 242 (unauthoritative)
>
>  100%[=====================================================================================>] 242           --.--K/s
>
>  16:00:39 (9.53 MB/s) - `/home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2' saved [242]
>
>  mkdir -p /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches
>  bzcat /home/tu/workspace/arm920_tool/buildroot/dl/u-boot-1.2.0-atmel-patch.tar.bz2 | tar -C /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches -xvf -
>  u-boot-1.2.0-atmel-patch/README.txt
>  u-boot-1.2.0-atmel-patch/README.txt~
>  touch /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches/.unpacked
>  toolchain/patch-kernel.sh /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-patches \*.patch
>  touch /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/.patched.u-boot-1.2.0-atmel-patch.tar.bz2
>  cp target/device/Atmel/at91rm9200ek/u-boot/at91rm9200ek.h /home/tu/workspace/arm920_tool/buildroot/build_arm/u-boot-1.2.0-atmel/include/configs/.
>  cp: cannot stat `target/device/Atmel/at91rm9200ek/u-boot/at91rm9200ek.h': No such file or directory
>  make: *** [/home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/u-boot-1.2.0-atmel/.configured] Error 1
>
> tu at Matt-LinuxDev:~/workspace/arm920_tool/buildroot$
>
>  Looking there seems there is no at91rm9200ek.h file
>
>
>  I did try and build the tool chain and UBoot for the Atmel AT91rm9260
>  dev board and it built without an error.
>
>  I think this means that build root support for the Atmel AT91rm9200ek
>  dev board is currently broken
>
>  I'm also thinking that Linux support for the Atmel AT91rm9200
>  microprocessor is spotty at best and maybe I should switch
>  parts?  (I feel after banging on this that without an experience
>  with a working tool chain built with buildroot, there is realy
>  no way I can fix this)
>
>  Mr Foo
>
>  -----
>
>
>
>
>  --- On Sat, 4/19/08, Pink Boy <dasfoo@yahoo.com> wrote:
>
>  > From: Pink Boy <dasfoo@yahoo.com>
>  > Subject: Re: [Buildroot] Buildroot for AT91RM9200EK?
>  > To: buildroot at uclibc.org
>  > Date: Saturday, April 19, 2008, 7:42 PM
>
>
> > Hi,
>  >
>  > One more thing, the command that seems to bomb,
>  >
>  >
>  > cp -pf
>  > /home/tu/workspace/arm920_tool/buildroot/project_build_arm/uclibc/linux-2.6.22.1/arch/arm/boot/uImage
>  > /home/tu/workspace/arm920_tool/buildroot/binaries/uclibc/at91rm9200ek-linux-2.6.22.1-20080420.gz
>  >
>  > The is no uImage in the source folder, there is however a
>  > zImage.
>  >
>  > ???
>  >
>  > Mr Foo
>  >
>  > _______________________________________________
>  > buildroot mailing list
>  > buildroot at uclibc.org
>  > http://busybox.net/mailman/listinfo/buildroot
>  _______________________________________________
>  buildroot mailing list
>  buildroot at uclibc.org
>  http://busybox.net/mailman/listinfo/buildroot
>

[Buildroot] Buildroot for AT91RM9200EK?

[Ulf Samuelsson]

> Found one issue,
[snip]
> 
> I did try and build the tool chain and UBoot for the Atmel AT91rm9260
> dev board and it built without an error.
> 
> I think this means that build root support for the Atmel AT91rm9200ek
> dev board is currently broken
> 

There is support in buildroot for AT91RM9200EK with dataflash.
You have to do "make at91rm9200dfc_defconfig"

Best Regards
Ulf Samuelsson

[Buildroot] r21760 and device_table.txt problems

[Christopher Taylor]

Hamish,

Yes, that makes sense. Unfortunately, I'm not even seeing regular files.
Your response leads me to conclude that while the device nodes are not
currently present in the fs tree (for the reasons you specified), they
_will_ be created during boot.

In a desktop environment this is usually handled by devfs or udev
these days.  How is it usually done within buildroot? I don't remember
enabling devfs or udev in buildroot or the kernel, but I'll look on
Monday.  If it's a static listings like before devfs, how should I go
about creating the device nodes for my target fs ... just mknod as
root before I make an image?

Would simply running make as root fix this and create the device nodes
per the deivce_table?

Respectfully,
Chris


>  Once the build finishes the device nodes will look like regular files.
>  This is because they are created within a fakeroot session. They can't
>  be real device nodes because that would require root priviledges at
>  build-time.
>
>  Hamish
>  --
>  Hamish Moffatt VK3SB <hamish@debian.org> <hamish@cloud.net.au>
>  _______________________________________________

[Buildroot] r21760 and device_table.txt problems

[Hamish Moffatt]

On Sun, Apr 20, 2008 at 01:17:34AM -0400, Christopher Taylor wrote:
> Yes, that makes sense. Unfortunately, I'm not even seeing regular files.
> Your response leads me to conclude that while the device nodes are not
> currently present in the fs tree (for the reasons you specified), they
> _will_ be created during boot.

Can you post the output of the final stages of make? You should see it
running genext2fs, mkfs.jffs2 or something similar for the output format
you are using, within a fakeroot environment which also includes running
the makedevs program.

In my case, I see a bunch of empty regular files left in the
project_build_$arch/$project/root/dev directory after the build, which
is what I expected.

> In a desktop environment this is usually handled by devfs or udev
> these days.  How is it usually done within buildroot? I don't remember
> enabling devfs or udev in buildroot or the kernel, but I'll look on
> Monday.  If it's a static listings like before devfs, how should I go
> about creating the device nodes for my target fs ... just mknod as
> root before I make an image?

The device table is supposed to be used to populate the static /dev in
the root image. If you want dynamic /dev I think you need to write this
into your rc script yourself. Probably mount tmpfs on /dev and run mdev
or udev to populate it.

> Would simply running make as root fix this and create the device nodes
> per the deivce_table?

I suspect not. It will depend on what you see from the build.


Hamish
-- 
Hamish Moffatt VK3SB <hamish@debian.org> <hamish@cloud.net.au>