[uClibc]Re: buildroot problems on SuSE8.1

JeeBak Kim jbkim at cs.pdx.edu
Sat Nov 23 04:48:24 UTC 2002


* Tsuyoshi GEMMA (tgem at ecodacs.or.jp) [021122 20:26]:
[snip]
> I am afraid that block-size is not sufficient. When I ran genext2fs
> manually with
> block size parameter increased, I got
>
> ./genext2fs: too big file system
[snip]
> First, I made a rough estimate of file size needed for root_fs
> by du command like this.
>
> du -k /home/tgem/workbench/buildroot/build/root/* | gawk
> 'BEGIN{s=0}{s=s+$1}END{print s}'

or just: du -ks /home/tgem/workbench/buildroot/build/root   ;-)

> The results was 108975. Then, I proceeded to try.
>
> dd if=/dev/zero of=/tmp/root_fs bs=1024 count=108975
> /sbin/mke2fs /tmp/root_fs
> /sbin/e2fsck /tmp/root_fs
> su root
> mount -o loop /tmp/root_fs /mnt/
> cp -a /home/tgem/workbench/buildroot/build/root/* /mnt/
> sync
> cd /mnt/
> chroot ./
[snip]
> Compiling by gcc was succeeded and ldd command showed that executable was
> linked against uClibc.
> These trial suggests that other components are successfully built.
> However, it seems that there are something weird in genext2fs and/or
> buildroot/make/ext2root.mk IMHO. But, I have no idea to fix this problem.
>
> Any suggestion are appreciated.

This is a limitation of genext2fs, the 8 meg (or whatever) max.  This
shortcoming is mentioned in ext2root.mk iirc.  Your method is what I
usually do too... with the addition of making the /dev files with a
script/mknod that parses the the device_table.txt (don't remember exact
name) file in the buildroot/sources dir.



More information about the uClibc mailing list