On 02/10/2010 05:08 PM, Gerardo Exequiel Pozzi wrote:
On 02/10/2010 04:49 PM, Gerardo Exequiel Pozzi wrote:
On 02/10/2010 09:35 AM, Dieter Plaetinck wrote:
On Tue, 09 Feb 2010 18:58:14 -0300 Gerardo Exequiel Pozzi<vmlinuz386@yahoo.com.ar> wrote:
Good!, the only problem for doing automatic builds is that in _random_ cases squashfs image of "overlay" is created corrupted (from the same data) :( I experimented this many times, and the way to test if all is good is: or doing a boot, or mounting the squashfs image on read some files. I can remember that also tested if the problem persiste with mksquashfs -processors 1 but the problem still persist. that's odd. can you investigate? maybe this is a bug in the squashfs tools? in the worst case scenario i add a bit of code to the buildscripts that mounts the image and tests it.
I can remember an email from GerBra, which these errors [#1]. But this is unrelated to errors messages about ioctl. The bug is on mksqashfs. I will look how can report it to devs ;) The bad image is created with lots of \x00 just after the header.
On mksquashfs the error is detected, but the program always return 0, the only difference is you can see this message: Lseek on destination failed because Bad file descriptor or: Write failed because Bad file descriptor Write on destination failed because Bad file descriptor
mksquashfs work/overlay lolcat.sqfs -noappend
adding one file inside (about ~100k), I can not reproduce the error. (I tried with different block sizes but is the same) Applying this patch solves the issue :)
Opening a bug report in FS ;)
Done: http://bugs.archlinux.org/task/18277 -- Gerardo Exequiel Pozzi ( djgera ) http://www.djgera.com.ar KeyID: 0x1B8C330D Key fingerprint = 0CAA D5D4 CD85 4434 A219 76ED 39AB 221B 1B8C 330D