IRC log for #elinux on 20080622

02:15.06*** join/#elinux Dr_Who (n=tgall@gentoo/developer/dr-who)
05:54.28*** join/#elinux linac (n=lin@117.24.2.98)
06:54.45*** join/#elinux emil|k (i=emil@e179235235.adsl.alicedsl.de)
08:11.43*** join/#elinux lyakh (n=lyakh@p57BD2536.dip0.t-ipconnect.de)
09:50.08*** join/#elinux emil|k (i=emil@e179235235.adsl.alicedsl.de)
11:32.26*** join/#elinux criccaSalentina (n=10408F82@d83-184-183-167.cust.tele2.it)
11:32.34*** part/#elinux criccaSalentina (n=10408F82@d83-184-183-167.cust.tele2.it)
12:10.03*** join/#elinux GPSFan (n=kenm@12.10.255.246)
16:21.57*** join/#elinux dijenerate (n=dijenera@72.22.128.194)
18:03.04*** join/#elinux Redhatter (n=vk4fsjl@2001:388:f000:0:0:0:0:279)
19:20.16*** join/#elinux pleemans (n=peter@d51A5E76A.access.telenet.be)
19:50.22*** join/#elinux GPSFan (n=kenm@12.10.255.246)
23:23.08*** join/#elinux ashes (n=ashes@modemcable123.78-80-70.mc.videotron.ca)
23:26.57*** join/#elinux gth (n=geir@195.0.160.170)
23:33.02gthHello, I have an odd problem with a PXA255 based board. I created a small system from scratch some years ago, with a linux 2.4 kernel/uclibc/busybox. Now I have been told to port the system to a 2.6 kernel, and update gcc, etc. I have used buildroot to create a new toolchain, and ported linux 2.6.25.x to the system. Everything looks ok, and the kernel boots and runs the old userspace just fine.
23:33.43gthNext is getting a new userspace. I am using buildroot to build a complete user space, but without luck
23:34.21gthThe kernel halts with: "Failed to execute /bin/sh.  Attempting defaults..."
23:35.06gthI have checked the root file system and everything looks ok
23:35.35gthI have tried every combination of softfloat/hardware floting point
23:36.07gthI have checked that libgcc is in place (I stuggled with that one last time I did this)
23:36.51gthI am sure the kernel finds /bin/sh, but for some reason it fails to run
23:37.27gthAny ideas for what to look for?
23:37.30ali_asCheck the file info to see if it's been compiled for the right cpu/OS enviroment?
23:38.28gthBuildroot build both the toolchain (used for the kernel) and userspace in one go, so I would expect the toolchain to be ok ..
23:40.16gth#file busybox
23:40.17gthbusybox: ELF 32-bit LSB executable, ARM, version 1, statically linked, stripped
23:40.21gthlooks ok
23:40.51ali_asAnd it's a little endian chip?
23:41.17gthYes, the intel xscale architecture is LE
23:41.22gtharm-linux-objdump -f busybox
23:41.22gthbusybox:     file format elf32-littlearm
23:41.22gtharchitecture: arm, flags 0x00000102:
23:41.22gthEXEC_P, D_PAGED
23:41.22gthstart address 0x00008730
23:41.37ali_asIs there a symbolic link for sh?
23:42.40gthyes, it is in place
23:44.11gthAnd since bb is statically linked, all dependencies should be ok ...
23:44.52ali_asI think you need someone who isn't me, but last dumb question - bb has been built with sh support?
23:45.56gthchecks /bin/sh support
23:48.26gthhmm. bb is build with a shell called ash.
23:49.20gthFEATURE_SH_IS_ASH=y so look ok
23:51.10gthali_as: those questions are not dumb....
23:54.28ali_asIt's mounting the root filesystem ok?
23:54.59gthIt looks ok from the console messages
23:55.27gthand the kernel boots the old userspace ok
23:55.41gthwith the same kernel messages
23:56.12ali_asOk and you have /dev/console and /dev/null present?
23:57.04ali_asAparently these should have been generated by buildroot, but you may need to be root for that to be successful.  I'm flicking through some posts with the same error.
23:58.02gthyes, but I do get a: "Warning: unable to open an initial console." before the "Failed to execute /bin/sh." message
23:58.41gthconsole is char, major=5 minor=1
23:58.52gthand it is the same on the old userspace

Generated by irclog2html.pl Modified by Tim Riker to work with infobot.