ltib
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Ltib] Re: Startup problem


From: Stuart Hughes
Subject: Re: [Ltib] Re: Startup problem
Date: Thu, 14 Aug 2008 09:26:47 +0100

Hi Mark,

This would cause the problem.

Let me explain how the libraries normally get onto the target.  The idea
is that we (by default) use the pre-built libraries from the toolchain,
this ensures compatibility with anything built by it.

The spec file that does this is called 'base_libs.spec', I've attached
the latest I have.  These rely on the toolchain being used to be laid
out a certain (standard way).  It copes with many "standard"
glibc/uClibc toolchains and also CodeSourcery built multi-libbed
toolchains.

Try this latest base_libs.spec, it might fix your problem.  If not
you'll either need to get hold of a suitable toolchain, or modify
baselibs to recognise the layout you have in your toolchain.

Regards, Stuart

On Wed, 2008-08-13 at 22:21 -0700, Mr Junk wrote:
> Stuart,
> 
> I might be on to a clue.
> 
> There are hardly any libs in the /lib directory.
> 
> On lot's of Googling I'm seeing a pattern that ld and as a link ld- 
> linux.so*  seems to be a culprit for this type of behavior.
> 
> Here is my /lib dump:
> 
> address@hidden:~/work/ltib/rootfs$ l lib
> total 11M
> drwxr-xr-x  3 root root 4.0K 2008-08-12 22:27 .
> drwxr-xr-x 17 root root 4.0K 2008-08-12 22:27 ..
> -rwxr-xr-x  1 root root  63K 2008-08-05 00:52 libgcc_s.so
> -rwxr-xr-x  1 root root  63K 2008-08-05 00:52 libgcc_s.so.1
> -rwxr-xr-x  1 root root 3.6M 2008-08-05 00:52 libstdc++.so
> -rwxr-xr-x  1 root root 3.6M 2008-08-05 00:52 libstdc++.so.6
> -rwxr-xr-x  1 root root 3.6M 2008-08-05 00:52 libstdc++.so.6.0.8
> lrwxrwxrwx  1 root root   15 2008-08-11 21:50 libtermcap.so ->  
> libtermcap.so.2
> lrwxrwxrwx  1 root root   19 2008-08-11 21:50 libtermcap.so.2 ->  
> libtermcap.so.2.0.8
> -rwxr-xr-x  1 root root  36K 2008-08-11 19:20 libtermcap.so.2.0.8
> drwxr-xr-x  3 root root 4.0K 2008-08-12 22:07 modules
> 
> That doesn't seem like everything I would need to run properly.
> 
> I certainly don't see an ld lib in there or anywhere.
> 
> All the libs in /usr/lib seem to be for the packages that are added.
> 
> Is there something I didn't do when I ported this processor over that  
> isn't triggering the full build of the lib directory.
> 
> Best,
> ml
> 
> P.S. The kernel I'm using was built with LTIB using my custom  
> toolchain and a config and kernel source that came with the board.
> 
> On Aug 13, 2008, at 9:34 AM, Stuart Hughes wrote:
> 
> > Hi Mark,
> >
> > Are you using the same kernel when you try the other rootfs that is
> > okay?
> >
> > If so I suspect you're running into problems with the rootfs you've
> > built with LTIB because of cross compiler and/or cflags settings.  My
> > suspicion is that maybe you're running into floating point exception  
> > (or
> > some other thing with the math unit) that cause these weird problems.
> > I've had this kind of issue before.  It may or maynot be the case, but
> > try building the rootfs from scratch with soft-float and enable math
> > emulation in the kernel.  At least this would eliminate one theory.
> >
> > Regards, Stuart
> >
> > On Wed, 2008-08-13 at 09:02 -0700, Mr Junk wrote:
> >> It even happens with nfs mount of the rootfs.  Weird.
> >>
> >> Thanks,
> >> ml
> >>
> >> On Aug 13, 2008, at 1:55 AM, Stuart Hughes wrote:
> >>
> >>> Hi Mark,
> >>>
> >>> If it only happens with LTIB build rootfs, then maybe suspect either
> >>> the
> >>> version of mkfs.jffs2 that ltib uses, or the command line it is
> >>> using to
> >>> build the image.
> >>>
> >>> The version can be found from:
> >>>
> >>> /opt/ltib/usr/bin/rpm -q mtd-utils  (Savannah version)
> >>> /opt/freescale/ltib/usr/bin/rpm -q mtd-utils (FSL version)
> >>>
> >>> To figure out the command line, run (you'll get a lot of output so
> >>> script it):
> >>>
> >>> $ ./ltib -D -v
> >>>
> >>> You can alter the command line by editing bin/Ltibutils.pm, look for
> >>> mkfs.jffs2
> >>>
> >>> I'd be interested if you find anything.
> >>>
> >>> Regards, Stuart
> >>>
> >>> On Tue, 2008-08-12 at 14:13 -0700, Mr Junk wrote:
> >>>> Vadim,
> >>>>
> >>>> Thanks for the idea.
> >>>>
> >>>> I had already checked that to make sure, and [ was where it was
> >>>> supposed to be and operational.
> >>>>
> >>>> After that, just to make sure, I had LTIB rebuild the rootfs from
> >>>> scratch to make sure everything was virgin.
> >>>>
> >>>> No good news yet.
> >>>>
> >>>> I tried copying a rootfs from another dev board onto the sd card  
> >>>> and
> >>>> that one worked perfectly with my kernel and everything else.
> >>>>
> >>>> It seems this is only happening with the rootfs generated by LTIB.
> >>>>
> >>>> Never seen anything like this before.
> >>>>
> >>>> Still diggin.
> >>>> ml
> >>>>
> >>>> On Aug 12, 2008, at 10:14 AM, Vadim Lebedev wrote:
> >>>>
> >>>>> Mark,
> >>>>>
> >>>>> Please check that you have /usr/bin/[    installed
> >>>>>
> >>>>> Thanks
> >>>>> Vadim
> >>>>>
> >>>>> Mr Junk wrote:
> >>>>>> Stuart,
> >>>>>> It is vanilla ext3 and can be be read and written fine by both
> >>>>>> the dev board and the host system when moving files to it.
> >>>>>>
> >>>>>> Very weird that the kernel can call /sbin/init to get busy box
> >>>>>> working but then can't span the fs to get other stuff.
> >>>>>>
> >>>>>> It does find and run rcS, the failure is finding and running [.
> >>>>>>
> >>>>>> If I comment out the line using [ the script run to the next  
> >>>>>> level.
> >>>>>>
> >>>>>> Weird.
> >>>>>>
> >>>>>> Hoped for some silver bullet.  I'll keep digging.
> >>>>>>
> >>>>>> Thanks for the response with all you email load.
> >>>>>>
> >>>>>> I hope you had a great time in the Alps, very beautiful and
> >>>>>> wonderful place to be.
> >>>>>>
> >>>>>> I miss them.
> >>>>>>
> >>>>>> Best,
> >>>>>> ml
> >>>>>>
> >>>>>
> >>>>>
> >>>>>
> >>>>>>
> >>>>>> On Aug 12, 2008, at 1:29 AM, Stuart Hughes wrote:
> >>>>>>
> >>>>>>> Hi Mark,
> >>>>>>>
> >>>>>>> The only educated guess I can give is that there's some kind of
> >>>>>>> problem
> >>>>>>> with the SD card driver related to the filesystem.
> >>>>>>>
> >>>>>>> Regards, Stuart
> >>>>>>>
> >>>>>>>
> >>>>>>> On Tue, 2008-08-12 at 00:20 -0700, Mr Junk wrote:
> >>>>>>>> Hello,
> >>>>>>>>
> >>>>>>>> My kernel compiles and runs fine.
> >>>>>>>>
> >>>>>>>> My rootfs compiles and builds fine
> >>>>>>>>
> >>>>>>>> My kernel starts up, finds the rootfs and starts init (busybox)
> >>>>>>>> fine,
> >>>>>>>> but then these errors occur in rcS:
> >>>>>>>>
> >>>>>>>> It seems that the "test" [ is not being found in /usr/bin. . .I
> >>>>>>>> added
> >>>>>>>> a path statement in rcS with no change.
> >>>>>>>>
> >>>>>>>> aiting 5sec before mounting root device...
> >>>>>>>> mmc0: new SD card on SPI
> >>>>>>>> mmcblk0: mmc0:0000 SA01G 995328KiB
> >>>>>>>> mmcblk0: p1
> >>>>>>>> kjournald starting.  Commit interval 5 seconds
> >>>>>>>> EXT3 FS on mmcblk0p1, internal journal
> >>>>>>>> EXT3-fs: mounted filesystem with ordered data mode.
> >>>>>>>> VFS: Mounted root (ext3 filesystem).
> >>>>>>>> Freeing init memory: 104K
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 14: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 24: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 29: [: not found
> >>>>>>>> /etc/rc.d/rcS: /etc/rc.d/rcS: 37: [: not found
> >>>>>>>>
> >>>>>>>> Has anyone ever seen this before?
> >>>>>>>>
> >>>>>>>> Any clues?
> >>>>>>>>
> >>>>>>>> Thanks,
> >>>>>>>> ml
> >>>>>>>>
> >>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> LTIB home page: http://bitshrine.org
> >>>>>>
> >>>>>> Ltib mailing list
> >>>>>> address@hidden
> >>>>>> http://lists.nongnu.org/mailman/listinfo/ltib
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>
> >>>
> >>>
> >>
> >
> >
> 

Attachment: base_libs.spec
Description: Text document


reply via email to

[Prev in Thread] Current Thread [Next in Thread]