Hello Dmitry,

Monday, September 24, 2007, 11:07:54 PM, you wrote:

[]

> Oh, yeah! It works.

  Great!

> Though it frightened me to death with a lots of
> following messages scrolling wildly through the screen while booting:
> ----------------------------------------------------------------------
> <5>[   91.270000] CLEANMARKER node found at 0x014bc000, not first node
> in block (0x01480000)
> <5>[   91.320000] CLEANMARKER node found at 0x014c4000, not first node
> in block (0x014c0000)
> <4>[   91.360000] Empty flash at 0x014c7fc4 ends at 0x014c8000

  This happens on other devices too, including ones w/o real flash
support (h4000). So, I assume this comes from mtdram devices, which
hosts ramdisk image. More investigation is of course needed.

> ...and so on
> ----------------------------------------------------------------------
> I've been patient and my patience was finally rewarded with Angstrom bootup
> logo and then working GPE. Amazing! I played with it a little - usb
> networking via cradle and bluetooth (hcitool scan) seem to work fine. 
> As a proof-of-concept it's simply brilliant, though in general it's
> not very attractive with the current set of applications - usually users
> want to see more than just gpe-terminal :)

  Well, that's what Angstrom x11 image offers, and that's by design.
For example, OPIE LiveRamdisk will offer more goodies in slightly less
space ;-). But the idea here is to be quick and easy demo for people
who otherwise can't bother themselves to do an installation per
current vague, inconsistent, and boring installation instructions.

  Of course, the right way to do that would be create adhoc "live"
image which would include mp3 player, and other catchy stuff, and
that's on my million-entry todo list, but I won't get to it anytime
soon myself...

> And this frightening messages (see above), which certainly caused by
> this line in fstab:
> /dev/mtdblock3  /               jffs2   defaults        1  1

> That's seem to be recurring problem for me every time I try fresh rootfs
> images - every time I need to change etc/fstab to reflect my real
> root device and filesystem (/dev/mmcblk0p1 and ext3 usually). I wonder
> if there's a generic way to avoid this problem. Detecting right root by 
> parsing
> kernel command line? Or maybe adopting RedHat's strategy with mounting
> filesystems by its labels? Or something else?

  For me, solution sounds trivial - there should be no hardcoded entry for root
in fstab, voila. That's what h4000 has and I never faced any probs
like that h2200's boring "hang during boot" issue. So, someone should
try that and RFC that on the devel list. If there's good reason for
the root entry being in fstab (heh, how fscking root would work
otherwise, right? ;-) ), fstab should refer to /dev/root, and that
should be created properly based on real root used.

> Dmitry "MAD" Artamonow



-- 
Best regards,
 Paul                            mailto:[EMAIL PROTECTED]


_______________________________________________
Angstrom-distro-users mailing list
[email protected]
http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/angstrom-distro-users

Reply via email to