Ubuntu

iBook SE 366 no desktop, only cursor

Bug #23378 reported by Mtrettenbach on 2005-10-07

This bug report is a duplicate of: Bug #23426: Gnome wont start if date is incorrect.. Edit Remove

Affects		Status	Importance	Assigned to	Milestone
	Ubuntu	Invalid	Medium	Daniel Holbach

Bug Description

A search for previous bug reports on the issue I encountered revealed only 7225,
which doesn't appear the same.

I am trying Breezy (5.10) release candidate live for PPC, on an iBook SE (366
MHz, 320 MB RAM -- i.e., NewWorld, 1st generation iBook, native 800x600). The
iBook has YDL 4.0 installed now, and previously ran on YDL 3.0.1, both of which
have run without problems (and YDL 4.0 still does).

After I press "c" with the 5.10 RC for PPC live in the drive, it runs through
the screens for language, keyboard, resolution, etc. fine, reports a green "OK"
on all lines that scroll past during boot (except for connecting to the
Internet, but that's because there's no connection, so it *should* fail), gets
to the chime sequence and gives me a dark brown desktop with an arrow cursor --
but nothing more. The touchpad makes the cursor move, but there are no icons on
the brown background to click on.

Doing ctl-alt-del brings up the Ubuntu login screen. There, I have tried all
the "session" type choices, including a failsafe console. But no matter which
type session I choose there, when the timer hits zero, the screen returns to the
dark, nearly black brown, with no icons (nor any console, when that's chosen as
the session type). I don't know whether this is really a bug in the RC for the
live PPC version or something that I could fix myself by getting to a console
some other way, if I weren't such an idiot.

Revision history for this message

Daniel Holbach (dholbach) wrote on 2005-10-07:

Thanks for your report. Could this be fixed with either the comments of
* http://bugzilla.ubuntu.com/show_bug.cgi?id=15690 or
* http://bugzilla.ubuntu.com/show_bug.cgi?id=15700 ?

Revision history for this message

Mtrettenbach (mtrettenbach) wrote on 2005-10-07:

Thanks for the references to 15690 and 15700, Daniel. 15690 doesn't sound
applicable, as he reaches and gets past a log-in screen, and has only some
grayed-out items. 15700 sounds much more like it for the greatest part: not P.
Salazar Mora's case, as he first did get a log-in screen, but migscabral's and
Michael Dorman's sounds similar, at least after they logged in. (In my case, I
don't even get to the log-in screen until I do ctl-alt-del from the blank brown
screen; and another difference is that the start-up sound plays harmoniously,
unlike one comment to 15700). I will try tonight when I return home (I'm
writing from work now, w/o the iBook and CD handy). In particular I will try to
get to a console so as to try some of the suggestions in 15700. I'd tried
before (as noted) to get to a console, and also tried to choose a failsafe
terminal type session from the login, but w/o success. Is there a boot argument
I can use on the live CD to start in terminal mode at runlevel 3 instead of 5?
(I *will* check for readmes on the CD, and try to think of why my choice of
failsafe terminal at the session-type screen isn't working, but thought I'd ask,
too.)

I should add that the md5sum matched, and that it was the 20051005 RC. To be
more specific about where in the sequence it produces the brown screen: it asks
for language, location, and keyboard (English, US), goes through "preparing for
live session" to 65%, when it warns the screen may blank, which it does but then
returns to the progress bar, it lists 3 resolutions of 1024, 800, & 640 to
configure x-server x-org, I usually delete the 1024 but leave the other two
(although I've done it with only 800 & only 640 w/no difference) & "continue";
the progress bar continues to 100%, the screen changes to the Ubuntu logo with
the reflection as though off a water surface, it begins listing messages such as
"disc parameters ok," "syncing clock to ubuntulinux" (which is the one that
fails, as there's no connection), finishes other messages with "ok," produces
the "X" form of a cursor, then the round "wait" circle form of a cursor, then
the arrow cursor, turns the screen brown, plays the startup sound -- and then
goes no further. I have tried every (I think) combination of the option, alt,
apple, & control keys together with the function keys F1-F7 in order to try to
get a console in order to try to learn something from the logs, but never
succeeded in getting a console. At the outset, I have tried "live video=ofonly"
and have tried "live-powerpc" (no image being asterisked), but with no different
results. I hope this extra precision on the sequence may help. I will try to
get to a console tonight to follow up on the info in 15700. Thanks!

Thanks for the references to 15690 and 15700, Daniel.  15690 doesn't sound
applicable, as he reaches and gets past a log-in screen, and has only some
grayed-out items.  15700 sounds much more like it for the greatest part: not P.
Salazar Mora's case, as he first did get a log-in screen, but migscabral's and
Michael Dorman's sounds similar, at least after they logged in.  (In my case, I
don't even get to the log-in screen until I do ctl-alt-del from the blank brown
screen; and another difference is that the start-up sound plays harmoniously,
unlike one comment to 15700).  I will try tonight when I return home (I'm
writing from work now, w/o the iBook and CD handy). In particular I will try to
get to a console so as to try some of the suggestions in 15700.  I'd tried
before (as noted) to get to a console, and also tried to choose a failsafe
terminal type session from the login, but w/o success.  Is there a boot argument
I can use on the live CD to start in terminal mode at runlevel 3 instead of 5? 
(I *will* check for readmes on the CD, and try to think of why my choice of
failsafe terminal at the session-type screen isn't working, but thought I'd ask,
too.)

I should add that the md5sum matched, and that it was the 20051005 RC.  To be
more specific about where in the sequence it produces the brown screen: it asks
for language, location, and keyboard (English, US), goes through "preparing for
live session" to 65%, when it warns the screen may blank, which it does but then
returns to the progress bar, it lists 3 resolutions of 1024, 800, & 640 to
configure x-server x-org, I usually delete the 1024 but leave the other two
(although I've done it with only 800 & only 640 w/no difference) & "continue";
the progress bar continues to 100%, the screen changes to the Ubuntu logo with
the reflection as though off a water surface, it begins listing messages such as
"disc parameters  ok," "syncing clock to ubuntulinux" (which is the one that
fails, as there's no connection), finishes other messages with "ok," produces
the "X" form of a cursor, then the round "wait" circle form of a cursor, then
the arrow cursor, turns the screen brown, plays the startup sound -- and then
goes no further.  I have tried every (I think) combination of the option, alt,
apple, & control keys together with the function keys F1-F7 in order to try to
get a console in order to try to learn something from the logs, but never
succeeded in getting a console.  At the outset, I have tried "live video=ofonly"
and have tried "live-powerpc" (no image being asterisked), but with no different
results.  I hope this extra precision on the sequence may help.  I will try to
get to a console tonight to follow up on the info in 15700.  Thanks!

Revision history for this message

Mtrettenbach (mtrettenbach) wrote on 2005-10-10:

Download full text (7.1 KiB)

After reading reports 15700 and 12276, I did an strace as suggested there and
here's the result:

ubuntu@ubuntu:~$ strace -p $(pidof esd)
Process 17651 attached - interrupt to quit
select(6, [4 5], NULL, NULL, NULL
<unfinished ...>

and after nothing more had happened for three by-the-watch minutes, I did a
ctl-c and got

Process 17651 detached

Should I have let it run longer? A return to ^-alt-F7 still had the blank
screen, and nothing had changed on ctl-alt-F3 or F4 either. By the way, I was
wrong earlier in saying I couldn't get to a command prompt from the brown
screen. I just hadn't pressed the function key quite long enough. Sorry; my
mistake.

I also did a

gdb /usr/bin/esd $(pidof esd)

and got back

attaching to program: /usr/bin/esd, process 17598

(this being a different time & so different pid for esd), and then a string of
alternating messages:

reading symbols from [...] (no debugging symbols found) ...
loaded symbols for [..same as above...]

(i.e., where [...] were different, but the "reading" and "loaded" lines
alternated, and the "reading line" always had "no debugging symbols found" in
it. And when I did, while in gdb, a "bt," it returned the following:

#0 0x0fe3b504 in select () from /lib/libc.so.6
#1 0x100050ec in clean_exit ()
#2 0x10003aa8 in clean_exit ()
#3 0x0fd8c80c in __libc_init_first () from /lib/libc.so.6
#4 0x0fd8c80c in __libc_init_first () from /lib/libc.so.6
#5 0x0fd8c80c in __libc_init_first () from /lib/libc.so.6
previous frame inner to this frame (corrupt stack?)

As for the terminal screens, ctl-alt-F1 didn't show much interesting: just the
same lines as appear in brown under the "reflected" logo:

* Setting the system clock... [ ok ]
* Synchronizing clock to ntp.ubuntulinux.org...
Error : Temporary failure in name resolution
[fail]
* Initializing random number generator... [ ok ]
* Setting up X server socket directory... [ ok ]
* Setting up ICE socket directory... [ ok ]
Last login: Sun Feb 14 00:26:44 1984 on tty2
Linux ubuntu 2.6.12-9-powerpc #1 Tue Oct 4 10:53:08 EST 2005 ppc GNU/Linux

and after that of course the free/no warranty blurb and a command line.

Ctl-alt-F4 is shown next, below. Does it indicate a problem with the CD I
burned? (I checked the md5 sums on the download, but can't recall whether I had
the burning s/w verify the CD. Usually I do, but I can't swear I did this
time.) But given how far the installer got, and given what I saw next on
ctl-alt-F4, I'm more inclined to suspect a locales issue.

Anyway, ctl-alt-F4 shows:

After reading reports 15700 and 12276, I did an strace as suggested there and
here's the result:

ubuntu@ubuntu:~$ strace -p $(pidof esd)
Process 17651 attached - interrupt to quit
select(6, [4 5], NULL, NULL, NULL
 <unfinished ...>
 
and after nothing more had happened for three by-the-watch minutes, I did a
ctl-c and got 
 
Process 17651 detached

Should I have let it run longer?  A return to ^-alt-F7 still had the blank
screen, and nothing had changed on ctl-alt-F3 or F4 either.  By the way, I was
wrong earlier in saying I couldn't get to a command prompt from the brown
screen.  I just hadn't pressed the function key quite long enough.  Sorry; my
mistake.

I also did a

gdb /usr/bin/esd $(pidof esd)

and got back

attaching to program: /usr/bin/esd, process 17598

(this being a different time & so different pid for esd), and then a string of
alternating messages:

reading symbols from [...] (no debugging symbols found) ...
loaded symbols for [..same as above...]

(i.e., where [...] were different, but the "reading" and "loaded" lines
alternated, and the "reading line" always had "no debugging symbols found" in
it.  And when I did, while in gdb, a "bt," it returned the following:

As for the terminal screens, ctl-alt-F1 didn't show much interesting: just the
same lines as appear in brown under the "reflected" logo:

* Setting the system clock...                  [ ok ]
 * Synchronizing clock to ntp.ubuntulinux.org...
Error : Temporary failure in name resolution
                                                [fail]
 * Initializing random number generator...      [ ok ]
 * Setting up X server socket directory...      [ ok ]
 * Setting up ICE socket directory...           [ ok ]
Last login: Sun Feb 14 00:26:44 1984 on tty2
Linux ubuntu 2.6.12-9-powerpc #1 Tue Oct 4 10:53:08 EST 2005 ppc GNU/Linux

and after that of course the free/no warranty blurb and a command line.

Ctl-alt-F4 is shown next, below.  Does it indicate a problem with the CD I
burned?  (I checked the md5 sums on the download, but can't recall whether I had
the burning s/w verify the CD.  Usually I do, but I can't swear I did this
time.)  But given how far the installer got, and given what I saw next on
ctl-alt-F4, I'm more inclined to suspect a locales issue.

Anyway, ctl-alt-F4 shows:

Feb 14 05:21:31 casper: Mounting snapshot...
Feb 14 05:21:31 kernel: [  246.064610] cramfs: wrong magic
Feb 14 05:21:31 kernel: [  246.079234] Unable to identify CD-ROM format.
Feb 14 05:21:31 kernel: [  246.085545] EXT2-fs warning: maximal mount count
reached, running e2fsck is recommended
Feb 14 05:21:32 casper: Scanning for swap devices...
Feb 14 05:21:32 casper: Found /dev/hda4
Feb 14 05:21:32 casper: Using swap devices: /dev/hda4
Feb 14 05:21:55 root: register-modules
 * Entering runlevel: 2
 * Starting system log daemon...                        [ ok ]
 * Starting kernel log daemon...                        [ ok ]
 * Starting system message bus...                       [ ok ]
 * Starting Hardware abstraction layer:                 [ ok ]
Starting pbuttonsd: pbuttonsd 0.7.1: iBook/G3 PB Pismo/G4 PB Titanium (PMU
version: 12)
 * Starting GNOME Display Manager...                    [ ok ]
 * Starting Common Unix Printing System: cupsd          [ ok ]
 * Starting HP Linux Printing and Imaging System        [ ok ]
 * Starting PCMCIA services...                          [fail]
 * Starting powernowd...
 *  CPU frequency scaling not supported
                                                        [ ok ]
 * Starting Bluetooth services...
 hcid sdpd                                              [ ok ]
 * Starting deferred execution scheduler...             [ ok ]
 * Starting periodic command scheduler...               [ ok ]
Last login: Sun Feb 14 00:26:44 1984 on tty3

& then the same ID/date of Linux, free and no warranty blurb, and a command
prompt.  I am copying it for you above just for completeness, and just in case
the cramfs "wrong magic" and "unable to identify CD-ROM format" messages are
relevant.  But what looks to untutored, merely guessing me like a more likely
problem is the following, from ctl-alt-F3:

insmod /lib/modules/2.6.12-9-powerpc/kernel/drivers/ide/pci/generic.ko
insmod /lib/modules/2.6.12-9-powerpc/kernel/drivers/ide/pci/cmd64x.ko
insmod /lib/modules/2.6.12-9-powerpc/kernel/drivers/ide/pci/aec62xx.ko
insmod /lib/modules/2.6.12-9-powerpc/kernel/fs/nls/nls_base.ko
insmod /lib/modules/2.6.12-9-powerpc/kernel/fs/isofs/isofs.ko
perl: warning: Setting locale failed.
perl: warning: Please check that your locale settings:
        LANGUAGE = (unset),
	LC_ALL = (unset),
	LANG = "C.UTF-8"
    are supported and installed on your system.
perl: warning: Falling back to the standard locale ("C").
locale 'fr_Fr.UTF-8' valid and available
perl: warning: Setting locale failed.
perl: warning: Please check that your locale settings:
        LANGUAGE = (unset),
	LC_ALL = (unset),
	LANG = "C.UTF-8"
    are supported and installed on your system.
perl: warning: Falling back to the standard locale ("C").
locale 'en_US.UTF-8' valid and available
Reading package lists...
Building dependency tree...
console-tools is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
[  545.609436] Linux Kenel Card Services
[  545.613428]   options:  [pci] [cardbus] [pm]

and then the last login time for tty1, Linux ubuntu ID and time, and free and no
warranty lines before the login.

The warning about the locale setting having failed seems odd, but it could
explain why no desktop is shown: which desktop should it display in what
language?  I saw that first when I chose English & US, so I tried again with
French, and then with Czech (those being the only other two languages whose
keyboards and words I know), with essentially the same results.  If it were an
install to the HD, I could edit etc/default/init; but of course the live CD
isn't writable.

What do you suggest as a workaround?  First stop xorg & gdm, and then just do a
setenv and then start a new session?  Hmmm, as I ask that, I get a sinking
feeling I did only half of what I should have done over the weekend: I didn't
try a locale command, nor a setenv LANG or setenv LOCALE command, either.  I'll
try those next, but won't delay this message to do so.

Fwiw, I also noticed that if I chose Czech, for example, as the language at the
outset, I would receive the progress messages in the blue and red %-bar screens
in that language -- e.g., nastava se sit pro DHCP, etc., *until* the "preparing
for live system" message -- which appeared in English, at 26% for "configuring
language" (in English); there was one more message in the chosen language
(setting up the keyboard -- nastavuje se klavesnice), but then it was all
English again.  I don't know that that's necessarily related to the failure to
set locale and language, but maybe it's relevant.

HTH.

Revision history for this message

Daniel Holbach (dholbach) wrote on 2005-10-11:

Thanks for your very detailed description. I just read another interesting
report (http://bugzilla.ubuntu.com/show_bug.cgi?id=17250) which indicated that
his problem was related to a wrongly set date. Could you have a look if your
time/date settings are ok? You mentioned "Feb 14" a couple of times, which
doesn't look right.

Revision history for this message

Mtrettenbach (mtrettenbach) wrote on 2005-10-11:

You're right; it's the same as 17250. I wrongly assumed the date was from the
live CD, as YDL still booted. But there the same date was in YDL, too, when I
looked. I did take out the battery a couple of weeks ago; I must not have
re-set the clock after putting it back in. I reset the clock now, did a reboot
to be sure the new date was held, then booted the live Ubuntu RC 20051005 CD --
and it's running fine. Mr. Zimmerman's "seen some other" comment in 17250 was
right on. Thank you both!

Revision history for this message

Daniel Holbach (dholbach) wrote on 2005-10-11:

REJOICE! :)))

This bug has been marked as a duplicate of bug 23426.

Report a bug

This report contains Public information

Everyone can see this information.

Duplicate of bug #23426 Remove

You are

Subscribing...

Edit bug mail

Other bug subscribers

Remote bug watches

ubuntu-bugzilla #17201 Edit

Bug watches keep track of this bug in other bug trackers.