Trying again, this time from a subscribed address (I hope)

On Wed, May 24, 2006 at 02:17:35PM +0100, Alan Lord wrote:
> Hi,
> 
> I am not really sure if this a problem with LFS or BLFS so I thought I'd 
> drop this one into this list:
> 
> I'm playing around with a newly built LFS (SVN 20060516) and am noticing 
> some weird things with the keyboard/screen...
> 
> During boot-up, as it gets to setting up the network interface using 
> dhcp the cursor stays on that line of the messages even when the login 
> prompt appears several lines below it (After going through the rest of 
> my init.d scripts). Login works, but the flashing cursor remains higher 
> up the screen...
> 
 I've seen that in the past, but I don't remember when.  More
recently, my G5 mac running as ppc64 (CLFS, obviously) is the only
box where I don't have a graphical login and yes, the cursor is
above the prompt - FWIW it uses a framebuffer with CONFIG_FB_NVIDIA.
I guess you are using a framebuffer ?

 It gets fixed if I switch to tty2, and back,  which was the same
fix for whatever triggered the problem in the past.  Probably I also
see it with 32-bit userspace, but at the moment that has a graphical
login.  I think it has only started happening on that box since
2.6.15, or even 2.6.16.

> Once logged in, entering ViM from the console makes the cursor disappear 
> altogether! You can still edit but just can't see where the cursor is. 
> It's a good guessing game but not very productive :-)
> 

 ouch!  Sounds like busybox's version of vi (to be fair, it's a long
while since I've used that, all I really remember is that it made me
wish for vim).

> I have built quite a bit of BLFS on top using another recent SVN book 
> and sometimes, in xterm, the bash interface seems to get stuck and 
> continuously scrolls as though someone is holding down the enter key and 
> it is just repeating... Simply hitting enter once stops this. 
> Occasionally xterm also stops spewing out [compiler messages for 
> example] as though someone has hit scroll-lock or ctl-s. Again a single 
> return starts things again...

 That sounds like a keyboard or kvm problem.  If not, I assume you
are running a newer kernel, and changes in the input subsystem are
doing this - if so, you might find the syslog has a lot of relevant
messages (in which case, technically it *is* a keyboard problem).

 Or, just possibly the X driver is responsible for the xterm
shenanigans - my G5 has an nvidia card, and all builds of X up to and
including 6.9, with kernels from 2.6.14 through 2.6.17-rc are
subjected to spurious vertical, and to a lesser extent horizontal,
scroll events in varying amounts (which makes highlighting for
copying, and sometimes even typing in spreadsheets, difficult) - it
also happens in ubuntu, so I have no feel for what is actually
responsible. (I do have a skeletal X on ppc64, just not on last
week's build, and a reasonably full X with ppc userspace).

 I'm assuming your previous build was somewhat old in terms of the
book's development - UTF-8 went in in about January, since then
we've probably moved to a newer glibc point release and perhaps a
newer gcc.  I don't know if you are using a newer X than in your
previous builds, but my best guess is kernel changes (there was a
thread on lkml the other week about trying to kill a log message of
'too many keys pressed' triggered by cheap keyboards, I think you
might have the same issue).

> 
> Now then, the main difference between this and my previous builds (of 
> which there have been many and none have exhibited these problems) apart 
> from the obvious software revision changes is UTF-8.
> 
> I have set in my i18n.sh
> 
> LANG="en_GB.utf8"
> 
> I was looking on the net for something else and found a blog where 
> someone said that the output of locale -a displaying xx_ZZ.utf8 is a BUG 
> and you should enter LANG="en_GB.UTF-8"
> 
> I have tried both ways and neither made any difference.
> 
 I remember Alexander telling me something similar, and now I use
en_GB.UTF-8 on all my boxes (including those running CLFS).  My mail
doesn't fully support UTF-8 at the moment (the server is a pure64
version of LFS-6.1, without ncursesw), but on the desktop I can
handle all European characters and such as «» (doubled <> marks)
on a clear day with the wind in the right direction, if I've
remembered to load my own Xmodmap addition so that I actually have a
compose key :)

Ken
-- 
das eine Mal als Tragödie, das andere Mal als Farce

----- End forwarded message -----

-- 
das eine Mal als Tragödie, das andere Mal als Farce
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-chat
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Reply via email to