Problems With Pdaxrom 1.1.0beta3 Sl-c3100

[ZDevil]

Thanks for the info.
If I use Meanie's image, should I mount it into the same place as the 3.4.6.img and use it the same way?
I'd really like to be able to compile QT programs as well.

Now I have to fiddle around with the rc.d scripts in order to be able to mount the image at bootup: when rc.sysinit wants to mount the image, ide3, where the image is located and linked to /home/root, is not yet available :-)

[div align=\"right\"][a href=\"index.php?act=findpost&pid=138212\"][{POST_SNAPBACK}][/a][/div]

Yes, just mount Meanie's zgcc in the same way as 3.4.6. The only thing you gotta change is to update the gcc version number in four lines in /etc/profile (e.g. 3.4.6 --> 3.4.5-4). Then source /etc/profile or restart X and you are ready to go.  

I also tried a bit before to get zgcc automount but failed. So I simply use a simple script to mount and umount.  (Note:unmount takes two steps, first "umount /opt/native/arm/..." and second "losetup -d loop0" (usually loop0 if no other loop devices are mounted).

[daniel3000]

You may want to try out the ttf-bit-stream-vera package from the feed. These fonts look good!

I have installed that package now.
mrxvt seems to use a font from that package (it is configured to use Bitstream vera Sans Mono).
the font indeed looks nice, but mc and mc -a still show a messed up layout in mrxvt using this font.

Well, it's not THAT important for me, but it would be nice to have a well-operable mc.

Any ideas?

thanks
daniel

[daniel3000]

Yes, just mount Meanie's zgcc in the same way as 3.4.6. The only thing you gotta change is to update the gcc version number in four lines in /etc/profile (e.g. 3.4.6 --> 3.4.5-4). Then source /etc/profile or restart X and you are ready to go. 

Well, if I don't change the mountpoint, i.e. path, I also don't have to change /etc/profile, because there only the path is used for several things.
Or does the path name somehow tell some binaries which compiler version it is, so I HAVE to change it to the actual version?
For now I have left it as it is and at least mounting and a ./configure work well (I haven't tested more yet).

I also tried a bit before to get zgcc automount but failed. So I simply use a simple script to mount and umount.  (Note:unmount takes two steps, first "umount /opt/native/arm/..." and second "losetup -d loop0" (usually loop0 if no other loop devices are mounted).
[div align=\"right\"][a href=\"index.php?act=findpost&pid=138217\"][{POST_SNAPBACK}][/a][/div]

I have now set up the new image for auto-mounting at bootup, works well.
These are the steps I went in order to make this work (from my diary-style setup notes):

Code: [Select]

Development stuff based on zgcc-3.4.5-4.squashfs (actually zgcc-3.4.6.img but 3.4.5-4 has more features, by Meanie)
grep zgcc /etc/rc.d/rc.sysinit
--> shows where the img has to go and where it will be mounted on reboot.
mv zgcc-3.4.5-4.squashfs zgcc-3.4.5-4.sfs
cp zgcc-3.4.5-4.sfs /mnt/ide3
ln -s /mnt/ide3/zgcc-3.4.5-4.sfs /home/root/
From /etc/rc.d/rc.sysinit cut section if..fi which mounts zgcc and paste it into the new file
/etc/rc.d/init.d/mount_devel
Add as the first line of mount_devel: #!/bin/sh
Search and replace in mount_devel:
zgcc-3.4.6.img --> zgcc-3.4.5-4.sfs
add the option "-t squashfs" to the mount command
chmod +x /etc/rc.d/init.d/mount_devel
ln -s /etc/rc.d/init.d/mount_devel /etc/rc.d/rc5.d/S99mount_devel
ln -s /etc/rc.d/init.d/mount_devel /etc/rc.d/rc4.d/S99mount_devel
ln -s /etc/rc.d/init.d/mount_devel /etc/rc.d/rc3.d/S99mount_devel
ln -s /etc/rc.d/init.d/mount_devel /etc/rc.d/rc2.d/S99mount_devel
ln -s /etc/rc.d/init.d/mount_devel /etc/rc.d/rc1.d/S99mount_devel
reboot
--> devel image is mounted on bootup and all developer tools are available.

[daniel3000]

Well, if I don't change the mountpoint, i.e. path, I also don't have to change /etc/profile, because there only the path is used for several things.
Or does the path name somehow tell some binaries which compiler version it is, so I HAVE to change it to the actual version?
For now I have left it as it is and at least mounting and a ./configure work well (I haven't tested more yet).

Not true what I wrote.
Indeed the mountpoint AND the paths in /etc/profile have to be changed from ...3.4.6... to ...3.4.5... (not ...3.4.5-4...!)

Something in the development system seems to look for includes only in the path reflecting the gcc version, which is 3.4.5 in this case. I had lots of compile errors with the wrong directory name.

Thanks for the hint!
daniel

[daniel3000]

I made a little script to fix the date:

Code: [Select]

/etc/rc.d/init.d/atd stop
sleep 3
rm -rf /etc/localtime
ntpdate pool.ntp.org
hwclock --systohc
hwclock --utc
ln -sf /usr/share/zoneinfo/SOME/ZONENAME /etc/localtime
/etc/rc.d/init.d/atd startI run it now and again when I'm on a network.

I just wanted to implement something similar.
However, one question:

I don't have the directory /usr/share/zoneinfo and no /etc/localtime.
What is the purpose of these and how do I get them?

thanks
daniel

[desertrat]

I don't have the directory /usr/share/zoneinfo and no /etc/localtime.
What is the purpose of these and how do I get them?

Install the "timezones" package (it's in the official feed) to get /usr/share/zoneinfo - this contains a bunch of files with timezone info. /etc/localtime is simply a symlink to the relevant timezone file in /usr/share/zoneinfo for your neck of the woods - you have to create it yourself after installing said package.

[daniel3000]

I don't have the directory /usr/share/zoneinfo and no /etc/localtime.
What is the purpose of these and how do I get them?

Install the "timezones" package (it's in the official feed) to get /usr/share/zoneinfo - this contains a bunch of files with timezone info. /etc/localtime is simply a symlink to the relevant timezone file in /usr/share/zoneinfo for your neck of the woods - you have to create it yourself after installing said package.
[div align=\"right\"][a href=\"index.php?act=findpost&pid=140223\"][{POST_SNAPBACK}][/a][/div]

Well, thanks. Now I have all I need it seems.
However, if I execute the script, following happens:

I live in Germany. It is 22:10 (10:10pm).
hwclock is set to 8:10pm by the script.
clock in tray then also shows 8:10pm (20:10).
OK, probably the UTC time, as it is here middle-european summertime (MEST).
Then I thought I could adjust it with the command
hwclock --hctosys
Result:
Clock in tray now shows 18:10 (6:10pm)!
What's up here?

I have linked /etc/localtime to /usr/share/zoneinfo/Europe/Berlin.

Something seems to be wrong here... ;-)
Any ideas?

Thanks
daniel

[desertrat]

... clock in tray then also shows 8:10pm (20:10).

If you're using matchbox, then I wouldn't try adjusting the clock too much - it has a tendency to lockup. Do your clock adjustments outside of X, outside of matchbox at least.

Then I thought I could adjust it with the command
hwclock --hctosys

You shouldn't use that command. The idea behind this whole thing is that the hwclock is set to UTC. and by creating the appropriate /etc/localtime, the date command will then show the correct
time.

Something seems to be wrong here... ;-)

The script works flawlessly for me  It's based on an earlier post (search for hwclock in this forum). Try running the commands manually and see whether that gives any clue (after each command run date)

[daniel3000]

If you're using matchbox, then I wouldn't try adjusting the clock too much - it has a tendency to lockup. Do your clock adjustments outside of X, outside of matchbox at least.

I have also noticed that.
That's why I modified my script slightly  

Code: [Select]

#!/bin/sh
echo This script reboots the Zaurus at the end!
echo Close all apps and then press Enter!
read
/etc/rc.d/init.d/atd stop
sleep 3
rm -rf /etc/localtime
netconnect.sh
sync
ntpdate pool.ntp.org
hwclock --systohc
hwclock --utc
ln -sf /usr/share/zoneinfo/Europe/Berlin /etc/localtime
echo Press enter to reboot now (all apps closed?)
read
reboot
Seems as if these lockups you mean are those I described in my "problems with pdaxrom 1.1.0beta3" thread. No real lockups but simply no response to keyboard or mouse events anymore.

This indeed seems to happen each time I adjust the clock AND after I close the Z without suspending and after a while opening it again to continue my work.
Could it be that there is also some clock adjustment action in those situations?


Okay, will try the script outside X and see if it works better then.
Thanks!

[desertrat]

That's why I modified my script slightly 

That's a bit painful having to reboot just to adjust the clock? Even Windows can do that without rebooting! Though I love the tagline "Your mouse has moved, Windows needs to reboot for the changes to take effect [YES]"

Seems as if these lockups you mean are those I described in my "problems with pdaxrom 1.1.0beta3" thread.

Possibly. Anyway adjusting the clock is a major cause of lockups on matchbox and has been widely reported in this forum? I'm sure I read somewhere that one of the workarounds implemented was to restart matchbox when the machine is resumed after a suspend of more than [some amount] of time, that is why resume sometimes takes a long time when you have a lot of apps opened.

No real lockups but simply no response to keyboard or mouse events anymore.

That is as good as a lockup  If you're connected to the network then you can sometimes ssh in and kill X thus regaining control of the machine without having to hard reset.

This indeed seems to happen each time I adjust the clock AND after I close the Z without suspending and after a while opening it again to continue my work.
Could it be that there is also some clock adjustment action in those situations?

I believe this is what happens (someone correct me if this is hopelessly wrong): when you suspend, the software/system clock is copied to hwclock on resume the hwclock is copied to system clock. Matchbox only uses system clock, so everytime you resume matchbox sees a jump in the system clock, if the jump is large (or larger than some indeterminate amount) it panics and commits suicide.

Which is why I use icewm