128 replies to this topic

[foxbob]

I apologize, but it worked.I connected the bluetooth keyboard and mouse.The mouse works very incorrectly.

Attached Files

•  1_0_1_835_15_FD_1680000_0_0_0_20200513193156.jpg   34.99KB   4 downloads
•  1_0_1_835_15_FD_1680000_0_0_0_20200513193244.jpg   36.57KB   4 downloads
•  1_0_1_835_15_FD_1680000_0_0_0_20200513193340.jpg   39.11KB   5 downloads
•  1_0_1_835_15_FD_1680000_0_0_0_20200513193355.jpg   65.63KB   5 downloads
•  1_0_1_835_15_FD_1680000_0_0_0_20200513193621.jpg   117.61KB   5 downloads

[WTE]

Maybe the reception to your bluetooth usbstick is bad?

By me the mouse works actually pretty good.

 

Second your still using linuxfb and I test actually with eglfs. Maybe ths can cause some smooth issues with rendering as well.

[WTE]

Some devs want to switch to eudev. Some (like me ) don’t really care. Problem with current thud version is that it has a big bug. Data on attached devices might be deleted. This bug needs to be fixed before we can switch to eudev and maybe also other things have to be done. Don’t exactly know the status.

 

Yes, I know again why this happens. I remove volatile-media This script will erase your devices with eudev !!!

[WanWizard]

Meaning the issue still exists?

[littlesat]

Just to understand it.... remove volition media solved the issue?

[WTE]

Just to understand it.... remove volition media solved the issue?

yes, eudev mounts on /run/media/* and mdev on /media/*

 

media-volatile set map /media as size 64KB

On some reason all mounts combined together cannot exceed 64KB so it will erase everything.

 

It seems when I adjust script to mount on media and not run/media that it won't been erased.

[WTE]

I do not know if other mounts for mediadevices like preselect by uuid in fstab is done outside the mountscript and so maybe still mount to /run/media and not /media which can cause erase of disk

 

For testing with zeus branch change busybox-mdev to eudev in https://github.com/O...openpli.conf#L7

 

Remove the stop command line in https://github.com/O..._%.bbappend#L15

Edited by WTE, 14 May 2020 - 18:55.

[WanWizard]

Merged.

[WTE]

That's fast

 

I think it works already very similar as mdev.

I have check eudev and didn't find any more /run/media mounts so it seems that it's now ok working.

 

It can been optimize as there are lot of values in udevadm and possible rules to add. I have added "usbhdd" so a disk with (s)ata connect on USB.

I assume we can symlink this as well like /hdd but then to /usbhdd. The record path it's maybe possible to check if symlink is activate /hdd if not try /usbhdd else no record disk path found and you should set it manual.

I think most have usbhdd or NAS for recording. So probably usbhdd is very often used but people need to set it now manual as it's in mdev not detect as /hdd

[littlesat]

So a python change in e2 is required for an external hdd?

[WanWizard]

There is more needed, this isn't finished yet.

 

Think about mounting a device in /media/<label>, automatically generating a samba share when a device is mounted, etc.

[WTE]

I was just thinking about a new feature to make it easier to set the recording path. This can been usefull for mdev as well. The recording path is set on /hdd and the mount to /hdd the disk need to been an internal HDD. Most I think have usbhdd a disk on USB so at moment you must set the osdmenu to expert and set it to your usb hdd device to get it working.

Just an idea if set recording path on auto then do like this by boot. Just quick not right code ofcourse just for idea

import os

if (check /tmp/ mount.hdd)

src = '/media/hdd'

elif (check /tmp/ mount.usbhdd)

src = '/media/usbhdd'

else

src = '/media/empty'

dst = '/hdd'

os.symlink(src, dst)

 

 

Mounting device with label (Usb stick) is already working. However not all devices have a label name and then it will been decide if it's mmc, hdd, usbhdd or just name it /dev/sdXx.

Samba and NFS I think this is set in fstab and I assume this will work as well. I have check previous from box to box and this was working fine. I will check a bit more and where this mounts are made. However I can do it early next week.

 

Initialize disks works now as well so from what I can see it works now very similar as Mdev.

Edited by WTE, 15 May 2020 - 08:49.

[Erik Slagter]

The proposed and applied changes look like reverts to changes we did to prevent eudev from being used and started. I guess it would be better to find the commits and revert them.

 

Also I think this should be in de thud branch, not develop or zeus. I'll have a look.

 

I think there is a flaw in the reasoning in the scripts of volatile-tmp. Apparently it will try to guard the volatile (ramdisk) file system from growing too large. But also apparently it does not use the system call the stat the fs (which would be good, like df), but it's measuring the files under the mountpoints and yes, with some mounts, it may run huge very quickly of course.

 

So it looks like there are two ways to fix this:

 - go and fix the volatile-tmp package, which would be preferrable but may be complex

 - stop mounting on directories in a tmpfs filesystem (WTE's approach)

 

I don't really see an objection against mouting on a flash directory, it's only not completely standard these days.

Edited by Erik Slagter, 15 May 2020 - 09:00.

[Erik Slagter]

This commit is "reverted" in thud:

commit ff50b867ac290d987165053f031d161b68acc124
Author: Harro Verton <wanwizard@openpli.org>
Date:   1 year, 4 months ago

    make sure udev doesn't start automatically

diff --git a/meta-openpli/recipes-core/eudev/eudev_%.bbappend b/meta-openpli/recipes-core/eudev/eudev_%.bbappend
index 3c739a7e..bc38362d 100644
--- a/meta-openpli/recipes-core/eudev/eudev_%.bbappend
+++ b/meta-openpli/recipes-core/eudev/eudev_%.bbappend
@@ -8,3 +8,5 @@ do_install_append() {
 }

 FILES_${PN} += " ${sysconfdir}/udev/rules.d/98-dbox-devices.rules "
+
+INITSCRIPT_PARAMS_eudev = "stop 04 S ."

It should be explicitly reverted there (in thud). Then the rebase of zeus will yield a conflict (because the parent directory is called udev in zeus, eudev in thud) but that's no problem.

 

Enabling udev was in my original sumo, but apparently it was skipped before merge, so there is no original commit for enabling udev.

[mrvica]

I assume it needs a bunch of MB´s in the flash, will it be an option to install it on a hdd or external drive like Kodi

Edited by mrvica, 16 May 2020 - 09:02.

[foxbob]

Collected with new changes and can not start.

And mounting an external drive shows that it is mounted,but nothing happens-enigma2-plugin-systemplugins-devicemanager.

root@et1x000:~# df -h
Filesystem                Size      Used Available Use% Mounted on
/dev/root                 3.5G    551.1M      2.8G  16% /
devtmpfs                144.9M         0    144.9M   0% /dev
tmpfs                    64.0K         0     64.0K   0% /media
tmpfs                   153.0M     68.0K    153.0M   0% /run
tmpfs                   153.0M     68.0K    153.0M   0% /var/volatile

Attached Files

•  1_0_1_835_15_FD_1680000_0_0_0_20200517084205.jpg   153.68KB   2 downloads

Edited by foxbob, 17 May 2020 - 06:52.

[athoik]

No /dev/shm directory?

 

Fixed on mdev by https://github.com/O...73ef3676200fda7

Edited by athoik, 17 May 2020 - 06:47.

[foxbob]

After mount 

root@et1x000:~# df -h
Filesystem                Size      Used Available Use% Mounted on
/dev/root                 3.5G    551.1M      2.8G  16% /
devtmpfs                144.9M         0    144.9M   0% /dev
tmpfs                    64.0K         0     64.0K   0% /media
tmpfs                   153.0M     68.0K    153.0M   0% /run
tmpfs                   153.0M     68.0K    153.0M   0% /var/volatile
/dev/sda1               915.9G    546.2G    323.1G  63% /media/hdd

[foxbob]

This is zeus-eudev branch.

[athoik]

This is not executed? openembedded-core/meta/recipes-core/udev/eudev/init ?

 

Because above eudev/init seems to create the required directories on /dev/

 

    [ -e /dev/pts ] || mkdir -m 0755 /dev/pts
    [ -e /dev/shm ] || mkdir -m 1777 /dev/shm