betacentauri

Hi pieterg,

 

I took technics patch (post #122) and adapted it, because it didn't run properly for me (perhaps I made somewhere a mistake...). I don't use a mutex, because I think it's not really necessary. I have added a new function isStream, because I didn't like to change valid function to return true even when connection is not established yet.

I hope you'll have some minutes to look at it. I think there are many people who don't like to wait 10 seconds if stream is not available

Hi,

 

here are 2 patches:

1. Timeshift didn't work for streams. This enables timeshift only for streams with service ref 1:0:... and not with 4097:0:...!!

  Problem was that timeshift file was opened with eHttpStream and not with eRawFile, because the main service ref is a stream.

2. If a URL only contains hostname and port (e.g. http://1.2.3.4:8888) it has to end with /. Without / it don't work. I changed it.

  And a HTTP redirect didn't work if the Content-Type is behind the Location.

  I mean something like this:

 

GET / HTTP/1.1
Host: w.x.y.z

HTTP/1.1 302 Found
Date: Sat, 30 Mar 2013 21:51:34 GMT
Server: Apache/2.2.15 (Fedora)
Location: http://a.b.c.d:8000
Content-Length: 207
Connection: close
Content-Type: text/html; charset=iso-8859-1

I know 2. patch has nothing to do with timeshift, but I didn't want to create a new thread for this little patch

 

 

EDIT: Sometimes sound was not available while watching timeshift (in streams). I'll try to find out, why sound disappears.

The latest commits are only visible in new repository.
I guess that the daily builds still use the old repository, because I made today an update and I cannot see the new file ScreenSaver.pyo (commit: http://sourceforge.n...02d4cd0d6d667e/)on my box.

and correct these errors http://openpli.org/f...post__p__337494

To 1: I'll try to test whether my suggestion works. If it works, I'll provide a patch.
To 2: As said before pieterg tried to tune it with no success: http://openpli.org/f...post__p__332900
To 3: If I have time (not very likely) I'll look at it.

C++ is not so difficult, if you have already programmer skills (but it's more work to build a binary and so on). And I also do little changes to existing code .

I don't care about the 3 lines. I only wanted to know what problem it is and whether I can fix it.

Ich bin leider ein bißchen ratlos. Wenn hier kein anderer antwortet, würde ich mal im englischen Teil des Forums nachfragen. Dort gibt es mehr Experten.

HDMI sollte kein Problem sein. Beim Satellitenkabel habe ich mal gehört, dass das sehr schlecht sein soll...
Welcher Receiver macht das Problem?
Hast du das richtige Image eingespielt (keine Ahnung ob man da etwas falsch machen kann). Wenn das vom Netz trennen nicht hilft, spiel mal ein älteres Image ein.
Per google habe ich das gefunden:
http://www.linuxques...-see-it-788546/
Da hört es sich nach einem Treiber Problem an. Also hoffentlich doch kein Hardwareproblem bei dir...

Hi,

ich bin kein Experte, aber das sieht aus meiner Sicht nicht gut aus. Hast du vielleicht das Satellitenkabel abgemacht ohne den Receiver vom Strom zu nehmen?
Hast du schon mal versucht den Receiver ohne Satellitenkabel zu starten (vorher unbedingt vom Strom nehmen). Ansonsten würde ich den Receiver für
mehrere Minuten vom Strom trennen.
Existiert /dev/fb0 wirklich nicht im filesystem?

Hi,

du könntest mal die folgenden Dinge anschauen:[list=1]
[*]Platz im filesystem: df -h
[*]Sind crash logs vorhanden? Wenn ja, bitte posten.
[*]Kannst du e2 manuell starten? Einfach enigma2 eintippen und return drücken.
[*]Es gibt doch zu allererst ein Xtrend Logo, dann kommt der blaue OpenPli Bildschirm und dann wieder kurz ein Xtrend Logo (wenn ich mich jetzt nicht ganz irre). Wo genau bleibt er hängen?
[*]Hast du irgendetwas besonderes gemacht? Änderungen an der Konfiguration?
[/list]

Hi,

sorry, I have found a new bug:
1. Start instant radio record for 30 minutes (WDR2).
2. After 12 minutes abort recording.
3. Play recorded file
-> Playing is stopped immediately, because it's 0 seconds long (but 30MB).

In the debug logs I see wrong calculated record length:

calcBegin not valid
calcBegin getFirstFrame==0
eMPEGStreamInformation::fixupPTS(offset=22862727 pts=3047188059)
fixupPTS empty
calcBegin m_futile 0, valid 1, off_beg 22862727, pts_beg 3047188059, pts_length 8540665976164183572
calcEnd: source_len 31950788
file size changed, recalc length
calcEnd: not valid
calcEnd: getLastFrame==0
eMPEGStreamInformation::fixupPTS(offset=22862727 pts=3047188059)
fixupPTS empty
pts_diff proc be 3047188059, end 3047188059
pts_diff proc diff: 0
pts_diff proc new length 0
calcEnd m_futile 0, valid 1, off_end 22862727, pts_end 3047188059, pts_length 0

I also have a recording which real length is about 17 minutes, but system calculates 7 minutes length.

If I delete the .sc file everything works fine!!
So I think that there's a problem with the .sc generating (First offset from sc file in calcBegin is 22862727. And offset in calcEnd is also 22862727).

If I do the same for TV recordings, I can't reproduce the problem.

I'll look into it the next days.

Hi openpli team,

I have found a solution for the bug with the black screen (immediately after zapping to a channel press pause -> screen is black and stays black).
If you start timeshift, 2 seeks are started (one from eDVBServicePlay::switchToTimeshift and another one from InfoBarGenerics).
After first seek m_futile in tstools.cpp is true.
-> At first m_end_valid (and not m_begin_valid!) gets true and the m_pts_length is calculated. But m_pts_length is wrong because m_begin_valid is false and m_pts_begin=0
-> m_pts_length is very big and filepush thread seeks to an unknown position.
(I can attach debug log and changed sources if you want to see them).

A fix is to use only 1 seek (the one from InfoBarGenerics).

So if you delete in servicedvb.cpp line 2599 (m_cue->seekTo(0, -1000); ), problem is solved.

Another solution would be to fix tstools....

@all: Already seven downloads. Did somebody test it and can confirm, that it's working??

@greatred: Here's the howto for my et9200. Your directories might be named in a different way.
1. Go to your build directory: <path_to_build_dir>/openpli-oe-core
2. Edit site.conf and put a # before last line:

#INHERIT += "rm_work"

3. Change to <path_to_build_dir>/openpli-oe-core/build-et9x00 directory
4. Execute this

source env.source; bitbake -f -c compile -b ../meta-openpli/recipes-openpli/enigma2/enigma2.bb

5. Enigma2 unpatched binary is now located in this directory:
<path_to_build_dir>/openpli-oe-core/build-et9x00/tmp/work/et9x00-oe-linux/enigma2-2.7+......./git/main
6. Edit this file: <path_to_build_dir>/openpli-oe-core/build-et9x00/tmp/work/et9x00-oe-linux/enigma2-2.7+......./git/lib/base/filepush.cpp
Goto line 141 and change

sendEvent(evtEOF);
if (m_stop)
  break;

into this

if (m_stop)
  break;
sendEvent(evtEOF);

And save file.
7. Change to <path_to_build_dir>/openpli-oe-core/build-et9x00 directory
8. Execute this

source env.source; bitbake -f -c compile -b ../meta-openpli/recipes-openpli/enigma2/enigma2.bb

9. If there were no error messages, binary is located here: <path_to_build_dir>/openpli-oe-core/build-et9x00/tmp/work/et9x00-oe-linux/enigma2-2.7+......./git/main
Please check file date of the binary. It must have the date of the second compile!!
10. You can strip binary to make it smaller (I should be around 40MB. After stripping it's only 2,8MB big). You don't need to do this:
To strip execute this:

<path_to_build_dir>/openpli-oe-core/build-et9x00/tmp/sysroots/i686-linux/usr/bin/mips32el-oe-linux/mipsel-oe-linux-strip<space><path_to_enigma2_binary>

11. Put the binary on your stb. E.g. in /tmp and do a chmod 755 enigma2
12. Do a init 4
13. Start enigma2 from /tmp with ./enigma2

Hi Pieterg,

thanks for looking into it.
I made several tests:
1. Playing recording with skipping before start of file or after file end. And watching until normal end of file -> no problems
2. Normal live tv -> as expected no problems
3. Normal timeshift -> the problem is fixed

My fix "only" sends no EOF, if there was a stop/pause request. So end of file recognition works as before.
If there is a bigger timeshift buffer, then also no EOF event is sent while skipping.

You and everyone else with an ET9x00 can test it. I have attached the enigma2 binary.

root@dm7025:~# mount -o remount, rw/boot
mount: can't find rw/boot in /proc/mounts

There must be a space between rw and /boot

mount -o remount,rw  /boot

@Betacentauri,

Can you please arrange this to have it by default in the repository?

Hi Pr2,

strange that this solves the problem...
Shouldn't be a problem to include it in the build_openpliPC.sh.
(but not today...)

Regards,
Betacentauri

Hi all,

I have added the copy command to build_openpliPC.sh script. The change is available in master and openpli_merge branch.

Regards,
Betacentauri