104 replies to this topic

[littlesat]

I prefer to wait on pieterg's advice here....

[betacentauri]

the subtitle rendering is probably missing a

m_dc->islocked()

check.
Each painter / dc operation is checking the lock, it seems.
That's how the issue should have been fixed I think.

 

Hi pieterg,

 

I have debugged it and found the error:

In lib/gdi/grc.cpp in procedure blitScale I have added a eDebug which shows that m_dc->islocked() is always 0. Then I tried to find out which islocked method is called. It's not the gFBDC::islocked! It's the virtual islocked method in gDC (grc.h).

Ok, then I thought, the variable m_dc must be of type gDC, otherwise it should be called gFBDC::islocked. But at runtime typeid(m_dc).name() says "5gFBDC". So it's type is gFBDC. Why is then the virtual method called?

 

I found the answer. It's because the islocked method in gFBDC has a const qualifier:

int islocked() const {....};

After removing const it works. A little bit strange.

 

Here is a patch.

I guess with it Littlesat can revert his tuxtxt subtitle workaround (I have tested with the subtitles and it works without the patch)

Attached Files

•  0001-Fix-framebuffer-locking.patch.txt   666bytes   16 downloads

[pieterg]

ok, good find.
Looks like it might have been working earlier this year

43ed3b0a (Mike Looijmans 2013-07-28 12:49:58 +0200 35) int islocked() const { return fb->islocked(); }

A reminder that we really should pay attention at the compiler warnings
(and enable -Woverloaded-virtual, if we didn't so already)

[betacentauri]

Hi Pieterg, Milo:

I guess instead of removing const we can also add const to gDC::islocked. Or?
What do you prefer?

[pieterg]

yes, that would be better

[betacentauri]

Ok, here is the patch.

Attached Files

•  0001-Fix-framebuffer-locking_v2.patch.txt   812bytes   13 downloads

[betacentauri]

Ok, I see the patch is applied and littlesat has reverted the workarounds. I guess you get now problems applying my patches (http://openpli.org/f...ndpost&p=388940). I'll rewrite them in the next days and post them here again.

[littlesat]

Ok at least thanks!

[betacentauri]

Ok, here are the patches.

 

As said before: With the patch you can assign every teletext functionality to almost every button you like via keymap.xml

Attached Files

•  enigma2_git_0001-Teletext-improvements.patch.txt   3.28KB   34 downloads
•  tuxtxt_git_0001-Teletext-improvements_v2.patch.txt   7KB   25 downloads

[quad]

I really like the idea of adding keymap.xml support to teletext (I would then add KEY_CHANNELUP and KEY_CHANNELDOWN for "page up" and "page down"). Unfortunately merging patches into files from the repository is exceeding my skills. I would really appreciate if this addition could be commited. What are the plans / schedule for that? Thanks..

Edited by quad, 25 January 2014 - 20:40.

[littlesat]

Isn't this still committed?

[betacentauri]

No, it's not committed. You wanted to wait what pieterg say to the patch.
http://openpli.org/f...ndpost&p=390836

Please post here a little info when you commit it!

Edited by betacentauri, 26 January 2014 - 00:03.

[littlesat]

I just tried... at this moment I'm not able to commit this patch...

[betacentauri]

Why? Error? Or other problem?

[littlesat]

$ git am -s ../tuxtxt_git_0001-Teletext-improvements.patch.txt
Applying: Teletext improvements
c:/######/tuxtxt/.git/rebase-apply/patch:212: trailing whitespace
.
#ifndef HAVE_TRIPLEDRAGON
error: patch failed: tuxtxt/plugin/plugin.py:16
error: tuxtxt/plugin/plugin.py: patch does not apply
Patch failed at 0001 Teletext improvements
The copy of the patch that failed is found in:
   c:/######/tuxtxt/.git/rebase-apply/patch
When you have resolved this problem, run "git am --resolved".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".

[betacentauri]

Patch is quite old. Perhaps something has changed. I'll take a look at it.

[betacentauri]

Strange! What did you do?

$ git clone git://git.code.sf.net/p/openpli/tuxtxt
Klone nach 'tuxtxt'...
remote: Counting objects: 263, done.
remote: Compressing objects: 100% (212/212), done.
remote: Total 263 (delta 129), reused 107 (delta 50)
Empfange Objekte: 100% (263/263), 173.72 KiB | 303.00 KiB/s, done.
Löse Unterschiede auf: 100% (129/129), done.
Prüfe Konnektivität... Fertig

$ cd tuxtxt
$ git am -s tuxtxt_git_0001-Teletext-improvements_v2.patch.txt 
Wende an: Teletext improvements
$

Important: Both patches must be applied! One in tuxtxt git and one in the e2 git.

[quad]

Betacentauri, are you saying that you are able to compile it correctly?

 

If so, should I send a message to pieterg asking him for a review and approval?

[betacentauri]

Hi quad,

 

that means that the patch could be applied without problems.

I didn't do another function test, but it should still work.

 

You can send pm to pieterg, but I guess he reads the forum and have no time to review it.

[littlesat]

I'll retry it later... Hold on...