[BLKMGK]

You know that you can monitor the CPU display with the title button, right? The little black bar with 4 (used to be 3) lines of info about the currently playing video?

Oh I know, I'm lazy! I actually sometimes have both sets of data on the screen, looks REAL ugly but that's okay. Now that both cores report separately it's even better. That said, I'm halted at 11702 for now and will hope for some help, this code that's screwing me up has got to be hitting others too somewhere...

[BLKMGK]

Log at Pastebin -> http://pastebin.com/m808dffc

[BLKMGK]

WooHoo! I've narrowed down the BUG!

Looking closely at the changes checked in for 11703 the changes made in XBMC/guilib/GraphicContext.cpp looked most suspect to me - line 26 references Xrandr which appears to have somethign to do with the crash since it occurs during resolution change. I'm NOT sure what's knocking it over but the deleted lines 584-585 and maybe 563 which seem to deal with resolutions could be it? Dunno', am not up to speed on the code or C really.

So anyway since this looked suspicious to me and I had a backup of 11702 that works for me. So I took the GraphicContext.cpp file from 11702, copied it to the test directory after an SVN up overwriting the existing file, compiled with build.sh NOUPDATE, crossed my fingers, got a GOOD build, and it RUNS!

I'm tempted to poke at it more but I don't know the developers intentions with the deletions and additions although it appears to be OSX based additions. Would a Dev please take a look at the changes introduced in this file for 11703 to try and track this down? If I poke at it deleting code and adding code I might munge it enough to work but I don't think that's exactly the best way to go

Thanks!!!
Edit: Okay, lines 584 and 585 are removed, 584 is moved to 552. 585 says m_Resolution = res; which is defining m_Resolution. One branch of the following IF statements does a compare against m_Resolution but because of 585 not being moved I do not see that variable being defined anywhere. I might try moving that second line up there too to see if that fixes it. I'd appreciate a second set of more skilled eyes though please. Core dump - oh well it was worth a shot

[madmax]

Hi BLKMGK,

seen the same thing here...I have the same problem...last working for me was 11702! I tried twice....everyting is fine including 11702, but 11703 I see this nasty crash and my hole desktop switches to that odd resolution...

Can't be just us 2 ? anybody else seeing this ??

madmax

[BLKMGK]

Hi BLKMGK,

seen the same thing here...I have the same problem...last working for me was 11702! I tried twice....everyting is fine including 11702, but 11703 I see this nasty crash and my hole desktop switches to that odd resolution...

Can't be just us 2 ? anybody else seeing this ??

madmax

I've corresponded with the Dev who made the changes in the file that's causing us issues. He's busy with real life right now but has agreed to have a look at the changes as soon as he's freed up.

For now I have two installs. One is 11702 copied to a "backup" directory. You can generate this by deleting all files in your main XBMC directory (make a backup of userdata) and doing an svn up -r 11702 (-r might be -revert). That will give you a Subversion pull for 11702, move in the userdata directory and do a ./build.sh NOUPDATE to get working again. To run the LATEST code (not tried today though) create a new XBMC directory, do the initial SVN pull like in the Wiki, but before building copy over the source file "XBMC/guilib/GraphicContext.cpp", then do ./build.sh NOUPDATE. See if XBMC runs.

This fixes the issue for me right now. I suspect that as soon as a change is made to XBMC/guilib/GraphicContext.cpp that has dependencies elsewhere we'll have problems but for now this is getting me by.

Curious - how does your hardware configuration compare to mine? If there are others running hardware like us that are NOT seeing issues it might be nice to try and figure out what is different. I'm on up to date Ubuntu 7.10, latest NVIDA release driver, 1920X1080 desktop rez (er whatever 1080i is - can't get a good "P" X config..)

[madmax]

ah, cool

i try your fix using the source code from 11702 with todays latest and report back here later...

I use 7.10 (up2date) and used envy to install nvidia drivers...V169.09

I run my this setup @ 1280x720, but get the same problem....

madmax

[BLKMGK]

I just updated to the latest as of this evening - working fine with my fix Once that file is copied over the SVN pull doesn't seem to touch it BTW so you may just need to do the copy nce. Pay attention to what files get updated though. Hopefully this works for you too!

I too have used ENVY to load the 169.09 drivers. Does it drop you to 720X5xx at the desktop? If you run at that point it should work fine but with crappy resolution. <shrug>

[madmax]

yes, i can confirm i pulled the latest a used the GraphicContext.cpp from rev. 11702 did a build.sh NOUPDATE and YES I get the latest e.g. DUAL core support and no odd crashes with resolution changes...

and yes, my hole desktop would switch to 720X5xx when the crash happend...

do you know if nvidias V169.09 is the latest ? I didn't try to update...

madmax

[BLKMGK]

169.09 is the latest. Sounds like we have the same issue, guess we'll just have to be patient for now. We've got a workaround so that helps at least...

[BLKMGK]

11820 APPEARS to have fixed this for me. I am updating to current right now to test that too. I was having compile issues so I blew away what I had and compiled all over again backing down to 11819 before realizing I needed a new package. After having gotten the new required package and seeing a clean compile I tried running without the old file I'd been using and it ran file - 11819 did not.

Okay, compiled clean and run 11828 appears to be fine, this bug looks to be stomped. Thanks!!