[kraqh3d]

DownloadBuildDeps.bat should fetch it and put it in the correct place. (It calls scripts/*_d.bat which contains libexpat_d.bat and fontconfig_d.bat.)

[Nuka1195]

i reran the builddeps.bat and now i have that file in system.

[kraqh3d]

If you still crash, its likely because of the problem in the screen resolution enumeration. In the log, you'll see the addons being added, and the last entry will be some thread message.

[taxigps]

check if you have a libexpat.dll in the system dir which cpluff.dll requires.

I have libexpat.dll and cpluff.dll in xbmc/system folder. that not the problem. 29461 is ok, 29462 and after crash on my laptop.

[kraqh3d]

@CrystalP:
The crash is occurring well before Xbmc ever gets to CWinSystemWin32::UpdateResolutionsInternal().

Code:

20:53:45 T:5364 M:182104064   DEBUG: thread start, auto delete: 0
20:53:45 T:6024 M:181805056   DEBUG: Entering:CWinSystemWin32::CreateNewWindow
20:53:45 T:6024 M:181821440   DEBUG: Entering:CWinSystemWin32::CreateBlankWindows

I presume its the same thing, where the window handle is being created with size 0,0 on screen 0. I'm going to add some more debugging code to confirm this.

** edit **

Yup, same thing. CreateNewWindow is getting called with those odd parameters. Execution dies in CreateBlankWindows as I don't get a message about entering ResizeInternal.

Code:

21:23:13 T:3188 M:253796352   DEBUG: Entering:CWinSystemWin32DX::CreateNewWindow
21:23:13 T:3188 M:253792256   DEBUG: Entering:CWinSystemWin32::CreateNewWindow
21:23:13 T:3188 M:253779968   DEBUG: width=0,height=0,screen=0,fullscreen
21:23:13 T:3188 M:253542400   DEBUG: Entering:CWinSystemWin32::CreateBlankWindows

** edit **

I found something else. When first entering CreateBlankWindows, m_MonitorsInfo contains nothing so m_hBlankWindows.reserve(BlankWindowsCount); crashes. I added "if (BlankWindowsCount > 0)" and now execution gets into ResizeInternal but not very far. And I cant find any way that execution can get into UpdateResolutions.

[CrystalP]

OK kraqh3d, something is very wrong here. You're building the Debug (DirectX) configuration, right?

CreateNewWindow is indrectly called from g_Windowing.CreateNewWindow() at line 638 in Application.cpp, after g_Windowing.InitWindowSystem() at line 573, which, among other things, calls CWinSystemWin32::UpdateResolutionsInternal().

The call sequence goes like this:
CApplication::Create() Line 573
CWinSystemWin32::InitWindowSystem() Line 60
CWinSystemBase::InitWindowSystem() Line 47
CWinSystemWin32::UpdateResolutions() Line 392
which calls UpdateResolutionsInternal.
(lines are from current SVN files, but I don't think they changed)

I don't see any tests that could derail this execution flow.

From what you're writing, the screens are not enumerated and that's what needs to be fixed. If you're running a headless PC, that's something else

[kraqh3d]

Yeah, I was just about to take back that last comment. I'm currently tracing through InitWindowSystem(). CWindowBase::UpdateResolutions is executing. Compiling again

** edit **
Ok I am a complete dumbass. Debug entries weren't being logged because settings haven't been read yet so Xbmc didn;t know it should log the debug entries. I believe this is what you're looking for:

http://pastebin.com/DiaQsPUd

Gimme a few minutes. I think I'm loosing info where the DWORD is getting truncated into a WORD. Lemme check.

[CrystalP]

Thanks, that's what I needed and I think I know what's happening.
Can you try this: on line 491, try successively 1, 2, 3 as second parameter to EnumDisplayDevices(), like this:
EnumDisplayDevices(ddAdapter.DeviceName, 1, &ddMon, 0)
I think the value 1 will fix your problem.

Nuka: do you also have Intel integrated video chipset?

[kraqh3d]

I noticed the active flag wasn't passing... bypassing that check, xbmc starts up.

http://pastebin.com/tTZxqBNE

Compiling with that test for you now, with the active flag check restored.

** edit **
Using a value of 1 works!

http://pastebin.com/amiN0hic

[Nuka1195]

just following your progress, and yes the computer that errors is intel integrated 945 gfx. the computer i compile on is nvidia integrated 7050 gfx.

[CrystalP]

Thanks for the help in tracking it down.

kraqh3d, Nuka1195, see r30138.

[kraqh3d]

I went right to bed, but synced early this morning, built, and tested it out before heading to the office. Everything is good. (I got into the office late, but I couldn't wait until this evening )

And fyi, the visualisations still have the wrong <type> in description.xml.

[taxigps]

now crash windows at r30153

log:
http://pastebin.com/cUhV3His

[CrystalP]

taxigps: to sum up, 29461 ok, 29462 not ok, >= 30138 not ok?
debug log & HW info please.

[kimp93]

I got crash on r30150 as well.
For me, r29957 run fine.
http://pastebin.com/FUYbcD7Y