Idea on Vista lockup on minimize

[canofworms]

Many Vista users are experiencing a problem where minimizing the frontend would cause it to lock up when restoring. I believe I might have found the cause, in a change in .28

• The game no longer uses 100% CPU when minimised. (An earlier fix in v3.6.20.1 was meant to do this, but was incomplete and did not work on all systems.)

Where it says game, I am presuming it means WA as a whole, and not just the actual play screen.

In order to test my theory, would it be possible for me to obtain a build of .28 without this fix so I can check this, or am I going down the wrong street/not possible to do that.

Thanks anyway :)

[Squirminator2k]

I can't honestly see the connection here, Can.

[canofworms]

I can't honestly see the connection here, Can.

Just the idea from a potential link I saw here (minimize change, minimize crash).

[CyberShadow]

Can you at least argument your idea?
Besides, that would mean 3.6.26.5 has to work and .28 not.

Also, post a minimize error log.

[canofworms]

Can you at least argument your idea?
Besides, that would mean 3.6.26.5 has to work and .28 not.

Also, post a minimize error log.

My argument is that WA is failing to start executing its code again once it has restored due to a bug somewhere. There is no error log as it just hangs (see a post elsewhere in here).

I am testing 26 right now. Edit: Made no difference, oh well, it was worth a try :( Sorry for any time wasted

Edit Number 2 (and this is an important one)
I went backwards and tested every version of WA I have access to, that is every beta patch and 3.0 with Silkworm 4.12.

I can report the following:
WA 3.0 (Vanilla-European) with Silkworm 4.12: WORKS!
WA 3.5beta1: I dont have this. Anyone have a copy?
WA 3.5beta2: Not really works but has a reasonable attempt - displays a grayish screen with a frame at the top for the logo, click where a button belongs and it crashes
WA 3.5beta3pre1 (The One That Was Not Released): As above
WA 3.6.x: All 3.6.x builds do not work.

I hope it helps, it would be great if someone could give me 3.5beta1 to test with.

And CyberShadow, many thanks for setting up worms2d.info, it is an excellent resource :D

Random bootnote: It seems nvidia wrote a decent driver for Vista, no slow frontend :D

[CyberShadow]

Does W:A use 100% of a CPU core when it freezes like that?

If so, use this (http://forum.team17.co.uk/attachment.php?attachmentid=30787&d=1192296057) to get a minidump for freezes.

[canofworms]

Does W:A use 100% of a CPU core when it freezes like that?

If so, use this (http://forum.team17.co.uk/attachment.php?attachmentid=30787&d=1192296057) to get a minidump for freezes.

It does not, uses 0% CPU, and roughly 23MB of memory. Edit: I didn't make it actually, I started after I posted this, and I really couldn't be bothered to wait 5mins for the dump since you dont really want it according to that.

[canofworms]

Okay, I have done two things. The first thing I did was run the crashdump WormKit module in the thread about the Czech thing. That produced crash dumps either on start, or just before minimize with an exception from the debugger on control C. I do not know where this happens because I cant run the debugger side by side to WA (lack of monitors).

The second did was run the minidump you advised earlier. I ran the minidump through Visual C++ 2008 (like I did with the previous), and the debugger broken on:
_KiFastSystemCallRet@0:
77749A94 ret
This is in ntdll.dll. I am finding it very strange that it stopped on a system call. This appears to be the result of WA calling something, then kernel32.dll sleeping then NtDelayExecution in ntdll.dll. The FastSystemCall may be a result of that, but I do not know as I am generally a Linux developer/one for my own OS and not NT.

Sorry if that made no sense, like I said I dont do NT. Also I lack the WA symbols too, which doesnt help. Regardless I have attached the dumps to this post. (I zipped it in 7zip with bzip2 compression, so if you cant open it, try in that).

Update: This appears to happen on non-NVIDIA cards too (happened on an ATI)