Re: Executing wine over make segfaults.

[Marcus Meissner]

On Mon, Nov 06, 2006 at 08:46:40AM +0100, Pavel Troller wrote:
> > This seems to be Wine-related problem (but not neccessary Wine bug) because 
> > everything else works fine with 2.6.18.1 kernel; I'm not alone who have 
> > strange problems with Wine under 2.6.18 kernel.
>
> No, you are not. I've posted a very similar problem with wine & 2.6.18
> a couple of weeks ago.
>   Just to recall, my experiences with the problem:
> - Not every windows app causes wine to segfault. There are fairly complex,
> networked apps, which work flawlessly (DC++), other ones, much more
> simple, cause wine to crash, like wine's itself "rundll.exe setupapi.dll"
> when it tries to create a fresh .wine directory. Please see my post to
> wine-devel dated Oct 11, with subject "wine segfaulting" for details.
> There is even a strace snippet.
> - It crashes almost identically on both i386 and x86_64 architectures, with
> the same applications.
> - As demonstrated in 1), it does so even in the .wine directory build
> process, which means that no DLL overrides or other user-broken things
> can be involved.
> - No user/system settings can change this. Tried to increase various
> ulimits, manipulate (disable) exec-shield etc. Just the only solution
> known to me is to boot 2.6.17 or less, which I cannot normally because
> of other features I currently need from 2.6.18.
> - I'm testing wine on the system it was built on (with 2.6.18). It should
> ensure maximum compatibility with its kernel (I'm using live kernel
> includes for <linux/*> and <asm/*>). However, it works when transferred
> to another system running 2.6.17.
>
> I'm ready to perform some debugging; however, currently I don't know where
> to start.

Can you get backtraces in gdb ?

I am using a 2.6.18 based openSUSE 10.2 Beta1 kernel and Wine works fine there.
(AMD64 however).

Ciao, Marcus