Re: wined3d: Initialize map_desc in wined3d_surface_map

[Marvin]

Hi,

While running your changed tests on Windows, I think I found new failures.
Being a bot and all I'm not very good at pattern recognition, so I might be
wrong, but could you please double-check?
Full results can be found at
https://newtestbot.winehq.org/JobDetails.pl?Key=2785

Your paranoid android.


=== w864 (32 bit device) ===
device.c:5409: Test failed: pBits not cleared on double lock.
device.c:5410: Test failed: Pitch not cleared on double lock.
device: unhandled exception c0000005 at 742814C8

=== w864 (64 bit device) ===
device.c:5409: Test failed: pBits not cleared on double lock.
device.c:5410: Test failed: Pitch not cleared on double lock.
device: unhandled exception c0000005 at 000007F84EDD90EB

wined3d: Initialize map_desc in wined3d_surface_map

[Stefan Dösinger]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 2013-10-15 20:34, schrieb Marvin:
> === w864 (32 bit device) === device.c:5409: Test failed: pBits not
> cleared on double lock. device.c:5410: Test failed: Pitch not
> cleared on double lock. device: unhandled exception c0000005 at
> 742814C8
> I'm surprised that the testbot executes those tests. Did you run those
> tests on a real Windows machine? If you want I can test them for you
> on a variety of windows versions and GPUs here.

If the tests pass on a real windows, I'm tempted to call the Windows 8
behavior broken. It's either a change in Windows 8 itself, or the
driver that is used in this VM.

If you believe that this is a problem in the test VM rather than your
patch, you can mark the behavior broken with a construct like this:

ok(locked_rect.pBits == NULL || broken(locked_rect.pBits == (void
*)0xdeadbeef), ...)

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/
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=/llD
-----END PGP SIGNATURE-----

wined3d: Initialize map_desc in wined3d_surface_map

[Lasse Rasinen]

Stefan Dösinger <stefandoesinger(a)gmail.com> writes:
> Am 2013-10-15 20:34, schrieb Marvin:
> > === w864 (32 bit device) === device.c:5409: Test failed: pBits not
> > cleared on double lock. device.c:5410: Test failed: Pitch not
> > cleared on double lock. device: unhandled exception c0000005 at
> > 742814C8
> > I'm surprised that the testbot executes those tests. Did you run those
> > tests on a real Windows machine? If you want I can test them for you
> > on a variety of windows versions and GPUs here.

I can have access to a Windows machine earliest on Friday, and even then
it won't be Win8. So yes please.
-- 
Lasse Rasinen
lrasinen(a)iki.fi

wined3d: Initialize map_desc in wined3d_surface_map

[Stefan Dösinger]

> > > === w864 (32 bit device) === device.c:5409: Test failed: pBits not
> > > cleared on double lock. device.c:5410: Test failed: Pitch not
> > > cleared on double lock. device: unhandled exception c0000005 at
> > > 742814C8
> > > I've tested your tests on two of my Windows 7 machines, one with a Nvidia
> > > Geforce 650M, and another with an AMD X1600. Both behave just like the
> > > testbot. I guess this means your original implementation of just not
> > > touching the output locked_rect was correct all along.

Volume::Lock clears pBits and both pitches on a failed lock in all failure
cases (resource not lockable, misaligned box, already locked, otherwise
incorrect box). I haven't tested d3d8 yet. As far as I can see surfaces
keep locked_rect untouched in all error conditions. I also haven't tested
Windows XP yet because my Visual Studio 2012 generated .exe files don't run
there any more :-\ .

Cheers,
Stefan