May 2008 - Wine-devel - list.winehq.org

winetest issue on w2k and below
by Paul Vriens 29 May '08

29 May '08

Hi, Lately I've been seeing an issue with winetest on Windows whereby the kernel32_test.exe file can't be deleted after the test. On NT4 I've also seen popups during the test with some remark about not being able to initialize user32.dll. This is when the process test is run for kernel32. Anyone else seeing this? Or better yet, anyone know the cause? -- Cheers, Paul.

2 2

Re: advpack: Fix buffer sizes for possibly quoted strings (try 2, including testcase)
by James Hawkins 29 May '08

29 May '08

On Thu, May 29, 2008 at 9:20 AM, Michael Karcher <wine(a)mkarcher.dialup.fu-berlin.de> wrote: > Increase buffer size to make quoted strings fit. This fixes the problem > Dan described in > http://www.winehq.org/pipermail/wine-devel/2008-May/065954.html which > is caused by quoted 'HKLM' not fitting into prefix. This patch does, on > request, *not* fix the bigger picture issue that made it possible for > this bug to slip for such a long time, which is missing error checking > on SetupGetStringFieldW calls in get_dest_dir. Valid .inf files should, > of course, not exercise the missing error check, so on valid .inf > files, this patch is sufficient for correct behaviour. > > In the testcase with quoted 'HKLM', SetupGetStringFieldW fails because > of unsufficient buffer size, which is *ignore* by the current code. As > from the previous destination parsed in the patch the local automatic > variable still contains the unquoted HKLM and by luck does not get > overwritten between invocations of get_dest_dir, the testcase passes > nevertheless except if compiled by gutsy's gcc 4.1 without optimization > which puts the local variable to a stack location that does not survive > between different invocations of get_dest_dir. > > The attached change to the testcase "pollutes" the local variable prefix > with HKCU, so the bug is always exposed and not just by bad luck. > --- > dlls/advpack/advpack.c | 10 +++++----- > 1 files changed, 5 insertions(+), 5 deletions(-) > You must've attached the wrong patch because there is no test case. Also, don't make a change to the existing tests. Please write a completely new test that exposes just the one bug you're trying to fix. They're called 'unit' tests for a reason. -- James Hawkins

1 0

Re: advpack: Fix buffer sizes for possibly quoted strings
by James Hawkins 29 May '08

29 May '08

On Thu, May 29, 2008 at 8:58 AM, Michael Karcher <wine(a)mkarcher.dialup.fu-berlin.de> wrote: > Increase buffer size to make quoted strings fit. This fixes the problem > Dan described in > http://www.winehq.org/pipermail/wine-devel/2008-May/065954.html which > is caused by quoted 'HKLM' not fitting into prefix. This patch does, on > request, *not* fix the bigger picture issue that made it possible for > this bug to slip for such a long time, which is missing error checking > on SetupGetStringFieldW calls in get_dest_dir. Valid .inf files should, > of course, not exercise the missing error check, so on valid .inf > files, this patch is sufficient for correct behaviour. > > In the testcase with quoted 'HKLM', SetupGetStringFieldW fails because > of unsufficient buffer size, which is *ignore* by the current code. As > from the previous destination parsed in the patch the local automatic > variable still contains the unquoted HKLM and by luck does not get > overwritten between invocations of get_dest_dir, the testcase passes > nevertheless except if compiled by gutsy's gcc 4.1 without optimization > which puts the local variable to a stack location that does not survive > between different invocations of get_dest_dir. > --- > dlls/advpack/advpack.c | 10 +++++----- > 1 files changed, 5 insertions(+), 5 deletions(-) > Add a test case please. -- James Hawkins

1 0

AppDB: More information about the test environment needed
by Groeschel, Volker 29 May '08

29 May '08

I just found two conflicting reports for Age of empires: http://appdb.winehq.org/objectManager.php?sClass=version&iId=10326 Both are on Ubuntu 8.04 and 1.0-rc1. One is Silver, one is Garbage. But it is just one example. My suggestion is to ask for more information about the test environment. To specify it I think it would be helpfull to know the grafics chip/card and the used driver. With the new free drivers around the block it could be even more helpfull to have this information. The same ATI card e.g. could use Radeon, RadeonHD or the Catalyst driver. The same is true for Nvidia with nvidia, nv and Nouveau driver. I often read that Wine developers recommend nvidia, but I actually have good experience with radeon. It is quite hard to get real informaltion on this topic.

2 1

Re: [1/2] tests: Add functions to make it possible to handle Windows misbehaviors that we don't want to reproduce in Wine.
by Alexandre Julliard 29 May '08

29 May '08

Francois Gouget <fgouget(a)free.fr> writes: > This patch introduces the buggy(), deprecated() and strict_wine > functions I mentioned in: > http://www.winehq.org/pipermail/wine-devel/2008-April/065322.html I like the idea, but I don't think this should copy the todo mechanism, as it doesn't scale well. For todo that's OK because the number of todos is supposed to remain small, and get smaller as Wine improves; but strict checking is something we'll want more and more of, so ultimately we'll have them all over the place. It gets even worse if we want to be strict for other platforms too. So I would suggest to not add a strict(platform), but make buggy() and deprecated() always do strict checking except when platform is "windows". If we need to have a buggy() check succeed on Wine for some reason it can always be wrapped in a todo_wine. -- Alexandre Julliard julliard(a)winehq.org

1 0

2k is the default version?
by Austin English 28 May '08

28 May '08

I was testing installing Virtualbox today and got an error that Windows XP is the minimum requirement. Funny I thought...we're emulating XP, it must want SP 2 or something strange. Fired up winecfg, and lo and behold, Windows 2000 is the default OS again. Thought I might be going crazy or botched my local tree, but checking out on a fresh tree shows the same thing. When/why did this happen? -Austin

4 7

wine video codec demo
by Dan Kegel 28 May '08

28 May '08

I put together a little script demonstrating how to install and use various video codecs and use them in a test application (media player classic). To try it out, do wget http://kegel.com/wine/winetricks wget http://kegel.com/wine/codecdemo sh codecdemo (after of course reading both scripts to make sure they don't do something evil.) To make this demo script easier to write, I added quicktime72 and mpc verbs to winetricks. (I also added a vlc verb, but that wasn't so useful as a system codec test, since vlc has its own embedded codecs, and doesn't use the system's.) The demo shows that Quicktime works more or less, but .wmv doesn't work (even with the codecs); it probably needs either native quartz.dll or Maarten's VMR9 implementation (which he is busily finishing up now; it seems to be good at finding video card driver bugs). Next step: Dirac.

1 1

Re: [ws2_32/tests] Fix crash on win98
by Kai Blin 28 May '08

28 May '08

On Wednesday 28 May 2008 10:08:49 Paul Vriens wrote: > Hi, > > As pi_size was uninitialized the value turned out to be the cause of this > crash. > > I don't know where that limit ( 2*sizeof(WSAPROTOCOL_INFOA) ) comes from > but tests show it has to be less then that on win98. Whoops, thanks for the catch. I had that initialized to 0 at one point in my test program, but it seems like I dropped it while reworking the tests. Cheers, Kai -- Kai Blin WorldForge developer http://www.worldforge.org/ Wine developer http://wiki.winehq.org/KaiBlin Samba team member http://www.samba.org/samba/team/ -- Will code for cotton.

1 0

advpack.c must be compiled with optimization?
by Dan Kegel 28 May '08

28 May '08

Oddly, if I compile the wine tree without -O2, the following test fails: ../../../wine advpack_test.exe.so advpack.c advpack.c:441: Test failed: Expected C:\Program Files, got C:\ advpack.c:443: Test failed: Expected size 17, got 4 make: *** [advpack.ok] Error 2 Compiling advpack/advpack.c with at least -O makes the problem go away.

3 3

Why I think it's worth fixing Valgrind warnings
by Dan Kegel 28 May '08

28 May '08

If you're wondering whether it's worth fixing Valgrind warnings (after all, they are just warnings, right?): First, some of the valgrind errors that show up when running the test suite might be real bugs. That might be enough reason right there. Second, clearing out the valgrind errors that show up in the test suite make it a lot easier to find Wine bugs when running real apps on wine on valgrind. (The lower the background noise level, the easier it is to see the real problem.) Third, if valgrind works well on wine, windows developers might start coming to Wine just to run Valgrind on their apps to find app bugs. (Don't laugh! It could happen!) This is why I hired John Reiser to improve Valgrind/Wine support, why I've been fixing a few valgrind warnings lately, and why I think it's important for other people to start doing it, too. That said, it's important when fixing Valgrind errors to really understand the code you're changing, lest you paper over a bug with another bug, or make things worse. (Here would be a good place for me to thank James for his code reviews :-) - Dan

4 5