Wine-Bug: https://bugs.winehq.org//show_bug.cgi?id=49202

Signed-off-by: Fan WenJie <fanwj@mail.ustc.edu.cn>

The table size typically keeps increasing such that in some cases the
size we got is already too small by the time we try to get the table
content in the next iteration.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54328

err must be set if the last attempt to allocate the tables fails.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54328

And get rid of the hacky casts.
Signed-off-by: Gabriel Ivăncescu <gabrielopcode@gmail.com>

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54298

On windows, we get some test failures as un-expected thread
create debug events are sent.
Try to filter them out in a couple of places.
Signed-off-by: Eric Pouech <eric.pouech@gmail.com>

Current test in debugger expects that all breakpoint debug
events appear before all the exit thread events.
On Windows, it happens that sometimes the breakpoint and exit
thread events are intertwined (making the test fail).

So, this patch:
- merges the exception events loop and the exit thread events loop
  into a single loop.
- detects the unordered sequence (mark it broken as windows only)
- extends the test so that we check that all exit thread events
  are received.
- introduce next_event_filter helper to not return some unwanted
  events.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=47874Signed-off-by: Eric Pouech <eric.pouech@gmail.com>

There are no errors when running chr(-1) e.g. in utf-8 locale.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54003

Trying to improve upon previous attempt: 2486f5a0

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53456Signed-off-by: Eric Pouech <eric.pouech@gmail.com>

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54319

Patch originally by Rémi Bernon.

Exposes an alignment issue under WoW64.

When free-ing an Alloc-ed pointer.

wined3d: Write into a mapped BO when replacing the entire destination in wined3d_context_gl_copy_bo_address().

Instead of using glBufferSubData(). In practice, this means that we will either
discard and rename the BO (if it's still in use) or write directly into it (if
not).

This improves performance drastically on NVidia GL drivers.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53408

wined3d: Map the destination BO with WINED3D_MAP_DISCARD when replacing the entire destination in wined3d_context_gl_copy_bo_address().

Analogous to the Vulkan backend.

This is not particularly impactful, though, since it only affects buffer-buffer
copies in the case where ARB_copy_buffer is not supported.

And a previously created context is passed, which may later be destroyed
on its own, causing a double free.

init_argv0_dir has a special case for FreeBSD where one branch led to
an unitialized variable and undefined behavior. Initialize properly.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53761

The tests are actually racy, and the async operation is cancellable in
a very short period of time after its creation, but before it started
executing.

This will be hard to test precisely and it's not related to dinput tests
in any way. Let's remove this to fix some spurious Windows failures.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=54283

The RunningObjectTableImpl_EnumRunning codepath returns a list allocated
from combase MIDL_user_allocate, which uses msvcrt heap.

To avoid double free when the caller releases the object.

It will later be freed in EnumSTATDATA_Release with HeapFree, and the
copy code path may allocate the member with HeapAlloc.

To match combase allocations, as for instance we're freeing pointers
returned from InternalIrotRevoke, which are allocated using combase
MIDL_user_allocate.