re4_tweaks relies on this function being hotpatchable [1].

[1]: https://github.com/nipkownix/re4_tweaks/blob/7cfcf5c4272f7d68d177661ffe84e9d1e0d2f383/dllmain/input.cpp#L903

SafeSEH is not applicable to architectures other than i386.

This fixes compiling with the clang ARM assembler, which cannot parse
".def @feat.00" since "@" is parsed as the start of a line comment.

Today, NtContinue() on ARM64 does not restore X16 and X17 from the
context.

This is because the values for X16 and X17 are overwritten when the
current thread returns to the "user mode" (PE side) via
__wine_syscall_dispatcher, which in turn uses them as scratch registers
for restoring SP and PC respectively.

We cannot avoid using scratch registers when restoring SP and PC.  This
is because ARMv8 does not have an unprivileged (EL0) instruction that
loads SP and PC from memory or non-GPR architectural state.

Fix this by making ARM64 __wine_syscall_dispatcher perform a full
context restore via raise(SIGUSR2) when NtContinue() is used.

Since raising a signal is quite expensive, it should be done only when
necessary. To achieve this, split the ARM64 syscall dispatcher's
returning behaviour into a fast path (that does not involve signals) and
a slow path (that involves signals):

- If CONTEXT_INTEGER is not set, the dispatcher takes the fast path:
  the X16 and X17 registers are clobbered as usual.

- If X16 == PC and X17 == SP, the dispatcher also takes the fast path:
  it can safely use X16 and X17 without corrupting the register values,
  since those two registers already have the desired values.

  This fast path is used in call_user_apc_dispatcher(),
  call_user_exception_dispatcher(), and call_init_thunk().

- Otherwise, the dispatcher takes the slow path: it raises SIGUSR2 and
  does full context restore in the signal handler.

Fixes: 88e33621

Signed-off-by: Nikolay Sivov <nsivov@codeweavers.com>

Fall through in create_link and allow SHGetFolderPathAndSubDirW to
create an empty folder in place if the alternatives are not available.

Creating a link back to $HOME makes a recursive directory structure,
which results in crashes in programs that naively search %USERPROFILE%
(e.g. Quicken).

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=55114

GL_MAP_WRITE_BIT does not mean that the buffer will be filled, unless an
INVALIDATE bit is explicitly set. The application is free to partially update
the buffer, even if it does not read from it.

Thanks to Aida Jonikienė for doing most of the debugging here.

Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=55045

Commit b1cc87cb ("msi: The line control has a height of exactly 2
device units.", 2007-12-16) open-coded the body of
(msi_)dialog_add_control() inside dialog_line_control() modulo the
height override.

Except the fixed height, line controls have nothing special compared to
other controls.  Thus, there is little merit in open-coding it.

Also, dialog_line_control() has bit-rotted over time; it already missed
a few changes that were done to any other controls.

Fix this by just using dialog_add_control().  Also, add a special case
logic just for line controls.

Fixes: 2c5bd492

Fixes: 76606eae
Fixes: 174b7b51

Today, CreateTextServices() sometimes triggers an assertion failure in
select_style().

When ME_MakeEditor() calls ME_MakeFirstParagraph(), the editor
(ME_TextEditor) is not in a fully initialized state.  For example, the
font cache (pFontCache) is not fully initialized, which sometimes makes
select_style() believe that the cache slots are fully occupied.

Fix this by delaying the call to wrap_marked_paras_dc() until the editor
is fully initialized.

Also, delay the call to ITextHost::TxReleaseDC() until after
wrap_marked_paras_dc(), since we need the device context a bit longer.

Fixes: b70eb32c

This lets ME_MakeEditor() reuse the device context throughout the editor
initialization process.