On Wed Aug 31 14:00:14 2022 +0000, Giovanni Mascellani wrote:
> For the future, please do code cleaning stuff (like renaming variables)
> in dedicated commits, so that the "meaty" commits are easier to read.
Okay,
I agree with this policy, now that merge requests have to be approved as a whole.
--
https://gitlab.winehq.org/wine/vkd3d/-/merge_requests/7#note_7348
Run C++ global/static destructors during DLL_PROCESS_DETACH while other
other dllimport functions they may want to call are still viable. While
windows imposes many restrictions on what may be done in such destructors
(given that the run inside the loader lock), there are lots of legal and
useful kernel32 functions like DestroyCriticalSection, DeleteAtom, TlsFree,
etc that are both useful and legal. Currently this does not work for builtin
modules because all the Win32 structures are discarded well before
NtUnmapViewOfSection finally does the dlllose.
Even for a winelib .dll.so module, it woudl be preferable for destructors
to execute during process_detach (before wine tears down the MODREF
and detaches dependant dlls), rather than be left until after the last
NtUnmapViewOfSection (when we finally reach dlclose)
Therefore, winegcc now always uses the DllMainCRTStartup entry point
unless you specify your own --entry=func. Previously it did this only for
PE modules using msvcrt. Making this default consistent matches cl.exe,
which also always defaults to _DllMainCRTStartup unless overridden by /entry:foo
https://docs.microsoft.com/en-us/cpp/build/reference/entry-entry-point-symb…
The ELF version of winecrt0.a now provides a DllMainCRTStartup which,
per the Itanium ABI that is in practice what is used by gcc and clang,
performs this this destruction by calling __cxa_finalize(&__dso_handle)..
This libc function is required to be idempotent, so it's OK that dlclose
still calls it again later (there will just be no further work to do).
Multiple calls to __cxa_finalize shall not result in calling termination
function entries multiple times; the implementation may either remove
entries or mark them finished.
https://itanium-cxx-abi.github.io/cxx-abi/abi.html#dso-dtor-runtime-api
This has two main effects; it moves ELF destructors earlier (before imports
are unmapped), and it moves them inside the Nt loader lock. Being earlier
was the intended goal, and moving them inside the lock seems fine. Any Win32
API calls in destructors are just being subjected to the same lock-hiearchy
rules as usual on windows (MSVC also runs destructors from DllMainCrtStartup)
https://docs.microsoft.com/en-us/cpp/build/run-time-library-behavior?view=m…
And any purely-ELF destructors that happen to also run earlier should never
call functions exported from wine (and thus don't care about ntdll's locks).
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/752
MSVC's delayimp.lib does not actually free delayload dependencies.
winecrt0's attempt to do so from ELF __attribute__((destructor))
is unnecessary and potentially harmful:
- When triggered naturally via LdrUnloadDll, this leads to recursive calls
to FreeLibrary, violating free_lib_count and missing DLL_PROCESS_DETACH
- when triggered by glibc's _dl_fini (at process exit), it leads to
use-after-free of the TEB (GetCurrentThreadID after the main thread is no longer Win32)
via FreeLibrary -> LdrLdrUnloadDll -> RtlEnterCriticalSection( &loader_section )
- double-free of the library itself, since the DLL_PROCESS_DETACH has
already been handled by LdrShutdownProcess
- Race against wineserver sending a SIGKILL from process_killed,
since all Win32 threads of the process have exited
Wine-Bug: https://bugs.winehq.org/show_bug.cgi?id=53032
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/750