May 2026 - Wine-GitLab - list.winehq.org

[PATCH 0/1] MR10882: dsound: Replace device reference counting with a stop event.
by Jon Koops (＠jonkoops) May 13, 2026

May 13, 2026

Follow-up to !10735. With device sharing removed, each DirectSoundDevice has exactly one owner and no longer needs reference counting. Replace the ref field with a stop event that the mixer thread waits on alongside the existing sleep event using WaitForMultipleObjects, and simplify DirectSoundDevice_Release to free unconditionally. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10882

2 1

[PATCH v2 0/5] MR10779: winex11: Keyboard layouts rework, part 2
by Matteo Bruni (＠Mystral) May 13, 2026

May 13, 2026

-- v2: winex11: Store Xkb keyboard layout LANGID. winex11: Keep a list of every known Xkb keyboard layout. winex11: Move Xkb layouts initialization to a separate function. win32u: Simplify layout name generation for base keyboard layouts. win32u: Tweak a couple of comments. https://gitlab.winehq.org/wine/wine/-/merge_requests/10779

3 15

[PATCH 0/5] MR10806: wineandroid: switch ioctl handling to socket-based transport (part of !10569)
by Twaik Yont (＠twaik) May 13, 2026

May 13, 2026

Part of !10569. This series converts wineandroid ioctl handling to a socket-based transport integrated with the Android main thread looper. It introduces reply_fd plumbing across all ioctl handlers, switches dispatch to a UNIX SEQPACKET socket, and migrates all ioctls to the new transport. The legacy unixlib/ntoskrnl ioctl path is removed as part of this transition. The final commit simplifies error handling by returning ioctl errors directly, matching the expectations of the socket-based transport and removing the need for additional conversion layers. Overall, this prepares the driver for further changes by unifying ioctl handling around a single transport and execution model. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10806

2 6

[PATCH 0/1] MR10867: wine.inf: Create Start Menu items for most GUI apps included with WINE
by Andrea Faulds (＠hikari_no_yume) May 13, 2026

May 13, 2026

-- https://gitlab.winehq.org/wine/wine/-/merge_requests/10867

2 4

Re: [PATCH v9 0/1] MR10528: oleaut32: Add fast paths for VarAdd and VarCmp with integer operands.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

@jacek do you need more real world evidence to pick this one up? -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10528#note_139681

1 0

[PATCH 0/4] MR10857: Draft: oleaut32: Bound GetFuncDesc impltypes walk by stored cbSizeVft.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

ITypeInfoImpl_GetInternalDispatchFuncDesc walked impltypes whenever any impl type was registered, surfacing inherited IDispatch slots ahead of own funcs in the index space. That matches MIDL-built TKIND_DISPATCH typelibs (where cbSizeVft includes inherited slots and cFuncs is inflated to match) but mismatches ICreateTypeLib2-built ones where cbSizeVft only spans own funcs and GetFuncDesc should expose only those. Skip the walk when the stored cbSizeVft has no room beyond own_count * ptr_size, so GetFuncDesc on ICreateTypeLib2 typelibs hits own funcs at indices 0..cFuncs-1 and returns TYPE_E_ELEMENTNOTFOUND past the end, matching native. This is needs !10856 merged first -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10857

2 6

[PATCH 0/2] MR10856: oleaut32: Don't include IDispatch slots in stored cbSizeVft for TKIND_DISPATCH.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

ICreateTypeInfo2_fnLayOut starts cbSizeVft at 7*ptr_size for TKIND_DISPATCH to position own funcs after the IDispatch vtbl, then increments per added func. The own funcs receive correct oVft offsets that way, but the resulting stored cbSizeVft is too large by exactly the IDispatch vtbl size: GetTypeAttr later derives cFuncs as cbSizeVft/ptr_size and so reports own_count + 7 instead of own_count. Subtract the prefix at the end of LayOut so cFuncs matches native, both in-memory and after SaveAllChanges + LoadTypeLibEx. Public cbSizeVft is unchanged: GetTypeAttr already overrides it to sizeof(IDispatchVtbl). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10856

2 4

[PATCH 0/2] MR10855: oleaut32: Don't fake IDispatch inheritance for TKIND_DISPATCH without an impl type.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

GetTypeAttr unconditionally inflated cFuncs by the IDispatch vtbl size and GetRefTypeOfImplType(0) returned dispatch_href, even for typeinfos created via ICreateTypeLib2 without AddImplType. In that case dispatch_href is -1, so callers that walked GetRefTypeOfImplType chains followed an invalid href. Only inflate cFuncs when an impl type is registered, and only return dispatch_href when the typelib actually has an IDispatch reference. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10855

2 4

[PATCH 0/2] MR10854: oleaut32: Return TYPE_E_BADMODULEKIND from ITypeInfo::CreateInstance for non-coclass.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

Native VBScript / oleaut32 returns TYPE_E_BADMODULEKIND when CreateInstance is called on a typeinfo whose typekind is not TKIND_COCLASS (e.g. a TKIND_DISPATCH built via ICreateTypeLib2). Wine currently returns E_INVALIDARG. -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10854

2 4

[PATCH 0/2] MR10853: oleaut32: Zero ITypeInfo::AddressOfMember out-pointer on failure.
by Francis De Brabandere (＠francisdb) May 13, 2026

May 13, 2026

For a TKIND_DISPATCH typeinfo built via ICreateTypeLib2, AddressOfMember has no DLL backing and returns TYPE_E_BADMODULEKIND. Native sets the out-pointer to NULL on this path; Wine currently leaves it untouched. Match native behavior: clear *ppv at entry so callers see NULL when AddressOfMember returns an error path (e.g. TYPE_E_BADMODULEKIND for typeinfos with no DLL backing). -- https://gitlab.winehq.org/wine/wine/-/merge_requests/10853

3 9