[PATCH v7 0/4] MR359: mmdevapi: copy ContainerID property from audio driver if available
26 Jul
2022
26 Jul
'22
8:30 a.m.
Some games with support for the haptic feedback and speaker features of the Sony DualSense
controller select the controller's audio output by filtering on the ContainerId
IMMDevice property to find one that matches the controller's HID's. This MR allows
this information to be accessible from the IMMDevice's property store when the audio
driver provides it (none for now, but I intend to add that feature to `winepulse.drv` and
maybe `winealsa.drv`)
This is made specific to containerId rather than supporting VT_CLSID on every property
because Wine currently stores VT_BLOBs directly in the registry value, which does not
allow us to safely disambiguate between VT_CLSID and VT_BLOB values when reading from
registry.
--
v7: mmdevapi: Invalidate ContainerID of unavailable audio devices.
https://gitlab.winehq.org/wine/wine/-/merge_requests/359
26 Jul
26 Jul
8:30 a.m.
New subject: [PATCH v7 1/4] mmdevapi: support VT_CLSID for containerId property in MMDevice_SetPropValue.
From: Claire Girka <claire(a)sitedethib.com>
CLSID is special-cased to this property because we can't safely differentiate
an encoded VT_CLSID from an encoded VT_BLOB.
---
dlls/mmdevapi/devenum.c | 12 ++++++++++++
1 file changed, 12 insertions(+)
diff --git a/dlls/mmdevapi/devenum.c b/dlls/mmdevapi/devenum.c
index bfeb3f3ecd8..48f4c736217 100644
--- a/dlls/mmdevapi/devenum.c
+++ b/dlls/mmdevapi/devenum.c
@@ -225,6 +225,18 @@ static HRESULT MMDevice_SetPropValue(const GUID *devguid, DWORD flow,
REFPROPERT
ret = RegSetValueExW(regkey, buffer, 0, REG_SZ, (const BYTE*)pv->pwszVal,
sizeof(WCHAR)*(1+lstrlenW(pv->pwszVal)));
break;
}
+ case VT_CLSID:
+ {
+ if (IsEqualPropertyKey(*key, DEVPKEY_Device_ContainerId)) {
+ BYTE value[24] = { VT_CLSID, 0, 0, 0, 1, 0, 0, 0 };
+ memcpy(value + 8, pv->puuid, sizeof(GUID));
+
+ ret = RegSetValueExW(regkey, buffer, 0, REG_BINARY, (const BYTE*)value,
24);
+ break;
+ }
+ /* If it's not containerId, fall through the default unsupported case as
we can't
+ ensure it will be decoded as CLSID. */
+ }
default:
ret = 0;
FIXME("Unhandled type %u\n", pv->vt);
--
GitLab
https://gitlab.winehq.org/wine/wine/-/merge_requests/359
8:30 a.m.
New subject: [PATCH v7 2/4] mmdevapi: decode ContainerId property to CLSID in MMDevice_GetPropValue.
From: Claire Girka <claire(a)sitedethib.com>
---
dlls/mmdevapi/devenum.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/dlls/mmdevapi/devenum.c b/dlls/mmdevapi/devenum.c
index 48f4c736217..10bf76057ee 100644
--- a/dlls/mmdevapi/devenum.c
+++ b/dlls/mmdevapi/devenum.c
@@ -189,6 +189,21 @@ static HRESULT MMDevice_GetPropValue(const GUID *devguid, DWORD flow,
REFPROPERT
break;
}
RegCloseKey(regkey);
+
+ /* Special case ContainerID as CLSID */
+ if(pv->vt == VT_BLOB && pv->blob.pBlobData &&
pv->blob.cbSize == 24 && pv->blob.pBlobData[0] == VT_CLSID &&
IsEqualPropertyKey(*key, DEVPKEY_Device_ContainerId)) {
+ GUID *guid = CoTaskMemAlloc(sizeof(GUID));
+ if (!guid) {
+ PropVariantClear(pv);
+ hr = E_OUTOFMEMORY;
+ } else {
+ memcpy(guid, pv->blob.pBlobData + 8, sizeof(GUID));
+ CoTaskMemFree(pv->blob.pBlobData);
+ pv->vt = VT_CLSID;
+ pv->puuid = guid;
+ }
+ }
+
return hr;
}
--
GitLab
https://gitlab.winehq.org/wine/wine/-/merge_requests/359
8:30 a.m.
New subject: [PATCH v7 3/4] mmdevapi: copy ContainerID from audio driver if available.
From: Claire Girka <claire(a)sitedethib.com>
Some games with support for the haptic feedback and speaker features of the
Sony DualSense controller select the controller's audio output by filtering on
the ContainerId IMMDevice property to find one that matches the controller's
HID's.
---
dlls/mmdevapi/devenum.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/dlls/mmdevapi/devenum.c b/dlls/mmdevapi/devenum.c
index 10bf76057ee..bf0fc4728d1 100644
--- a/dlls/mmdevapi/devenum.c
+++ b/dlls/mmdevapi/devenum.c
@@ -391,6 +391,8 @@ static MMDevice *MMDevice_Create(WCHAR *name, GUID *id, EDataFlow
flow, DWORD st
MMDevice_SetPropValue(id, flow, (const
PROPERTYKEY*)&DEVPKEY_DeviceInterface_FriendlyName, &pv);
MMDevice_SetPropValue(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_DeviceDesc, &pv);
+ set_driver_prop_value(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_ContainerId);
+
pv.pwszVal = guidstr;
MMDevice_SetPropValue(id, flow, &deviceinterface_key, &pv);
--
GitLab
https://gitlab.winehq.org/wine/wine/-/merge_requests/359
8:30 a.m.
New subject: [PATCH v7 4/4] mmdevapi: Invalidate ContainerID of unavailable audio devices.
From: Claire Girka <claire(a)sitedethib.com>
Some games, like Deathloop, enumerate all unavailable audio devices, including
unavailable ones, and stop at the first one matching the ContainerID.
Depending on how ContainerIDs end up being attributed, an active device may end
up sharing a ContainerID with an unplugged device that was previously plugged
in the same physical port. Depending on the audio backend, the same audio
device plugged in the same port may be seen as a different audio device by Wine
depending on things like in which order devices were discovered.
In those cases, a game might find an inactive matching device before the active
one, and thus fail to open the audio output.
By invalidating the ContainerID of unavailable devices, we this issue can be
avoided.
---
dlls/mmdevapi/devenum.c | 29 ++++++++++++++++++++++++++++-
1 file changed, 28 insertions(+), 1 deletion(-)
diff --git a/dlls/mmdevapi/devenum.c b/dlls/mmdevapi/devenum.c
index bf0fc4728d1..09fb3c44875 100644
--- a/dlls/mmdevapi/devenum.c
+++ b/dlls/mmdevapi/devenum.c
@@ -207,6 +207,26 @@ static HRESULT MMDevice_GetPropValue(const GUID *devguid, DWORD flow,
REFPROPERT
return hr;
}
+static HRESULT MMDevice_DeletePropValue(const GUID *devguid, DWORD flow, REFPROPERTYKEY
key)
+{
+ WCHAR buffer[80];
+ const GUID *id = &key->fmtid;
+ HRESULT hr;
+ HKEY regkey;
+ LONG ret;
+
+ hr = MMDevPropStore_OpenPropKey(devguid, flow, ®key);
+ if (FAILED(hr))
+ return hr;
+ wsprintfW( buffer, propkey_formatW, id->Data1, id->Data2, id->Data3,
+ id->Data4[0], id->Data4[1], id->Data4[2], id->Data4[3],
+ id->Data4[4], id->Data4[5], id->Data4[6], id->Data4[7],
key->pid );
+ ret = RegDeleteValueW(regkey, buffer);
+ RegCloseKey(regkey);
+ TRACE("Deleting %s returned %lu\n", debugstr_w(buffer), ret);
+ return hr;
+}
+
static HRESULT MMDevice_SetPropValue(const GUID *devguid, DWORD flow, REFPROPERTYKEY key,
REFPROPVARIANT pv)
{
WCHAR buffer[80];
@@ -391,7 +411,14 @@ static MMDevice *MMDevice_Create(WCHAR *name, GUID *id, EDataFlow
flow, DWORD st
MMDevice_SetPropValue(id, flow, (const
PROPERTYKEY*)&DEVPKEY_DeviceInterface_FriendlyName, &pv);
MMDevice_SetPropValue(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_DeviceDesc, &pv);
- set_driver_prop_value(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_ContainerId);
+ /* The mechanism we use to attribute Container IDs is not very robust and
could end up making
+ an active device share a ContainerID with inactive devices, and some games
enumerate even
+ inactive devices, stopping at the first matching one.
+ To avoid issues, invalidate the ContainerID of devices that are not
present. */
+ if (state & DEVICE_STATE_ACTIVE)
+ set_driver_prop_value(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_ContainerId);
+ else if (state & DEVICE_STATE_NOTPRESENT)
+ MMDevice_DeletePropValue(id, flow, (const
PROPERTYKEY*)&DEVPKEY_Device_ContainerId);
pv.pwszVal = guidstr;
MMDevice_SetPropValue(id, flow, &deviceinterface_key, &pv);
--
GitLab
https://gitlab.winehq.org/wine/wine/-/merge_requests/359
9 Mar
9 Mar
1:14 a.m.
This seems nifty, any updates? @huw
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/359#note_26478
10 Mar
10 Mar
10:01 a.m.
The issue here is that Wine isn't currently saving binary data in the correct format
(it looks like the data should be prefixed by the appropriate VT_ type and possibly a
count value). So the problem becomes how to fix the format while updating any existing
saved data.
Special casing one specific property (as this MR does) isn't the right way to do
this.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/359#note_26622
10:23 a.m.
(it looks like the data should be prefixed by the
appropriate VT\_ type and possibly a count value).
Or more likely just the binary dump of the whole PROPVARIANT.
--
https://gitlab.winehq.org/wine/wine/-/merge_requests/359#note_26624
420
days inactive
647
days old
7 comments
4 participants
participants (4)
-
Claire (＠ClearlyClaire) -
Claire Girka
-
Huw Davies (＠huw)
-
Olivia/Isabelle Mahadevan (＠doubledge-spec)