On Thu, 23 Nov 2023 at 19:04, Sui Jingfeng <sui.jingfeng@xxxxxxxxx> wrote: > > Hi, > > > On 2023/11/23 16:08, Dmitry Baryshkov wrote: > >>> The host can not specify the > >>> DRM_BRIDGE_ATTACH_NO_CONNECTOR flag, it will cause a warning here. And > >>> it can not omit the flag (as otherwise the first bridge will create a > >>> connector, without consulting the second bridge). > >> The semantics of DRM_BRIDGE_ATTACH_NO_CONNECTOR flagare implement-defined, > > No, they are not. Semantics are pretty simple: do not create the > > drm_connector instance. Pass the flag to the next bridge in the chain. > > > >> for our case, I could just ignore it if their > >> don't have a signal(DT or ACPI) tell me that there are multiple bridges > >> in the chain. This depend on community's attitude. > > Ignoring a flag is a bad idea. > > > Can you also read the code in the bridge/lontium-lt8912.c please? > when flags == 0 is true, the lt8912 driver will just create > a drm_connector instance in the drm bridge drivers. The behavior > is similar with this patch in the perspective of spirit. > > And the most important thing is that no matter what the flag the upstream > port is passed, lt8912 just always pass the DRM_BRIDGE_ATTACH_NO_CONNECTOR > flags to the next bridge. Does this count as a kind of ignore? or > > This is to say that both the lt8912 and the tfp410 drm bridge drivers are > allowing create a drm_connector manually in drm bridge drivers. They didn't > being asked to move the drm_connector related code to display controller > driver. I don't know why I can't follow this way? This is called 'legacy'. > > Do you really read the code before do comments or I failed to understand something? -- With best wishes Dmitry
