[enhancement] [docs] How to write a driver that supports Mir

Bug #1193263 reported by Daniel Holbach
10
This bug affects 2 people
Affects Status Importance Assigned to Milestone
Mir
Triaged
Wishlist
Unassigned
mir (Ubuntu)
Triaged
Wishlist
Unassigned

Bug Description

It'd be good to document this. This is blocked on bug 1118909 though.

Changed in mir:
status: New → Triaged
importance: Undecided → Medium
kevin gunn (kgunn72)
Changed in mir:
importance: Medium → High
Revision history for this message
Daniel Holbach (dholbach) wrote :

Looks like the dependent bug is closed now. Do we have any information on this already or concepts we want to explain or code examples we can link to?

Revision history for this message
Daniel van Vugt (vanvugt) wrote :

I don't think bug 1118909 was adequately resolved to start on this one.

In particular, we don't have a framework for supporting vendor drivers yet. And if vendors were to wedge their code into Mir to make it work, we'd all have licensing problems. Partly because replacing the buffer management code involves GPL. We need to design a solution for this before it gets documented.

Revision history for this message
Alan Griffiths (alan-griffiths) wrote :

I agree with Daniel, we haven't even set the success criteria for such design work let alone started on it.

Revision history for this message
Alan Griffiths (alan-griffiths) wrote :

I guess I should clarify the above comments: they reflect concern about trying to document a moving target, not a reluctance to provide documentation.

At present there is no bright line in the Mir code that separates driver related code from the rest of Mir. The closest approximation that exists is libmirplatformgraphics.so - which is GPL code and consumes some C++ interfaces from libmirserver.so while implementing other C++ interfaces. Nothing is currently in place to ensure these interfaces remain stable.

Work has yet to be completed confirming the required licence, API and ABI stability requrements for driver support, then we need to implement and document a solution to these concerns.

summary: - [docs] How to write a driver that supports Mir
+ [feature] [docs] How to write a driver that supports Mir
tags: added: feature
summary: - [feature] [docs] How to write a driver that supports Mir
+ [enhancement] [docs] How to write a driver that supports Mir
tags: added: enhancement
removed: feature
Revision history for this message
kevin gunn (kgunn72) wrote :

i think what is needed is likely an update to our docs reflecting the use of the key integration tests to determine if a ubtunu-touch port of hw is ready for mir integration. we have a draft for this, just a matter of following up.

Revision history for this message
Alan Griffiths (alan-griffiths) wrote :

I think we need to clarify or possibly split this bug.

Mir has the concept of "platform" plugins to support different driver models. It comes with "mesa" and "android" platforms.

This bug could be about:

1. validating a driver against either the mesa or android platform; or,
2. writing a new "platform" for an drivers that support neither mesa nor android.

(or both)

Changed in mir:
importance: High → Wishlist
Revision history for this message
Michał Sawicz (saviq) wrote :

Syncing task from Mir.

Changed in mir (Ubuntu):
importance: Undecided → Wishlist
status: New → Triaged
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.