OpenChrome VIA KMS Has A Goal For This Summer

Last week a status update was issued concerning the plans for OpenChrome kernel mode-setting (KMS) support for VIA hardware under Linux. There's finally a goal set for a release this summer.

James Simmons, the independent developer that's near single-handedly been bringing kernel mode-setting support to VIA Technologies hardware via the OpenChrome project, issued a new project status last Thursday.

Simmons has been working on VIA KMS support along with in-kernel memory management (GEM/TTM) for more than one year, but it's been going at a slow pace with him being basically the only active VIA developer. Meanwhile, VIA Technologies really isn't doing much after they effectively abandoned their open-source / Linux strategy and their hardware really isn't going anywhere.

Last June was when James Simmons said the OpenChrome KMS support was nearly done, but there's still a lot of random work left before he'll push for mainline inclusion.

The xf86-video-openchrome 0.2.905 driver was released earlier this month as the first update in a while, but it's still KMS-less. The status update from James indicates he has a goal to do the next release as v0.3.0, which is currently the "kms_branch" of the driver. He wants this DDX driver released by the first of June.

The targets for this OpenChrome DDX driver release is RandR support as well as KMS support, i.e. so the DDX driver can play well when the mode-setting is being done in the kernel and not by the X.Org driver.

As far as when the KMS bits will land in the tree of Linus Torvalds, James says, "After the xorg driver starts funneling into distros we can then think about merging the drm-openchrome branch, but that is a further down the road." So James sounds like he plans to wait to push for mainline inclusion of the VIA KMS/DRM driver bits until the OpenChrome 0.3.0 X.Org driver has been around for a while... This is because if an older UMS-only OpenChrome X.Org driver hits a new VIA-KMS-enabled kernel there would be problems. (Or he could just push the new code into staging and disable it by default...) But this will mean that the VIA KMS support probably won't be mainline until at least the Linux 3.6 or 3.7 kernel time-frame; until then, those wanting VIA KMS will need to pull his drm-openchrome kernel.

As far as what this lone developer has been working on lately, he's been working on various KMS and KMS bits. He's also addressed outstanding bugs so that the new xf86-video-modesetting driver can work with the VIA KMS kernel. "Last night the generic modesetting driver was running and I managed to even use RandR with no major problems. Also the hardware cursor ass implemented for the kernel tree so the modesetting driver no longer has problem with the cursor code. With these improvements I will be able to very soon implement KMS support in the openchrome xorg driver."

On the user-space mode-setting side in OpenChrome, James fixed up several bugs as well ranging from cursor problems on VX900 laptops to IGA mapping to RandR fixes.