Zack's Kernel News

Status of HGA Framebuffer Driver

The HGA Framebuffer driver is no longer maintained. Roland Kletzing confirmed that the official maintainer, Ferenc Bakonyi, has not had the relevant hardware since 2001. Thus, the MAINTAINERS entry is completely out of date and Roland has posted a patch to remove it.

Fourth-Party Drivers?

Michael Buesch has been designing and building his own hardware devices and recently asked whether he should submit drivers for that hardware or just keep them as a separate patch. The overwhelming response was, "Send in the drivers!" Apparently, as long as Michael – or anyone else – is willing to document how folks can build the hardware for themselves, the drivers for that hardware will be a welcome addition to the kernel. With that kind of encouragement, hopefully we'll start seeing a variety of unique hardware drivers appear.

Removing Firmware from the Kernel Tree

David Woodhouse wants to remove all third-party firmware out of the kernel source tree. David is in favor of letting the kernel load arbitrary blobs of firmware code (provided the resulting binary can be distributed legally under its license), but he doesn't think such things belong in the kernel. He's working on creating a separate git tree for all of them.

A number of replies dissented from this. Most folks were in favor of isolating firmware into a single location, but taking it out of the tree seemed too much because it created issues of competing firmware packages and other requirements that some folks don't want to tackle.

The most vocal opponent of David's idea was David S. Miller, who has been fighting against this idea for a while, specifically regarding the tg3 driver, which he felt needed to keep its firmware in the main tree. David Miller said that if distributions like Debian wanted to avoid any binary-only data in their kernel, it was up to them to write and maintain the necessary patches out-of-tree.

David Woodhouse responded that he'd already taken steps to ensure that even without removing the firmware, his changes would still be useful, but he did feel it would be better to take them out entirely. At this point, folks turned the discussions into a technical consideration about how the idea might be implemented. Whether some of the ideas were intended as actual suggestions or just as ruminations on what might be possible if such a thing were desired was unclear. What was clear is that a lot of folks had a variety of opinions, with many different reasons for them. In terms of what might happen with David Woodhouse's patches, I'd expect some kind of compromise. Removing the firmware is something that free software purists prefer, but it also might be something preferable for practical, legal reasons. In that case, folks like David Miller could lose the battle, but hopefully without too much inconvenience being forced on them.