Hacker News new | ask | show | jobs
by jessikat 1697 days ago
There's a difference between drivers being upstreamed to the Linux kernel, and a datasheet being available. The source code is not the datasheet, and if there are bugs in the driver, where is the datasheet to compare the implementation to?

When it comes to hardware, source code availability is not sufficient.

Other OS platforms can have different driver architectures, different kernel ABIs and other platform quirks in comparison to Linux, and having to reverse engineer a driver for another platform from a Linux kernel driver is not very open. And then there are licensing considerations to be taken into account as well.

Linux is not the only opensource operating system in existence :-/ And for a project like Haiku, we don't have the people-power to reverse engineer Linux drivers, and again, the question of the GPL also applies when our kernel & drivers are MIT licensed.

1 comments

As a software maintainer/distributor, I don’t think we at Red Hat are the ones responsible to publish data sheets. That’s the job of the hardware manufacturer/vendor. When they decide to only offer it in non-public ways but allow us to help with getting the drivers upstream AND open source, it is a deal you might not like, but I, as a pragmatist can accept. We’re not the only Linux distributor working this way. Ubuntu/Canonical and SUSE don’t work that different from us on this. Again. You may not like it, and I understand that, but it has lead to making Linux a first class OS on Lenovo with an upstream first goal. I like it that way, even if it can become even better and more open.
> As a software maintainer/distributor, I don’t think we at Red Hat are the ones responsible to publish data sheets.

I don't think the parent commenter suggested you are responsible for publishing the material. However, it would be great to have vocal support from distro maintainers to encourage hardware manufacturers to publish them in order to provide better support across all systems. If only, to have more transparency about what parts of "supported hardware" are actually supported by the distro drivers, and what specific datasheet references could be requested through other venues (such as directly asking the manufacturer).

Of course not. It just gets more to the point of the original article. Whilst publishing and upstreaming driver source code for Linux is awesome, it's still only a half-measure in the overall picture of things being truly open, and we should try to demand more from hardware manufacturers. But we don't really have much sway in those regards, and it's disappointing as an OS-enthusiast, and a Haiku developer.