2 ms·
This is very welcome. ARM has tended to be somewhat difficult to port to - vendors have multiple versions of the instruction set out there depending on device
by zdw 16y ago
This is very welcome.
ARM has tended to be somewhat difficult to port to - vendors have multiple versions of the instruction set out there depending on device size/power requirements, with different floating point units (NEON, VFP) etc.
The result is we end up with situations like OpenEmbedded supporting multiple kernel trees in it's build engine just to take care of the wide range of hardware it supports.
Hopefully this will make the situation somewhat better.
- mycroftiv 16y agoWhat exactly is the "this" you are referring to? This article is pretty much an official acknowledgment that a problem exists and needs an investment of developer resources. The situation won't get better until/unless stakeholders make an investment of resources. Clearly someone is going to have to pay for a lot of hours of time spent by developers who are competent to work on arm arch stuff in a "global" way. I'm not too familiar with the details of the various multi-stakeholder alliances and how they get their funding and decide who to hire to do what, but it sounds like the Linux Foundation (or whoever) needs to support a few more developers with a job description more focused on the kernel itself rather than just relying on the companies contributing what they did to get specific devices ready for shipping.
- rbanffy 16y agoAnything that increases functionality/LoC in the kernel is good. In any complex software project - and the Linux kernel is about as complex as sanity and present technology will allow - you have to do periodical codebase clean-ups. This is one case where the pain to maintain is forcing a cleanup. And, as always, it's discussed openly, so, whoever depends on it doesn't get surprised by the next release.