On Thu, 26 Jul 2018, [email protected] wrote:
Why not now?
Problem I am seeing is that there aren't enough abstraction frameworks, and creating these might take years. I saw Linus argued against ARM developers who kept coming with their SoC-unique patch-sets, and told them to go home and come back with something more generic. This actually happened.
We have *not* seen this for packet accelerators for instance. I have advocated Linux Foundation to do this, they know the issue, it's something we all want to fix, but it needs customers to push OEMs to push SoC vendors to do this. It's a classical "everybody benefits in the long run, but short term there is no immediate benefit" kind of thing.
Your example of PCI is classic, I see lots of Open* innitiatives but they're typically quasi-open, the source is open, but the work is done behind closed doors and code is thrown over the fence.
To me it's amazing that Linux kernel has gotten as far as it has, and actually has a so-so working community around it, with lots of companies funding work going in there. Things could be a lot worse.
-- Mikael Abrahamsson email: [email protected] _______________________________________________ Cerowrt-devel mailing list [email protected] https://lists.bufferbloat.net/listinfo/cerowrt-devel
