04-17-2016, 08:17 PM
(This post was last modified: 04-17-2016, 08:19 PM by MichaelMeissner.)
(04-17-2016, 01:18 PM)peterz Wrote: What about signing Non Disclosure Agreement? Would be that an option? Ummm, in general, you can't sign a NDA and contribute to an open source project based on information in the NDA, since your code would violate the NDA. Of course, it really depends on the exact NDA.
Time limited NDA's that are used before a chip is shipping and you agree not to put in code to public sources until until it is shipping is perhaps doable. But the Allwinner chips are shipping now, and any time limit would mean users would have to wait for a fix until the time limit expires.
For example, I work for an employer adding GCC support for their processors, and I have signed various NDAs for what is in future machines. However, I can't contribute code for that new machine until it has been announced. On the GCC side, there are windows where you can contribute new features, and windows where you concentrate on fixing bugs (right now, we are in such a window), so my team has to concentrate on getting things announced and released in a timely fashion. This way NDAs can work with open source projects. But there has to be a timeout to allow things to be contributed.
(04-17-2016, 08:17 PM)MichaelMeissner Wrote: (04-17-2016, 01:18 PM)peterz Wrote: What about signing Non Disclosure Agreement? Would be that an option? Ummm, in general, you can't sign a NDA and contribute to an open source project based on information in the NDA, since your code would violate the NDA. Of course, it really depends on the exact NDA.
Time limited NDA's that are used before a chip is shipping and you agree not to put in code to public sources until until it is shipping is perhaps doable. But the Allwinner chips are shipping now, and any time limit would mean users would have to wait for a fix until the time limit expires.
For example, I work for an employer adding GCC support for their processors, and I have signed various NDAs for what is in future machines. However, I can't contribute code for that new machine until it has been announced. On the GCC side, there are windows where you can contribute new features, and windows where you concentrate on fixing bugs (right now, we are in such a window), so my team has to concentrate on getting things announced and released in a timely fashion. This way NDAs can work with open source projects. But there has to be a timeout to allow things to be contributed. I know this is open source, but perhaps desperate times needs desperate solutions. 
04-18-2016, 06:44 AM
(This post was last modified: 04-18-2016, 06:44 AM by tkaiser.)
(04-18-2016, 05:45 AM)peterz Wrote: I know this is open source, but perhaps desperate times needs desperate solutions.
No, it's not open source and that's the whole problem: https://github.com/ssvb/u-boot-sunxi/com...92fb8f4058
We need the DRAM initialisation code licensed as GPLv2 and then everybody would be happy again. Releasing something under the GPL is quite the opposite of signing NDAs.
04-18-2016, 10:45 AM
(This post was last modified: 04-18-2016, 11:01 AM by tllim.)
(04-17-2016, 01:18 PM)peterz Wrote: (04-17-2016, 09:14 AM)tllim Wrote: (04-17-2016, 09:06 AM)rahlquist Wrote: Why are they so unwilling? Are they hiding something?I don't get that.
They worry about their China competitor access to their info. What about signing Non Disclosure Agreement? Would be that an option?
This is an option but I don't think open source community will agree.
(04-18-2016, 06:44 AM)tkaiser Wrote: (04-18-2016, 05:45 AM)peterz Wrote: I know this is open source, but perhaps desperate times needs desperate solutions.
No, it's not open source and that's the whole problem: https://github.com/ssvb/u-boot-sunxi/com...92fb8f4058
We need the DRAM initialisation code licensed as GPLv2 and then everybody would be happy again. Releasing something under the GPL is quite the opposite of signing NDAs. I have spent a lot of time on this boot0 topic and already make 4 trips to China on this discussion with Allwinner. Hopefully will see a good result soon, keep my finger cross.
(04-18-2016, 10:45 AM)tllim Wrote: (04-17-2016, 01:18 PM)peterz Wrote: (04-17-2016, 09:14 AM)tllim Wrote: (04-17-2016, 09:06 AM)rahlquist Wrote: Why are they so unwilling? Are they hiding something?I don't get that.
They worry about their China competitor access to their info. What about signing Non Disclosure Agreement? Would be that an option?
This is an option but I don't think open source community will agree.
This is not an option as you should know. It's about getting the libdram source code licensed in an approriate way so that it can be used together with other GPL licensed code.
The real reason why Allwinner protects this boot0 BLOB is that they not only do DRAM initialization there but also check the SoC ID to do marketing differentiation between 3 otherwise identical SoCs: A64, R18 and H64 most likely only differ by a single bit that gets checked by boot0 (at least that's what happens with A83T, R58 and H8).
(04-19-2016, 02:15 PM)tkaiser Wrote: (04-18-2016, 10:45 AM)tllim Wrote: (04-17-2016, 01:18 PM)peterz Wrote: (04-17-2016, 09:14 AM)tllim Wrote: (04-17-2016, 09:06 AM)rahlquist Wrote: Why are they so unwilling? Are they hiding something?I don't get that.
They worry about their China competitor access to their info. What about signing Non Disclosure Agreement? Would be that an option?
This is an option but I don't think open source community will agree.
This is not an option as you should know. It's about getting the libdram source code licensed in an approriate way so that it can be used together with other GPL licensed code.
The real reason why Allwinner protects this boot0 BLOB is that they not only do DRAM initialization there but also check the SoC ID to do marketing differentiation between 3 otherwise identical SoCs: A64, R18 and H64 most likely only differ by a single bit that gets checked by boot0 (at least that's what happens with A83T, R58 and H8).
Your assumption about SoC ID protection may not stand.
|
