On Wed, Mar 5, 2014 at 12:19 PM, Dmitriy B. <rzk...@gmail.com> wrote: >> That is not in the category of "what Linaro does". The code written at >> Linaro is open and open lists discuss it. > > > How do I discuss it when I cant even run it? > >> >> It is the vendor pushing >> those NDAs, not Linaro. Joining Linaro is not going to automatically >> change Allwinner's position on keeping Cedar source secret. Do we >> even know for sure if Allwinner has the ability to open source it and >> that Cedar is free from a third party licensing mess. > > > Cedar is described well enough right now at linux-sunxi and if Linaro wants > - they will try to use that. If not - we will see another implementation of > unusable kernelside V4L2 dec/enc with userspace part that is distributed by > NDAs (look at Coda/Samsung MFC/whatever is kicking there right now. For MFC > it took couple months to get it running well enough with XBMC for Exynos4). > >> BTW, Hisilicon has never been open source friendly so why did you buy >> their stuff? > > > I'm happy with my android there. But, its just the fact that HiSilicon spend > time and money on getting mainline and 3.10.x android kernels that bothers > me. Why cant I use it? No, lets say different, for exactly *who* it was made > opensource? For a nice line in press release "we are now opensource"? > "Not usable" - that is the picture of Linaro that I mean here as a HiSilicon > device owner. Sorry if it sounds too raw, but thats the facts. :) Hope it > will be different for Allwinner. > >> By any way you slice it, I believe Allwinner joining Linaro will >> improve their software offerings, not hurt them. > > > Actually, If you had experience working with Linaro, could you please hint > there that we could enjoy CSI drivers for A1x/A20, since you had/have a
The CSI driver in sunxi is an ancient version. Recent support for the CSI unit is in their Android kernel. Nobody has ported this new code into the sunxi kernel. I started to do the port but there has been a lot of divergence between the two kernels and the port is more of a rewrite. There are more complications to this since Android has modified the VL4 interface. AFAIK there isn't any missing info preventing this port. My current understanding of the Android CSI driver is that I don't know of any obvious problems with it. Some of the image sensors don't work right but those are problems in the image sensor drivers. What I initially thought were problems in the CSI code are actually problems in the h.264 code. In general not all features of the CSI unit are supported. For example the CSI unit supports Transport Stream input but I've never seen a driver or hardware for it. Unclear if anyone has written a driver for it yet. What isn't working right is h.264 encoding. I can't get Cedar to compress more than around 10:1 and x86 software encoders can do 50:1 pr more. Not being able to achieve high compression ratios means that 720P consumes too much bandwidth to be used on the Internet. I have been unable to get it to generate streams less than 10Mb/s while my desktop has no problem generating 500Kb/s streams. My next step is to put together a bunch of sample encodings to demonstrate to Allwinner what it should be doing. If they are given examples showing exactly what they need to do I suspect they will fix the h.264 blob. > project in that field? As I said - everything simple was already done, even > ARM PSCI is working with help from ARM employee. But things like kernelside > Cedar, CSI and other multimedia stuff is missing > http://linux-sunxi.org/Linux_mainlining_effort#Left_to_be_done It would be > nice for Linaro to start from there, and not with rewriting what is already > done. > > -- > You received this message because you are subscribed to the Google Groups > "linux-sunxi" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to linux-sunxi+unsubscr...@googlegroups.com. > For more options, visit https://groups.google.com/groups/opt_out. -- Jon Smirl jonsm...@gmail.com -- You received this message because you are subscribed to the Google Groups "linux-sunxi" group. To unsubscribe from this group and stop receiving emails from it, send an email to linux-sunxi+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
