On Mon, Aug 4, 2014 at 11:12 PM, Maxime Ripard < [email protected]> wrote:
> On Mon, Aug 04, 2014 at 02:14:13AM +0300, Simos Xenitellis wrote: > > Hi All, > > > > Nicolas (from ARMDevices.net) has conducted quite a few interviews with > > Chinese hardware companies around Shenzhen. > > There are several videos with Allwinner, such as > > > http://armdevices.net/2014/07/23/allwinner-64bit-armv8-processor-announced/ > > > > I think it would be a good opportunity to interview Allwinner about > issues > > with source code development. At > > https://plus.google.com/u/0/+charbax/posts/6uqUutxjQiw (see comments) > he is > > OK to either mail Allwinner or arrange to visit them for a video > interview. > > My preference would be an interview on camera and I believe it should be > > feasible. > > There may not be immediate results out of this, however it would be great > > to have some official response. > > > > What's needed is to describe to Nicolas what questions to ask. > > I am not familiar with all important questions that can be asked so it > > would be good to help add to the list, and explain to Nicolas so that he > > can discuss them at ease. > > > > Here is my attempt with a question. Feel free to correct me. > > > > 1. The Linux kernel holds now the hardware description of SoCs in a data > > format called Device Tree (DT). There are more details about DT at > > > http://events.linuxfoundation.org/sites/events/files/slides/petazzoni-device-tree-dummies.pdf > > For example, here is the DT file for the Rockchip 3188, > > > https://github.com/torvalds/linux/blob/master/arch/arm/boot/dts/rk3188.dtsi > > > > While many manufacturers have provided DT files for their products, there > > are none yet from Allwinner. > > Here we can ask for Allwinner to provide them for all SoCs, or we can ask > > specific details that will help to produce those files. Do we have a > > preference? > > The only thing we need to write those are a good technical > documentation and board schematics. Allwinner is only really involved > in the former. > For the A20, there is this document for EVB schematics, https://github.com/OLIMEX/OLINUXINO/blob/master/HARDWARE/A20-PDFs/A20_PAD_STD_V1_1.rar?raw=true and http://dl.linux-sunxi.org/A20/A20%20User%20Manual%202013-03-22.pdf for technical documentation. Are these what you are referring to? If we want to make a proper question/request, we need a table with the SoCs and what's missing for each one of them. Can someone make such a table? > > 2. Source code in mainline Linux. We explain why it is important, etc. > > > > There is a list of items (mainly drivers) at > > http://linux-sunxi.org/Linux_mainlining_effort > > Are these drivers without any released source code? Has the source code > > been released but it needs lots of work to add to mainline? What should > we > > ask Allwinner to do? > > I'm not sure to get what you mean, but the drivers that got merged > were either: > - rewritten from scratch (GPIO, clocks, SPI, etc.) > - an adaptation to the Allwinner SoCs of already existing drivers > (GMAC, I2C, SATA, USB, etc.) > - Allwinner source code cleaned up (EMAC, MMC, etc.) > This is very specific and useful. > > As for the things Allwinner should do, they should move to using > standard Linux API. They improved a lot that aspect when developping > the A23 BSP, and hopefully will continue to do so. > > With Linux API you mean that Allwinner should use much as possible from what is offered at https://www.kernel.org/doc/htmldocs/kernel-api/ like the available data manipulation functions rather than re-implementing them? Can you give some specific examples and perhaps somewhat quantify the extent of the issue? What went well with the A23 BSP? Simos > Maxime > > -- > Maxime Ripard, Free Electrons > Embedded Linux, Kernel and Android engineering > http://free-electrons.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 [email protected]. For more options, visit https://groups.google.com/d/optout.
