[Letux-kernel] [Lenny400] JZ4730/Minibook framebuffer driver updates

H. Nikolaus Schaller hns at goldelico.com
Sun Sep 10 19:31:44 CEST 2017

Hi Paul,

> Am 10.09.2017 um 19:18 schrieb Paul Boddie <paul at boddie.org.uk>:
> On Sunday 10. September 2017 18.50.29 H. Nikolaus Schaller wrote:
>> Fine! I haven't applied yet but there is 4.13.1 soon so I will do and
>> provide a complete build asap.
> I should try and merge from an updated upstream.
> [...]
>> Both commands create SD cards that look good when mounting and inspecting
>> elsewhere...
>> The only thing I don't know is if the jz4730 can handle big SD cards. I.e.
>> bigger than 2 GB?
> David has installed a Debian root filesystem on a 8GB card, and I think it 
> works OK.


> Although the Minibook is quite old, I'm not sure whether it would be 
> its own limitations or those of the kernel that would introduce card size 
> limits. And given that the Ben NanoNote seems to use 8GB microSD cards without 
> problems, I would imagine that the SoC family was capable enough at that time.

Well, I think the kernel itself does not introduce special card size limitations.
And for ARM it works with 256GB cards (I have one on a Letux Cortex 8 board).

Let's think about where limitations could be...

1. the total card size may be limited by file system but FAT and ext can handle big file systems
2. the raw MMC driver treats the card as single big file with partitions - but there I am not aware of limits
3. the partition manager might ot be able to handle big partitions - but this works well on ARM - why should MIPS be different?
4. there might be a limit in the MMC controller inside the JZ4730 and the raw MMC driver that it can't send big block addresses automatically.
   It would only a problem if the MMC controller has e.g. a 16 or 24 bit register for block address and there is a "read block" command
   and everything else is done by DMA. Then, an old controller can never send "extended" addresses going beyond the card's capabilities
   to the card.

Another topic of SD, SDHC, SDXC cards is that they allow higher speed and draw more energy than provided.
Since this must be negotiated with the card first, we might limit it by some special mmc DT attributes.

AFAIK even SDXC cards can be run in the slowest and lowest energy mode. So we we are certainly
not able to access them in 96MBit mode, but they should work...

>> But in any caseI have some still working low-capacity SD cards... The
>> smallest one with 128 MB :)
> For some people these are like gold, so you might have a high-margin card 
> retailing business in the future if you have enough of them. :-)
>> So we are close to be able to set up a test scenario.
> Great! I haven't really tried to test anything yet.


More information about the Letux-kernel mailing list