[RFC 0/5] rockchip_sfc: add support for Rockchip SFC

林鼎强 jon.lin at rock-chips.com
Fri Jun 4 15:42:18 CEST 2021

Hi Chris:

It's my honor to read about spi-rockchip-sfc driver from you, and your code has made a big difference to me.

Recently, we happen to be willing to submit the spi-mem rk sfc drivers, and we have tried some of them
in the internal process. Compared with the RK internal code, I have the following thoughts on your code:
1) Only support spi-nor, not support spi-nand
2) Only support rx_dual, tx_quad, not support tx_dual, tx_quad
3) Some of the code may be further simplified, such as "rockchip_sfc_register_all" 
4) Some of the logic may be further simplified, such as "rockchip_sfc_init" 
5) Doesn't support SFC ver4 ver5 althought witch is compatibled with lower version

In order to support all these features, I adjust the drivers on the basis of your code witch is attach to these mail, these
drivers have been test in linux 5.10 rk3568 board, and I'm still doing more strict tests.

So I suggest that we complete it together, or we may consider transferring it to me for submission If I have the honor.

I'm still unfamiliar with the community culture. I hope I didn't offend you. If I have, I'll say sorry in advance. 
Best wishes.

jon.lin at rock-chips.com
From: Kever Yang
Date: 2021-06-02 09:27
To: Chris Morgan
CC: u-boot; heiko.stuebner; vigneshr; jagan; sjg; Chris Morgan; 赵仪峰; 林鼎强
Subject: Re: [RFC 0/5] rockchip_sfc: add support for Rockchip SFC
Hi Chris,
On 2021/6/2 上午12:54, Chris Morgan wrote:
> On Tue, Jun 01, 2021 at 08:22:09PM +0800, Kever Yang wrote:
>> Add Yifeng from rockchip.
>> Hi Chris,
>>      First of all, I think you should remain the origin author info in the
>> signed-off.
> Okay, I can do that. Please note that since I submitted this I was
> asked to redo the upstream linux proposed driver to use the spi-mem
> framework. I think for now honestly I'd like to abandon this patch
> and resubmit a little later with one that is more or less the same
> (using the spi-mem framework) as the Linux driver.
This sounds better, then we can wait for driver with new framework.
- Kever
> https://patchwork.ozlabs.org/project/linux-mtd/patch/20210528170020.26219-2-macroalpha82@gmail.com/
>> Hi Yifeng,
>>      Please help to review this driver.
>> Thanks,
>> - Kever
>> On 2021/5/26 上午5:49, Chris Morgan wrote:
>>> From: Chris Morgan <macromorgan at hotmail.com>
>>> Requesting comments for a proposed patchset for adding the Rockchip
>>> serial flash controller to u-boot. The goal of these patches is to
>>> enable it for the Odroid Go Advance so that it may eventually boot
>>> exclusively from the SFC on mainline U-boot (I have tested this and
>>> it works).
>>> The specific help I need with this patch is:
>>> 1) I don't know the best way to upstream the XTX25F128B flash chip.
>>> This chip uses a continuation code for the manufacturer ID, however I
>>> cannot seem to find any way to actually read the continuation code.
>>> There is a risk of this driver, used as-is, to collide with another
>>> chip which has the same manufacturer ID with a different continuation
>>> code.
>>> 2) The Rockchip SFC driver itself (as it is mostly as-is from the BSP
>>> U-Boot sources) supports SPI NAND and chips of varying sizes, but my
>>> implementation only permits me to test with a single 128Mb flash chip.
>>> The driver itself does some checking on the bitlen in the routine
>>> rockchip_sfc_xfer() which is what is called for the dm_spi_ops.xfer.
>>> I'm not sure if there is a better way to do this. Additionally, I have
>>> to bit-shift the address written to the SFC as I suspect the value is
>>> meant to be left justified, but I never tested it further.
>>> Additionally, it might be worth mentioning but I noticed the Rockchip
>>> BROM will only boot the TPL/SPL off of the SFC if I write it to address
>>> 0x10000. This is not documented and different than the address looked
>>> at for SD card booting (512 * 64 = 0x8000 for SD Card booting). Also,
>>> like the SD card driver I can confirm that if DMA is enabled at the SPL
>>> stage A-TF seems to fail silently, then when Linux loads it hangs.
>>> There is an ifdef to force FIFO mode only in the SPL stage.
>>> Tested: Read (works)
>>> Write (works if you write to an erased sector)
>>> Erase (works)
>>> SPL Read (works if you edit the u-boot,spl-boot-order)
>>> Chris Morgan (5):
>>>     spi: rockchip_sfc: add support for Rockchip SFC
>>>     rockchip: px30: Add support for using SFC
>>>     rockchip: px30: add the serial flash controller
>>>     mtd: spi-nor-ids: Add XTX XT25F128B
>>>     rockchip: px30: add support for SFC for Odroid Go Advance
>>>    arch/arm/dts/px30.dtsi                     |  38 ++
>>>    arch/arm/dts/rk3326-odroid-go2-u-boot.dtsi |  10 +-
>>>    arch/arm/dts/rk3326-odroid-go2.dts         |  22 +
>>>    arch/arm/mach-rockchip/px30/px30.c         |  64 ++
>>>    drivers/mtd/spi/Kconfig                    |   6 +
>>>    drivers/mtd/spi/spi-nor-ids.c              |   8 +
>>>    drivers/spi/Kconfig                        |   8 +
>>>    drivers/spi/Makefile                       |   1 +
>>>    drivers/spi/rockchip_sfc.c                 | 652 +++++++++++++++++++++
>>>    9 files changed, 926 insertions(+), 1 deletion(-)
>>>    create mode 100644 drivers/spi/rockchip_sfc.c
-------------- next part --------------
A non-text attachment was scrubbed...
Name: spi-rockchip-sfc_modified_20210604.c
Type: application/octet-stream
Size: 16869 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20210604/0cbb168b/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: spi-rockchip-sfc.c
Type: application/octet-stream
Size: 21684 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20210604/0cbb168b/attachment-0001.obj>

More information about the U-Boot mailing list