From: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
To: Rob Kramer <rob@teegee.cloud>, Jack Mitchell <ml@embed.me.uk>
Cc: u-boot@lists.denx.de
Subject: Re: DDR timing for vendor board
Date: Fri, 9 Dec 2022 19:25:55 +0100 [thread overview]
Message-ID: <fc0a9051-e4a3-7c2e-b91a-c2ea820a2e85@gmail.com> (raw)
In-Reply-To: <d6746a8a-fe27-4e03-89e3-878c61efd5e9@teegee.cloud>
Hi Rob,
On 12/9/22 12:10, Rob Kramer wrote:
> Hi Jack,
>
> Thanks for your suggestion, I hadn't thought of using just the rkbin ddr file together with u-boot SPL. My biggest objection was the rkbin miniloader that makes assumptions on partition layout.
>
> It seems to work, but now I'm in for some DT pain :)
I don't know anything about Rockchip but from experience with other
SoC's and DDR2/3 controllers I would say that the reason for this
closed-source loader is likely some proprietary dynamic tuning algorithm
for some memory controller parameters (e.g. DQS) to adapt to
manufacturing tolerances of the board. If that's the case, the static
setting of memory size and timings via DT won't be enough to run the
board stable and Jack's approach would be more reliable.
>
> Cheers,
>
> Rob
>
> 9 Dec 2022 00:49:30 Jack Mitchell <ml@embed.me.uk>:
>
>> On 08/12/2022 04:01, Rob Kramer wrote:
>>> Hi all,
>>>
>>> I have a RK3288 board from a Chinese display vendor that came with the
>>> usual giant Rockchip tarball that they patched here and there to make
>>> the board work. It seems to be based on a rk3288-evb, since that is what
>>> they patched in the kernel. The kernel is a 4.4 kernel with Android
>>> stuff in it (i.e. fiq-debugger) and a large amount of Rockchip patches,
>>> u-boot is 2017.09, with rk patches.
>>>
>>> It turns out that u-boot TPL/SPL won't boot because the DDR timings are
>>> incorrect, and the Chinese vendor uses the Rockchip
>>> rk3288_ddr_400MHz_v1.09.bin loader. I'm using u-boot 2022.01 for now,
>>> and I've tried to naively modify the timing in
>>> arch/arm/dts/rk3288-evb.dts, but it doesn't work at all, with varying
>>> errors on boot.
>>>
>>> The Rockchip loader provides the following info when booting:
>>>
>>> In
>>> Channel a: DDR3 400MHz
>>> Bus Width=32 Col=10 Bank=8 Row=15 CS=1 Die Bus-Width=16 Size=1024MB
>>> Memory OK
>>> OUT
>>> Boot1 Release Time: Apr 11 2018 10:32:58, version: 2.36
>>> ChipType = 0x8, 232
>>>
>>> I've tried various DDR3 (not LPDDR3) settings from other boards, for
>>> example for a Firefly (666 MHz DDR3):
>>>
>>> U-Boot TPL 2022.01 (Jan 10 2022 - 18:46:34)
>>> Col detect error
>>> DRAM init failed!
>>> ### ERROR ### Please RESET the board ###
>>>
>>> It was expected that 666MHz doesn't work, but if I just change the
>>> frequency in the dts, that also fails (error -22).
>>>
>>> How can I support the DDR for this board? I can't even see what the ID
>>> on the chips is, because the heatsink is blocking sight and seems to be
>>> attached with some sort of thermal glue.
>>>
>>> Is there a way to read back the DDR timings (phy-timing, sdram-params)
>>> from the kernel/SoC on a board that is booted using the proprietary loader?
>>>
>>> Cheers!
>>>
>>> Rob
>>>
>>
>> Hi Rob,
>>
>> You could be in for a world of hurt here as Rockchip are very poor at
>> supporting different DDR init configurations in u-boot. In the past with
>> awkward boards I've used the Rockchip DDR init blob as the TPL binary
>> for u-boot then skip the DRAM init in the u-boot SPL.
>>
>> From my travels in this area I've found that single channel RAM boards
>> are particularly difficult to get working as the majority if not all
>> mainlined board use dual channel RAM.
>>
>> Sorry I can't be more help, but trying to use the DDR blob is a good
>> starting point to get you going on mainline.
>>
>> Good Luck!
>>
>> --
>> Jack Mitchell, Consultant
>> https://www.tuxable.co.uk
--
- Daniel
prev parent reply other threads:[~2022-12-09 18:26 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-12-08 4:01 DDR timing for vendor board Rob Kramer
2022-12-08 16:48 ` Jack Mitchell
2022-12-09 11:10 ` Rob Kramer
2022-12-09 18:25 ` Daniel Schwierzeck [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=fc0a9051-e4a3-7c2e-b91a-c2ea820a2e85@gmail.com \
--to=daniel.schwierzeck@gmail.com \
--cc=ml@embed.me.uk \
--cc=rob@teegee.cloud \
--cc=u-boot@lists.denx.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox