[U-Boot] [PATCH 1/2] arm: imx6: Add DDR3 calibration code for MX6 Q/D/DL

From: Eric Nelson <eric@nelint.com>
To: u-boot@lists.denx.de
Subject: [U-Boot] [PATCH 1/2] arm: imx6: Add DDR3 calibration code for MX6 Q/D/DL
Date: Thu, 17 Dec 2015 08:48:01 -0700	[thread overview]
Message-ID: <5672D931.9060505@nelint.com> (raw)
In-Reply-To: <CAJ+vNU17DYcRe+eWoY7gV8J11tzevhiHh_GMN8kUg3G8yUGSKQ@mail.gmail.com>

Hi Tim,

On 12/17/2015 08:39 AM, Tim Harvey wrote:
> On Wed, Dec 16, 2015 at 7:00 AM, Eric Nelson <eric@nelint.com> wrote:
>> Hi Marek,
>>
>> On 12/16/2015 07:40 AM, Marek Vasut wrote:
>>> Add DDR3 calibration code for i.MX6Q, i.MX6D and i.MX6DL. This code
>>> fine-tunes the behavior of the MMDC controller in order to improve
>>> the signal integrity and memory stability.
>>>
>>
>> I'm glad to see that others are interested in this.
>>
>> I've been working on something similar, but struggling to have time
>> to finish and clean it up:
>>         https://github.com/ericnelsonaz/u-boot/tree/memcal-pass1
>>
>> My aim is/was a bit different though, and aims to be a replacement for
>> the DDR stress tool, which is cumbersome to use.
> 
> Eric,
> 
> I would love to see a series posted that adds IMX6 MMDC calibration
> and stress test to U-Boot. 

Will do. Some other demands on my time will push this into next
week (and Christmas) though.

I'll try to get an RFC version out before the holiday though, (after
reviewing Marek's patch).

> I agree the Freescale code is very difficult to use and I don't trust
> what its doing.

It's always tough to trust code that you can't see and discuss.

Because it's a pain to run, I also suspect that many (most) boards
are running with calibration gathered from a small set of boards,
and I've seen lots of board->board variation.

Something that can be run using imx_usb can make the process of
gathering data much easier.

The board->board variations hint that Marek's on the right track and
that calibration really should be done at run-time.

> I'm currently seeing the Freescale DDR3 stress test
> fail on 8Gb density memory and I am not convinced its not a problem
> with their code (yet I haven't had time to pick through it with a fine
> toothed comb and compare with how I setup the MMDC in U-Boot).
> 

I haven't seen any issues with 8GiB densities, but have only tested
on a small set of board designs (primarily Nitrogen6_max).

Regards,

Eric