On Sat, Sep 10, 2016 at 04:40:47PM -0700, Eric Nelson wrote:
>On 09/06/2016 07:15 AM, Tom Rini wrote:
>> On Tue, Sep 06, 2016 at 07:00:56AM -0700, Eric Nelson wrote:
>>> On 09/06/2016 06:40 AM, Tom Rini wrote:
>>>> On Fri, Sep 02, 2016 at 10:53:58PM +0200, Petr Kulhavy wrote:
>>>> Would it be possible to implement having the next stage also be sent via
>>>> imx_usb_loader? ie there's examples today of doing u-boot.imx + kernel
>>>> + initrd via imx_usb_loader, so what would be needed for SPL +
>>>> u-boot.img (+ kenrel + initrd) via imx_usb_loader? Thanks!
>>> SPL+u-boot.img could be bundled into a single image through the
>>> use of plugins which would require:
>>> - updates to mkimage to support plugins, and
>>> - Makefile updates to produce a third output (u-boot.imx?), and
>>> - an update to SPL startup on i.MX to check for the plugin flag
>>> and return to the boot ROM after startup (instead of loading
>>> U-Boot) if set.
>It appears that the mkimage support is present in a patch against v2015.04:
>Peng, can you up-stream this?
I would like see your points on this. I tried to uptream plugin before, but
Since more people would like to see this in, do you agree to let plugin
>>> Troy implemented the key bits back in 2012, and I provided
>>> some links here:
>> Ah yes, this. I really would like to see this come in as I think it'll
>> be required to really drop the old style u-boot.imx binaries for cases
>> like factory programming. As well as the use cases outlined before too
>> about supporting multiple boards more easily.
>I'd also like to see this get pulled into main-line, and I just ran across
>another reason for supporting plugins.
>The i.MX7 LPSR mode resumes from a "power-off" state by going through
>the boot ROM, and it's up to the code loaded by the boot ROM to recognize
>the condition, take DRAM out of self-refresh and jump back into IRAM:
>Some additional commentary is in this post on NXP community:
>For this use case, SPL will be too heavyweight and plugin support is
U-Boot mailing list