[Skiboot] [PATCH] astbmc: Handle failure to initialise raw flash

[Stewart Smith]

"Andrew Jeffery" <andrew at aj.id.au> writes:
> On Wed, 20 Mar 2019, at 18:58, Vasant Hegde wrote:
>> On 03/20/2019 11:46 AM, Andrew Jeffery wrote:
>> > Initialising raw flash lead to a dead assignment to rc. Check the return
>> > code and take the failure path as necessary. Both before and after the
>> > fix we see output along the lines of the following when flash_init()
>> > fails:
>> 
>> If flash_init() fails, then we are sure we are going to crash at later 
>> stage.
>> May be we should assert() if falsh_init() fails. But that will change 
>> current flow.
>> 
>> Between I think this should go to Stable # 6.2.x. I will grab this patch for 
>> next stable release.
>
> Stewart and I were debating whether it should go to stable. If you're
> happy to pick it up I'm fine with that; we had concluded that it was
> somewhat cosmetic as things stand now.

I've picked it up for master as of
6225d11924933f0541fd3b1a6192a7fcda279f1f.

The problem I mentioned on internal slack with Mambo and STB payloads
was actually coming from 8d8a9ca30453bc92977409966e6006f8c18f4be5 and
some poor bisecting / makefile dependencies. So I fixed it in merging
that and drank heavily as I remembered exactly how we work out where on
earth the payload is and how we jump to it.

I'd be deeply keen for us to simplify that code path and be very very
very predictable (maybe standardise on things being in the DT if we're
not loading from flash?). I think the biggest users of "kernel is
already in memory" are sim and BML.

-- 
Stewart Smith
OPAL Architect, IBM.