[U-Boot-Users] [DNX#2007012542000021] [PATCH] Abort booting if the DTB version is incomp [...]

[Jerry Van Baren]

Wolfgang Denk wrote:
> Dear Timur,
> 
> in message <45F853ED.2060005 at freescale.com> you wrote:
>> I don't want to be annoying, but if you're going to apply this patch, I think it needs to 
> 
> Unfortunately there is a difference between the things I want to  do,
> the things I am capable of doing, the things I can do, and the things
> I'm actually doing.
> 
>> be applied now.  There's a patch for the device tree compiler (DTC) that updates it to 
>> version 17, which claims to be compatible with V16.  Soon, we're going to have V3, V16, 
>> and V16 DTBs flying around, so U-Boot needs to make a stand on what it considers 
>> compatible and what it doesn't.
> 
> If I'm realistic, I have to admit that I won't  find  much  time  for
> merging patches in the next days or maybe even weeks. Sorry.
> 
> Maybe we can find a custodian for the device tree related things?
> 
> Best regards,
> 
> Wolfgang Denk

Hi Timur, Wolfgang,

If I understand this, currently there is _no_ version checking and bootm 
will mysteriously fail if it is fed a version 3 blob and just as 
mysteriously work if fed a v16 (or v17) blob.

Version 17 is 100% compatible with version 16, it adds a new field in 
the header that is reserved (0x00000000) in v16 and sets the 
"comp_version" field to 17 instead of 16 ("last_comp_version" remains at 
16).

Version checking is good.  Having said that, if the patch is or is not 
applied, I don't see how anything changes materially from where we are 
right now, other than it will detect and abort if a v3 blob is passed in 
(which is good).  Currently u-boot will accept and properly process a 
v16 or v17 blob (theoretically - I have not had a chance to run a v17 
blob, but I've looked quite hard at the change).

I've been working with fdt blobs, implementing a "fdt" command.  It 
currently can print individual properties, recursively print nodes 
starting at any point in the tree, and set property values (but if the 
size is the same).

I would consider being a fdt-related custodian, but I don't have a lot 
of time to work on it so I'm not sure my response time would be an 
improvement over Wolfgang's. :-(

gvb