[igt-dev] [PATCH i-g-t 00/23] tools/intel_vbt_decode: Improve the parser

[Jani Nikula]

On Wed, 15 Jun 2022, Petri Latvala <petri.latvala at intel.com> wrote:
> On Wed, Jun 15, 2022 at 09:44:26AM +0300, Jani Nikula wrote:
>> On Wed, 15 Jun 2022, Ville Syrjala <ville.syrjala at linux.intel.com> wrote:
>> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
>> >
>> > Change the igt VBT parser to make copies of the BDB data blocks
>> > (just as we do in the kernel parser now). This makes it less
>> > hazardous/confusing if your block definition don't happent to
>> > match reality. 
>> >
>> > Eg. with the current scheme we could very easily misinterpret
>> > random bits from other blocks/etc. as our block data by just
>> > using an overly large structure for decoding. With the block
>> > copies being  at least as big as out struct definitions the
>> > worst that can happen is that we just decode some extra zeroes
>> > rather than some random other bits.
>> >
>> > The other thing I'm bringing over is the data pointer generation
>> > stuff for modern VBTS (which lack the LFP data table pointes block).
>> > That lets us keep decoding the LFP data block in exactly the
>> > same way on old and new VBTs.
>> >
>> > And naturally we have massive gaps in what data the tool actually
>> > parses from the VBT. So lots of patches to just parse more stuff.
>> > I didn't do an exhaustive study on this topic, so still probably
>> > missing some useful things.
>> 
>> I glanced through all this, and it LGTM, but can't claim I did a
>> detailed, thorough review. I also don't think it should really be
>> necessary for this particular tool. If the changes work for you, it's
>> good enough for me.
>> 
>> Cc: Petri, good to go with that?
>
> Yeah, ack.

Ville, I think you can just merge.

You can put

Reviewed-by: Jani Nikula <jani.nikula at intel.com>

on it, but with the caveat above.

-- 
Jani Nikula, Intel Open Source Graphics Center