On Thu, 2022-07-21 at 14:43 +1000, Dave Airlie wrote: > > +Users switching to a newer kernel should *not* have to install newer > +firmware files to keep their hardware working. At the same time updated > +firmware files must not cause any regressions for users of older kernel > +releases. That seems sane, and certainly something we've done in wireless in the past. > +* Firmware files shall be designed in a way that it allows checking for > + firmware ABI version changes. It is recommended that firmware files be > + versioned with at least a major/minor version. It is suggested that > + the firmware files in linux-firmware be named with some device > + specific name, and just the major version. The firmware version should > + be stored in the firmware header, or as an exception, as part of the > + firmware file name, Eh, I went to write a whole paragraph here and then read it again ... Maybe this should say "[t]he _full_ firmware version", to contrast with the previous sentence mentioning the "major version". > in order to let the driver detact any non-ABI typo - 'detect' > + fixes/changes. The firmware files in linux-firmware should be > + overwritten with the newest compatible major version. > That's also a bit confusing IMHO - did that mean "minor version"? Or something? I mean ... if you overwrite a file that has the major version in the filename then by definition it is the same major version? > + This means no major version bumps without the kernel retaining > + backwards compatibility for the older major versions. Strictly reading this might require aeons of support for firmware version, if you have a release cadence of them like every 6 weeks for a new _major_ version (yes, because APIs change), then that's rather harsh. In practice we've often done this, but I think some reasonable cut-off could/should be there, such as dropping support after a reasonably long time frame (say a year?) Often though that's less a question of "does it still work" and rather one of "do I still support that" and the answer for the latter is obviously "no" much quicker than the former. johannes
