Revert "soundwire: initial work"

[xiulipan]

Reverts #1244

ABI changes in this PR caused a blocking regression now. As no decision is made to update the kernel to align the ABI now.

Revert and have more discussion is a good option.

[xiulipan]

@plbossart @lgirdwood @tlauda @jajanusz
Please have some decision and action about the recent blocking regression.
I think a simple revert first and discuss later is a good idea to make other feature and validation continue.

[jajanusz]

@xiulipan Please specify what do you mean by regression. All things look like they have no impact on driver that is not even aware of soundwire. We bumped minor version what means that there are changes indeed but it should be backward compatible.

[xiulipan]

@jajanusz
Regression means all platform now is not able to load tplg with the dev branches
thesofproject/linux#822

[jajanusz]

IMO it works as designed on FW side. You just have to make driver to correctly follow ABI versioning ideology.

[lgirdwood]

NAK, this is a kernel bug. Kernel should not reject minor ABI differences (only reject major ABI differences)

[plbossart]

NAK, this is a kernel bug. Kernel should not reject minor ABI differences (only reject major ABI differences)

It doesn't help to just close bugs and push the issues under the rug. It's not a minor change and we need a process for ABI changes. I completely disagree with just waving things through on the grounds that things are minor.

[lgirdwood]

@plbossart thesofproject/linux#822 is showing only a minor ABI difference unless we are meaning another issue ?

[plbossart]

@lgirdwood exactly my point, the PR 822 is showing a minor ABI difference but I don't think it is. The classification looks wrong since by adding a DAI type you can end-up with a configuration that does not work. It's different from a field that gets ignored.
And it didn't have a matching kernel PR so there is still no process to align.