On 4/30/19 9:54 AM, Vfiinod Koul wrote:
On 30-04-19, 08:38, Pierre-Louis Bossart wrote:
On 4/30/19 3:51 AM, Vinod Koul wrote:
On 15-04-19, 08:09, Pierre-Louis Bossart wrote:
Signed-off-by: Pierre-Louis Bossart pierre-louis.bossart@linux.intel.com
drivers/soundwire/Kconfig | 2 +- drivers/soundwire/bus.c | 87 ++++++++-------- drivers/soundwire/bus.h | 16 +-- drivers/soundwire/bus_type.c | 4 +- drivers/soundwire/cadence_master.c | 87 ++++++++-------- drivers/soundwire/cadence_master.h | 22 ++-- drivers/soundwire/intel.c | 87 ++++++++-------- drivers/soundwire/intel.h | 4 +- drivers/soundwire/intel_init.c | 12 +-- drivers/soundwire/mipi_disco.c | 116 +++++++++++---------- drivers/soundwire/slave.c | 10 +- drivers/soundwire/stream.c | 161 +++++++++++++++--------------I would prefer this to be a patch per module. It doesnt help to have a single patch for all the files!
It would be great to have cleanup done per logical group, for example typos in a patch, aligns in another etc...
You've got to be kidding. I've never seen people ask for this sort of detail.
Nope this is the way it should be. A patch is patch and which should do one thing! Even if it is a cleanup one.
I dislike a patch which touches everything, core, modules, so please split up. As a said in review it takes guesswork to find why a change was done, was it whitespace fix, indentation or not, so please split up based on type of fixes.
With all due respect, you are not helping here but rather slowing things down. I've done dozens of cleanups in the ALSA tree and I didn't go in this sort of details.
Thats fine, it is upto people, everyone has different views, mine is different from Takashi's. We all know for example networking has different stable and code style rule. That is how it is and I dont think we would have one rule for all kernel.
All I ask is to be able to review and split up accordingly, I guess that is a fair request
The fact that the series was tagged as Reviewed by Takashi on April 11 and we are still discussing trivial changes tells me the integration model is broken.
Is it? you got feedback on 15th (that too after my 2 week conf/vacation break) and I got called crazy for that, not helping!!
It's not just me the patches related to runtime-pm from your own Linaro colleagues posted on March 28 went nowhere either.
Does it matter it was a Linaro colleague or not, a patch was posted, feedback given (similar to cadence one) we agreed that the fix is not correct and so patch was not applied. I don't think Srini cried over it!
Moving forward, I suggest we merge SoundWire-related patches through the sound tree. There will be dependencies in the coming weeks between SOF and SoundWire and it makes no sense to have separate maintainers and make the life of early adopters more complicated than it needs to be. If we have 3-week delays for trivial stuff, I can't imagine what the pace will be when I publish the next 20-odd patches I am still working on, and the code needed for the SoundWire audio device class being standardized as we speak. Things were fine up to now since no one was actually using the code, we are in a different model now.
I disagree and wont accept it. I dont think you understand that you are not the most important person in the whole world, the 20 patches series you are cooking would sure be greatest ever, but that is not the point. The kernel has a process, you got a feedback, please fix that and post v2 rather than cribbing, complaining and calling crazy. The energy would have been better spent on fixing the feedback provided.
Dependencies are _always_ there in kernel development and we know how to deal with it. Am sure Takashi, Mark and me can come to reasonable agreement, I wouldn't worry about that!
What we dont do is create new model for your 20 patches.
And I guess I dont have anything more to say on this thread, so I wont bother replying, please feel free to post v2 and I shall review.
Friends have disagreements. We remain friends and I will provide a v2.
I still believe it makes no sense to split the integration of SoundWire-related patches in two different trees. The only rationale for it might be that SoundWire is a 'bus' than could be used in other areas. Except that for now and the foreseeable future (2022+) it's only for audio as a replacement of HDaudio, so the pragmatic way of dealing with SoundWire is to merge the code through the audio tree. And given that the code is not in a usable state at the moment, dealing with the audio tree would not have any negative impact on anyone.