AMD Sensor Fusion Hub Support Is Not Coming With Linux 5.8

[AmericanLocomotive]

This is still nothing that relates a picture of a turtle with AMD hardware.

He's saying AMD takes their sweet time getting things done in regards to Linux. He's not wrong either. Zen APUs have been out in some form for almost 3 years now, and the Sensor Hub still isn't supported.

But, like I said in other threads: Large companies move slow. A company the size of AMD has a lot of bureaucracy. In their bad years, they laid off most of their Linux devs. Now that they're competitive (especially in the HPC space), they are going to start hiring a lot more Linux developers. But it takes time to find the right talent, figure out what positions you actually need to hire, and then it will take months for driver-writers to get up to speed on the hardware.

Well, I got my own fair share of work experience (and had to deal with some eccentric people along that way) hence I disagree with that judgment, it was a short summary of: "There is a lot more work to do." And I see nothing wrong with his phrasing, you need to communicate negative feedback somehow and I see no reason for sugarcoating the fact that he regards that work as insufficient. I have also seen far harsher language being used on the LKML, I wouldn't consider this to come anywhere near that.

I didn't say the Intel guy needed to sugarcoat anything and pretend it was good. The problem is that comments like that are not constructive, it's condescending. The AMD engineer asked for a code review - he's looking for criticism and feedback. There is literally nothing helpful about saying "TL;DR: It needs work", it provides no useful feedback, no helpful information, no anything. Especially with preceding: "You asked for a review, here it is". It just comes off as rude and condescending.

A better response would have been "Hi Sandeep, I had time to review the patch and there are many significant problems that need to be addressed before this can be mainlined. I've outlined some of my concerns below:"

The Open-Source community likes to project an image of collaboration, and openness and working together, but the more I read and learn about the community, it becomes more and more clear that verbal-abuse rudeness and hostility is pervasive throughout the community. It even almost seems encouraged by some.

[jrch2k8]

It's very disrespectful, especially in a professional context. It's not constructive in any possible way. It doesn't offer any suggestions on what is wrong, or how improvements could be made.

Imagine going to a coworker and saying "Hey, I just finished drafting up the plans for this project, could you review for any potential problems and suggest improvements" and then he replies "You asked me for a review, well here it is: It's bad, try again". That is literally what that Intel engineer did. It doesn't help anyone, and it's just him purposely trying to be a jerk.

Who gives a damn, LKML is not some safe space, if your code is crappy and you have the guts to ask for mainline then be prepared, if you think the code is good enough and you can prove it(all that matters in this instance) then follow the process, take input and fight your way in.

This is science not feelings based, i don't care if the guy bitch slap him across the table as long as the technical reasons are accurate and measurable because all it matter is the quality of the code. period. It does not matter if you are an asshole, a neo nazi, a furry, republican, democrat, an spawn of satan himself, if your code is measurably better i want it in the kernel, if your code is bad i don't care if you are the holiest man that have ever lived take your POS out and try again.

If this guy can't take some rough comments now and then in LKML he should look for another job(doesn't seem the case tho)

[JustinTurdeau]

Waiting for the next Intel engineer that reviews the AMD code to reply: "TL,DR" like the earlier Intel engineer did.

...the snark from the Intel reviewer, while looking cute, is completely unprofessional and disrespectful towards the original submitter IMHO. So much "Code of Conduct" in Linux development, eh?

I think you need to improve your reading comprehension. "TL;DR" doesn't always have to mean "your post was too long, so I didn't read it". It can also mean "here's the short version of my reply".

The post you linked to seems to fall into the latter category, since he goes on to do a line-by-line review and thus clearly did read the whole thing.

[ms178]

I didn't say the Intel guy needed to sugarcoat anything and pretend it was good. The problem is that comments like that are not constructive, it's condescending. The AMD engineer asked for a code review - he's looking for criticism and feedback. There is literally nothing helpful about saying "TL;DR: It needs work", it provides no useful feedback, no helpful information, no anything. Especially with preceding: "You asked for a review, here it is". It just comes off as rude and condescending.

A better response would have been "Hi Sandeep, I had time to review the patch and there are many significant problems that need to be addressed before this can be mainlined. I've outlined some of my concerns below:"

The Open-Source community likes to project an image of collaboration, and openness and working together, but the more I read and learn about the community, it becomes more and more clear that verbal-abuse rudeness and hostility is pervasive throughout the community. It even almost seems encouraged by some.

There is a great legal rule in our German Civil law about interpretation, it is all about the "recipient horizon" and how an objective third-party of the target audience must understand it, which means in this case what an average developer on the LKML would objectively think when reading this.

I can see your point, but It seems that we both interpret and judge his words very differently. And in my view you are reading way too much negativity into his wording (also he did provide a thorough technical review). I side with JustinTrudeau and jrch2k8 in general - it is about the quality of the code first and foremost and although I would like to not get yelled at for my mistakes, I don't need to take that personally. Certainly there might be other examples which cross the line of decent behavior, but that specific posting does not cross that threshold - Linus Torvalds could provide far better examples of bad language, not that I would care much if his technical arguments hold true.

And this particular posting is a great example why a Code of Conduct might be a bad idea overall, it empowers people to play moral and language police, wrongly accusing developers who don't adhere to their own subjective standards. In my view this is more annoying than it helps. You also can keep an entire staff occupied to judge every other posting but the energy and resources could be put into better code instead.

[JustinTurdeau]

Imagine posting such long-winded diatribes because someone in the thread misunderstood a perfectly reasonable LKML post. Imagine if people used half the time they spend bloviating to instead just read and understand the original subject matter.

[AmericanLocomotive]

There is a great legal rule in our German Civil law about interpretation, it is all about the "recipient horizon" and how an objective third-party of the target audience must understand it, which means in this case what an average developer on the LKML would objectively think when reading this.

I can see your point, but It seems that we both interpret and judge his words very differently. And in my view you are reading way too much negativity into his wording (also he did provide a thorough technical review). It is about the quality of the code first and foremost and although I would like to not get yelled at for my mistakes, I don't need to take that personally. Certainly there might be other examples which cross the line of decent behavior, but that specific posting does not cross that threshold - Linus Torvalds could provide far better examples of bad language, not that I would care much if his technical arguments hold true.

I agree that the quality of the code is important - and in this case I do not think what was said was that egregious. I re-read it a few times, and it's honestly not that bad - but having a TL;DR: in a "professional" post targeted towards an employee of a public company is still not wonderful.

My point is you can address the code being bad without attacking or being condescending to the person themselves. The response I wrote: ""Hi Sandeep, I had time to review the patch and there are many significant problems that need to be addressed before this can be mainlined. I've outlined some of my concerns below:" Is short, succinct, doesn't "sugar coat" anything and addresses the problem without in any possible way being rude.

Who gives a damn, LKML is not some safe space, if your code is crappy and you have the guts to ask for mainline then be prepared, if you think the code is good enough and you can prove it(all that matters in this instance) then follow the process, take input and fight your way in.

I never implied it should be a "safe place". What I'm saying is that it's easy to be respectful, while still getting to the root of whatever particular issue is occurring. I've worked with and for businesses where you had to "fight your way in" through the disrespect. What ends up happening is you create a "good 'ol boys club". You end up fostering group-think, where the same stale old ideas are recycled over and over and little actual progress is made. I saw many extremely intelligent, hard-working and talented people pushed out of my last place of work because they couldn't take the abuse and disrespect.

This is science not feelings based, i don't care if the guy bitch slap him across the table as long as the technical reasons are accurate and measurable because all it matter is the quality of the code. period. It does not matter if you are an asshole, a neo nazi, a furry, republican, democrat, an spawn of satan himself, if your code is measurably better i want it in the kernel, if your code is bad i don't care if you are the holiest man that have ever lived take your POS out and try again.

This where you are wrong. Until we have AI algorithms generating all of our code start-to-finish, there will always be a human and emotional element to programming. Especially collaborative programming like the Open Source community. The place I work for now understands this. If I am doing something wrong - my peers will not hesitate to tell me. But they do it in a respectful manner, and offer suggestions and improvements. They don't call me a "autistic idiot" - because it's not conducive to actually fixing the problem.

I'm part of many tech and engineering forums. I can tell you without a doubt that best forums are the ones that are heavily moderated. Absolutely no personal attacks are allowed at all, and anyone who even thinks about hurling an insult gets banned. The result is extremely high quality, in-depth discussion about highly technical topics. The signal-to-noise ratio is through the roof. Constant belittling and bickering just clogs things up and slows thing down - when you help people actually do better, the whole group is more productive. The forums I'm on with very lax moderation basically have every single thread eventually turn into a flame-war.

And to re-iterate: Being respectful doesn't mean "sugar coating" and "creating a safe space". You can tell people their code is awful, and is no way fit to mainlining - but just be respectful about it. Offer suggestions, tell them what you see wrong.

The autism level of posters on this forum is mind blowing.

...and that exactly proves my point. What does calling people who think "you should just be respectful to others", "autistic", do for the conversation? Does it add anything to the discussion? Does it offer a solution to the problem? It does nothing at all except try to discredit the opinion of others by insinuating "they're autistic, so what they say doesn't matter".

What's ironic, is that in general autistic people don't even pickup on such issues. I work with and interface with some autistic people as part of my job. One of defining characteristics that autistic people frequently have is an inability to express empathy and sympathy. They often inadvertently say rude and condescending things without really understanding (or sometimes caring) why or how something they said could make someone else upset.

[duby229]

AMD with software support on Linux

It's because their in-house development model sucks ass. They throw a massive amount of (highly abstracted) code into a tarball and then release the whole shebang once it's gotten so out of control that nobody can follow it. AMD releases code with -layers- of abstraction such that you can't figure out what that code does and it can't be bisected so there is no method to dig through those -layers- of abstraction.

If AMD would just develop on upstream repositories with a good VCS like git.... Wishful thinking, I know...

EDIT: Anyway, AMD goes through this every single time they release something that needs to be upstreamed. First code style gets fixed, then -layers- of abstraction get removed, and then finally the code gets understood well enough that upstream devs are comfortable excepting it. And it usually takes many, many reviews.

Like I said, If AMD would just develop their code upstream in the first place.... Wishful thinking.....

[JustinTurdeau]

What's ironic, is that in general autistic people don't even pickup on such issues.

That's not "ironic", it's exactly the point.

Also, having expectations that other people accommodate your shortcomings works both ways. If you expect other people to be patient with autists, then autists themselves ought to think long and hard before they write autistic diatribes. How many misunderstandings do they have to bump their heads against before they start to notice a pattern forming?

Anyway, the point here is that the original LKML post meant absolutely no ill will whatsoever. 80% of this thread is just people whining about something that didn't even happen.

[AmericanLocomotive]

EDIT: Anyway, AMD goes through this every single time they release something that needs to be upstreamed. First code style gets fixed, then -layers- of abstraction get removed, and then finally the code gets understood well enough that upstream devs are comfortable excepting it. And it usually takes many, many reviews.

I feel that a lot of this is because of the limited size, and resources their open-source team has. They're likely sharing a lot of code with other teams, and doing their best to get the necessary updates and patches released. As they expand their open source teams and hire more people, there is undoubtedly going to be a lot of clashing styles and preferences. I'm thinking that as long as the CPU team continues to execute well (Zen 3, Zen 4) and they secure more HPC and server wins, you will see a gradual turn-around on their open source development efforts.

That's not to say they aren't trying now, as they clearly have a bunch of very dedicated developers. But things are clearly chaotic over there. It's only been 3 years since Zen was released. I really wouldn't expect too much improvement for at least another two years. It takes a long time to assemble a good team and get everyone on the same page.

That's not "ironic", it's exactly the point.

Yes, it is ironic.

In your post, you claimed that people who "complain" and "whine" about being respectful to other people's feelings and emotions are autistic. That is ironic, because one of more notable characteristics of people with autism is an apparent lack of empathy, sympathy or any kind of real awareness of what their words or actions cause others to feel or think. So someone with autism would most likely not be writing a "diatribe" about being respectful and kind to others. They would most likely be the ones writing material that (not necessarily intentionally, mind you) comes off as disrespectful or rude to others.

So yes, the statement you made (that you later edited out of your post) fits the definition of "irony" perfectly.

[duby229]

I feel that a lot of this is because of the limited size, and resources their open-source team has. They're likely sharing a lot of code with other teams, and doing their best to get the necessary updates and patches released. As they expand their open source teams and hire more people, there is undoubtedly going to be a lot of clashing styles and preferences. I'm thinking that as long as the CPU team continues to execute well (Zen 3, Zen 4) and they secure more HPC and server wins, you will see a gradual turn-around on their open source development efforts.

That's not to say they aren't trying now, as they clearly have a bunch of very dedicated developers. But things are clearly chaotic over there. It's only been 3 years since Zen was released. I really wouldn't expect too much improvement for at least another two years. It takes a long time to assemble a good team and get everyone on the same page.

Years ago I had a conversation (argument) with Bridgman about this very topic. That conversation was on this forum, I could probably find it for you if you're down read it. But my point is AMD already knows that their coding methods cause huge amounts of problems with upstream projects. But, of course, all AMD wants to do is play "pass the buck", or "the blame game".

EDIT: AMD had adopted an open source driver policy in 2007. Almost 13 years ago. To -this- very day AMD still doesn't use any kind of version control or release management. Come on now, we're talking about the linux kernel here, -THE- project that git was invented for.... AMD ran out of excuses 1 or 2 years later, it's now been -13- years.... How in the fuck do they expect to contribute efficiently with upstream when they refuse to implement release management. Well, they play "pass the buck" and "the blame game"...