Intel IFS Ready To Weed Out Faulty Silicon With Linux 6.2

[coder]

Intel ought to put this capability in all their products. Imagine if distros could include a 2-monthly on-reboot timer that would run the test and set a flag to warn the user that their CPU was going bad, and it was time to use the warranty or back off the overclock.

I don't want it only on reboot, though. If I can have a monthly SMART self-test of my HDDs, then why not my CPU? The fact that they're merging this into the kernel suggests you don't need a reboot, to use it.

When you've got a server, and a lot of us have little servers at home (often built with non-server hardware), you want to know as soon as it starts failing. They tend not to get rebooted often, so only testing on boot is far from ideal.

[marlock]

That depends on the sort of test and detectiin Intel was able to implement. Some of it may be able to monitor the CPU health passively, some may run actively but cause negligible load and live well with other processing tasks, some may need to stop everything else for a significant time on a specific core, etc.

I'm acrually quite curious now to learn what it's really doing.

[lowflyer]

So the post this time was perfectly correct by formal writing rules. Now maybe we need to revise this on technical places like this ...

My post was not meant as a critique of you in any way. You eloquently pointed out: "Its very common in fact for acronyms to collide." But you go off on a tangent. The "formal writing rules" are not a solution for this public forum. Few people here know about these rules and even less care about them. Also, there are few places less "formal" as this. The frequent use of acronyms and even more the on-the-fly *creation* of acronyms is a CANCER in any discussion! This is even valid in more formal places. I'm not alone with this opinion, look at what exactly Linus Torvalds criticizes in his latest Intel-bashing.

[oiaohm]

My post was not meant as a critique of you in any way. You eloquently pointed out: "Its very common in fact for acronyms to collide." But you go off on a tangent. The "formal writing rules" are not a solution for this public forum. Few people here know about these rules and even less care about them. Also, there are few places less "formal" as this. The frequent use of acronyms and even more the on-the-fly *creation* of acronyms is a CANCER in any discussion! This is even valid in more formal places. I'm not alone with this opinion, look at what exactly Linus Torvalds criticizes in his latest Intel-bashing.

No Linus complaint is not anti using acronyms.

I know, I know, NIH and all that, but at least as long as we are
limiting ourselves to regular US-ASCII, we really only have 17576
TLA's to go around, and at some point it gets not only confusing, but
really quite wasteful, to have everybody make up their own
architecture-specific TLA.

https://en.wikichip.org/wiki/arm/tbi

Issue there is TBI already existed in different platforms supported by the Linux kernel yes with two different decodes "Top-byte ignore" and "Top-bits ignore". "Top-bits ignore" is basically the same thing as Intel LAM and AMD UAI just implemented by cray a long time ago. The generic code between platforms for this task really does need a single name.

Linus Torvalds is not against the usage of acronyms. But he is against multi-able creation for the same thing with no good reason. With LAM what Intel did was the same as everyone knows this thing as a wheel and then calls it "circular transportation facilitation device" and using zip ties to hold the wheel on instead of bolts.

Maybe that's not how LAM works on x86, but your changes to untagged_addr() are *not* x86-specific.

This bit by Linus starts explaining why he started getting mega annoyed. One thing to have a unique individual acronym inside your own driver(Linus is known to let this slide). Its another thing when this starts bleeding out into generic cross platform code that already has a different acronym for the same thing. This has been a long time known thing to get on Linus bad side.

IFS being "in-field scan" is not a intel unique term it first appears with Laser field scanners(yes those things that 3d map area) this includes the feature of being able to compare formal plans of area to current scan of area and show operator the difference. I don't know if Intel happens to get this right by luck or not. Thinking Intel uses some of those laser field scanners with the IFS feature in their factories its a bit hard to tell.

NIH(not invented here) with acronyms bad but is so is NIH with feature descriptions. Linus big issues is NIH with acronyms and descriptions as it makes his job harder and comments in code worse particularly when you end up with generic code areas using multi different acronyms/descriptions for exactly the same feature. Yes the thing Linus attempts to enforce is in fact a "formal writing rule" where when writing a document and there is multi acronym/descriptions for the same thing you are meant to come to agreement to use 1 though the complete document even if you reference all the different ones in like glossary.

The "formal writing rules" are not a solution for this public forum. Few people here know about these rules and even less care about them.

Quite a few public forums do use formal writing rules sometimes this is enforced by force. Like Linus refusing to merge stuff because the documentation/comments in code put on LKML(linux kernel mailing list) were not to the formal writing rules.(yes he has done this quite a few times and normally about 4 times a year)

Knowing Linus like for the formal writing rules what was Intel developers thinking with LAM comments in generic code that already had a term for the same thing. The answer not thinking and basically stuck a kick me sign on their backs.

Lowflyer do notice that the written articles here obey formal writing rules on acronyms and the administrators here writing comments in the forum also obey formal writing rules on acronyms. We are guests here maybe everyone should follow the hosts lead in behavior in this area. Yes polite behavior rules observe your host and attempt not to do anything they would not class as acceptable.

[lowflyer]

oiaohm , you're over-engineering a trivial issue. A little bit less use of acronyms, a little bit less of "trying to sound intelligent" and a little bit more of "say what you mean" would go a long way.

No Linus complaint is not anti using acronyms.

I never claimed that. But read his statement again but very slowly:

Is it too late to ask Intel to call this "Top-Bits-Ignore", and
instead of adding another crazy TLA, we'd just all agree to call this
"TBI"?

His full post mainly addressed a technical issue. But he was obviously quite upset by the careless introduction of new acronyms.

Quite a few public forums do use formal writing rules sometimes this is enforced by force. Like Linus refusing to merge stuff because the documentation/comments in code put on LKML(linux kernel mailing list) were not to the formal writing rules.(yes he has done this quite a few times and normally about 4 times a year)

Are you aware that this here is not the LKML? I do not think that Linus would have been happier if Intel had followed the "formal writing rules of acronyms".

Lowflyer do notice that the written articles here obey formal writing rules on acronyms and the administrators here writing comments in the forum also obey formal writing rules on acronyms. We are guests here maybe everyone should follow the hosts lead in behavior in this area. Yes polite behavior rules observe your host and attempt not to do anything they would not class as acceptable.

Well, no. Although, some articles follow the acronym's writing rules. But, to be honest, you are one of the offenders here as well. Excessive use of acronyms - irrespective of whether they follow the rules or not - is unpolite at least. It is indeed quite offensive.

People write a lot of crap with the help of acronyms. Following the "formal writing rules" does not remove the crap from the scribble.

[oiaohm]

His full post mainly addressed a technical issue. But he was obviously quite upset by the careless introduction of new acronyms.

Key word here is careless. Every year there are many platform/device unique acronyms added to the Linux kernel that Linus does not get upset to because they are restricted to platform only code or driver only code. Yes not in generic code.

Form that Linus post.

Maybe that's not how LAM works on x86, but your changes to untagged_addr() are *not* x86-specific.

You find "changes to [some function or area of generic code] are *not* [something]-specific" in almost every single post that Linus is getting upset about acronyms. So if you don't cherry pick 1 out alone instead look over all the cases of attempted added acronyms and added acronyms to the Linux kernel there is very exact trend that is almost 100 percent.

Are you aware that this here is not the LKML? I do not think that Linus would have been happier if Intel had followed the "formal writing rules of acronyms".

Formal writing rules do say if you are adding to a document be aware of the existing acronyms and don't cause duplicate. Duplicate in this case is when you have two acronyms in the same document that in reality mean exactly the same thing. Yes generic code is in it own .c file compare to driver unique or platform unique code so own documents.

Yes if intel had followed the rules in "formal writing rules" Linus most likely would not have been upset. Arm in the last 10 years has added a lot of unique platform specific acronyms to Linux kernel comments and not got on the bad side of Linus. Please note ARM kept those unique platform specific acronyms out of generic code and doing that is following formal writing rules when it comes to the use of acronyms.

That arm and others have got away with adding acronyms without issue and Intel gets yelled at by Linus there had to be a difference. Yes when you look at the other ones that get yelled at there is a clear trend why.

Linus is still not the best person at expressing exactly why he is annoyed with a person submitting code at times.

Well, no. Although, some articles follow the acronym's writing rules. But, to be honest, you are one of the offenders here as well. Excessive use of acronyms - irrespective of whether they follow the rules or not - is unpolite at least. It is indeed quite offensive.

All the articles of phoronix in fact follow formal writing rules when it comes to acronyms..

https://www.phoronix.com/news/Torvalds-Bashes-Intel-LAM

Is it too late to ask Intel to call this "Top-Bits-Ignore", and instead of adding another crazy TLA [Three Letter Acronym], we'd just all agree to call this "TBI"?

Is that the exact quote of what Linus wrote the answer is no.

Is it too late to ask Intel to call this "Top-Bits-Ignore", and instead of adding another crazy TLA, we'd just all agree to call this "TBI"?

Notice the difference how it was corrected. What Linus wrote was not exactly to formal writing rules so it was corrected to make it more readable to those who don't use LKML/Linux kernel standard acronyms. Yes the example your choose happened to have exact evidences how Michel and any other author for phoronix operates.

lowflyer Linus himself is guilty of using acronyms in way outsides would not understand. Here you attempting to say Linus is against Excessive use of acronyms. Linus complaints about excessive use of acronyms is very conditional. Interesting that you just called the person you were using as example for your case unpolite most likely without knowing.

The majority of my posts don't use acronyms. When I say majority its over 98% don't have acronyms in them. That is also because I am obeying formal writing rules. One of the formal rules is that you don't use acronym for a single use except for special cases instead you use the long form its one of the most common formal writing rules with Acronyms broken. Special cases are like where you have had to quote someone who had used the single usage or if the acronym happens to be name(LKML) would be this.

lowflyer majority of the problems with acronyms go away if people obey the formal rules. Problem is lot of people are not taught them in school. Formal rules around acronyms limit how damaging to readability using acronyms can be and is a decent trade off compactness of writing and readability.

[lowflyer]

What is your problem? You create an elephant out of a small remark.

lowflyer Linus himself is guilty of using acronyms in way outsides ... / ... Interesting that you just called the person you were using as example for your case unpolite most likely without knowing.

You're picking apart a broad-brush remark of mine. I did not criticize Linus, I criticize *every* excessive use of acronyms.

The majority of my posts don't use acronyms. When I say majority its over 98% don't have acronyms in them. That is also because I am obeying formal writing rules. One of the formal rules is that you don't use acronym for a single use except for special cases instead you use the long form its one of the most common formal writing rules with Acronyms broken. Special cases are like where you have had to quote someone who had used the single usage or if the acronym happens to be name(LKML) would be this.

You really seem to like the Formal Rules of Acronyms (FRA). Do you have a Philosophiae Doctor on FRA?

lowflyer majority of the problems with acronyms go away if people obey the formal rules. Problem is lot of people are not taught them in school. Formal rules around acronyms limit how damaging to readability using acronyms can be and is a decent trade off compactness of writing and readability.

Sorry to be direct. Which country are you from? You expose a blind belief in FRA.

No, the problems with acronyms don't "go away" by following FRA. No set of rules will ever solve the problems. Acronyms are always problematic. That's exactly the reason why FRA were introduced. The help mitigate the problem - but they don't solve it. Zombie-like adherence to FRA makes it even worse. Banning acronyms by a rule like "Do not ever use acronyms" is bad too. Acronyms are sometimes useful.

The only rule that most people overlook but the one rule that actually helps is "use acronyms sparingly". Look it up, it's also one of the FRA. With a few exceptions Michael does it quite well. He uses acronyms sparingly. And no, I'm not going to check your posts for adherence to FRA - that's your job.

[oiaohm]

Sorry to be direct. Which country are you from? You expose a blind belief in FRA.

Australia but that makes really makes no difference.

Formal writing rules are a global thing. Interesting point is all the formal writing rules on acronyms are duplicated many times over by different groups. Be it newspapers or academic writing there are common writing rules these are the formal writing rules.

No, the problems with acronyms don't "go away" by following FRA. No set of rules will ever solve the problems. Acronyms are always problematic. That's exactly the reason why FRA were introduced. The help mitigate the problem - but they don't solve it. Zombie-like adherence to FRA makes it even worse. Banning acronyms by a rule like "Do not ever use acronyms" is bad too. Acronyms are sometimes useful.

That right you cannot ban Acronyms out right instead have to manage the fall out.

The only rule that most people overlook but the one rule that actually helps is "use acronyms sparingly". Look it up, it's also one of the FRA. With a few exceptions Michael does it quite well. He uses acronyms sparingly. And no, I'm not going to check your posts for adherence to FRA - that's your job.

"Avoid alphabet soup" can be a good thing to search. Yes "avoid alphabet soup" most common way "use acronyms sparingly" are written in different English writing guides. Of course they could not write "use acronyms sparingly" that would be too straight forwards and would google translate well. Yes different languages have their own equals to "Avoid alphabet soup" that don't translate well either.

The Associated Press Stylebook has acronym quoted a lot in other formal writing stylebooks used by the press. "NNCBTPSNBRTASTSAFW" (Names not commonly before the public should not be reduced to acronyms solely to save a few words.) That is a public publication one.

lowflyer the formal rules of acronyms I have mentioned you find written in some form in every single press Stylebook/styleguide this not limited by language or country.

This is the horrible part these formal writing rules exist there is 100% sure agreement on most of these formal writing rules by press and academic writers and other parties doing formal writing on how the completed documents should look but they don't agree on how to word the rules tell you to-do that so all their stylebooks/styleguide have the same rules worded differently. There is about 150 different ways in English

Formal writing rules exist but we do have two problems.
1) lack of agreement on the text of formal writing rules. This is like the mass translations of the christian bible.
2) Generally you only start getting taught formal writing rules when you go to university todo the right courses or get a press job.

Finally not all university course class teaching formal writing as important. Yes the programmers at Intel being USA can have done a complete USA programming course and done zero formal writing training. Yes what Linus is upset with with LAM stuff is a breach of USA collage stylebooks/styleguides of course a programmer out of USA collage may not have even read a single line of the collage stylebook/styleguide. In fact you could swap USA with most countries around the world.

Linus gets upset with the same things with acronyms over and over again. Rules over acronym usage never making it into the Linux coding coding-style.rst or the "How to write kernel documentation". Yes the section of the Linux kernel documentation that is the Linux kernel documentation styleguide is called "How to write kernel documentation" and it absolutely bare bones..

One thing about formal writing rules you expect them put in styleguides to prevent repeating problems.

lowflyer I know you say formal writing rules of acronyms will not fix everything this I agree. Operating rule less mess is 99% of a time be worse than formal writing rules produced documents.

Linus complains about acronyms in limited cases. But he never formalizes the rules to prevent it from happening again. But that in a lot of ways should not be Linus job but should be the Documentation writers job to update the styleguide. One problem Linux kernel is always way short of documentation writers.

[lowflyer]

"Avoid alphabet soup" can be a good thing to search. Yes "avoid alphabet soup" most common way "use acronyms sparingly" are written in different English writing guides. Of course they could not write "use acronyms sparingly" that would be too straight forwards and would google translate well. Yes different languages have their own equals to "Avoid alphabet soup" that don't translate well either.

The word "sparingly" translates well to all languages I know. "Alphabet soup" is an analogy that doesn't translate well. You find the exact sentence "should be used sparingly" here, here, here and here. You'll find this word in most other writing rules. Most sources list the "use acronyms sparingly" as the first rule *before* all the other rules. The UN manual even mandates that acronyms "should not be used at all" at least for general assembly resolutions. I think the UN manual is more important than newspapers' style-guides as they need to make sure that *everybody* understands each other.

The Associated Press Stylebook has acronym quoted a lot in other formal writing stylebooks used by the press. "NNCBTPSNBRTASTSAFW" (Names not commonly before the public should not be reduced to acronyms solely to save a few words.) That is a public publication one.

AP is not a "formalizing body". Even if their work is published by many others.

This is the horrible part these formal writing rules exist there is 100% sure agreement on most of these formal writing rules by press and academic writers and other parties doing formal writing on how the completed documents should look but they don't agree on how to word the rules tell you to-do that so all their stylebooks/styleguide have the same rules worded differently. There is about 150 different ways in English

Formal writing rules exist but we do have two problems.
1) lack of agreement on the text of formal writing rules. This is like the mass translations of the christian bible.

It's either 100% agreement or it's an agreement on most rules. It cannot be both. "100% sure agreement on most" is a bit of a misnomer, but I get what you mean.
I see you often prefix "writing rules" with the word "formal". Here is the problem:

• the writing rules are not formal

It does not matter how often your repeat the word "formal" - and I get that you want these rules to be formal - they are not. Even if they are repeated in all the style-guides on the planet. The writing rules are at best "house rules". A publishing company or a university may reject works that don't follow their rules. But that's all they can do.

There are reasons why these rules are not formalized. Lack of agreement on the wording is even acknowledged by you. But the disagreement goes further. There is lack of agreement on the rules itself, and lack of agreement whether to "have" formal rules. And then there is the issue that it is *not* an issue of "not following the rules". It is an issue of "lack of communication skill".

No amount of rules can cover all cases. Rules will be broken. Rules *can* and *are* misused. Less rules are easier to follow. Acronyms can be (mis-)used to intimidate, deceive or to sound more "knowing". Some people will interpret the fact that *there are* rules as "yes, we are allowed to just sprinkle your text with as many and as unusual acronyms we see fit".

2) Generally you only start getting taught formal writing rules when you go to university todo the right courses or get a press job.

Well, not in my country. I would say for most of Europe. But this is besides the point. Being "taught formal writing rules" does not help someone to express himself better. Being "taught to write" is better than "being taught to follow rules". It's an issue of acquiring the skill to write.

... Rules over acronym usage never making it into the Linux coding coding-style.rst or the "How to write kernel documentation". Yes the section of the Linux kernel documentation that is the Linux kernel documentation styleguide is called "How to write kernel documentation" and it absolutely bare bones..
../..
... But that in a lot of ways should not be Linus job but should be the Documentation writers job to update the styleguide. One problem Linux kernel is always way short of documentation writers.

General writing rules do not belong into coding styles. Putting more rules into more style-guides will make the problem of having not enough documentation writers only worse.