Linux 5.15-rc5 x86 Changes Aim To Fix "Yet Another Hardware Trainwreck"

[schmidtbag]

So you want me to link the article again or...? Are you just dumb and using generic responses or something?

I want you to quote the exact place in the article that you think so obviously explains why NobodyXu is wrong. Not only am I not able to find anything that so concisely supports your notion, but, I was able to find something in the article that favors the whole backward-compatibility theory:
"In theory HPET support should be removed altogether, but there are older systems out there which depend on it because TSC and APIC timer are dysfunctional in deeper C-states."
But, I am also speaking generally. I deliberately pluralized "posts" in my first response to you, because this isn't the first time you've acted like this. I, and presumably the people who have been upvoting my comments, are sick of it. Clearly, your behavior isn't working as well as you thought it would.

What makes you think he's gonna read a link to the article when he commented before reading it in the first place. You're so clueless it's not even funny.

At this point, do you think anybody here feels you actually read the article? You're spending so much time acting like NobodyXu is a total idiot yet you can't spend 30 seconds to prove why. And I'm the one who is clueless?
The ones who do nothing but talk shit and laugh are the ones who know the least.

Yeah it's a thing called shame and embarrassment. You can't appeal to these people with facts, they're irrational emotional specimens.

You are making rash assumptions that "these people" aren't open to a civil discussion. What evidence do you have that a member as new as NobodyXu is irrational?
Y'know what's irrational? Making baseless claims.

Facts are all that matters? How cute of you. If facts were all that matters we wouldn't have such bullshit in the first place. But yes, facts are all that matters in a perfect world where people are not morons. I don't insult those type of people. I insult the people who think "every opinion is equally valid, even if they are factually wrong" which is THIS case to begin with.

Here's the thing:
If someone is factually wrong, it's not a matter of opinion anymore. Every opinion is equally valid, when it actually qualifies as an opinion. Sometimes, something can be an opinion until proven otherwise. Case in point: NobodyXu had an opinion of why it's so hard to fix HPET issues at the hardware level. According to you, there are disagreeing facts. Should those supposed facts be presented in a tactful manner, then NobodyXu ought to be mature enough to drop the opinion.

These kind of clowns don't give a shit about facts, but embarrassing them works... sometimes. At least better than zero.

In my experience, people shut up pretty quickly when you explain how they're wrong and leave the insults out. They're more likely to shut up when you express an understanding of their thoughts.
I've actually made friends with someone on these forums, over an argument, who I talk to regularly. Y'know why? Because we kept things civil. There are about 3 other people I can think of from these forums (and others) who I have had pleasant and meaningful conversations with that stemmed from a strong disagreement.
Who would have thought that sticking with facts and not being a prick about it would yield so much productivity!

[Weasel]

I want you to quote the exact place in the article that you think so obviously explains why NobodyXu is wrong. Not only am I not able to find anything that so concisely supports your notion, but, I was able to find something in the article that favors the whole backward-compatibility theory:
"In theory HPET support should be removed altogether, but there are older systems out there which depend on it because TSC and APIC timer are dysfunctional in deeper C-states."

That has nothing to do with it. Hardware can be broken or have bugs, that's not unusual (in fact, many times it's even inferior quality to software). You find "older systems" everywhere since everything ages in the same way.

That's like saying you drop support for older systems without AVX-512, which is an extension to the architecture, and call it a "backwards compatibility issue with the arch"? What the actual fuck?

If anything, an architecture that CANNOT add new features over time is an actual joke.

But, I am also speaking generally. I deliberately pluralized "posts" in my first response to you, because this isn't the first time you've acted like this. I, and presumably the people who have been upvoting my comments, are sick of it. Clearly, your behavior isn't working as well as you thought it would.

Or they're the butthurt trashes with facebook opinions and can't stand the fact they got their dreams and hopes shattered by facts.

Remember, public embarrassment is the important thing here. It's good for them to be mad about it, goal achieved.

Here's the thing:
If someone is factually wrong, it's not a matter of opinion anymore. Every opinion is equally valid, when it actually qualifies as an opinion. Sometimes, something can be an opinion until proven otherwise. Case in point: NobodyXu had an opinion of why it's so hard to fix HPET issues at the hardware level. According to you, there are disagreeing facts. Should those supposed facts be presented in a tactful manner, then NobodyXu ought to be mature enough to drop the opinion.

Except baseless opinions, like his, deserve no respect.

The issue is with the hardware. It's like saying the .zip file format is flawed because a specific software using it is full of bugs and does it wrong and corrupts archives that you have to "work around" with hacks.

But I mean, this is literally explained by the guy who wrote the patch.

tl;dr hardware tends to be of shit quality and full of bugs, even worse than bloated software. And the problem is that it falls down to the software to workaround its issues.

[schmidtbag]

That has nothing to do with it. Hardware can be broken or have bugs, that's not unusual (in fact, many times it's even inferior quality to software). You find "older systems" everywhere since everything ages in the same way.

Oh hey finally you're actually contributing something to the conversation without being obnoxious. See? That wasn't so hard.
To my understanding, HPET wasn't always broken. According to the article, it gradually became unmanageable due to feature creep.

That's like saying you drop support for older systems without AVX-512, which is an extension to the architecture, and call it a "backwards compatibility issue with the arch"? What the actual fuck?

I interpret it more as saying you drop AVX-512 because the complexity it brings isn't worth the benefits it yields, but leave AVX and and AVX2 alone. Seems like HPET in and of itself isn't the problem, it's all the crap added on top of it.

Or they're the butthurt trashes with facebook opinions and can't stand the fact they got their dreams and hopes shattered by facts.

You mean the facts that you still haven't presented?

Remember, public embarrassment is the important thing here. It's good for them to be mad about it, goal achieved.

You're too caught up in your own ego if you think that actually worked. Nobody here is siding with your behavior, so, you didn't embarrass anyone.

Except baseless opinions, like his, deserve no respect.

It's not baseless, because even I understood, from the article, how he came to that conclusion. Even if he's wrong, the statement objectively wasn't baseless.
As of right now, your claims are as good as baseless.

The issue is with the hardware. It's like saying the .zip file format is flawed because a specific software using it is full of bugs and does it wrong and corrupts archives that you have to "work around" with hacks.

Not really, because ostensibly Windows also has issues with such things. I wouldn't be surprised if other OSes like the BSDs don't even have a proper/complete HPET implementation. So, it's more like saying "the .zip format is flawed because all software using it encounters bugs or needs awkward workarounds".

[Weasel]

Oh hey finally you're actually contributing something to the conversation without being obnoxious. See? That wasn't so hard.
To my understanding, HPET wasn't always broken. According to the article, it gradually became unmanageable due to feature creep.

Because they never bothered to fix it with new features. They = the hardware vendors who ship it in the hardware. You know, it's a physical thing right?

Not really, because ostensibly Windows also has issues with such things. I wouldn't be surprised if other OSes like the BSDs don't even have a proper/complete HPET implementation. So, it's more like saying "the .zip format is flawed because all software using it encounters bugs or needs awkward workarounds".

Yeah but it's not the format that's the issue, it's the (hardware) implementation of it.

Actually why is this even a problem? Just disable it at kernel command line.

Oh wait, your (or the other guy's) issue isn't the backwards compatibility of the arch, it's the backwards compatibility of the LINUX KERNEL, since it wants to support systems where HPET is the only timer they can use in deeper C-states (or whatever he mentioned). This isn't going to go away even if you "fix" this in new hardware or add a proper timer source. Old hardware isn't going to magically get fixed.

[Azrael5]

As far as the HPET argument, should be verified how what worked in old kernels, no longer works in recent kernels.
Another matter deals with hardware makers. If they would apply diagnostic softwares, such as fwts, they could fix any error checked by them. Often these tools are used on working machines but they should test the machine before its launch.

[schmidtbag]

Actually why is this even a problem? Just disable it at kernel command line.

Right, because that totally has no effect on anything else...

Oh wait, your (or the other guy's) issue isn't the backwards compatibility of the arch, it's the backwards compatibility of the LINUX KERNEL, since it wants to support systems where HPET is the only timer they can use in deeper C-states (or whatever he mentioned). This isn't going to go away even if you "fix" this in new hardware or add a proper timer source. Old hardware isn't going to magically get fixed.

Personally, I don't really care either way, since this isn't the sort of thing that affects me. Though the way I understand it, HPET was good, but then Intel and AMD kept meddling with it and now it's causing kernel issues (again, not just the Linux kernel). So, it's more that old hardware is fine and new hardware is what's facing issues. I find that to be a valid complaint.