Linus Torvalds Rejects RISC-V Changes For Linux 6.17 For Being Late and 'Garbage' (phoronix.com)
- Reference: 0178628016
- News link: https://linux.slashdot.org/story/25/08/09/2147224/linus-torvalds-rejects-risc-v-changes-for-linux-617-for-being-late-and-garbage
- Source link: https://www.phoronix.com/news/Linux-6.17-RISC-V-Rejected
> Only on Friday were the RISC-V code updates submitted for the Linux 6.17 merge window. The Linux 6.17 merge window is expected to wrap up on Sunday with the Linux 6.17-rc1 release... [T]his pull request has been rejected by Linus Torvalds for Linux 6.17 on the basis of being late in the merge window especially with his international travels this week being known. And he's unhappy with some of the code included as part of this merge request. .
Here's the text of [2]Torvalds' response ...
> RISC-V Patches for the 6.17 Merge Window, Part 1
No. This is garbage and it came in too late. I asked for early pull requests because I'm traveling, and if you can't follow that rule, at least make the pull requests *good*.
This adds various garbage that isn't RISC-V specific to generic header files.
And by "garbage" I really mean it. This is stuff that nobody should ever send me, never mind late in a merge window.
Like this crazy and pointless make_u32_from_two_u16() "helper".
That thing makes the world actively a worse place to live. It's useless garbage that makes any user incomprehensible, and actively *WORSE* than not using that stupid "helper".
If you write the code out as "(a
In contrast, if you write make_u32_from_two_u16(a,b) you have not a f%^5ing clue what the word order is. IOW, you just made things *WORSE*, and you added that "helper" to a generic non-RISC-V file where people are apparently supposed to use it to make *other* code worse too.
So no. Things like this need to get bent. It does not go into generic header files, and it damn well does not happen late in the merge window.
You're on notice: no more late pull requests, and no more garbage outside the RISC-V tree.
Now, I would *hope* there's no garbage inside the RISC-V parts, but that's your choice. But things in generic headers do not get polluted by crazy stuff. And sending a big pull request the day before the merge window closes in the hope that I'm too busy to care is not a winning strategy.
So you get to try again in 6.18. EARLY in the that merge window. And without the garbage.
Torvalds' message drew a conciliatory [3]response from the submitter of the patches. "I'll stop being late, and hopefully that helps with the quality issues."
[1] https://www.phoronix.com/news/Linux-6.17-RISC-V-Rejected
[2] https://lore.kernel.org/lkml/CAHk-=wjLCqUUWd8DzG+xsOn-yVL0Q=O35U9D6j6=2DUWX52ghQ@mail.gmail.com/
[3] https://lore.kernel.org/lkml/mhng-655602B8-F102-4B0F-AF4A-4AB94A9F231F@Palmers-Mini.rwc.dabbelt.com/
Couldn't read source... mirrors? (Score:2)
Sadly the LKLM is hidden behind some ridiculous "protection". Does anyone know a mirror that's usable?
Re: (Score:1)
It's a mailing list. You subscribe to it, and you get the content through e-mail.
Re: (Score:2)
A mailing list that requires you *subscribe* to it *before* you start being sent messages? What kind of crazy world is this?
(hopefully that sets off your sarcasm detector...)
Re: (Score:2)
> Sadly the LKLM is hidden behind some ridiculous "protection".
Try searching for the LKML instead ;D
In the not too far future you will be able to get it from the internet archive, but they don't index it very often.
Actual quote from Linus, mangled in summary (Score:3, Informative)
Mangled summary:
> If you write the code out as "(a
> In contrast, if you write make_u32_from_two_u16(a,b) you have not a f%^5ing clue what the word order is. IOW, you just made things *WORSE*, and you added that "helper" to a generic non-RISC-V file where people are apparently supposed to use it to make *other* code worse too.
Apparently EditorDavid is new to Slashdot and doesn't know how to escape metacharacters. The actual text in TFA, quoting Linus:
> If you write the code out as "(a << 16) + b", you know what it does and which is the high word. Maybe you need to add a cast to make sure that 'b' doesn't have high bits that pollutes the end result, so maybe it's not going to be exactly _pretty_, but it's not going to be wrong and incomprehensible either.
> In contrast, if you write make_u32_from_two_u16(a,b) you have not a f%^5ing clue what the word order is. IOW, you just made things *WORSE*, and you added that "helper" to a generic non-RISC-V file where people are apparently supposed to use it to make *other* code worse too.
Re: (Score:2)
Somebody mod this informative. The summary now makes sense and why Linus was upset.
Re: (Score:1)
> Somebody mod this informative. The summary now makes sense and why Linus was upset.
Yeah also
> +b
could be
> +b & 0xff
E
How DO you escape meta characters? (Score:1)
Is there an FAQ for Slashdot? One would think they would have enabled their editor to accept characters without having to do funky weird stuff by now.
I hope (Score:2)
that the RISC-V silicon is better than the code.
Re: (Score:2)
Linus is not particularly impressed with it either.
He considers it a new rehashing of old mistakes.
Re: (Score:2)
> Linus is not particularly impressed with it either.
Linus has said he doesn't know Verilog and knows little about FPGA and ASIC design.
> He considers it a new rehashing of old mistakes.
RISC-V is an upgraded MIPS, which was a pretty good architecture.
Re: (Score:2)
Linus is opposed to anything RISC. He prefers Intel's ISA, probably because he learned it first.
Re: (Score:2)
I hope he's come to terms with us all being on AMD's ISA then :)
Re: (Score:2)
RISC ISAs are tied to a very specific approach about how the hardware underneath should be implemented. If the hardware isn't implemented the way the RISC instructions imply, then the entire implementation becomes far less efficient than modern non-RISC designs. (The entire concept of RISC are fixed format, fixed timing, instructions that can be decoded with a simple logic based decoder instead of microcode. But that design has performance limits, to get higher performance you need an approach closer to AMD
Re: (Score:2)
> The entire concept of RISC are fixed format, fixed timing, instructions that can be decoded with a simple logic based decoder instead of microcode. But that design has performance limits, to get higher performance you need an approach closer to AMD/Intel/Transmeta's "Translate multiple instructions into microcode and identify areas where multiple instructions can run simultaneously in real time"
I'm not sure what you're trying to say here. Are you trying to say that RISC CPUs can't be pipelined?
THE REAL CODE (Score:1)
Linus had a good code example, and Phoronix quoted it. However, Slashdot turned it into ")a".
The original code... and no I can't get it to display the two characters either using tags or backslashes...
"( a TWO-GREATER-THAN-SIGNS-FOR-BINARY-SHIFT-LEFT 16 ) + b"
E
Re: (Score:3)
Here it is:
> If you write the code out as "(a << 16) + b", you know what it does and which is the high word. Maybe you need to add a cast to make sure that 'b' doesn't have high bits that pollutes the end result, so maybe it's not going to be exactly _pretty_, but it's not going to be wrong and incomprehensible either.
Re: (Score:1)
> Here it is:
>> If you write the code out as "(a << 16) + b", you know what it does
>> and which is the high word. Maybe you need to add a cast to make sure
>> that 'b' doesn't have high bits that pollutes the end result, so maybe
>> it's not going to be exactly _pretty_, but it's not going to be wrong
>> and incomprehensible either.
First, THANKS. I know the default is to criticize everyone but you just taught me something valuable, and based on google searches, not just me. Again, thanks.
"(a << 16) + b"
For those wondering what he did, he took the LESS-THAN signs (not GREATER-THAN -- mea culpa) and used < instead.
So the quote (with extra spaces) is
> ( a & l t ; & l t ; 16 ) + b
I enclosed all those with LESS-THAN quote GREATER-THAN and then closed the quote with LESS-THAN SLASH quote GREATER-THAN like so:
> quoted-stuff-here
E
Re: (Score:2)
> For those wondering what he did...
And for those who'd like to know HOW to see the markup of another post, just reply and click "Quote Parent". You don't actually need to post it.
Re: (Score:2)
for the record, you can use some HTML entities here on slashdot. My favorite is — but obviously & is pretty good too for obvious reasons
Re: (Score:2)
> "( a TWO-GREATER-THAN-SIGNS-FOR-BINARY-SHIFT-LEFT 16 ) + b"
[cough] .. "two *less*-than signs"
Re: (Score:2)
(jeez ... no wonder we can't get back to the Moon)
NUCULAR REACTOR ON THE MOON (Score:1)
> (jeez ... no wonder we can't get back to the Moon)
September 13, 1999.
Re: (Score:2)
> Linus had a good code example, and Phoronix quoted it. However, Slashdot turned it into ")a".
> The original code... and no I can't get it to display the two characters either using tags or backslashes...
> "( a TWO-GREATER-THAN-SIGNS-FOR-BINARY-SHIFT-LEFT 16 ) + b"
(a << 16) + b.
What I typed to get that was "(a << 16) + b".
That's why Linux wins. Quality. (Score:2)
And it wins where it matters. Where waste of any type matters (Lean, Toyota). On servers, where the company cares and efficiency translates to income. Also on desktop, where the individual cares and efficiency translates to work satisfaction.
Re: (Score:2)
> Where waste of any type matters (Lean, Toyota).
Pity about Toyota's code, which we know to be trash after the code reviews (not NASA's worthless one, but the good one from the Barr Group) revealed that they not only don't follow industry best practices, they don't even follow their own documented guidelines.
We've got the solution you need (Score:2)
Are you behind schedule, work piling up faster than you can keep up? Well, we've got just the thing for you. It's the latest and bestest little vibe friend that can help you dig out from under all that late work. Just fire up Dr. Philonous' latest all around best friend you could even want. It will help you bang out all that code in a jiffy so that you can be the star you've always wanted to be. No need to check, it's all good.
Year Of Linux On The Desktop (Score:3)
Sounds like everything is on schedule for desktop dominance by the year 2140.
Re: (Score:2)
It's been on my desktop since 1997.
Re: (Score:2)
With WSL2, it's on my Windows 11 desktop and has been for a few years, working side-by-side. I get the steak AND the salad.
Re: (Score:1)
- Looks at UID.
You were alive in 1997?
Re: (Score:2)
> It's been on my desktop since 1997.
Gotcha beat. Had a DEC Alpha AXPpci 33 running Linux in 1994. Installed from 30 or so floppies.
Re: (Score:3)
30 floppies in the mid-1990s... are you sure you're not remembering Windows 95, grandpa?
Re: Year Of Linux On The Desktop (Score:2)
Did a floppy install in '94 too. CD-Rom support was sketchy back then. On PCs it was all proprietary interfaces, no IDE standard.
Re: (Score:2)
I was installing FreeBSD from two floppies and a network connection. 30 floppies is tragic.
Re: (Score:2)
> I was installing FreeBSD from two floppies and a network connection. 30 floppies is tragic.
It was faster to use 30 floppies if you had to install on several computers or wanted/needed to re-install many times. Downloading everything over dialup every time was tragic. You could put the 30 floppies content on a local NFS server for even faster installs although if that's what you meant although.
Re: (Score:2)
Why'd ya wait so long, a solid year behind the early adopters.
Re: (Score:2)
I don't consider that a "kind and gentle" response, but if he described the code accurately, it does seem warranted.
Re: (Score:2)
I have a kind of mindset when deep in coding for weeks, I don't tolerate any BS and I get terse. I forget to be polite, and almost forget that it is humans that I am talking to. I'm guessing he had a moment like that.
Re: (Score:2)
No F bombs, no anatomical suggestions. Unlikely to leave nothing but a smoking pair of shoes if said in person...
Re: (Score:2)
What, One of the few O/S'es you can have on RISC-V is linux. Windows not so much.
Re: (Score:2)
I just meant that with Linus ranting like a rabid whippet, it is no wonder that Linux is not yet dominant in the desktop consumer space.
Re:Year Of Linux On The Desktop (Score:4, Insightful)
No OS user has a problem with Linus being rough with people who try to push lousy code into the kernel. Most of them have no idea what a kernel is to begin with, or who Linus is, much less read the LKML or any coverage of what happens there.
The reason Linux isn't dominant in the desktop space boils down to two primary causes. Microsoft's strongarm tactic, and most people buying a computer the same way they buy a toaster; as an appliance which they replace when it starts smelling funny. Most people wouldn't even consider reinstalling the OS on their computer, much less replace it with something else. Of those who do, a huge chunk are already using Linux.
Re: (Score:2)
> No OS user has a problem with Linus being rough with people who try to push lousy code into the kernel.
Agreed, many systems running Linux are used in high reliability and safety critical applications. There is no place for crappy code in the kernel or important run time libraries, so we are definitely relying on people like Linus being tough on people who let problematic code into PRs.
Re:Year Of Linux On The Desktop (Score:4, Insightful)
On the contrary, I would suggest that MS could have used a few code reviews like that over the years.
Re: (Score:2)
> I just meant that with Linus ranting like a rabid whippet, it is no wonder that Linux is not yet dominant in the desktop consumer space.
Indeed, Linus should have followed the example of Bill Gates, who famously gained dominance in the desktop market by being such a very nice guy.
Re: (Score:3)
1) No Linus is mad because someone did the exact thing he told them not to do: submit code late. 2) Linus is not concerned with Linux on desktop as much as you. In past interviews he has said as much. Considering Linux has a larger server market share than Windows and Android destroys Windows in the portable and phone market, I do not think Linus is worried.
Re: (Score:2)
He's doing quality assurance. Every single time I've dug into code changes he's critical of, I have agreed with him. I don't want to suck his dick or anything, but he is obviously an extremely competent software architect and programmer.
Re: (Score:2)
Yeah, if he really wants to dominate the desktop, he should [1]be throwing chairs [wired.com] out of rage and anger.
[1] https://www.wired.com/2011/11/cloud-foundry/
Re: (Score:2)
> it is no wonder that Linux is not yet dominant in the desktop consumer space.
What's the connection? 1) No consumer at all will be in contact with Linus. Consumers don't even know he exists. 2) How is your rationale consistent with the fact that linux is the dominant OS in many professional applications?
Re: (Score:3)
You think nobody ever gave a code review at Microsoft like that?
The only difference is that you can see it here.
Re:Year Of Linux On The Desktop (Score:4, Informative)
Sounds like everything is fine, since the other side has responded and the response is.
OK, sorry. I've been dropping the ball lately and it kind of piled up
as taking a bunch of stuff late, but that just leads to me making
mistakes. So I'll stop being late, and hopefully that helps with the
quality issues
That is, the content of the Linus comment is acknowledged, people have planned to rectify the issues, there are apparently not hurt feelings and the RISC-V community and everyone else will get a safer update one cycle later.
Shit happens sometimes even to the best, which may well have been the case here.
What's not to your liking?
Re: (Score:2)
Phoronix posts some decent stuff but a lot of it is clickbait, trying to taint Linus as some sort of dictatorial sociopath. The kernel has survived more than three decades because its founder still gives a damn about code quality.
A couple of dozen contributors, one big merge and the maintainer probably didn't write or review the helper function that Linus got annoyed about. This right before the release.
Just because it compiles and 'works' doesn't mean you should ship it and as a maintainer, he'll do a more
Re: (Score:2)
Yeah, I agree, it appears to be a thing blown out of all proportion that will be forgotten next week with the fortunate side effect of less bad code in the kernel.
Re: (Score:2)
> "What's not to your liking?"
Can't speak for who you are replying to, but Linus' post was unnecessary long, repetitive, and hostile. He is right, of course, but that doesn't mean it needs to be so caustic.
Re: (Score:2)
Aw, come on, he's tamed a lot with age, you surely remember what it was like in the 90s and the early 00s...
Seriously though, probably the reason for this minor tantrum is the frustration coming from him expecting these problems and warning against it, and then them appearing anyway.
[1]https://www.techspot.com/news/... [techspot.com]
I haven't followed LKML closely for too many years to have an opinion if there is a better way to deal with code quailty problems reappearing in every new architecture. Given the track record, I'
[1] https://www.techspot.com/news/103830-linus-torvalds-worries-risc-v-repeat-same-mistakes.html
Re: (Score:2)
It IS worth noting that he did not once insult the programmer, he kept his opinion strictly to the code and its lack of quality. That's a big (and meaningful) step! And honestly, it looks like it worked.
It WAS garbage code. Even good programmers/people write bad code sometimes, and I'd rather someone tell me that the code is bad than not so I can fix it.