Cover photo
Verified name
Linus Torvalds
Works at Linux Foundation
Attended University of Helsinki
Lives in Portland, OR


Linus Torvalds

Shared publicly  - 
Hey, either Macs don't count much on the desktop, or we may have to finally lay the "year of the Linux desktop" joke to rest.
Google's low-cost Chromebooks outsold Apple's range of Macs for the first time in the US recently. While IDC doesn't typically break out Windows vs. Chromebook sales, IDC analyst Linn Huang...
Michael Shigorin's profile photoGreg Miernicki's profile photoBrian Gray's profile photoMathias Hasselmann's profile photo
No, not even remotely there yet. Back then "Linux Desktop" was a handy code for getting a free, libre desktop - in contrast to the proprietary, closed systems Microsoft, Apple, Sun, ... and the like were offering.

Today we even have fallen behind that old dream: This fenced garden platforms (iOS, Android, Windows Phone/10) and also the Cloud are even more restrictive than anything we complained about in the late 90er's, much more closed and restrictive than we could have imagined. Looking back Ballmer and Gates appear like saints.

The revolution failed. Linux and Android simply are used because they are free as in beer. Not because of any liberty. Not because of any technical supercity.
Add a comment...

Linus Torvalds

Shared publicly  - 
The new parallel lookup and readdir code from Al Viro just landed in -git.

This is a big deal: we've always serialized accesses to the same directory using the directory inode mutex, and while cached lookups (or lookups in different directories) scaled fine, that per-directory serialization could keep you from getting good performance on some loads.

I remember looking at some nasty samba load that +Andrew Tridgell had, where the serialized readdir caused tons of problems. That was over a decade ago. And it should finally be fixed now, although I suspect samba has lots of workarounds to just avoid the issue.
Niall Douglas's profile photodigi owl's profile photoMichael Simms (lgpmichael)'s profile photoNikita Danilov's profile photo
+Tom Fogal, +Robin Humble, most likely this won't help Lustre. Bad Lustre "ls -l" performance (officially known as its "Achilles heel"), is the price it pays for the scalable IO path. Because Lustre clients do writes directly to IO servers, bypassing meta-data server, only io servers collectively know exact file size. As a result, each stat(2) call incurs (in the worst case) an rpc to each io service on which the file is striped, so "ls -l" incurs tons on rpcs.

Add a comment...

Linus Torvalds

Shared publicly  - 
There's a fascinating (and pretty depressing) story over at LA Times about OxyContin.

The article itself is depressing, with he fixation on the dosing schedule apparently being so important from a marketing standpoint. Interesting.

The fact that the top comment (at least right now) then blames the addicted people makes it more depressing still, and wasn't nearly as surprising ;(
Lucas Isreal's profile photoКорень Зла's profile photoRoberto Migliorati's profile photoSvein Engelsgjerd's profile photo
I have a family member who have to use oxycontin. Not addicted yet, but what is worse constant severe pain or becoming addicted? What I find interesting by the article is there higher dose recommendation instead of more frequent dosage. Thanks for posting this article - very interesting! 
Add a comment...

Linus Torvalds

Shared publicly  - 
Some days are more productive than others.

I just spent 15 minutes researching heavy duty dog poop bags on Amazon. And I was seriously considering just ordering multiple different brands to compare when I finally came to my senses.
Ri Snyman's profile photoCameron VanNatta's profile photomoth rock's profile photolarysa fabok's profile photo
Aww, you post a good thing. If I broadcast that I pooper scoop, it just wouldn't have the same impact.
Add a comment...

Linus Torvalds

Shared publicly  - 
I'm enjoying the Youtube mixup of the SpaceX CRS-8 first stage landing with Lonely Island's "I'm on a boat" song. Congrats to everybody involved.

I'm not very hip, so the only other time I've heard that song is when somebody in my poker group is mangling it (horribly). I think it works better for Elon Musk, but maybe that is because I'm not losing money.
Jackie Moon's profile photoNilesh Kokane's profile photoArinerron -'s profile photoSee Saenghan's profile photo
 ·  Translate
Add a comment...

Linus Torvalds

Shared publicly  - 
Hey, I have a new math problem for G+ people.  This worked really well back a few years ago when I did the "figure out number of bytes from a bitmask efficiently", so let's see if somebody on G+ has the background to do a good job on a totally different math problem for a totally different open-source project.

This time, it's about gas compressibility in scuba diving, obviously for the +Subsurface project.

The issue is that scuba cylinders end up having a (small) gas compressibility factor difference from the ideal gas law, and it's noticeable enough that some people care.

For example, one of the most common cylinders, the AL80, fits 80 cubic feet of an ideal gas, but only about 77.4 cubic feet of actual air, because air at 3000 psi (the fill pressure of an AL80) has a "compressibility Z factor" of about 1.03.

And it turns out that there are many ways to get those compressibility factors, but they all have some problems.

 (a) ignore them. That's actually quite reasonable, but at higher pressures - but still relevant to scuba - the 3% error ends up being more like 10%.

 (b) use a table lookup. That's what we did for a while, and it works fine, for air. Take the air compressibility table at 300K from Wikipedia, do linear interpolation between points for pressure, and you get quite good results. But only for air.

 (c) use actual functions used by physicists. This is the next step we did, because one of the people involved is an actual physicist. The idea is that you can get the compressibility right for other gases too. Sadly, the common functions actually do pretty badly for the usual scuba gases and pressure ranges, probably because these are not ranges that most physicists care about.

 (d) use a least-square polynomial fit of the Wikipedia values. This gets you much prettier code, and good approximations, but gets us back to "just air", but at least in a prettier form than my disgusting table lookup.

 (e) something else.

My current idea for (e) is to do the same least-square fitting for the three relevant gases (Oxygen, Nitrogen and Helium) using the table at

and then just doing a linear interpolation of the compressibility of the different gases based on the actual gas mix.

Anyway, I realize this is pretty esoteric, but maybe there is somebody around on G+ that can tell whether that linear mix is a good idea or just horrible. 

Don't tell me about van der Waals equations or Redlich-Kwong. Our resident physicist already tried those. In the range we care about (temperature at about 20°C, pressure in the 1-350 bar range), they aren't actually sufficiently close to the experimental values to work..
Guitong Geng's profile photoArinerron -'s profile photo
+Linus Torvalds​ Thank you for making my life so much better by developing Linux, fellow Oregonian. I remember you saying respect is only earned. Well, you have definately earned my respect! :)
Add a comment...

Linus Torvalds

Shared publicly  - 
The Who were in town...
Olivier Caron's profile photoOrjan Sinclair's profile photoAri Lemmke's profile photoMatt Quackenbush's profile photo
The Who are the greatest live band ever. As an enormous fan, I have to say that those pics are totally awesome, but they make me miss John Entwistle even more.

Pino Pallendino is a most excellent bassist, but the fact that Pete brought in his little brother (Simon) on guitars and two additional keyboardists is a testament to just how massively difficult it is to fill the gaping hole in the sound that exists without John.
Add a comment...

Linus Torvalds

Shared publicly  - 
The neighbors had a small tree fall overnight (I think it's a golden chain, but my arboreal knowledge is non-existent).

Neither of us having anything as manly as a real chainsaw, I ended up helping him cut off the branches with my circular saw and a couple of long extension cords instead.
Andreas Pashalidis's profile photoAndreas Leitgeb's profile photoBrynjard Øvergård's profile photoPaul Frederick's profile photo
+Brynjard Øvergård
it is not fear, it is accepting reality. Which is power tools have more power than we do. A well conditioned athlete might be able to generate a half a horsepower sustained. Your common run of the mill out of shape slob more like a tenth of a horsepower. That's less than a 100 Watt light bulb. A 13 amp circular saw is 2.1 horsepower. So you're a man down from having a basketball team at the end of your arm with that saw in your hand.
Add a comment...

Linus Torvalds

Shared publicly  - 
I've lived with my cheap UHD Dell TN panel that I got for black Friday a year and a half ago. It was really cheap and really not very good, but I wanted the resolution and I felt it would last me until the good IPS panels came down in price.

Today is that day.

I had delayed and kind of hoped that I would have an OLED monitor by now (since I look at text all day, infinite contrast really would be nice), but I suspect that will be another year and a half wait for that next step.

But UHD IPS panels are pretty reasonable these days, and boy what a difference it makes. Much better color, and the electronics inside have improved too. No longer do I need to turn off dpms just because the old Dell panel would take ages to wake up and often not wake up reliably at all.

It's an LG 27" IPS panel in case anybody cares. It got delivered just an hour ago, but so far so great.
Linus Torvalds's profile photoMaan Abou Hatab's profile photoHerman Bovens's profile photoKalin Kozhuharov's profile photo
There is no coming back after I started using 4K 40" displays at home and work... Cheap, Philips BDM4065UC/11 via DP on skylake i7 (I don't need 3D accel) works great. I can crank fonts up to 24pt and still have enough text, perfectly visible from probably a meter. Or go down to 12 and see tons of long log lines or CSV, etc. DSLR photos also work great, at least the one I managed to take good (slightly out of focus or blurred look horrible though!)
Add a comment...

Linus Torvalds

Shared publicly  - 
I got my 23andme report back.

I was kind of hoping for something interesting, but I guess this is one of those "it's better when it's boring" things. No interesting genetic disorders.

But now I know that I have blue eyes and very low chance of unibrow. Whew. All those mirrors weren't lying to me all those years, and it's not my wife sneakily plucking my brow while I'm asleep.

I'm also apparently 99.7% European, and likely to drink more coffee than average. Yeah, that didn't come as a big shock either.
Свежий «Freshman47» Человек's profile photopeter scoggins's profile photoChet Gurevitch's profile photoMarkus Åberg's profile photo
+Linus Torvalds On paper we are 8th cousins + some other more distant connections but due to lack of genetic diversity in Ostrobothnian villages centuries ago that kind of "adds up" in the genes I expect 23andme to classify us as something like "3rd-6th cousins" or at least "4th to distant". My 23andme profile is public so all 23andme users should be able to see me in the "DNA relatives" list
Add a comment...

Linus Torvalds

Shared publicly  - 
So I wonder what the excuse is from the apologists that think we don't need ECC in the consumer space this time around?

If you are selling CPUs, and you "differentiate" your products around ECC, you should be ashamed of yourself.  It's not acceptable.

Yeah, I've complained about lack of ECC for over a decade now (you know who you are). There's always some f*cking excuse for why some SKU or other doesn't really "need" it.

Stop it with the excuses already. 
New research finds "bitflipping" attacks may pose more risk than many admit.
Tokollogo Maluleke's profile photoGuitong Geng's profile photoRyan Voots's profile photoYuhong Bao's profile photo
I should also mention that DDR3 non-ECC SO-DIMMs and DDR3 ECC SO-DIMMs have different pinouts! Fortunately this is fixed in DDR4.
Add a comment...

Linus Torvalds

Shared publicly  - 
Went to the Bay Area for Daniela's gym meet over the weekend, and watched the local evening news on Saturday.

I'm convinced the discussion in the newsroom before their weather segment went something like this:

 "Should we kick some puppies onstage tonight?"

 "No, we did that last weekend. How do you feel about pentagrams and human sacrifice?"

 "Wait, I know - there are people from Portland in town. How about we make the whole weather segment be about how we're finally getting some much-needed rain? Let's look extra happy about it!"

 "Mwhahahahhhahh! Perfect!"

... and then they went on to do exactly that.

Sean Roark's profile photoAndreas Leitgeb's profile photoSuraj Mishra's profile photo
+Linus Torvalds Sir, don't have enough words in any Language :-|
Add a comment...
Creator of Linux and git
  • University of Helsinki
Basic Information
SW Engineer
  • Linux Foundation
    SW Engineer, present
  • OSDL
  • Transmeta Corp
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Portland, OR
Helsinki, Finland - Bay Area, CA
I've been to Fish 'n fins twice now, once during off-season (June), and once during high season (November). Very professional. They clearly get very crowded during high season, but seem to successfully juggle having multiple boats out. Good diving, good people.
Public - 2 years ago
reviewed 2 years ago
1 review