B1 Intermediate 3119 Folder Collection
After playing the video, you can click or select the word to look it up in the dictionary.
Loading...
Report Subtitle Errors
>> Hi everybody. Since we started a little bit late, Iím going to keep this nice and
short unlike our guest, Greg, who is quite tall. And--yeah. This actually--it rose out
of a discussion we were having between some Kernel folks that I know, and Bryan [INDISTINCT]
and right over here, about sort of the nature of the embedded system--what's that?
>> It might not [INDISTINCT] >> Well, no, I mean, like from outside the
company, inside the company like--donít--I'm speaking here. All right?
>> Go ahead. >> Anyway, and so, you know, at the end of
this today, I want you to come out and speak. It would be fun for us and fun for you and
stuff. So, here we are, Greg. >> HARTMAN: Thanks. Is this working? All right.
I'm Greg Kroah Hartman, Iím a Linux Kernel developer. Iím a sub-system maintainer of
the Kernel for USB, driver Core, I used to do PCI, I like to see all the weird things.
I'm one half of the stable Kernel maintainer group which I'll talk about later. So here,
we're going to talk about the Kernel, the process, who's doing it, how itís being done,
why you should care and why Google should care and things like that. Please tackle--ask
questions in the middle, I like it. Make fun of me. I'm going to make fun of you. So, beat
them. So, first thing. This is the most important thing, donít get anything. Realize this.
This is what the Linux Kernel do for the past year and a half from the 2.6.20 to 2.6.25
Kernel. This is what was changed per day in the Kernel. This is a stable Kernel. This
is a stable development process. This is freaking insane. It is the fastest moving so--project--so--software
project known to anybody. Largest project of its kind known to anybody but it's also
created a result thatís never been done before and we have an operating system that supports
more devices than any of our operating system ever has. We support more processors than
any other operating system ever has in the history of computing. And we're doing it in
an open manner, in a very chaotic manner. Moving at a rate, that fast; shipping something
thatís stable, that everybody, including yourself, uses everyday. We broke all the
software development rules and we're continuing to break it. If that doesnít make you guys
scared, it should. The biggest thing to that make you guys--should make you guys scared
is, if you keep your code outside the Kernel, thatís going to pass you by. Do you think
you can keep up with that kind of rate of change, figure out what's going on with that
rate of change? If you get your code into the Kernel, that rate of change will happen
to you automatically. If you donít, it will blow you by. So, get your code in the Kernel.
It's the main thing I want to talk about. So, thatís it. Thatís crazy. And also, this
is the first question people ask like, "Oh, thatís jus drivers, right? Drivers are changing
that fast.î No. It's the whole Kernel. I looked at it, Iíve broke it down by the proportion
of the code--size of the Kernel. The Kernel is about nine million lines of code. The Core--the
Kernelís five percent, driverís a little over 50%, 25%. If you look at the rate of
change at each individual portion, it's exactly proportion--itís exactly proportional to
the amount of code thatís in the Kernel for that section. So, the Core of the Kernel is
changing at 5% of that rate. The driver section is changing at 55% of that rate. The architecture
specific spot is going up--what is that? I donít know, it's going at 30% of that rate.
So, it's uniform. It's flat across the whole thing. So, donít think that just--that the
drivers are changing, it's the whole thing; it's everything. So, and along with that rate
of change, it breaks down to this. It's going really fast. Three point sixty nine changes
an hour, 24 hours a day, seven days a week. During the 2.5, the 2.6 development cycle,
we're running about 2.5 changes an hour. And that was unheard of; no way that Iíd ever
thought we could do that. Since then, we've been going up. I donít know why. We're getting
faster. We're getting faster. We're doing more rate of change and it's getting bigger.
So, this is some of itís staple. So, here's our graph. If you want to see all the--that
lines of code that are added and modified, diluted over time, I just did for the 26 [INDISTINCT]
and long. If you look--if you used that, it was a little bit less. So, it's--itís really--itís
kind of staying around the same but at lines out, it keeps going up. Thatís interesting.
Thatís not really trend worthy but it is continuously going up. But itís trend worthy
is the changes per hour. It's continuing to go up. 2.6.25 changes per hour is amazing
. Well, that rate of change does not included moving files around. Moving files around the
Kernel doesnít affect it at all. It does not change at all. So, we're going faster,
doing more rate of change per hour per release. Everybody happy? Nobodyís questioning that?
Oh, come on. All right. So, how we do this? Hereís how we do this. Extensively, it starts
off at the developer on the bottom to make a change to a file, to--they find a bug, they
want to add a new driver, do something. Then it takes that change, the path, should they
send it off to the maintainer or the driver of the file. In the Kernel, we have about
600 unique maintainers that are listed on who owns what. You canít find what--who owns
what, you can look in the file using git and see who modified it last. Whoever touched
the floppy driver last is the maintainer. Never touch the floppy driver. So, that personís
responsible for looking at it and then they see it, they review it and then they sign
off by it. They say, ìThis looks good to me. I think itís okay.î And they add their
signature called Sign Off. And Iíll talk a little bit more about that later. So, the
developer says, the sign off, itís good. They say itís good. They send it off on to
the subsystem maintainer. Subsystem maintainer like USB, PCI, BFS, security, Core, things
like that. They consolidate all the patches from all data people and then they hold on
to them. They create this little mini trees everywhere. These little mini trees are either
in gits, using quilt, somebody--I think thatís all everybody uses, some tarballs maybe. And
then all that, every night gets pulled into the Linux-next Tree. Itís from Steven Rothwell
in Australia. Which--for IBM, the Aus labs. He pulls it in every night, the Linux-next
Tree merges them all together and sees if it builds or not. Sends you angry emails or
you break the bill or you break the merge, itís great. Itís awesome. Itís testing
all the subsystem maintainer stuff for all our development things. Andrew at the same
is creating--is now relying on Linux next and sucking everything in the host tree, about
once a week or so. And then also, he pulls things from the Linux Kernel Mailing List,
a sort of mailing list that happen to get drop by subsystem maintainers or whatnot,
he gets--he sucks those under his tree and everybody--and then he does a release every
couple of weeks. So thatís testing, this is the developmental stuff. So all the crazy
stuff the subsystem maintainers are holding on to, all the stuff thatís coming across
the mailing list but Andrew picks it up, gets pulled to these releases. Then when Linus
says itís the merge tree--windows open. Iíll talk more about that later. All the subsystem
maintainers hit Linus. We got two weeks to hit Linus with all the changes you want. We
all go to Linus. Andrew sends a bunch of stuff to Linus after the subsystem maintainers get
there and then we go on, the cycle releases. Iíll talk more about releases later. So,
nice little tree structure. So, Linus and Andrew and the next tree. Does that make sense?
Nice hierarchy, right? Nice and pretty. So, Iíd try to graph this. Looking at the relationship
between people and the work--way patches flow to the tree. I think itíd be a nice pyramid
or whatnot. It turned out, it isnít. I made a graph that was about 40 feet long and 4
feet tall, of one Kernel release. It was a mess. Itís a big giant spaghetti mess. It
looks like an internet routing system. And which is good because it means that, if this
person doesnít pay attention, we route the packets around them, Andrew picks it up and
then sends it to subsystem maintainer, it sends it to Linus or this person sends it
directly to there and goes around the maintainer or this person disappears for a week. So,
itís very good. Itís lossy. We can handle people disappearing, handle people come--reappearing
and it works out really well. It isnít a strict hierarchical development system which
is good and it shows how we can keep up this massive rate of change so fast. Itís kind
of scary. If you own a file or a driver, you need to accept the fact that other people
are going to be changing it, extensively, without you knowing it but hopefully you should
be paying attention and you just have to accept that so other people can do that. You can
always change it back if you didnít like it. But thatís the way we get things done.
Thatís how we can work so fast. Make sense? Cool. So, hereís a release cycle. So, Linus
does the release based on something. We say itís good. So then, we have two weeks. We
can throw anything we want at Linus. All the subsystem maintainers throw everything we
want. Extensively, itís been tested in the Linux-next Tree, before that I have been testing
the MM3 from Andrew. So, we hopefully expect that the code has been tested, it builds,
it hopefully compiles sometimes and all merges. So, thatís when the hundreds of thousands
of patches go in at once, two weeks. Linux releases an RC1 Kernel. Then the subsystem
maintainers that--were on vacation or forgot about the merge window being open, wake up
and say, ìOh, no. I forgot my stuff.î They send it to Linus again so, then goes an rc2
a couple of weeks later. Hopefully, thatís getting better. People are slowing down. Everything
happens between rc1 and rc2. Then, its bug fixes. We start doing bug fixes, we track
bug fixes, regressions, we do RC1, 3, 4, 5, whatever. Until we think things are slowing
down and good enough and boom when it says, ìOkay. This oneís good. Out the door.î
And then we start the whole cycle over again. There is no stable and unstable development
trees anymore. You step odd was unstable and development even was stable, not anymore.
For the past four years, weíve been doing this model. When we finally realize that this
model really worked good, we said, we should try and do release every two to three months,
right? If you look, weíre releasing on an average of every two and three quarter months,
for the past three to four years, like Clockwork, you can trust that a new Kernel will come
out between two and three months. This is a really, really good thing. Unless the up
stream just rose, figure out in their schedules when things are going to be done within a--within
a reasonable time frame that can plan when things are going to come out. The [INDISTINCT]
can figure that out, developers can figure that out, ìI miss this merge window, whenís
the next one going to come around?î You don't have to worry that their selves are going
to get drop for forever. It works really, really well. One thing that happened though
is we realized during these two to three month time, what's happens about security bugs?
These are people that actually use the Kernel.org--Kernel. What happens then? Thereís bugs, security
problems happen, what are we going to do? So, we formed--two people formed the stable
tree which is me and Chris Wright who works for Red Hat and we do security updates or
bug fixes; 0.1, 0.2, 0.3, 0.4, 0.5 and we have a real strict list of rules of what we
will accept, what we won't accept. It's pretty easy. It has to be bug fixes only and security
fixes. The patches have to be in the upstream tree already, unless thereís some really,
really odd thing, things like that. It must be small, obvious. We do those releases every
couple of weeks, whenever we get around to it. Sometimes we're slacking, sometimes we're
not. Security updates, we can do it usually if we got an inbox. We can do it in couple
of hours. So, we release things as soon as possible if we can. And that's working out
really, really good. The fun things is, after the--we drop them, after 2.6.20 comes out,
it might be one more 2.6.1.9 stable and then we stop. We don't have to do it. Weíll have
to maintain that for forever. You can maintain it for forever if you want to. Somebody picked
up the 2.6.1.6 Kernel, decided to maintain it on his own for a long, long time. We all
thought he was crazy. He got a job out of it. So, maybe he wasnít so crazy at all.
So that he could do things over time. So, it worked out really well. If you want--if
anybody wants the 2.6.2.4 Kernel to maintain it for a while, let me know. Weíll give it
to you. We don't care. So, that's how it works. If itís stable, releases happen, development
and releases happen and then we drop stable trees. So, we don't have to maintain them
for forever, which given our rated change previously is a good thing to do. Makes sense?
Everybody happy with that? Is it working? All right.
>> So, you guys don't have an open analysis [INDISTINCT] every blogger really preferred
at least [INDISTINCT] >> HARTMAN: We don't have a notion of a preferred
release anymore, no. Every release is a preferred one, the latest one. Based on the rate of
change, if you report a bug--a bug problem on 0.2--on 0.2 Kernels ago, it's given how
fast Kernelís often changing. We're going to ask you, can you reproduce this on the
current version? It's the first thing we're going to ask you. So, yeah. So, everythingís
preferred. The last one is preferred. Now, note when the distros do releases, we know
what Kernels are going to release. So, sometimes those Kernels hang around a little bit longer.
Fedora and OpenSUSE are based on 2.6.2.5 right now. Odds are that Kernelís going to hang
around for a little while because Chris Wright works Red Hat and I work for SUSE. So, weíll
probably want to hold on to that Kernel for a little while longer, make sure bug fixes
get in there and it's good. But, if you want the new features and things like that, you
should go to the latest one. [INDISTINCT] Yeah.
>> Why use the two-week merge window, why not the [INDISTINCT]
>> HARTMAN: Why two-week window, merge [INDISTINCT] what?
>> Why accept merges all the time? >> HARTMAN: Why Linus accepts merges all the
time? Because we want to be able to throw a bunch of things and we wanted to shake things
out and stabilize them. We want you to test and feel good that we can--at the very end,
we don't want to accept the massive bunch of USB changes that nobody has really tested
or run for the past three months--three weeks or so, you know, we want to at least some
kind of software--good engineering practices. It looks like it something, we know what weíre
doing here. Yeah. >> Thereís never a period of where thereís
no merging or [INDISTINCT] their merging over to something newer and then the RC what windows,
youíre shutting off with, is that correct? >> HARTMAN: So thereís no...
>> [INDISTINCT] gaps where... >> HARTMAN: Thereís no gaps where were not
accepting any merges, yes. Well, we think of, yeah. When we--we use gits to send things
to Linus. So, he pulls from the--the subsystem maintainers trees and put--merges them in.
So, that's where we merge things but it should be bug fixes only. From RC2 on, unless it's
a really odd thing. We should have bug fixes only are going in, into the merges. I think
of merging as a new feature, I'm sorry. Itís a wrong terminology. I use--you can use git
to send this stuff to Linus. Linus came here and talked about git. I use quilt. Quiltís
really good too. So, yes. >> So, how do we test them?
>> HARTMAN: How do we test? >> Yes.
>> HARTMAN: One thing about the Kernel is, there is no good way to test it except running
it. You canít run like unit test--you can write unit test that extensively test some
of the external features, such as the system call interface, it hasn't broken some odd
things. But really it is testing--the operating system is running on your machine. There are
so many hundreds of thousands of permutations or devices and interactions with timing and
everything else that we wanted to run on all your machines. So, we rely on you to test.
We rely on the developers to test and that's the only way we can obviously test. Like--classic
example is, I had a problem when you were--you were plugging in two USB devices at once and
yanking one out, the data was being streamed out too. You canít unit test that unless
I want to put--I have a big box of USB devices that I do test a lot, but it doesn't make
for an automated system. So, we ask you to test it yourself. We rely on the development
community to test. That's all we can do. There isn't a same way to write a unit test for
an [INDISTINCT] it just doesn't work because we're relying on so many external inputs.
People keep trying and that's great and we want as many test. We want as many functional
tests as we can get. But like performance--and we performance test also. People are trying
to test it but it's a real hard thing to do that.
>> So there is no formal--letís say, we have between--like, demos in the 2.6.25 and the
last one out, [INDISTINCT] they were waiting for someone to do something for it. Is this
a formal procedure or does somebody does [INDISTINCT] and what is something that you [INDISTINCT]
>> HARTMAN: What if--so, what was holding it up and what was the formal procedure? We
do track regressions. So, like people record a regression, ìHey, this worked on 2620 and
2.6.21 RC whatever, itís broken.î Regressions retract. So, towards the end of the release
cycle we say, have we fixed all the regressions? Have we fixed all the bugs that have been
reported? Some people are trying to track the bugs that have been reported. If it isnít
a regression and it's just as a bug that always been there or something that new happens or
for something that's been added that's brand new and newer support [INDISTINCT]. We don't
really treat that as important as a regression from the--actually machines working. We want
to make sure we never break a machine that was working previously. We ripped whole features
out because we broke the machine that was working properly. Which is tough, ACPI and
some interaction and some low level platform code is nasty. So, we change that and it breaks
whole models of laptop, thereís the IBM series of laptops that was notorious for breaking
whenever we change anything in ACPI. Itís tough. So, we try and track those and that's
what's keeping us, ìLook, we know that thereís this one class of bugs that a bunch of people
are hitting, letís try and fix that up before we do get between RC 7 and the final release.
So, we do keep track of that. So, if you have a problem, let us know. So, as a developer
and as a user, if you want to help Linus out, test these RC releases. Test them and let
us know if there's problem, you have regressions or not and they're good. I mean, all these
are run on everybody's machines. You don't have to worry too much about these--these
things. I wouldnít run it on something thatís production wise but definitely run it on your
local machines--around on my laptop, all the other ones. If you want to test what's--if
you want to help us test what's going to happen next, test the Linux-next Tree, test Andrew
Mortonís tree. That shows that crazy developmental stuff thatís still under heavy development
that were trying to merge in and figure out what is good enough to go into the next merge
window. We got features in there that aren't really good enough to go. There might be a
couple releases away, things like that. What's the question? Okay. All right. Move on. Like
I said, we're doing this every two and three quarter months. So, we're actually doing what
we said we were going to do and that's a good thing, that's a time-base release cycle; itís
turned out to be a really nice thing. It works really well. A lot of Open Source projects
are kind of switching towards it, it works. You know that all of [INDISTINCT] this feature--this
window. Two and three quarter--three months later, itíll get in. It's a good thing to
do. 9.2 million lines, we've been growing at a rate of 10% increase in code size per
year since I started tracking it back to 260. It's a linear graph. Itís just pretty much
straight which is amazing. We're still growing. We're still getting bigger. That's a pretty
big chunk of code. Now, not all machines run at 9.2 million lines of code, that's funding.
You write in the Core and a few drivers around it, not working and things like that. So,
that's everything. That's all the different platforms, everything else. And in the last
year and a half, this is the number that I find the most impressive. We had 2,399 unique
contributors to the Kernel and that's amazing. That's showing we're doing really, really
well. Now, half of these contributors only contributed one patch. Half the half contributed
two, half the half the half contributed three; well, exponential curve there. But the top
of the curve is getting flatter. When we did the 2.6, the 2.5 or 2.6, the top 20 people
did 80--8%--80% of the work. Now, the top 30 people did 30% of the work. Just by quantity
of patches. So, we are flattening out and weíre spreading the load out a lot more evenly.
I looked--it was 200 people--the top 200 people, do they--no. Two hundred people did 8%? Something--I
was thinking of the top 200 people that did a lot of work. I was spreading a lot better.
So, that's good. I mean, here's the--the backgroundís graph. I did the crazy graph. It's on kernel.org,
you can try and load it in your browser. It's a good SVG test. It broke [INDISTINCT] machine,
crashed Mackintosh, it only--I got somebody from HP to print it out, he had printers in
his barn. Big roller printers in his barn, he printed out that. You can go through him.
So, thatís our number of developers, thatís good. And our number of individual developers,
per Kernel release is going up also. A number of companies that are actually sponsoring
developers is also going up which is a really good thing, showing more companies are getting
involved as a spread of Linux goes bigger. So, letís talk about people, quantity. We
can track people like quantity, this is just quantity not quality. I'm not going to talk
about quality. So, this is the number of changes that got into the Kernel. Top 10 people for
the past year and half; number one, Adrian Bunk, there's a lot of janitorial patches.
Al Viro of BFS. Thomas Gleixner, the X86--no, i386 next 86 arch maintainer along with Ingo
Molnar. David Miller networking Core. Bart, IDE. Paul Mundt is SH. Ralf Baechle is MIPS
maintainer. Ingo again, Patrick McHardy is net filter and Tejun is Lab ATA, disk drive
stuff. So, two of these drive guys are top 10. Thatís quantity. Thatís a number of
changes that they got in the Kernel. Yes, [INDISTINCT] So, signed off by. Remember how
I mentioned every person along the hierarchal tree signed off on it actually reviewed it.
Here's the top signed-off, this shows who are the gate keepers and how--who's doing
what. Andrew is the top. Linus is next. Now, the funny thing is, Andrew has more than Linus,
so that everything goes to Linus. Linus doesnít sign off on patches he pulls from sub-maintainers,
we're using git because he isnít looking at them. Heís trusting those maintainers
to do the thing. So, I make up a git tree for Linus to it pull from. He says, ìGreat.î
He pull from it and he doesnít sign off on those. That's why his number is lower than
Andrew's. So actual the real number patches that got in is less than the sum of those
two, itís kind of funny. So again, David Miller in networking. Jeff Grazik is in network
drivers. Ingo, architecture for Intel. Me, Thomas are in Intel. Mauro does video for
Linux, little webcam stuff. He does that in his own spare time. Works for a telecom company
in Brazil, awesome stuff. Paul Mackerras is Power PC. John Linville is wireless drivers.
So, you can kind of see what areas are the code, Kernel is doing a lot of activity. So
again, video for Linux is a very active area, wireless is a very active area. Interesting.
So, let's see who is funding these companies. So, we can see the people, receive the number
of changes it does. So, who's doing this work? So again, for the past year and half, Red
Hat, IBM, Novell, Intel, Oracle, Consultants, these are individuals working for consultancy,
they are being paid by other companies or working for consulting companies themselves.
Academia, I'm very happy to see Academiaís actually going up, at numbers going up. And
Renesas, Renesas? I think there an SH maintainer--developer. Which is really interesting, they just pop
in the top 10. So, you might want to guess what's number one to five? Come on. Canonical?
I'll tell you where Canonical is later. Well, I'll tell you where they are now. They're
about 300th, I think. Canonical had six changes in the past five years. Canonical does not
give back to the community, that's what they do.
>> Linux foundation. >> HARTMAN: Linux Foundation, no. Linux Foundation
is not on the list anymore because they only have one developer.
>> They [INDISTINCT] dock book. >> HARTMAN: They log to--I'm sorry log what?
Dock book? Docks has to do with others and that goes in the Kernel. Anyway, come on.
>> SUSE? >> HARTMAN: SUSE is Novell. So, Novell, SUSE
same to go. HP, no. HP [INDISTINCT] all right, Iíll just go. You guess. Amateurs, number
one. These are people working for themselves. We have asked them who they work for and say
I'm a hobbyist and doing something on time. And the number five is, unknown. They are
either, we have no idea who they're working for or that they wish to remain unknown. Some
people say, ìIím not allowed to say who I work for.î Hereís my [INDISTINCT] So,
we track this stuff. So, between the--some of those two, itís almost 25, almost one
quarter of the number of changes that go into the Kernel. Who's being done by people that
are not being sponsored to do it. And another thing about these unknown individuals, none
of them--well, I think--I think about five of them are contributing more than 10 patches.
So, there are lots and lots of people in the tail that we donít know who theyíre working
for. So, itís not a large--they're not large contributors. So--or you can look at the other
way, 75% of the development thatís on the Kernel is paid for, is sponsored. Because
these companies care about what happens to Linux. If you want to see what's going to
happen to the Kernel, what's going to work for you, you need to influence it and realize
what's going to be coming in Kernel? What's next? It's whatever you make it to be. Whatever
you want it to be lots and lots of companies are working to--are relying on Linux. You
guys rely on Linux. Lots of other companies, Iíve given this talk to a number of companies.
Only one company Iíve ever given this talk to is actually on this list. What? Google's
not here. >> [INDISTINCT]
>> HARTMAN: Where is Google? Googleís 13th. If you don't count Andrew; so, without Andrew...
>> We count Andrew. >> HARTMAN: You count Andrew. I count Andrew
too. Andrew does great work. But--so, I was trying to think of a metric, do you--do you
do the mean of a company's developers, you know, the look of a massive body or do you
count the net quantity but in smaller companies of computer developers, how do you rank them?
I don't know how. I'm trying to figure out some fun stuff with statistics. My wifeís
in Graduate School, she happens to be taking statistics class right now. Since she has
lots of ideas, I can use her. So, yes. Without Andrew, you guys are 40th. You guys rely on
Linux. This either means one or two things; that the way Linux is going and the Kernelís
going--and working, itís just fine for you. And youíll like it and youíll love it and
you're happy. That's great. Lots of people are that way but the other thing is, do you
trust these companies to develop the Kernel in a way that matters to you. I would say
no. >> Why?
>> HARTMAN: You trust these guys? They don't know your business as well as you do. You
only know the loads that you do as well as you do, so if it's going as well then fine,
I'm happy and I'm not going to complain about it. Yes, [INDISTINCT]
>> No, I mean, one thing that happened to us on the Andrew type thing is, they will
move support for--I know a trial [INDISTINCT] this is [INDISTINCT]
>> HARTMAN: Yes. >> If we invented more active earlier on,
something might be, you know, and check it. I think they mention only one thing in the
Kernel that uses this and is just not working. >> HARTMAN: Yes.
>> Maybe if we were in there, we might have been involved in this setting. They might
has still been removed but... >> HARTMAN: At least you should know. I mean,
if there's things you care about, that's fine. And that's fine. If you care about--talk about
this thing, that's great. When I come is I'll talk to you and say, ìWell, company ranking
so and so is doing our work for us,î which is great too. All these companies, I know
IBM or Red Hat and Novell. We both do consulting for other companies and do work for them.
But again, their overworked companies and there's no reason why the individual company
yourself can't get involved. We make it very easy to get involved in the Kernel element
process based on our--number of people who are contributing, it's kind of working so
there's no excuse for you guys not to contribute. So, that's my reason. So, let's pick-out some
and then weíll [INDISTINCT] so, 27 people are in the list. Twenty-seven of you guys
contributed a code [INDISTINCT] and here you are. So, Andrew number 11, awesome. Great.
Very happy. David, number 149th, see here, that's almost coming. So, he is a perfect
example of what I like to say is a great person to have. He started of working at [INDISTINCT]
University doing janitorial Kernel work. He looked at the Kernel, here's a fix, I can
do, start running into it, got hired by you guys, I don't know what he does, and has contributed
now to the Core Kernel, the, BF--to the--what does he do? NM stuff, out of memory stuff,
Core Kernel guy, he's contributing pretty well, 67. That's in the top 8% of all Kernel
developers like quantity out there right now, great job. Paul, what does Paul do? I looked
at CPU scheduling, that's good. Ken does what? He used TLB and other sort of stuff. Nice
Core Kernel work, really good. Then there's Matt, is Matt here? Okay. Matt puts everybody
else in this company to shame. Matt contributed spelling fixes to the Kernel, which is awesome,
and I will not say--and I love him and that's a really good to have but Matt beat out all
your other Kernel guys, he's number five out of the whole company contributing changes
that fix spelling. So if you're a developer, I donít--you should feel a little ashamed
about that. I mean, I gladly take the spelling change. I hope they contribute to more in
the future but come on. All right, guys. And then I didnít--I gave up on looking at [INDISTINCT]
what you guys do. >> How does the 27 compare to the other companies?
>> HARTMAN: On quantity? Number of developers? >> Number of developers.
>> HARTMAN: It's pretty good. There's some companies with like 90 some, which is amazing
like that--but then there's smaller companies with like five developers that are very high
in the list too. So it's like I can't look at pure quantity of unique contributor. I
just can't figure out a number. Do I look at the mean, the contributing one? I look
at the mean here, you're way down to seven, you know. I think you guys want me to do that.
So Martin, where is Martin? Martin disappeared. So the joke is when anybody ever goes to work
for Google, they enter the black hole. Never to be seen from again, Martin, you still actually
contribute to the Kernel. I know Martin, so I can make fun of him. Here's some more, two
with patches, two patches, two patches and we start getting to one. Al, thatís cool
to see Al works here. He used to contribute a lot; used to and then there's the rest.
>> Just that there are so many interesting things.
>> HARTMAN: There are so many interesting--and thatís great and thatís awesome and I have
nothing against that. But if you can--if you--and I know you guys do a lot of development in-house
and have your own custom set-up and certain things like that, which is great and fine.
But if you think you can keep your changes outside the main Kernel tree and keep up with
that rate of change, I think you're going to have a problem. It's really, really hard
to do even with our good tools that handle merging, git and things like that. I know
this really doesnít show and none of that Android stuff has landed, hopefully there's
Android shuffled in. >> [INDISTINCT] know, your work is passion
able. >> HARTMAN: Okay. Four [INDISTINCT] already,
I can handle it. >> Do we have a locker room? I'm hoping to
get that [INDISTINCT] >> HARTMAN: Hopefully, yeah. I've heard--I've
heard that before. But yes, I trust you guys. I trust you guys will gather the stuff but,
so. >> [INDISTINCT] trust yourself this [INDISTINCT]
>> HARTMAN: Trust yourself just to maintain it.
>> There's a lot of talk about how easy it is to contribute and your mileage can vary
sometimes. >> HARTMAN: Your mileage varies sometimes
and how--yes, and how easy it is. If you ever have a problem with getting things in, talk
to me, talk to Andrew, talk to the other people. We will help make this happen. I have developers
that are willing to help out with cleaning stuff up, we've take--we take code from other
companies that are not the proper coding style, we'll clean it up for you. We'll maintain
and we'll get it in the Kernel. We do this kind of stuff all the time so it shouldnít
be here. So, my main point here, you guys rely on Linux, there's no reason why you guys
can't get involved. You guys should contribute and help and drive it in a way that you feel
is going to work best for you. Otherwise, it's going to work--good then I hope--and
I hope it still works, so. Everybody asks what's new. Here are some of the things that
happened in 2.6.26. It's a joke to tell you everything that happened new. It's a lot of
stuff happened, a lot, a lot of the stuff. So any questions? That was fast. In the back,
yeah. >> When you talk about the growth of the number
of lines imposed with the Linux, I think they probably have 95% of that is device drivers
orÖ >> HARTMAN: Well, maybe the talk of the code--95%
of these drivers, no. I think you missed in the beginning, 50% of the code in the Kernels
is drivers--55% of codes in the Kernels are drivers. The rate of change is proport--is
exactly proportional to the amount of code that is in the Kernel. So those 4,000 some
lines added, 55% of that is in the driver's subsystem. So 5% of that is in the core Kernel.
It's proportional across the way. So the majority is drivers by a small bit, after driver's
is architecture specific stuff then networking but that's it, so it isnít all drivers. But
it's scary in a way, that means your Core of the Kernel is changing that fast. The Core
of the Kernel is changing at a rate that blows all other operating systems away just by looking
at the map or just touching the core or the whole subsystem, think about that slurs. Yes?
>> So what would you say is driving the change to the Core Kernel? Is it because people just
keep getting up better ways of doing things or because the requirements are constantly
changing? >> HARTMAN: So, what's driving the change--the
thing in the--what's driving the real change in the core? It is the fact that Linux is--I
mean, Linux always jokes--Linux is not--is not intelligent design, it's evolution which
means we react to stimuli thatís happening in the world. Just like, "No, we have to support
this new architecture which means we have to change the core to do this now." We donít
over plan things. We take what's happening right now. So, yes, new features happen and
we add and we figure out how to change the core again to work. There's people who come
up with, "Hey, I have good idea that we can get this--if we rewrite these--we rewrite
them--this locks to new [INDISTINCT] instead of spin locks, we can get the RT stuff in
there and get real-time in the core. The real-time stuff shows that you can take these crazy
out of the tree patches and merge them into the Core Kernel and now we have a real-time
where Kernel of Linux--some of the people said that it can never be done, which is pretty
amazing from the [INDISTINCT] so, it's everything. Itís just responding to external stimuli.
So we're not over planning, we're reacting at the time that it needs to be done. [INDISTINCT]
a lag a little bit, so some hardware companies would come up with a new device and like,
"Oh, crap, we you rewrite this." Luckily we're getting better and these hardware companies
are getting involved in the process. Look at Intel in that list. Intel's very high.
They're very big and Linux because Linux runs on all those chips they want to sell chips
they make sure work's really good on their chips. [INDISTINCT] penguin picture. Yeah.
>> MALE: So why donít you guys don't we pay more attention to consistency?
>> HARTMAN: Why don't we pay more attention to consistency...
>> MALE: [INDISTINCT] >> HARTMAN: Something's starting to change
rapidly and our interface would do change rapidly. Hopefully, we never change the user
space interface. So, we have the boot line changing for them to do stuff. I don't know
maybe they had a really good reason to do that. Hopefully they did and that sounds like
a foolish thing to do. We hopefully never change--we change the internal Kernel API
all the time. Everybody was like, "You have to have a frozen. You have to have a state
API." Then you say, "What API?" We have 8--9,000 symbols, we have 12,000 different structures.
I mean that are passed around inside of the Kernel. You can't freeze that because it doesnít
let's you ever change anything. So inside the Kernel we change all the time, we change
rapidly, we have really good reasons for doing that. But outside the Kernel user space API's,
we should never change. I can run binaries ever built for 0.696 just flowing today on
the Kernel, 2.6 Kernel today, cisco interfaces don't change. For the most part, there's a
little gray area of some interfaces that were on the Kernel and user space like the proc
file system that--end up changing every once in a while but hopefully we catch all the--the
users around that. But we're trying to document that. We're documenting that user space API
now. There's documentary [INDISTINCT] ABI, that we document that. The command line stuff--we
do document that although change was odd, I don't know why. It shouldn't have. You should
complain to developers. Maybe they didn't realize the broke it. Sometimes that--like,
"Oh, I didn't realize I broke that." You know, like, "That's a book. I need to go fix it."
Regressions are always going to happen and hopefully we fix this. In the back.
>> MALE: Once a upon a time, you advocated the distribution [INDISTINCT]
>> HARTMAN: Yes. >> MALE: [INDISTINCT]
>> HARTMAN: Yes. >> MALE: [INDISTINCT]
>> HARTMAN: So the question is awhile ago, I said that enterprise Kernel should just
grab the Kernel all the time for there's enterprise Kernels. While Andrew keeps complaining that
the user space that are quality going down the reporting. I think Andrew's wrong. I see
every single Kernel bug that comes in the three different districts or to Kernel.org
and to two districts, gen2 and [INDISTINCT]. And over time that number reported is flat.
Red Hats bugs, their bugs require for the Kernel--is actually going down. Now, either
that means we have a fixed number of testers and people reporting problems, this can be
true. But we're increasing our size of our code; we're increasing a number of users on
our code. So, obstensively, I'd like to think that we are doing better because our code
size is going in a number of bugs reported is flat. Andrew sees a lot of--a lot of bugger
reports. He sees a lot of stuff and a lot of stuff all the time. I see a lot of stuff
all the time too. But if you look at our rate of change for what's happening I would argue
that we are doing better or doing good. If we break the user space API, let us know and
we'll fix it. I mean, regressions always happen. I can say that. So as far as the enterprise
version, I've said it before publicly I still think the enterprise Kernel model is wrong.
I get paid by enterprise Kernel Company. I think it's wrong. Because if you look at what
they do so, take SUSE or their [INDISTINCT] SLED 10--SUSE or enterprise ten. You take
2616 Kernel froze it in time and so, okay, here's what we're going to do. And they're
fixing bugs and make sure they pass all regression test and ship it. A year later companies come
along and sort of say, "We're going to do a service pack up date because we need to
support new hardware." Problem is new hardware comes out that changes the guts of the Kernel
based on the fact that we didn't know those things are going to happen. We didn't know
that ATA was going to happen. We didn't know these new devices were going to happen. We
know the wireless stack, what was going to happen there. So you go back and you kind
of shoehorn all these stuff into there. If you look at what we have--we have I think
3000, maybe 4000 patches to 2616. It's not a 20616 Kernel anymore. It's 2616 with a bunch
of bug fixes that we know were reported to it. And that's scary because we know a lot
of bugs have been fixed between 2616 and what we have today, 26.5 that werenít reported,
that werenít fix or that we didnít know about it and we didnít [INDISTINCT] so, extensively
the Kernel, the user interface should be stable and stay the same. We donít change our version
number to make people feel good that are used to the old Unix Model, things moving very
slowly. The windows model things move very slowly that I think, "Theyíll stay with API."
no, they just kind of slow API. All those Kernels--all the other operating system, they
change, they break their API, they break their driver API. I'm on the windows driver development
mail list. It's the best mailing and stuff. You just see the quotes for the Microsoft
developers. Oh, yeah, we did something change there so we donít remember what change are
we going to have to test, a lot of angry developers all the time. So, just like users, it's just
like Linux. We change our APIs all the time. They donít have any fixed API. So, for inside
the Kernel, outside the Kernel they do, hopefully. Outside of our Kernel we do. So, I still think
that you should move forward. Now there's a model people that take a box, an enterprise
box, put it on it and throw in the corner and donít touch it for five years. And then
enterprise model that only gets security fixes and backwards to that that's great but only
about 20% of the market out there because if you buy a box to shove in the corner and
you're like a year from now you want to buy a new box, the new box is new hardware, and
yet this [INDISTINCT] a new hardware on that old version of Kernel and that doesnít work,
for desktop it just doesn't work. So, I still think the model isnít going to move faster.
So, inside I'm trying to argue that, [INDISTINCT] is trying to argue that, see what works, you
know, it's an interesting dilemma. So, now we're still making money so I'm happy. That's
it? You guys are easy. >> MALE: So, where are you going to go next,
let take... >> HARTMAN: Where are we next? Well, what's
next? Yes, well, like I say is whatever you wanted to be. Things that I think are cool,
KVM. KVM, is really, really cool. It's a running a virtual machine it's a hyper visor, turns
Linux into a hyper visor. I think it's going to win other of this and model, two different
or the VMware model. Like, what VMware does it's actually Linux running there. So all
the big embedded or big hyper visor guys are really running Linux--cant find a watch. But
so KVM School, KVM can afford the S390 main frames, port IO 64 for the arms coming and
there're some other accessory on port too. So KVM is really cool. It's a really neat
technology, just new virtual machines. So there's another way, if you have an enterprise
[INDISTINCT] did you want to run fixed, you can run on top of KVM or a virtualized machine,
you would change the hardware, you get the new drivers for your virtualized machine for
your low level stuff than your virtualized machine and you can still render old version
window just fine because at API between the two is fixed. And that's a new idea. So, you
can run your applications to run your old app--the old things, they're running on this
old API especially you're C++ libraries enterprise have a certain C++ library and it never want
to change that--things like that. I know of some big shops, Linux shops, they're stuck
using [INDISTINCT] and [INDISTINCT] three because the C++ developers do not want to
ever change the compiler. It's what those guys do as they strip out the Kernel underneath
and replace he Kernel with the latest enterprise for the latest Kernel--old kernel and then
usually above that now even knows it happens to use them, they just know a raw metal stuff.
I give [INDISTINCT]. PSP3 is coming hopefully sometime in the future and of all without,
that should be good [INDISTINCT]. Yeah. >> MALE: So, I'm wondering [INDISTINCT] has
a great [INDISTINCT] pretty dynamic thing. I was, kind of, curious, you know, like the
[INDISTINCT] on the particular subsystem they [INDISTINCT] with that and it comes to like
on particular individual. And what's your plan at certain [INDISTINCT]?
>> HARTMAN: So, as the load as we... >> MALE: Or clients want to settle down.
>> MALE: [INDISTINCT]. >> MALE: You know what I mean?
>> HARTMAN: Yes, [INDISTINCT], resettle down. >> MALE: Settle more.
>> HARTMAN: Settle more. The substance and maintenance what's our load and as originally
the--our load has gone out. I know my personal load I dropped, I donít maintain the PCI,
I mean, [INDISTINCT] more because I wasnít paying enough attention to it. I know some
of this is going to get more attention is now taking over. So, hopefully we can or smart
enough to recognize among ourselves that when things are going wrong and we can change.
We are nothing but introspective--I love talking about this. Every Kernel, someway we always
talk about the process. On the middle is we argue about the process. As our process good
or not and we changed things all the time, we have this table tree because people like
our process really isnít covering up. Let's see how that works. The Linux next off, hey
this process isnít working some system maintainers are burning out. Let's try this. So, we're
nothing if not keep trying things. So, hopefully we just keep trying things and as soon as
you're working really, really well, well better than we ever thought it would be. We havenít
hitting max limits yet so far. Every time we do a Kernel is like I was the fastest ever.
The 2625 Kernel blows away all those--those are average number. She's a 25 as the fastest,
most amazing rated change, most developers, most individual companies and anything we've
ever done. It's crazy and now, probably the next one will be more faster. So I don't know.
Hopefully--I don't know where we're going top stop. We've already broken all the records
so I don't know. We're making up as we go. Don't feel good about Linux now. Here. Here.
>> So I have a sort of strange question--I mean, the set of layers kind of suck right
now. How is it going with that? >> Set of layers of suck you, that's news.
>> [INDISTINCT] >> HARTMAN: They use more suits, then let
the developers know. >> Okay.
>> HARTMAN: I mean that--that's, you know, new to me so let them know. If they don't
know these problems then that's a big thing. If you find the problem let us know that's
the best thing about our development model. You can find the person responsible for the
code and make fun of him. >> Or give me a patch.
>> HARTMAN Or give me a patch. Well, that's sign off by everything. So that's really--we
like it--call it the path of blame. So you see a bug, "Hey look, all you guys miss this,
fix it." So, yes if it's using too much CPU let us know. We'll work on it. Tashion is
really good, really responsive [INDISTINCT] really good at that. So, they're good at this
stuff. All the stuff has really weird hardware. Some of the hardwares written so that it's
requiring more CPU power to do some of these things just because some of those manufacturers
are also CPU manufacturers. So, they want you to buy new CPU. I don't know. One thing
about KBM it doesn't make you buy a new CPU that's why AMD and Intel. I think itís a
virtualization stuff. >> Does KBM [INDISTINCT] to [INDISTINCT]
>> HARTMAN: KBM is the--or KBM makes Linux be the hyper visor. The way it has a Kernel
developer it is the hyper visor. If you look at the way Zen model is, they have to have
to their own scheduling, their own memory management or their power management so and
then have Linux running on the side to handle the drivers because nobody wants to write
drivers. Linux is always used for drivers. Herd uses Linux for drivers. Yeah, I'll take
that. So, Linux--so why not do--so Zen is having a duplicate all the numerous stuff,
all the power management never run in--on the laptop, it will suck your power dry. So,
why not just use Linux as the core, because the core of the Linux is equal the size of
the core of Zen. Actually--Linux Kernel is actually developed by more people. Maybe they
know more things. They can do more things faster so use--so KBM models reuse Linux and
then instance of the--the guest operating system runs in there's a little driver. There's
KBM driver and just runs up there as a--it runs as a process. So, the really cool thing
about that is real time. You can now real time schedule these guest operating systems
and the real time is a real new thing. Some people can measure network latency and latency
of processes in dollars. And they can do that in Dollars and then--so, if they can reduce
that. They can make it predictable. They can throw some really, really big machines at
it and make their cost go down and make more money. So real time really matters for those
guys. Real time might matter for you guys for network latency. Other big companies would
do a lot of web services really care about the network latency they want to be able to
schedule. They donít want to be waiting for another process to come in and finish thing.
They want to be able to know what's going to happen to determine it holistically. So,
anyway--so KBM can handle that. KBM can then real time schedule your guest operating systems
Zen can't do that. Zen also did their development often in the side of the corner and never
played nice with the Kernel developers. They're trying to catch up now. I don't know if they're
ever going to catch up which is sad. That's it. You guys are easy.
>> Will there be a 2.7 Kernel? >> HARTMAN: Will there be a 2.7 Kernel? We
said a long time ago, "Oh, if we ever do four little page tables or if we ever replace the
timers of the systems, maybe we'll need a development branch that we're here to do this."
We've done those things. The really cool thing about that development model and the one problem
of the developed models we burn engineer's time. We waste engineers. We make you break
your changes up into tiny discrete units that can be tested and obviously fit--obviously
reviewed. So we make you take your, "Oh, I'm going to rewrite the timer sub system into
a lot of individuals steps over time so which you'll never know if things go in." So, we've
done these major things we thought we're going to have to happen already and people didn't
notice. Well, some of the performance guy notice four label page tables [INDISTINCT].
[INDISTINCT] reformats some place. They fix that since then. So, I don't know. We don't
need a two set. I mean, the numbers are marketing. You know, so I don't think so. Anyway I think
we're running out of time. >> I just want to--let me--put the mic here.
Is this mic active? Good. I just want to remind all of you that our outgoing open source patch
process is very, very low ceremony. If you got a patch right now you can ship it out
within the hour. So we're not holding you back as a company, so...
>> HARTMAN: Spelling fixes this all. >> Consider spelling fix. Anyone in this crowd
can do a spelling fix so I'll encourage you to make Linux the best spelled happening system
in the world. Oh, yeah. Now, we're certainly fast when it comes to shipping out patches.
We only ask--will you look at your first couple and that's mostly to make sure that you're
not insane and not from a--not from a technology stand point but actually from a craziness
stand point, so. And after that we ended up trusting you and that's how it works so please
do it. >> Okay. Well, thank you all for having me
talk. >> Not that I think that 13 is bad. I think
it's pretty good for a company of our type. >> HARTMAN: It is so.
>> It's a decent place to start. >> HARTMAN: It's okay.
>> But it'd be great if we were nine. >> HARTMAN: Yes.
>> I could make that first slide. >> HARTMAN: Yes, everyone wants to be on that
first slide. >> Great.
>> HARTMAN: Great. Well, thank you very much.
    You must  Log in  to get the function.
Tip: Click on the article or the word in the subtitle to get translation quickly!

Loading…

Greg Kroah Hartman on the Linux Kernel

3119 Folder Collection
dnhwang published on November 12, 2013
More Recommended Videos
  1. 1. Search word

    Select word on the caption to look it up in the dictionary!

  2. 2. Repeat single sentence

    Repeat the same sentence to enhance listening ability

  3. 3. Shortcut

    Shortcut!

  4. 4. Close caption

    Close the English caption

  5. 5. Embed

    Embed the video to your blog

  6. 6. Unfold

    Hide right panel

  1. Listening Quiz

    Listening Quiz!

  1. Click to open your notebook

  1. UrbanDictionary 俚語字典整合查詢。一般字典查詢不到你滿意的解譯,不妨使用「俚語字典」,或許會讓你有滿意的答案喔