Kernels

Name
Email
Subject
Comment *
File	Select/drop/paste files here
Password	(Randomized for file and post deletion; you may also set your own.)
* = required field	[▶ Show post options & limits] Confused? See the FAQ.

Flag
Oekaki	Show oekaki applet (replaces files and can be used instead)
Options	Do not bump (you can also write sage in the email field) Spoiler images (this replaces the thumbnails of your images with question marks)
Allowed file types:jpg, jpeg, gif, png, webm, mp4, pdf Max filesize is 16 MB. Max image dimensions are 15000 x 15000. You may upload 3 per post.

File (hide): b4e8a1e9f821f52⋯.jpg (32.55 KB, 711x387, 79:43, os2.jpg) (h) (u)

File (hide): ba07d017cb9f805⋯.jpg (52.9 KB, 500x571, 500:571, BoyWithKittenOnHeadAww.jpg) (h) (u)

▶Kernels Anonymous 03/31/18 (Sat) 16:59:09 No.891389>>891439 >>891989 [Watch Thread][Show All Posts]

Let's talk kernels, /tech/!

So let's start this off with a common argument i've seen from BSDfags and some others. "The Linux Kernel is bloated!"

But honestly, I don't see how that can be possible. I'll give ya the coreutils. Those should definitely be reworked, or better yet, fully replaced by busybox. But the kernel HAS to be as big as it is in order to have driver support for all the hardware that it can run on. If it was smaller, it wouldn't be able to have the amazing level of compatibility that it does, far superior to *BSD, and only beaten by Windows due to manufacturers being kikes.

I mean, if we're talking about microkernels, that's a bit different because you'd want the kernel to be as small as possible and all the drivers to be in userland, but for any monolithic kernel, it makes no sense to start crying bloat when there's a perfectly acceptable reason for it.

▶Anonymous 03/31/18 (Sat) 17:04:39 No.891391>>891984 >>897782

File (hide): 693f1bb7cd014ac⋯.webm (6.06 MB, 266x180, 133:90, To the kernel and beyond.webm) (h) (u) [play once] [loop]

▶Anonymous 03/31/18 (Sat) 17:05:02 No.891392>>891437 >>891668

>"We're getting bloated and huge. Yes, it's a problem"

Linus Torvalds

▶Anonymous 03/31/18 (Sat) 17:22:15 No.891398

File (hide): 48491375488ae1d⋯.jpeg (80.72 KB, 600x600, 1:1, american-corn-kernels-600….jpeg) (h) (u)

Love me some popped corn kernels OP.

▶Anonymous 03/31/18 (Sat) 18:27:45 No.891426>>891442 >>891458 >>891743

File (hide): bd2e701d56c8323⋯.png (16.57 KB, 443x220, 443:220, 2018-03-31-142626_443x220_….png) (h) (u)

>microkernel

meme

     - Microkernels are the way to go
          False unless your only goal is to get papers published.
          Plan 9's kernel is a fraction of the size of any microkernel
          we know and offers more functionality and comparable
          or often better performance.

>linux

http://harmful.cat-v.org/software/operating-systems/linux/

I do have a heavily stripped down linux kernel on my gentoo rig, with no where near as much support as the OpenBSD kernel and also no module support its about the same size. But even torvalds agrees that the linux kernel is disgustingly large at this point.

▶Anonymous 03/31/18 (Sat) 18:54:31 No.891437

>>891392

>"I made a quote"

No link.

▶Anonymous 03/31/18 (Sat) 18:56:42 No.891439

>>891389 (OP)

>you'd want the kernel to be as small as possible and all the drivers to be in userland, but for any monolithic kernel, it makes no sense to start crying bloat when there's a perfectly acceptable reason for it.

The only kernel space drivers should be core hardware anything on a mobo except networking, video (not including basic vesa output drivers), and CPU microcode.

Everything else should be user space.

Anything with closed source firmware bins should be user spaced. Nearly all Intel, all video cards, nearly all NICs, etc

Every DMA capable device should be sandboxed.

▶Anonymous 03/31/18 (Sat) 19:06:23 No.891442>>891449 >>891452 >>891669 >>891743

>>891426

does plan9 support btrfs?

does plan9 support anything except sshing into it from a real OS?

▶Anonymous 03/31/18 (Sat) 19:26:38 No.891449>>891679

>>891442

Why would a kernel from the 90s, designed to support two filesystems, support btrfs?

▶Anonymous 03/31/18 (Sat) 19:28:35 No.891452>>891677 >>891679

File (hide): deb3b1c88da12c5⋯.png (953.82 KB, 500x380, 25:19, 9man01.png) (h) (u)

File (hide): 4d69cd3ebd6a8fc⋯.png (256.23 KB, 473x356, 473:356, nintendo.png) (h) (u)

File (hide): 0d66195975d9e17⋯.png (219.33 KB, 500x667, 500:667, 9decisions.png) (h) (u)

>>891442

>does plan9 support a broken filesystem

of course not

>does plan9 support anything except sshing into it from a real OS?

It can browse the web with mothra or abaco, it can program in the purest form of C, it can run emulators for old game systems, it can play audio/video files, basically anything one would want to do with a computer. It even has a linuxemu

▶Anonymous 03/31/18 (Sat) 19:44:15 No.891458>>891676

>>891426

Funny how security never gets brought up here, which is the real advantage.

These are in reference to MINIX, but are ideas common across microkernel designs.

>In monolithic operating systems, device drivers reside in the kernel. This means that when a new peripheral is installed, unknown, untrusted code is inserted in the kernel. A single bad line of code in a driver can bring down the system. This design is fundamentally flawed. In MINIX 3, each device driver is a separate user-mode process. Drivers cannot execute privileged instructions, change the page tables, perform I/O, or write to absolute memory. They have to make kernel calls for these services and the kernel checks each call for authority.

>A special process, called the reincarnation server, periodically pings each device driver. If the driver dies or fails to respond correctly to pings, the reincarnation server automatically replaces it by a fresh copy. The detection and replacement of nonfunctioning drivers is automatic, without any user action required. This feature does not work for disk drivers at present, but in the next release the system will be able to recover even disk drivers, which will be shadowed in RAM. Driver recovery does not affect running processes.

More at:

http://wiki.minix3.org/doku.php?id=www:documentation:reliability

▶Anonymous 04/01/18 (Sun) 07:59:26 No.891668

>>891392

https://www.theregister.co.uk/2009/09/22/linus_torvalds_linux_bloated_huge/

▶Anonymous 04/01/18 (Sun) 08:01:04 No.891669>>891692

>>891442

What are you talking about, Plan 9 doesn't even support cp -r.

▶Anonymous 04/01/18 (Sun) 08:30:46 No.891676>>891759

File (hide): 61050fb8b4993d4⋯.pdf (470.85 KB, Hacking Blind.pdf) (h) (u)

>>891458

Automatically restarting stuff that crashed is bad for security. If it crashed, that means it's buggy, and maybe it got exploited. That's why OpenBSD tries its best to crash early, so bugs can be noticed and fixed.

Read up on ROP and BROP. These attacks pretty much depend on being able to retry over and over. If a subsystem conveniently gets restarted, that makes their attack feasible.

https://youtube.com/watch?v=FzJJbNRErVQ

▶Anonymous 04/01/18 (Sun) 08:31:51 No.891677>>891742

>>891452

Any recommended hardware to get started on plan9/9front?

▶Anonymous 04/01/18 (Sun) 08:41:52 No.891679>>891742

>>891449

>>891452

>Plan 9's kernel is a fraction of the size of any microkernel we know and offers more functionality

▶Anonymous 04/01/18 (Sun) 09:26:12 No.891683

>But the kernel HAS to be as big as it is in order to have driver support for all the hardware that it can run on.

But the majority of that hardware is massively shit, AKA x86, so it is bloat. Linux was designed for x86, therefor its always and has been intrisically terrible. PPC era OS X was way better.

▶Anonymous 04/01/18 (Sun) 10:15:04 No.891692>>891734 >>898284

>>891669

>this meme again

Its called cpdir

cp -r is a hack that does basically just that

▶Anonymous 04/01/18 (Sun) 13:03:34 No.891734>>891960

>>891692

>Plan9 is so bloated it has two programs to copy files and directories instead of one

wow. this is the reason why i use arch linux.

▶Anonymous 04/01/18 (Sun) 13:44:56 No.891742>>891743

File (hide): f8424079903714d⋯.pdf (3.96 MB, dash1.contentsmaintainedst….pdf) (h) (u)

>>891679

And? You have yet to provide any evidence to counter that point. The post you are replying to doesn't even correlate to to the post that you quote. One is talking about software that the operating system as a whole supports and the other is talking about the kernel. Of course I can come up wit equally retarded questions about the Gnee/Lunix OS such as:

does lunix support anything except serving files to the windows and mac operating systems?

Or spefically related to kernels:

Let 'x' = some arbitrary filesystem that kernel 'A' supports and that kernel 'B' doesn't support

does 'B' support 'x'?

>>891677

It supports a surprisingly large amount of hardware. Lots is documented in PDF-related. Basically just dd the image to a flash drive and see if it boots. Let me warn you though, its quite different to any Unix-like you have tried before. The first time I tried it I couldn't figure out how to get past a certain stage in the first boot, and the second time I couldn't figure out what the command to poweroff was.

▶Anonymous 04/01/18 (Sun) 13:48:34 No.891743>>891751

>>891742

>And? You have yet to provide any evidence to counter that point.

already did that here: >>891442

>The post you are replying to doesn't even correlate to to the post that you quote.

The post that I'm replying to is a reply to >>891442 which is a reply to >>891426

>One is talking about software that the operating system as a whole supports and the other is talking about the kernel.

So where is the Btrfs support???

>offers more functionality

Protip: it doesn't

▶Anonymous 04/01/18 (Sun) 14:25:44 No.891751

>>891743

Congrats on proving your low reading comprehension skills.

But I'll give you benefit of the doubt. Am I supposed to assume that all microkernels support btrfs? You've provided no source. What about everything that the plan9 kernel supports and that microkernel 'A', 'B', etc don't support? You have provided 1 piece of minor evidence, it hardly disproves the quote I posted. Of course you also have to show that whatever microkernel you choose is also smaller than the plan9 kernel since that is one of the requirements in the quote I posted.

▶Anonymous 04/01/18 (Sun) 15:14:30 No.891759

>>891676

In security-critical systems, it's a good practice to crash early.

In safety-critical systems, you can either shut down early and crash ( not a good practice in nuclear reactors since byproducts make it impossible to restart before next week , not good choice in aeroplane systems either) or you can restart the crashed process ( in factories, this would make more damaged products so it's better to just shut down ). Maybe there are more options, these are just a few examples. Ofc I'm no expert.

▶Anonymous 04/01/18 (Sun) 22:12:33 No.891960>>898284

>>891734

GNU/Linux is so bloated it puts these two functions into one binary

▶Anonymous 04/01/18 (Sun) 23:20:47 No.891984

>>891391

That was beautiful.

▶Anonymous 04/01/18 (Sun) 23:39:20 No.891989>>892225 >>892268

>>891389 (OP)

Is there anything between kernel and userland

▶Anonymous 04/02/18 (Mon) 16:11:13 No.892225

>>891989

Nope. Userland is in reference to anything that's not in the kernel.

▶Anonymous 04/02/18 (Mon) 17:48:54 No.892268>>892296 >>892362 >>892364 >>892373

File (hide): c4dbd05cb36106b⋯.webm (4.89 MB, 500x281, 500:281, 1468494345108.webm) (h) (u) [play once] [loop]

SeL4 is the future. In a better world GNU/Hurd would be wrapped around it and we would have a fast, bloat free, mathematically secure proof microkernel with a floss userland driver/memory manager which is easy to extend. At least dreams are free, anons.

>>891989

GNU/Hurd

▶Anonymous 04/02/18 (Mon) 18:18:13 No.892287

>bsd fags having opinions on Linux

My sides

▶Anonymous 04/02/18 (Mon) 18:28:38 No.892296>>892362

>>892268

>mathematicaly secure

FALSE.

SeL4 is mathematically proven to conform to specification. It is not proven to be secure, work correctly if the spec was incorrect, or be bug free

▶Anonymous 04/02/18 (Mon) 22:51:44 No.892362

>>892268