FreeBSD Catching Up To Linux DRM Graphics Drivers, In Sync With Git

[starshipeleven]

better hardware support does not make it more advanced.

He said "more advanced" AND "more hardware support. Two different statements.

Concede for now. Until FreeBSD11 next month.

Plain copying Linux drivers makes this a nonsense comparison. Same code is same code.

Care to comment why Facebook was looking for Linux kernel engineer back in 2014. Stated as "seeking a Linux Kernel Software Engineer to join our Kernel team, with a primary focus on the networking subsystem. Our goal over the next few years is for the Linux kernel network stack to rival or exceed that of FreeBSD"

Would be nice to have a bit more proof than a job announcement meant to gather some attention cheaply (by adding the obvious troll they ensured EVERYONE in the field would get the message that they were hiring, without paying a dime).

Indeed? Why exactly not? Statement but zero reasoning.

Because of all reasons already stated duh.

Btw, with file systems, there is always FUSE option (not fusefs). You can handle ext4/ntfs whatever with it. It presumes reading documentation first.

FUSE has crap performance because it runs in userspace.

Linux does not have native production ready filesystem with deduplication and file compression. Btrfs does not count. It's not "ready". It could only use it's implementation of ZFS for it.

which is why Pawlerson said "its only strong side is ZFS", as that's something that works in btrfs only so far, and btrfs isn't mature as everyone knows.

IF it has the capability (I did not bother to check).

How about you do your own fucking homework?

[aht0]

a)No, I'm just claiming that you should post citations when you make such claims, or at the very least provide them when asked (I posted a [citation needed] above)
b)Simple logic too --> more stuff supported and more features --> more bugs.
c)Also simple logic --> more usage outside of classical server role --> more testing on features not used in such roles --> more bugs discovered.
d)Simple logic again --> no note-worthy development of BSD --> waay too much bugs.

a) not even understanding what do you mean.
b/c) you are utterly forgetting or ignoring general habit of Linux's distributors/kernel developers for inlcuding software/code that's barely tested or not tested at all, just to get "feature xxx" out and announced. Includes also "non-classical" roles.

With such behaviour, claim that Linux has more bugs because it has got more "stuff and features" supported, becomes laughable. Kernel is full of "DANGEROUS" and "EXPERIMENTAL" features. Some of which have been in such a state for a loooong time. Logic from there cuts both ways of course. It also somewhat supports your claim. But what's the point of such abandonware?

But lets leave off off "simple" security bugs and concentrate our attention to regressive bugs: amount of this type of bugs in Linux kernel is even worse.. just check the bugzilla.kernel.org.

Tell me now that cause of Linux bugs is still "more features and stuff" but not at all sloppy coding without giving much forethought to the results.. Regressive bugs caused by unnecessary API changes introduced by kernel patches are especially fitting example. Author of the patch is mostly not going to care about the issues he caused downstream. It's left to the package maintainers to adjust and fix..

Most of the stuff on that page was invented by third parties, dumbass. As I said, that's not "stuff invented by OpenBSD" like you think it is.

Did you ever checked it for more than for couple items or it's fully informed claim? Did you check if "third parties" happened to be openbsd devs at the time?

And my complaint was not a "you newbie should do it yourself", but a "if you want to make a fair comparison about systemd please fix distro issues first, since you are clearly able to".
I also assumed you were a newbie as veteran users would have fixed this on first boot, so I provided the answer to the newbie question "why openSUSE takes so long booting, I see systemd talking about network", so that even newbie users can fix it themselves by point and clicking around (something that is possible mostly in SUSE/OpenSUSE due to YAST).
That "dived under the hood" is something you pulled out of your ass, please refrain from doing so.

And I explained that with full-blown distro, I don't have least interest in fixing shit that's supposed to work in the first place. I do not use PC-BSD much for the same reason. Theoretically nice "distro", supposedly "ready for desktop use" but time to time I would be running into issues I am not willing to search fix for. Like having wifi-manager not discovering shit. Reason: distros tend to do stuff by their own way and I do not like learning the specifics and differences of every single one. My head is not a trash bin.

A graphical DE is required in minimal desktop installations, and OpenSUSE installs quite a bit more stuff than FreeBSD (as you might have noticed due to longer installation times).
All this stuff going on in forums is mostly a text-warping-and-information-finding-based game. Nothing more. Someone's game is trolling on shit, or fanboying on shit, someone else's is bashing them until they ragequit.
You said it already, chances one party convinces the other is 0%, I answered I do it for fun. It's more rational than answering that I do it for a better tomorrow or for kittens or whatever.

And I, yet again, remarked that extra shit that OpenSUSE installed besides KDE, does not really matter. Because it's not going to be loaded/read from drive, until you actually LOG IN into desktop. Difference that those few services running in backgroud would do to boot times are infitesimal. I measured boot time until KDM login screen came up. Which means I did not log IN. KDE standard packages are more or less the same in Linux and BSD. And how many different packages one or other OS contained, that were NOT loaded during boot, is not relevant.

FYI: if you want to sound more believable you should stop bashing the other side so hard (i.e. systemd for example), or trying to find random bs reasons to shit on Linux.

As long as ppl refrain from bs reasons to shit on other OSes. And I comment on other threads as well, often pure-Linux topics. And I like bashing systemd. It's unnecessary bastard conceivance in my eyes. :P I still haven't got single argument, how it's "init portion" is better than OpenRC.

Like that they hurt BSD or something by going their own way and making their own choices not keeping in mind BSD's interests too.
No really? Is that even wrong? Where is stated that Linux development should behave in a convenient way for BSDs?

So, the fact that systemd makes harder to use various software on non-systemd Linuxes as well, is nothing? Suddenly such distros have to start looking for workarounds to get some piece of software made artificially dependent (Poettering successfully nagging on devs for example) on systemd running again.

Frankly I can't suggest any good reason to stay on BSDs for a desktop user (for servers and networking they do offer neat features and they are competitive outside the hardcore container things that are systemd-only stuff), but I suggest you to find some and state them instead of random bashing and trying to make BSDs look like poor and needy bullied by Linux.

Since you are not using any BSD ofc you cannot. But it has never stopped you being opinionated about it.

How about.
+ Once set-up, it generally runs without a hitch.
+ On need, you can encrypt your WHOLE disk, not just some partition(s).
+ Clear difference between OS and userspace. Safer and more stable approach.
+ It's far less bloated.
+ and you don't have to worry about system binary upgrade killing it unexpectedly.
+ BSD "security" and "permissions" is not such a bloody mess. You are able to figure it out yourself, if you happen to run into an issue.
+ Shit you need, webbrowsers, smart-card support, chat programs, media players etc, are all there and generally work.
+ Probability of getting infected while being in net is smaller compared to Linux desktop. Yeah, there is malware targeting Linux desktops. And it's amount is increasing. And no, there is functional commercial antivirus that would work. I tried finding one about 6 months a go.
+ Repositories of the binary package Linux distros do not contain all available open source software. Libraries conflicts do not allow the luxury. Compiling on your own is pain in the butt. It's lots easier on FreeBSD. Either by using ports or just direct download and compile. Unless piece of software contains too many 'linuxisms', it generally works.
+ Old applications rarely work in contemporary Linux. (glibc incompatibilities, errors, incompatible libraries etc). Probability is better in BSD using Linux ABI.
+ BSD audio support is much better. Especially OpenBSD's. Helps that there are several audiophiles among devs.


- You have to pick your hardware. With some fore-knowledge, it's not an issue .
- Newest versions of software are not always available. There are "bleeding-edge" repos for those who want it.
- You do not get easy 'click-next' GUI, have to figure out yourself how to make base BSD install into functional desktop system. Documentation is contemporary and complete. Use it.
- Getting filesystems mounted is not as automatic as on "linux". FUSE exists however, you can manually mount pretty much anything you daily need, from NTFS to EXT4.

[aht0]

He said "more advanced" AND "more hardware support. Two different statements.

Be it so. I would say "more advanced and more half-assed hardware support". Which would also be truth, as insulting as it is.

Plain copying Linux drivers makes this a nonsense comparison. Same code is same code.

You can't at least no longer point out that certain hardware is not working..
I would think 1:1 copy would not work in any case..

Would be nice to have a bit more proof than a job announcement meant to gather some attention cheaply (by adding the obvious troll they ensured EVERYONE in the field would get the message that they were hiring, without paying a dime).

It looks like FB kept in mind ipv6-related development. Also solving of performance/stability issues. If they experienced those issues on a level they saw need to hire expensive engineer and express it, then what more proof do you expect? FB is not a candy store around the corner.

http://www.theinquirer.net/inquirer/news/2359272/facebook-wants-linux-network-stack-to-rival-or-exceed-freebsd

FUSE has crap performance because it runs in userspace.

non-argument. You do not need any special performance with mostly usb2 interfaces... Had to look pretty hard to find USB3 memory stick. Most of the stuff I've ever needed to attach to the PC has had USB2.

which is why Pawlerson said "its only strong side is ZFS", as that's something that works in btrfs only so far, and btrfs isn't mature as everyone knows.

I could add list out of hand but I've done it already multiple times, no point.

How about you do your own fucking homework?

I did not have the least interest. I needed to bring out the possibility though.

[starshipeleven]

Be it so. I would say "more advanced and more half-assed hardware support".

Yeah, it's so half-assed that BSD is making a shim to port graphics drivers whole, and there are or were projects that port drivers for other stuff too.

You can't at least no longer point out that certain hardware is not working..

A small consolation for the defeat of having to milk the same cow they milk to get most of their desktop userspace.

I would think 1:1 copy would not work in any case..

You would think wrong, they devised a new interface in their kernel so they could plug drivers from linux directly. That's the entire point, and why it's so great for them.

It looks like FB kept in mind ipv6-related development. Also solving of performance/stability issues. If they experienced those issues on a level they saw need to hire expensive engineer and express it, then what more proof do you expect? FB is not a candy store around the corner.
http://www.theinquirer.net/inquirer/...exceed-freebsd

More random words without proof. I asked for proof, not for more news articles that provided publicity to that same thing.
Facebook lives off social stuff, they know how to troll like pros to get every site and every newsfeed mirror their announcement.

non-argument. You do not need any special performance with mostly usb2 interfaces... Had to look pretty hard to find USB3 memory stick. Most of the stuff I've ever needed to attach to the PC has had USB2.

Ah, it's the "I don't use a feature most people in developed world find essential, so I declare they don't really need it" time of the week again.

This time is "people don't need USB 3.0 speeds". How about you go back in your cave and start your fire for cooking your raw meat by rubbing sticks or something?

I could add list out of hand but I've done it already multiple times, no point.

All you posted so far has little supporting evidence or is plain bs.

ZFS is the only point where BSDs have a clear edge on Linux (but with the ZFS port this is getting moot) and there are proofs.

I suspect too that some BSD has better networking than linux, especially for the ultra-high end stuff, but proofs of that are hard to get, and your random bs isn't helping.

I did not have the least interest.

Yeah, this is apparent from the stuff you said above too. It's apparent that you "troll from the hip", (i.e. you spout random stuff without knowing what you are talking about).

[aht0]

Yeah, it's so half-assed that BSD is making a shim to port graphics drivers whole, and there are or were projects that port drivers for other stuff too.

for some reason I need to turn off powersaving of my Android phone to use it's web browser without having phone crashing randomly. Does not matter which version of Android, kernel version is same for all of them.
for some reason I had to jump trough various loops to have my Asus laptop boot into Linux without it having keyboard and touchpad both become inoperative in about 30% of boots.. Issue present only in Linux.
For some reason I keep bumping into random shit like this, like Murphy's Law.. Wanted to put Linux into a computer in my parents house, installation GUI freezed at some point. Had no other disk with me, did not want to kill off their mobile internet limit, so they still have windows.
I even remember Ubuntu boot media which for some reason could not recognize USB devices.. like what the fuck... what do you think I should think then? It's clearly half-assed approach. Minimal testing or no testing at all.. Just churn out more code and let users find bump into faults..

A small consolation for the defeat of having to milk the same cow they milk to get most of their desktop userspace.
You would think wrong, they devised a new interface in their kernel so they could plug drivers from linux directly. That's the entire point, and why it's so great for them.

Linux does the same. Milking the available userspace. Linux itself is technically only a kernel. GNU Hurd should feel insulted.

More random words without proof. I asked for proof, not for more news articles that provided publicity to that same thing.
Facebook lives off social stuff, they know how to troll like pros to get every site and every newsfeed mirror their announcement.

You would not accept any "proof" anyway. Even if some engineer would come behind your door claiming it. Such trolling and claims could very well backlash Facebook if happen to be groundless. Richer the company, more probable someone is going want to sue it just to make money off it.

Ah, it's the "I don't use a feature most people in developed world find essential, so I declare they don't really need it" time of the week again.
This time is "people don't need USB 3.0 speeds". How about you go back in your cave and start your fire for cooking your raw meat by rubbing sticks or something?
All you posted so far has little supporting evidence or is plain bs.

Do not substitute your worldview to a worldview of everyone or even most. Majority of consumer devices manufactured these days are still using plain old USB2 interface. Plug in your digital camera and check. Or some Android smartphone. Or car's dash cam. Not to mention USB3 memory sticks, you have to specifically search for one. 90%+ are still USB2 old crap.
Only thing relevant I can think of would be external USB3 drive. Have no such animal, can't test it. Sorry.

For more advanced users, adding disks into PC, copying shit off and over - which would need more performance - it does not qualify as "most people in developed world find essential" so cut your own bs. I cant even remember when was the last time I copied files from "foreign" disk to disk directly, except when cloning whole disks.

You are not even sure if BSD userspace has the same crappy performance as Linux had with fuse. Might very well be worse or better. Doesn't stop you from bitching about it.

[starshipeleven]

for some reason I need to turn off powersaving of my Android phone

Android uses blob drivers and hacked/device-specific linux kernels, and it's well-known that many firmwares tend to be sucky. Not mainline Linux's fault.

Does not matter which version of Android, kernel version is same for all of them.

That's because of above. The hardware manufacturer (the company making the SoC) gives SDKs to the OEMs (companies making the device), the SDK contains a hacked linux kernel and the blobs that go with it.
The most a OEM can do is update the stuff over it (if possible, not always possible), any update of that must come from the hardware manufacturer first, as a new SDK release.

For some reason I keep bumping into random shit like this, like Murphy's Law.. Wanted to put Linux into a computer in my parents house, installation GUI freezed at some point. Had no other disk with me, did not want to kill off their mobile internet limit, so they still have windows.
I even remember Ubuntu boot media which for some reason could not recognize USB devices.. like what the fuck... what do you think I should think then?
It's clearly half-assed approach. Minimal testing or no testing at all.. Just churn out more code and let users find bump into faults..

Still blaming the wrong guy (linux as a whole in this case) because some shit you use isn't configured properly? How about you stop using that distro and try with something for pros instead?
Debian installer for example NEVER EVER failed on me or anyone else around me and we install/installed tons of systems as Debian is the base for servers or VOIP or other stuff we install to clients.

Same for RHEL (or CentOS).

OpenSUSE installer seems to be good too, but I don't have so long experience.
A neat thing is that it is also configured for serial install (i.e. it will properly recognize a serial interface used to communicate with a headless system on boot, and will show up there so I can control it over serial).
Debian installer isos cannot do that unless you open, reconfigure, repack them.

Linux does the same. Milking the available userspace.

For linux this isn't an issue anyone cares about.
I mentioned this because BSDs have that pride of being "the most unix system", so all the stuff in their distro must be Unixy and BSD-licensed or permissive-licensed for the very least or they don't feel ok.
Some BSDs even write pages where they claim "innovations" when it's stuff that is made by third parties.

GNU Hurd should feel insulted.

GNU hurd isn't even 64bit, it cannot really be insulted.

You would not accept any "proof" anyway. Even if some engineer would come behind your door claiming it.

I need numbers, not authority. Some tests that can be repeated also by me possibly, but anything that can be independently checked is ok.

Such trolling and claims could very well backlash Facebook if happen to be groundless. Richer the company, more probable someone is going want to sue it just to make money off it.

That's not a public statement, there is no risk of backslash.

Majority of consumer devices manufactured these days are still using plain old USB2 interface.

Because they don't need more than that. All devices that need more speed than that are on USB 3.0. For example all wifi ac dongles are on USB 3.0 (I doubt that they actually need the bandwith, they likely need more power than what USB 2.0 can give).

Not to mention USB3 memory sticks, you have to specifically search for one. 90%+ are still USB2 old crap.

Uhm, that's probably because you're in a second-world country (you said you were in north-east europe?).
Here in the smug first world countries all flash drives are on usb 3.0 since at least a couple years, people in US (the smuggest first world country, and proud of it) can get a 16GB USB 3.0 flash drive for 2$ (not kidding), while I can get one for like 5-6 euros at a store.

Only thing relevant I can think of would be external USB3 drive. Have no such animal, can't test it. Sorry.

Same as above. Hard drives in the shops here migrated to usb 3.0 like 3 years ago.

For more advanced users, adding disks into PC, copying shit off and over - which would need more performance - it does not qualify as "most people in developed world find essential" so cut your own bs. I cant even remember when was the last time I copied files from "foreign" disk to disk directly, except when cloning whole disks.

It's called making backups or "the internal drive of this PC is full and I'm not able to change it or add another internal one". People sometimes do that and don't like to have to wait like 2 hours to fill a 2TB 2.5inch hard drive.

Of course I'm only counting those that use the PC for more than *office, internet and email, as those can do without USB 2.0 (or without the whole PC) too.

You are not even sure if BSD userspace has the same crappy performance as Linux had with fuse. Might very well be worse or better. Doesn't stop you from bitching about it.

I also deployed FreeNAS in the past so yes I know. All non-native filesystems are ok for copying stuff off a drive every now and then, but for extended usage, no. Just no.

[aht0]

Android uses blob drivers and hacked/device-specific linux kernels, and it's well-known that many firmwares tend to be sucky. Not mainline Linux's fault.

Difference of opinion. Lets leave it at that..

Still blaming the wrong guy (linux as a whole in this case) because some shit you use isn't configured properly? How about you stop using that distro and try with something for pros instead?
Debian installer for example NEVER EVER failed on me or anyone else around me and we install/installed tons of systems as Debian is the base for servers or VOIP or other stuff we install to clients.

It was Fedora installation medium. Last summer/autumn I think. Not a single distro comes with a message that "shit here is not configured properly, avoid me". How the fuck I should know to avoid it. I shall keep it in mind and try Debian next time.

Same for RHEL (or CentOS).
OpenSUSE installer seems to be good too, but I don't have so long experience.
A neat thing is that it is also configured for serial install (i.e. it will properly recognize a serial interface used to communicate with a headless system on boot, and will show up there so I can control it over serial).
Debian installer isos cannot do that unless you open, reconfigure, repack them.

OpenSUSE installer is generally just fine-working. OpenSUSE's sole downside is having to download 4,5+ GB when you don't happen to have it in hand.

I need numbers, not authority. Some tests that can be repeated also by me possibly, but anything that can be independently checked is ok.
That's not a public statement, there is no risk of backslash.

When some statement reaches bunch of newssites, gets distributed and amplified but is not public statement, then what the hell is?

Because they don't need more than that. All devices that need more speed than that are on USB 3.0. For example all wifi ac dongles are on USB 3.0 (I doubt that they actually need the bandwith, they likely need more power than what USB 2.0 can give).

No, ac dongles do not need whole USB3 bandwith, USB2 would limit it though pretty severely. Did not think on wifi dongles, I usually use internal card of some sort if at all possible.

Uhm, that's probably because you're in a second-world country (you said you were in north-east europe?).
Here in the smug first world countries all flash drives are on usb 3.0 since at least a couple years, people in US (the smuggest first world country, and proud of it) can get a 16GB USB 3.0 flash drive for 2$ (not kidding), while I can get one for like 5-6 euros at a store.
Same as above. Hard drives in the shops here migrated to usb 3.0 like 3 years ago.

hf comparing prices https://www.photopoint.ee/malupulgad . or http://arvutitark.ee/est/tootekataloog/Arvutid-ja-lisad-Valised-andmekandjad-USB-malupulgad//?p=5-717 When interface type is not marked, it's USB2. Small market I guess and/or higher taxes.

I also deployed FreeNAS in the past so yes I know. All non-native filesystems are ok for copying stuff off a drive every now and then, but for extended usage, no. Just no.

Concede. No experience to say yea or no.

[starshipeleven]

Difference of opinion. Lets leave it at that..

Lolnope. instability caused by drivers NOT developed in the linux kernel project isn't the fault of the linux kernel project.
The same as instability caused by VERY dirty hacks to the kernel done by someone else (as they don't have the time to do a good job) are not the fault of the linux kernel developers.

I know about routers, and I've seen some weird shit that basically hardcodes stuff from the kernel code (that should remain hardware-agnostic) so that it can use some SoC-specific hardware accelerator for NAT. There is no way stuff like that would get mainlined, too bad bad bad practice.

Not a single distro comes with a message that "shit here is not configured properly, avoid me". How the fuck I should know to avoid it.

Haha, you got it reversed. You need to look at distros that boast about being "rock-solid", not about the ones that boast about being "bleeding edge".
As I said with Debian (stable) and CentOS you can't go wrong as they are aimed at server usage.

But yeah, for hardware support you probably need a newer kernel that what they have, and also Debian by default does not install non-foss stuff, so it's annoying as you need to add it yourself.

I'd say for Debian to aim for Linux Mint Debian MATE as it pre-builds a desktop environment and ships it pre-configured with flashplayer, codecs, firmwares and stuff, plus provides ports of the softwares that you want to keep updated more often (firefox, thunderbird), I have no idea about CentOS, most of its derivatives are server stuff usually and I've never been a fan of RHEL anyway.

You probably still need to enable backport repositories for Debian to get a more recent kernel for hardware support, but with Mint Debian you have Software Sources (a graphical tool) to do it so it's just a copy-paste of the repo link in there.

What I've found that is decently recent but stable enough for normal use is openSUSE +KDE. It just needs normal configuration and has Yast that deals with it without having to open text files around.


But yeah, in general on Linux there is this "bleeding edge but more likely to be broken" vs "stable but usually horribly outdated" (still safe as it is still supported and receives security patches, but old software with old features).

On BSD there is only the latter. Most of their software trickles slowly in their repos.

OpenSUSE installer is generally just fine-working. OpenSUSE's sole downside is having to download 4,5+ GB when you don't happen to have it in hand.

They offer netinstall disks tho, that's 250mb or so.

When some statement reaches bunch of newssites, gets distributed and amplified but is not public statement, then what the hell is?

Public statement is something that is said on behalf of the company. Like if Zuckerberg or some other power figure comes out and claims that.

[aht0]

Lolnope. instability caused by drivers NOT developed in the linux kernel project isn't the fault of the linux kernel project.
The same as instability caused by VERY dirty hacks to the kernel done by someone else (as they don't have the time to do a good job) are not the fault of the linux kernel developers.

Phone in question is Samsung Galaxy S3 i9300, from what I found, using Exynos chipset. I did sent it into warranty all those eons a go and it was returned with a statement that there is nothing wrong with it. Yeah it's using firmware blobs but it seems to be the case only for GPS, Bluetooth and WiFi. SoC itself should be in linux kernel, fully.

As long as I can keep buying quality replacement batteries at need (Anker or Extremecells, not fake Chinese crap), turning off powersave is not too troublesome and phone itself seems to last better than newer Galaxy S'es.

Haha, you got it reversed. You need to look at distros that boast about being "rock-solid", not about the ones that boast about being "bleeding edge".
As I said with Debian (stable) and CentOS you can't go wrong as they are aimed at server usage.

Victim on marketing I suppose. "The Fedora Workstation Live image allows you to make media for your computer that provides a complete, running Fedora Workstation you can run right away." Live media actually worked as advertised. What did not work, was installing complete system using it.

What I've found that is decently recent but stable enough for normal use is openSUSE +KDE. It just needs normal configuration and has Yast that deals with it without having to open text files around.

Leap 42.1? Last time I experimented with it's KDE5, resizing and moving bunch of widgets around the desktop space made it crash pretty quick..When I get time, I am going to try Leap again.

On BSD there is only the latter. Most of their software trickles slowly in their repos.

There are bleeding-edge repos. Even with KDE5 in it. How much people are using them or how stable the software is, I am unable to tell. I am using default repos.

Public statement is something that is said on behalf of the company. Like if Zuckerberg or some other power figure comes out and claims that.

Not CEO's job for sure. Not with a company in FB's size. Maybe job for someone who is controlling over engineers themselves or personnel department...

[aht0]

They offer netinstall disks tho, that's 250mb or so.

Not really an option while using cellular internet at their place. Depletes quite a bit of the limit. I'd rather download what I can in advance, using broadband at home.