AMD Radeon Graphics Driver "AMDGPU" Ported To DragonFlyBSD

[phoronix]

Phoronix: AMD Radeon Graphics Driver "AMDGPU" Ported To DragonFlyBSD

The "AMDGPU" kernel graphics driver has been ported and pulled into the DragonFlyBSD operating system...

https://www.phoronix.com/scan.php?pa...d-DragonFlyBSD

[Developer12]

Once again asking for *any* of the BSDs to start patching Spectre v1 gadgets. There's virtually no performance impact and there are now free/open-source software tools to scan for gadgets. You don't have the excuse of needing a copy of Synopsys Coverity anymore.

[kbios]

I really think it would be better for the BSDs to have a common kernel and differentiate themselves in other aspects, porting drm drivers from linux *multiple times* sounds like a nightmare

[NateHubbard]

I really think it would be better for the BSDs to have a common kernel and differentiate themselves in other aspects, porting drm drivers from linux *multiple times* sounds like a nightmare

Hmm... In that case, they could easily achieve that by just using the linux kernel.

[kpedersen]

I really think it would be better for the BSDs to have a common kernel and differentiate themselves in other aspects

They are fairly different in other aspects. It is almost like suggesting that Haiku and FreeBSD should share the same kernel.

The BSDs aren't like Linux distros. They are meant to diverge. When you look into the nuts and bolts of things, they are actually substantially different. For example FreeBSD has a shim layer to abstract much of the Linux DRM/KMS code. OpenBSD does not. If anything, I think Haiku and FreeBSD are *more* similar in terms of the graphics layer than OpenBSD.

Not just graphics either. Audio systems are different too.

[kpedersen]

Once again asking for *any* of the BSDs to start patching Spectre v1 gadgets. There's virtually no performance impact and there are now free/open-source software tools to scan for gadgets. You don't have the excuse of needing a copy of Synopsys Coverity anymore.

Wasn't OpenBSD renowned for being the first to solve the spectre issue?

https://why-openbsd.rocks/fact/meltdown-spectre/

[Black_Fox]

Wasn't OpenBSD renowned for being the first to solve the spectre issue?

Based on data in the following article, I doubt that they had any chance of being the first:

https://malcont.net/2018/01/dont-lik...-bugs-handled/

[Developer12]

Wasn't OpenBSD renowned for being the first to solve the spectre issue?

https://why-openbsd.rocks/fact/meltdown-spectre/

Even *if* they were, there are large components still completely unsolved. Spectre v1 for example remains *completely* unmitigated for ALL BSDs, with no apparent plans to change that.

You do know that there are *lots* of spectre variants right? New ones are found every few months, including during this year.

[microcode]

I really think it would be better for the BSDs to have a common kernel and differentiate themselves in other aspects, porting drm drivers from linux *multiple times* sounds like a nightmare

The primary thing they are differentiated on is the kernel though. OpenBSD has an audited, relatively small kernel; FreeBSD has more drivers and features; NetBSD is small and portable; DragonFlyBSD has weird filesystems and exotic scheduler functionality.

[jabl]

Probably impossible. You have to understand that the main opensource BSD variants exist because of conflicting goals and irreconcilable differences. FreeBSD focused on x86, NetBSD was about portability to many arches, OpenBSD focused on security, and DragonflyBSD was initially about SMP scaling and personality conflicts. For various reasons, in BSD-land, forks are sometimes acrimonious and stay divergent. Perhaps due to Linus' successful management, Linux forks tend to try to rejoin the mainline (or die). Things may be friendlier in BSD-land these days, and I believe there has been some userspace sharing, though.

Unfortunately the BSD's have all inherited the same original dysfunctional governance model. There's this small core group that decides stuff, and if you disagree, well, then you fork the entire OS, kernel, userland, the whole shebang. Which leads to a lot of wheel reinvention, further splitting the already small developer pool.

The big success of the Linux ecosystem, by comparison, has been that it has been much more accepting of outsiders and their ideas. Sure, as the BSD fans like to point out, the Linux world is less coherent as a result, but look who runs the world and who's a small niche player?