@bison

Yep.

Spectre affects also PowerPC CPUs. (more: http://tenfourfox.blogspot.fi/2018/01/is-powerpc-susceptible-to-spectre-yep.html )

Meltdown affects intel (and possibly one ARM flavor). (so far)

AmigaLikeOS devices are "affected" by not having memory protection at all.

(+ ... it's not more "dangerous" to use intel for home computing vs Amiga because of Meltdown or Spectre ...
+ untill we have memory protection we do not need performance affecting meltdown -fix on intelAROS either
+ but hey: AROS becomes 0,5%...30% faster than before vs mainstream on intel HW (because Win+Linux slows down) )

Last edited by KimmoK on 05-Jan-2018 at 07:30 AM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK

Of course protection that is possible to bypass during some circumstances is still "better" than no protection at all.

But personally I've always liked the 3DO approach for protection in an Amiga-like OS: write protection only. Still would be possible to read passwords and such but in an Amiga-like OS one can assume the users know what they are doing not letting any crap into the system. ;)

@KimmoK

Looks like they updated their blog about PowerPC vulnerability to Spectre:

http://tenfourfox.blogspot.com/2018/01/more-about-spectre-and-powerpc-or-why.html

Quote:

KimmoK wrote: @bison Yep. Spectre affects also PowerPC CPUs. (more: http://tenfourfox.blogspot.fi/2018/01/is-powerpc-susceptible-to-spectre-yep.html ) Meltdown affects intel (and possibly one ARM flavor). (so far) AmigaLikeOS devices are "affected" by not having memory protection at all. (+ ... it's not more "dangerous" to use intel for home computing vs Amiga because of Meltdown or Spectre ... + untill we have memory protection we do not need performance affecting meltdown -fix on intelAROS either + but hey: AROS becomes 0,5%...30% faster than before vs mainstream on intel HW (because Win+Linux slows down) )

@thread

This is a really good explanation of the problem:

https://www.raspberrypi.org/blog/why-raspberry-pi-isnt-vulnerable-to-spectre-or-meltdown

Update:

Matthew Dillon said this:

http://lists.dragonflybsd.org/pipermail/users/2018-January/313758.html

And Linus said this:

https://lkml.org/lkml/2018/1/3/797

Update 2:

Matthew Dillon said this:

http://lists.dragonflybsd.org/pipermail/users/2018-January/335626.html

This all sounds really grim. All my systems are affected, with the exception of RPi3.

Last edited by bison on 08-Jan-2018 at 10:35 PM.
Last edited by bison on 05-Jan-2018 at 11:36 PM.

_________________
"Unix is supposed to fix that." -- Jay Miner

@themamboman

Thanks! So G3 is safe bet now.

ARM Cortex-A53 and A55 are also unaffected because they are not OOO. So it turns out going with a Raspberry Pi 3 as my general home server and a cheap ARM powered NAS for storage wasn't such a bad idea :P

Developers are not happy...

meltdown:
"Us kernel programmers have spent 20+ years reducing system call and interrupt overhead, and Intel blew it all up in one day. To say that we are all pissed would be a grave understatement."
"We kernel programmers have to implement the horrible performance destroying mitigation to workaround meltdown on Intel CPUs."


Spectre:
"Intel is also trying to push all sorts of crap onto the programming community. They are pushing hard to implement horrible hacks in GCC and other compilers and are trying to push horrible hacks to indirect procedure calls as a mitigation for spectre. THIS WILL NEVER WORK!!!!!. 30,000+ applications would have to be recompiled with the changes and kernels would have more horribly hacked code pushed into them just to obtain a PARTIAL mitigation."
"Spectre can only be completely fixed in hardware"


I have not yet seen tests on embedded PPC CPU (440, 460, P1022, P5020, T208x...) vs which ones are affected by spectre.
(so far G3 and 7400 are safe and 7450, G5 not safe)

People who plan implementing memory protection on amigalike systems:
A-do it the safe way on the first try
B-alternatively, wait for spectre immune CPUs

Last edited by KimmoK on 08-Jan-2018 at 09:46 AM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

>which ones are affected by spectre.

I have just recompiled the Spectre Poc for AmigaOs4 : tested on a sam460 it dont seems to works (dont get the message) but I cant conclude nothing about that for the moment as it may be the compilation that was wrong

Based on this:
http://nanard.free.fr/spectre-PowerPC.tar.gz

Just found a ppc_intrinsics.h on the net

and changed the beginning accordingly to let it compile on my cross compiler

#define __POWERPC__ 1
#define __ppc__ 1

#if defined(__POWERPC__)
#include "ppc_intrinsics.h"
#else


Anyway as already said Spectre is not needed to hack the simple AmigaOS

>Us kernel programmers have spent 20+ years reducing system call and interrupt overhead

And those coding gurus never imagined that having kernel space and user space mapped together may cause a problem once the supervisor mode will be hacked ...

Last edited by thellier on 08-Jan-2018 at 01:06 PM.

Spectre is particularly interesting. I wonder what other side channel attacks are possible on modern architectures?

My phone was rooted via a rowhammer exploit. It makes me wonder if the hardware designers are aware that such things are theoretically possible but choose to ignore it or whether we are just in an age in which the sheer complexity of everything exposes an attack surface so large that things like spectre are considered too remote compared to more easily leveraged attacks to be worth the effort. Until it's demonstrated and published, that is.

_________________
Doing stupid things for fun...

@Karlos

>Until it's demonstrated and published, that is.

So it is the experts that expose the poc that create the menace...
It is a self realising stuff "we experts create an exploit that may be dangerous if used this way" .... "ooh sorry we exposed how to use it so it is now really dangerous" "doooh"

As those "experts" create a financial prejudice to the industry (that now will need to update OSes & processors) so what is the difference with hackers ;-P

Today by removing the -pipe option in the makefile I recompiled natively the spectre poc on a Sam440 and tested it : didnt worked too

@thellier

Also recompiled on X1000, didn't work either

_________________
Cuida tus piedras gordas.

A1200/030 32Mb
A4000D
A1-X1000.

cool, first time people are happy to discover that a simple recompile of some code doesnt work on their systems? wonder if i should check if these would work on my real amigas..

Last edited by wawa on 09-Jan-2018 at 01:52 PM.

@wawa
>happy to discover that a simple recompile of some code doesnt work

And I just tested on the X5000 and it dont works too...
So tested on Sam440 Sam460 X5000 and works nowhere

@thellier

someone desperatively needs to port that proper! it doesnt work on up to date amigah system!

Last edited by wawa on 09-Jan-2018 at 06:07 PM.