Personal computing discussed

Moderators: renee, Flying Fox, morphine

  • 1
  • 4
  • 5
  • 6
  • 7
  • 8
 
captaintrav
Gerbil First Class
Posts: 178
Joined: Thu Dec 12, 2013 12:51 pm
Location: Saskatchewan, Canada

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 4:20 pm

Microsoft is claiming Windows 7 has more of a performance hit than Windows 10 due to kernel design differences. But.... PCID acceleration is missing on the Windows 7 version of the patch. I don't want to spread FUD, but whether by necessity or by choice, whether you have a Skylake or newer isn't going to help the performance hit on Windows 7. Yet another unfortunate circumstance for those of us stuck waiting to validate applications on Windows 10 now.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 4:56 pm

PCID and INVPCID would require significant work to the Windows 7 kernel. So if the patch lacks the hardware acceleration path, I would not be surprised. Windows 7 is outside of mainstream support, Microsoft not gonna waste their time tweaking the kernel.

Server 2008 and Server 2012 don't even have a patch currently.
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 7:42 pm

https://youtu.be/W-AHuwfyLoo

I made a video that shows you how to manage the Meltdown and Spectre patch, e.g. status, enable, disable, in Windows.

For those of you with legacy systems, how to disable the KPTI changes is most likely of interest.
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
ludi
Lord High Gerbil
Posts: 8646
Joined: Fri Jun 21, 2002 10:47 pm
Location: Sunny Colorado front range

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 9:14 pm

Ryu Connor wrote:
https://youtu.be/W-AHuwfyLoo

I made a video that shows you how to manage the Meltdown and Spectre patch, e.g. status, enable, disable, in Windows.

For those of you with legacy systems, how to disable the KPTI changes is most likely of interest.

Is that your voice-over? Remove the southern accent and you could probably wake up Ok Google on my phone.
Abacus Model 2.5 | Quad-Row FX with 256 Cherry Red Slider Beads | Applewood Frame | Water Cooling by Brita Filtration
 
DragonDaddyBear
Gerbil Elite
Posts: 985
Joined: Fri Jan 30, 2009 8:01 am

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 9:30 pm

I just read the Ars article explaining just why there is a performance impact. It stems from the TLB design. For you smarter gerbils, it's TLB especially hard to design (relative to the complexity of designing processors)? AMD had a bug related to the same thing in the Phenom processors.

https://arstechnica.com/gadgets/2018/01 ... rformance/
 
chuckula
Minister of Gerbil Affairs
Posts: 2109
Joined: Wed Jan 23, 2008 9:18 pm
Location: Probably where I don't belong.

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 9:51 pm

DragonDaddyBear wrote:
I just read the Ars article explaining just why there is a performance impact. It stems from the TLB design. For you smarter gerbils, it's TLB especially hard to design (relative to the complexity of designing processors)? AMD had a bug related to the same thing in the Phenom processors.

https://arstechnica.com/gadgets/2018/01 ... rformance/


Ah the translation look-aside buffer... the bane of many many designers' existences.
A lot of this stuff comes down to how processors need to manage memory where there's the literal physical memory that holds stuff and then there's the virtual memory space that all your software actually sees. The TLB is vital in making the translation between virtual <--> physical memory as fast as possible.

As the article points out, one part of a bunch of the patches (especially meltdown) includes nuking the contents of the TLB more often to prevent the leakage of information from the kernel space to the user space. Every time the TLB gets flushed, a puppy cries! And the TLB needs to be flushed noticeably more often when KPTI is turned on.
4770K @ 4.7 GHz; 32GB DDR3-2133; Officially RX-560... that's right AMD you shills!; 512GB 840 Pro (2x); Fractal Define XL-R2; NZXT Kraken-X60
--Many thanks to the TR Forum for advice in getting it built.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 10:15 pm

ludi wrote:
Is that your voice-over? Remove the southern accent and you could probably wake up Ok Google on my phone.


Are you saying I sound like you with a southern accent?
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
ludi
Lord High Gerbil
Posts: 8646
Joined: Fri Jun 21, 2002 10:47 pm
Location: Sunny Colorado front range

Re: Intel Processor bug incoming?

Thu Jan 11, 2018 11:56 pm

Ryu Connor wrote:
ludi wrote:
Is that your voice-over? Remove the southern accent and you could probably wake up Ok Google on my phone.


Are you saying I sound like you with a southern accent?

Close enough, yes. One of us could get fired from a voice acting job and the audience wouldn't notice when the other one got hired as a replacement.
Abacus Model 2.5 | Quad-Row FX with 256 Cherry Red Slider Beads | Applewood Frame | Water Cooling by Brita Filtration
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 12:20 am

DragonDaddyBear wrote:
I just read the Ars article explaining just why there is a performance impact. It stems from the TLB design. For you smarter gerbils, it's TLB especially hard to design (relative to the complexity of designing processors)? AMD had a bug related to the same thing in the Phenom processors.

Well, based on historical evidence, I'd say if it isn't difficult, CPU designers still somehow seem to manage to screw it up an awful lot. Intel's i860 (early '90s RISC and kind-of-VLIW-ish CPU, think of it as a precursor to the Itanium) also had some nasty TLB design flaws. Been there, seen that, have the battle scars...
Nostalgia isn't what it used to be.
 
cegras
Gerbil First Class
Posts: 193
Joined: Mon Nov 05, 2007 3:12 pm

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 9:32 am

 
Waco
Maximum Gerbil
Posts: 4850
Joined: Tue Jan 20, 2009 4:14 pm
Location: Los Alamos, NM

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 11:07 am

cegras wrote:

Yep. Desktop users may not care but this is causing absolute hell in performance-sensitive applications with lots of syscalls.
Victory requires no explanation. Defeat allows none.
 
DragonDaddyBear
Gerbil Elite
Posts: 985
Joined: Fri Jan 30, 2009 8:01 am

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 11:29 am

There's a lot of moving stuff on this. Does the patch affecting performance hit Ryzen as well?
 
captaintrav
Gerbil First Class
Posts: 178
Joined: Thu Dec 12, 2013 12:51 pm
Location: Saskatchewan, Canada

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 1:16 pm

Ryu Connor wrote:
PCID and INVPCID would require significant work to the Windows 7 kernel. So if the patch lacks the hardware acceleration path, I would not be surprised. Windows 7 is outside of mainstream support, Microsoft not gonna waste their time tweaking the kernel.

Server 2008 and Server 2012 don't even have a patch currently.


Ah, if Windows 7 didn't support PCID in the first place that makes sense.
 
captaintrav
Gerbil First Class
Posts: 178
Joined: Thu Dec 12, 2013 12:51 pm
Location: Saskatchewan, Canada

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 1:31 pm

Waco wrote:
cegras wrote:

Yep. Desktop users may not care but this is causing absolute hell in performance-sensitive applications with lots of syscalls.


I wonder if for whatever reason (kernel design?) Linux is getting hit harder performance wise. Or it could be just the workloads in question? What's worse is the current kernel enables KPTI to address Meltdown, but patching Spectre on Linux is still coming. So possibly more performance hits looming?
 
cegras
Gerbil First Class
Posts: 193
Joined: Mon Nov 05, 2007 3:12 pm

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 1:34 pm

https://translate.google.de/translate?h ... chmarks%2F

If the frametimes are considered, the differences become clearer: The threshold, over which 99 per cent of the frametimes lie, falls with active Meltdown Patch by four (720p) and / or three (1080p) per cent, the active Specter countermeasures cost again four and / or three per cent , All in all, the performance losses in this discipline are therefore very much in line with what could have been observed when switching to the previous CPU generation.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel Processor bug incoming?

Fri Jan 12, 2018 1:52 pm

captaintrav wrote:
Waco wrote:
cegras wrote:

Yep. Desktop users may not care but this is causing absolute hell in performance-sensitive applications with lots of syscalls.

I wonder if for whatever reason (kernel design?) Linux is getting hit harder performance wise. Or it could be just the workloads in question? What's worse is the current kernel enables KPTI to address Meltdown, but patching Spectre on Linux is still coming. So possibly more performance hits looming?

I suspect it is very much a function of workload. Server applications that do a lot of I/O will be hit hard because they make a lot of system calls.
Nostalgia isn't what it used to be.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Intel Processor bug incoming?

Sat Jan 13, 2018 12:24 am

Speculative Execution Side Channel Mitigations

https://software.intel.com/sites/defaul ... ations.pdf
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
nerdrage
Graphmaster Gerbil
Posts: 1314
Joined: Thu Mar 06, 2003 2:49 pm
Location: Raleigh, NC

Re: Intel Processor bug incoming?

Mon Jan 29, 2018 10:27 am

The latest news from Saturday is we now need an emergency out-of-band Windows update to roll back the kernel mitigation for Spectre variant 2 because it may cause data loss and/or corruption in addition to the known rebooting issue. On top of that, Intel says to stop patching microcode/firmware updates until further notice. Dell and HP have pulled their firmware updates. No wonder Torvalds is so pissed.

What a fubar'ed mess.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel Processor bug incoming?

Mon Jan 29, 2018 11:49 am

Yeah, we appear to be progressing beyond "dumpster fire", and approaching "smoking crater" status.
Nostalgia isn't what it used to be.
 
Glorious
Gerbilus Supremus
Posts: 12343
Joined: Tue Aug 27, 2002 6:35 pm

Re: Intel Processor bug incoming?

Mon Jan 29, 2018 12:53 pm

nerdrage wrote:
The latest news from Saturday is we now need an emergency out-of-band Windows update to roll back the kernel mitigation for Spectre variant 2 because it may cause data loss and/or corruption in addition to the known rebooting issue. On top of that, Intel says to stop patching microcode/firmware updates until further notice. Dell and HP have pulled their firmware updates.


These would appear to be all the same thing. That microsoft update is cagey, but I can't see how what they are talking about isn't a reversion of the microcode they must have slipped into a previous update.

I don't know why they aren't straightforward about that, but I can't figure what the heck else they are saying.

Ubuntu did the exact same thing about a week ago.
 
Ryu Connor
Global Moderator
Posts: 4369
Joined: Thu Dec 27, 2001 7:00 pm
Location: Marietta, GA
Contact:

Re: Intel Processor bug incoming?

Mon Jan 29, 2018 5:03 pm

Glorious wrote:
nerdrage wrote:
The latest news from Saturday is we now need an emergency out-of-band Windows update to roll back the kernel mitigation for Spectre variant 2 because it may cause data loss and/or corruption in addition to the known rebooting issue. On top of that, Intel says to stop patching microcode/firmware updates until further notice. Dell and HP have pulled their firmware updates.


These would appear to be all the same thing. That microsoft update is cagey, but I can't see how what they are talking about isn't a reversion of the microcode they must have slipped into a previous update.

I don't know why they aren't straightforward about that, but I can't figure what the heck else they are saying.

Ubuntu did the exact same thing about a week ago.


The OOB Patch sets a registry key that disables the Spectre kernel mitigation. This is meant to help people who have already updated their firmware, have the defective Intel microcode, and are now suffering with instability.

So if you never updated your firmware, you don't need this OOB patch.

Link
All of my written content here on TR does not represent or reflect the views of my employer or any reasonable human being. All content and actions are my own.
 
Glorious
Gerbilus Supremus
Posts: 12343
Joined: Tue Aug 27, 2002 6:35 pm

Re: Intel Processor bug incoming?

Mon Jan 29, 2018 5:09 pm

Ah thanks, I wish I had found that before I posted.
 
bfg-9000
Gerbil Team Leader
Posts: 242
Joined: Tue Mar 01, 2016 9:17 pm

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 5:05 am

I don't claim to understand this at all, but apparently full software protection from these vulnerabilities requires not only microcode and OS patches, but application compiler changes as well--and Paul Kocher saw a 60% performance hit from protecting every conditional access this way.

I read this as the current software fixes only being quick-and-dirty mitigations for just the specific vulnerabilities mentioned in the original paper, intended to keep performance loss to a minimum until fixed hardware can be produced in quantity. Then we may begin to see more severe fixes to drive customers to purchase the new hardware. Truly a win-win for both Intel and Microsoft.

More software mitigations may be needed soon as someone at nVidia just discovered you can use other cores to trigger the cache coherence protocol too.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 9:03 am

OK, the thing to keep in mind with Spectre is that exploitation requires that the attacker already have the ability to run malicious code on your machine. So protecting against every single possible attack vector, in every piece of code, is paranoid and counter-productive. I hope the compiler and application vendors don't do this; it's like using a bazooka to take out a rat. Way too much collateral damage. In most use cases a typical home or office user is likely to encounter (I'll discuss the biggest exception below), if you're being attacked via the Spectre vulnerability, it's already "game over". Your system is already compromised and Spectre is pretty far down on your list of worries.

The biggest risk by far is outdated/unpatched web browsers. Your web browser implicitly runs untrusted code all the time (in the form of JavaScript on web pages). This opens the possibility of malicious code on a web page using the Spectre vulnerability to leak information from outside the JavaScript sandbox (i.e. anything in your web browser's memory) to the attacker. As an end user, the first (and most important) line of defense is to make sure you keep your web browser updated; Microsoft, Google, and Mozilla are already putting mitigations in place. The JavaScript compilers in the web browsers are being updated with the appropriate patches, and the sandboxing of data associated with different web sites is being strengthened.
Nostalgia isn't what it used to be.
 
Bauxite
Gerbil Elite
Posts: 788
Joined: Sat Jan 28, 2006 12:10 pm
Location: electrolytic redox smelting plant

reality check

Fri Feb 16, 2018 9:58 am

Javascript should be default deny surfing the web, if the computer you're using has anything of real value on it. Only whitelist domains you need to accomplish a purpose. Decide how much you really give a damn about websites that still don't work when you enable their root and cdn domains. The browsers try hard, but keep in mind for many of their vulnerability updates, they learned about and patched that particular issue only because someone else already got owned by it.

Honestly though the best policy is you have computer(s) just for surfing and it can't connect in any meaningful way (shares, casual file transfer, etc) to computers you care about. Its hard because you have to change your habits. You are the biggest security threat.
TR RIP 7/7/2019
 
Glorious
Gerbilus Supremus
Posts: 12343
Joined: Tue Aug 27, 2002 6:35 pm

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 10:39 am

Bauxite wrote:
Javascript should be default deny surfing the web, if the computer you're using has anything of real value on it. Only whitelist domains you need to accomplish a purpose. Decide how much you really give a damn about websites that still don't work when you enable their root and cdn domains. The browsers try hard, but keep in mind for many of their vulnerability updates, they learned about and patched that particular issue only because someone else already got owned by it.


I'm not entirely sure what default deny really accomplishes over simply blocking it altogether: do you actually proof-read all that minified JS? If not, you're clicking "run" in ignorance anyway.

And if you block altogether, I mean, this is basically just not using the internet. Are you going to do the full RMS thing and wget/Curl sites to read offline?

And whitelisting, wow. That is a lot of work to accomplish very little: Cisco put out a report, I don't know, FIVE YEARS AGO, that completely disputed the naive idea that malware only comes from "bad" websites. In fact, they found the opposite: most of it came from "legitimate" sites.

https://www.techrepublic.com/blog/it-se ... 0rn-sites/

And, in 2018, since the major browser vendors have features where they keep records of known-bad actors and warn or prevent you from accessing them, there is every reason to believe that this is only *MORE* true today. That's why MS keeps nagging about how Edge is the "more secure" browser, those guys compete on the basis of those lists. Virtually all of the potential value of what you think you are doing is baked in already by people who do this 24/7 and professionally.

---

I'm sorry, but when your recommendations are either "basically don't use the internet" or "do an incredible amount of work for minimal-to-zero return" this is just empty security theater in which the actors are overworked.

Bauxite wrote:
Honestly though the best policy is you have computer(s) just for surfing and it can't connect in any meaningful way (shares, casual file transfer, etc) to computers you care about. Its hard because you have to change your habits. You are the biggest security threat.


Unless you have real network segmentation, like using the advanced features of a"smart" network switch, that compromised computer *can* connect in a meaningful way to the rest of your network.

In fact, the point isn't anything you did on the outer-edge computer, but rather how much work you did to secure *EVERYTHING ELSE* on the rest of your LAN.

I mean, if you have a NAS, how is it configured? Are you totally sure it was appropriately locked down? EDIT: (and you never-ever "just this once" accessed it from the browsing computer? That's an incredible amount of discipline to expect from anyone).

If you have a server, does the IPMI has a default password (supermicro, as of just 2 years ago, yes! :roll: )?

What about everything else on the network/wifi that isn't strictly a regular computer? Webcams? Speakers? IoT isn't exactly known for the best security in the world, is it? :wink:

Again, security theater. Just because you don't currently have a mapped network share doesn't mean you're immune. At all.
 
just brew it!
Administrator
Posts: 54500
Joined: Tue Aug 20, 2002 10:51 pm
Location: Somewhere, having a beer

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 10:49 am

Disabling JavaScript may have been a viable option 15 years ago. Doing so today renders most of the web useless, or at best very inconvenient to use.

Yes, these vulnerabilities are a big deal for service providers, because they can result in leakage of data between customers who are running on shared infrastructure. For the typical end user, nothing has really changed. Keep your OS and browser current on security patches, and try to avoid the dark alleys of the internet.
Nostalgia isn't what it used to be.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 11:01 am

just brew it! wrote:
Disabling JavaScript may have been a viable option 15 years ago. Doing so today renders most of the web useless, or at best very inconvenient to use.

Yes, these vulnerabilities are a big deal for service providers, because they can result in leakage of data between customers who are running on shared infrastructure. For the typical end user, nothing has really changed. Keep your OS and browser current on security patches, and try to avoid the dark alleys of the internet.

And steer clear of reddit and 4chan. I hear they are these "dark alleys" JBI speaks of (/s but seriously stay away from these toxic places)

In other news, Intel has been hit with at least 32 lawsuits/SEC complaints from Meltdown/Spectre. What a joy.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
Aether
Gerbil First Class
Posts: 154
Joined: Sat Dec 20, 2014 8:50 pm

Re: Intel Processor bug incoming?

Fri Feb 16, 2018 12:36 pm

just brew it! wrote:
The biggest risk by far is outdated/unpatched web browsers. Your web browser implicitly runs untrusted code all the time (in the form of JavaScript on web pages). This opens the possibility of malicious code on a web page using the Spectre vulnerability to leak information from outside the JavaScript sandbox (i.e. anything in your web browser's memory) to the attacker. As an end user, the first (and most important) line of defense is to make sure you keep your web browser updated; Microsoft, Google, and Mozilla are already putting mitigations in place. The JavaScript compilers in the web browsers are being updated with the appropriate patches, and the sandboxing of data associated with different web sites is being strengthened.


I appreciate your posting this. I had gotten this impression from articles I have read regarding Spectre, but this is the first time I have seen it clearly stated.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Intel Processor bug incoming?

Thu Mar 15, 2018 9:09 am

Intel is announcing that they have developed hardware fixes for both the Meltdown and Spectre v2 vulnerabilities, which in turn will be implemented into future processors. Both the next version of Intel’s Xeon server/HEDT platform – Cascade Lake – as well as new 8th gen Core processors set to ship in the second half of this year will include the mitigations.


https://www.anandtech.com/show/12533/in ... ium=social
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
  • 1
  • 4
  • 5
  • 6
  • 7
  • 8

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On