Intel processor security flaw requires OS kernel level fix

[MrJim]

No drop in games performance then, nice.

Not on the 8700k, at least. I guess it's possible other Intel CPUs may be more seriously affected...

[Biscuit]

Not on the 8700k, at least. I guess it's possible other Intel CPUs may be more seriously affected...

Thats what I was thinking. Be interesting to see how Core2 and Gen1-Gen3 Core i chips have been affected

[Lanky123]

Turn around on silicon is slow. Months to prove a fix, then 3 months for production silicon at the new mask rev level to turn up in the market place.

This got me thinking - anyone have a reasonable guess as to how easy this will be to fix in new processors? New stepping, new model or entirely new architecture to sort it out? Mainly wondering how the Intel roadmap is likely to be affected - can they afford to release their next batch of CPUs (Cannonlake?) with the problem still there on the basis that OS updates have kind of, sort of fixed it anyway?

[Tabbykatze]

This got me thinking - anyone have a reasonable guess as to how easy this will be to fix in new processors? New stepping, new model or entirely new architecture to sort it out? Mainly wondering how the Intel roadmap is likely to be affected - can they afford to release their next batch of CPUs (Cannonlake?) with the problem still there on the basis that OS updates have kind of, sort of fixed it anyway?

I would suspect because they're already in their final stages, Intel will continue with their deployment because in actuality these issues don't affect the commercial market too much anyway. However, on the Server and Corporate side of things, I expect a large lashback against Intel and Intel may have to actually hold back their server arm to fix these flaws. No company wants to buy a processor with a potential 30% speed dump in their applications or run a security risk.

[cheesemp]

This got me thinking - anyone have a reasonable guess as to how easy this will be to fix in new processors? New stepping, new model or entirely new architecture to sort it out? Mainly wondering how the Intel roadmap is likely to be affected - can they afford to release their next batch of CPUs (Cannonlake?) with the problem still there on the basis that OS updates have kind of, sort of fixed it anyway?

From my very limited understanding of Meltdown (the intel only issue) it should be fixable using some improvements to the speculative execution architecture. Not a minor change at a guess but not a complete rewrite (based as I say on my limited understanding although I do have a degree in computer engineering so not completely armchair - just slightly as I've become software focused since my degree). As for the more general 'Spectre' issue I'm not as sure. Its been implied Spectre can be mitigated by software design however that then leaves you at the mercy of devs...

[DanceswithUnix]

This got me thinking - anyone have a reasonable guess as to how easy this will be to fix in new processors? New stepping, new model or entirely new architecture to sort it out? Mainly wondering how the Intel roadmap is likely to be affected - can they afford to release their next batch of CPUs (Cannonlake?) with the problem still there on the basis that OS updates have kind of, sort of fixed it anyway?

This stuff is deep into the execution pipelines. Clearly Meltdown can be fixed because AMD don't have the problem, but I can't see it being easy and could harm performance as the problem here seems to be how aggressively Intel are speculatively executing instructions vs how aggressively they are checking the permissions. Spectre sounds nastier, but I'm sure defences will be constructed.

It strikes me that execution has to switch from top address bit not set to top bit set (at least in AMD64 which is all people really care about these days and architecturally implements a split instruction space for OS vs applications) and detecting a single bit flip to tread carefully might lower the pain. But who knows other than Intel, I have never seen the code for their pipelines.

[scaryjim]

Just been tacked onto another complication with Windows patches via twitter:

https://support.microsoft.com/en-us/...dates-released

Microsoft has identified a compatibility issue with a small number of anti-virus software products.

The compatibility issue is caused when anti-virus applications make unsupported calls into Windows kernel memory. These calls may cause stop errors (also known as blue screen errors) that make the device unable to boot. To help prevent stop errors caused by incompatible anti-virus applications, Microsoft is only offering the Windows security updates released on January 3, 2018 to devices running anti-virus software from partners who have confirmed their software is compatible with the January 2018 Windows operating system security update.

Yep, your anti-virus may be making unsupported calls to kernel memory that will make your device unbootable if you apply a patch to fix a security vulnerability affecting kernel memory....

[Ozaron]

Yep, your anti-virus may be making unsupported calls to kernel memory that will make your device unbootable if you apply a patch to fix a security vulnerability affecting kernel memory....

I haven't even updated my 6600K for the IME vulnerability yet, someone please slow down this trainwreck.

[peterb]

Yep, your anti-virus may be making unsupported calls to kernel memory that will make your device unbootable if you apply a patch to fix a security vulnerability affecting kernel memory....

Well, at least that ensures your system is no longer vulnerable! Result!

[CAT-THE-FIFTH]

https://www.theregister.co.uk/2018/0...s_annotations/

We translated Intel's crap attempt to spin its way out of CPU security bug PR nightmare
As Linus Torvalds lets rip on Chipzilla

LOL.

[marshalex]

https://www.theregister.co.uk/2018/0...s_annotations/



LOL.

It's a pretty accurate summary of the Intel PR BS that has been spouted this morning.

[CAT-THE-FIFTH]

Also someone has shown how they can use the vulnerability to steal passwords too:

https://www.reddit.com/r/sysadmin/co..._in_real_time/

[CAT-THE-FIFTH]

It's a pretty accurate summary of the Intel PR BS that has been spouted this morning.

I can understand Intel doing it,but not hardware enthusiasts on some other forums defending it and then trying to make AMD sound like they have the issue to the "same" level,which is hilarious.

[CAT-THE-FIFTH]

No drop in games performance then, nice.

Computerbase.de did some benchmarks too.







However,I would like to see more I/O bound games tested,especially many true open world games(Skyrim,FO4,etc).


Also,this sounds a bit dodgy:

http://uk.businessinsider.com/intel-...18-1?r=US&IR=T

Intel was aware of the chip vulnerability when its CEO sold off $24 million in company stock

[Xlucine]

Looks like as much of an issue for games as ryzen's "1080p performance"

[CAT-THE-FIFTH]

Just been tacked onto another complication with Windows patches via twitter:

https://support.microsoft.com/en-us/...dates-released



Yep, your anti-virus may be making unsupported calls to kernel memory that will make your device unbootable if you apply a patch to fix a security vulnerability affecting kernel memory....

Avira have a patch coming soon.