nodes crash using SEP collectors after Meltdown/Spectre patches

[Ronald_G_1]

After patching our OS for Meltdown we're seeing nodes crash during amplxe-cl collection using the HW counters (any of the non-PIN collections)

PIN collections work ( simple hotspots )

Also, using the older 2017 Update 1 VTune and collectors there are no issues.

But 2018 Update 1 with event counters crashes the node almost immediately after start of collection on any application. 

SEP drivers were not rebuilt after the Meltdown kernel patches - perhaps that is a test to do?

Anyone else seen this behavior?  Nodes crash hard and have to be power cycled.

[McCalpinJohn]

Are you seeing "double fault" error messages in the system logs? 

We have seen this a couple of times with 18.0.2 (build 525261), but we don't yet have a concise, reliable reproducer....

[H__Dane]

I also encountered this just now when doing an Advanced Hotspots trace, googled it and found this thread.

The second after clicking the "Play" button on VTune 2017, my system instantly crashed without a BSOD.

After upgrading to VTune 2017, it hanged and gave a BSOD saying something to the effect of "A clock interrupt was not received on a secondary processor".

My system is Windows 7 Enterprise, with the Meltdown patches just applied a couple of days ago.

[Jackson_M_Intel]

It looks like your issue may be related to this article. Please take a look:

https://software.intel.com/en-us/articles/intel-vtune-amplifier-impact-of-recent-os-security-updates  

As the article states, we're still investigating this.