[Beowulf] cursed (and perhaps blessed) Intel microcode

[Mark Hahn]

Hi all,
You may have seen that Intel has an updated "guidance" doc
on CPU microcode:

https://newsroom.intel.com/wp-content/uploads/sites/11/2018/03/microcode-update-guidance.pdf

Recall that in January, they released microcode for a number of models,
then later told people not to use it. we updated a big new cluster
to the new 0xb000025 - and it's not causing problems,
but would like to get the new "production" version xb00002A
mentioned in this document. we have Broadwell EX chips
(406f1 - aka 06-4f-01 or in linux terms, family 6, model 79, stepping 1)

There *is* an updated microcode data file:
https://downloadcenter.intel.com/product/873/Processors
which seems to correspond to the document above.
But alas, it contains buggy 06-4f-01 microcode 0xb00002A.

Has anyone managed to get a clear idea of where things stand?
Attempting to engage Intel just got me a "I'll ask my boss whether
I can tell you about that" kind of response.

thanks,
Mark Hahn | SHARCnet Sysadmin | ***@sharcnet.ca | http://www.sharcnet.ca
| McMaster RHPCS | ***@mcmaster.ca | 905 525 9140 x24687
| Compute/Calcul Canada | http://www.computecanada.ca
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit

[Mikhail Kuzminsky]

Post by Mark Hahn
https://downloadcenter.intel.com/product/873/Processors
which seems to correspond to the document above

How I believe, it's for correction of general defect practically for
all CPUs.
Then microcode update may give also decrease of performance. May be
Intel have idea to improve this update and do not recommend use this
version ?

Mikhail Kuzminsky
Zelinsky Institute of Organic Chemistry
Moscow
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.or

[Mark Hahn]

Post by Mark Hahn
https://downloadcenter.intel.com/product/873/Processors
which seems to correspond to the document above

How I believe, it's for correction of general defect practically for all
CPUs.
Then microcode update may give also decrease of performance. May be Intel
have idea to improve this update and do not recommend use this version ?

I don't quite understand what you're saying. Yes, it's expected that
the mitigations for meltdown and spectre have performance costs.
meltdown is pretty much taken care of by the way newer kernels avoid
mapping all physical memory. afaikt, the microcode changes are supposed
to mitigate the exfiltration risk for spectre, in combination with kernel
tweaks (retpoline, some extra patterns of fence instructions).

my appeal is for information about microcode updates that Intel claims
are now recommended for production use, and which only appeared in March.

regards, mark hahn.
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit ht

[Chris Samuel]

Post by Mark Hahn
But alas, it contains buggy 06-4f-01 microcode 0xb00002A.

Are you sure? I see this instead with the current version:

***@quad:/tmp$ iucode_tool -L microcode-20180312/intel-ucode/06-4f-01
microcode bundle 1: microcode-20180312/intel-ucode/06-4f-01
001/001: sig 0x000406f1, pf_mask 0xef, 2017-03-01, rev 0xb000021, size 26624

All the best,
Chris

[Mark Hahn]

Post by Chris Samuel

Post by Mark Hahn
But alas, it contains buggy 06-4f-01 microcode 0xb00002A.

microcode bundle 1: microcode-20180312/intel-ucode/06-4f-01
001/001: sig 0x000406f1, pf_mask 0xef, 2017-03-01, rev 0xb000021, size 26624

sorry, typo. yes, it contains the buggy 0xb000021,
from January, not the expected/desired 0xb00002A.

can I take it that you're looking for the update too (and don't have it)?

thanks, mark.
_______________________________________________
Beowulf mailing list, ***@beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.or

[Chris Samuel]

Post by Mark Hahn
sorry, typo. yes, it contains the buggy 0xb000021,
from January, not the expected/desired 0xb00002A.

Oh! Well that's not good... :-(

I'll try and nudge a person I know there on that...

Post by Mark Hahn
can I take it that you're looking for the update too (and don't have it)?

Currently waiting to see if it percolates down through the distro chains
(we're on RHEL, but I saw that Ubuntu has passed it on).

However, we've only got Skylake and KNL on the system I'm working on, though
we may inherit older nodes at some point when the older system gets
decommissioned.

All the best,
Chris

[Christopher Samuel]

Hi Mark,

Post by Chris Samuel
I'll try and nudge a person I know there on that...

They did some prodding, and finally new firmware emerged at the end of
last month.

/tmp/microcode-20180425$ iucode_tool -L intel-ucode-with-caveats/06-4f-01
microcode bundle 1: intel-ucode-with-caveats/06-4f-01
01/001: sig 0x000406f1, pf mask 0xef, 2018-03-21, rev 0xb00002c, size
27648

note the *with-caveats* part.

The releasenote file says:

-------------------8< snip snip 8<-------------------

-- intel-ucode-with-caveats/ --
This directory holds microcode that might need special handling.
BDX-ML microcode is provided in directory, because it need special
commits in
the Linux kernel, otherwise, updating it might result in unexpected system
behavior.

OS vendors must ensure that the late loader patches (provided in
linux-kernel-patches\) are included in the distribution before packaging the
BDX-ML microcode for late-loading.

-------------------8< snip snip 8<-------------------

Here be dragons..

Good luck!
Chris