Well, more information: Report for the same issue on Fedora: https://bugzilla.redhat.com/show_bug.cgi?id=1353103
All reports of this issue being "fixed" by a firmware update (in Debian, Arch and Fedora) are for firmware updates that come with microcode 0x8a already, effectively rendering the microcode update unecessary (which is detected by the kernel, and the microcode update is *not* attempted). There are no reports of issues on Skylake-U K-1 (the "somewhat fixed" hardware stepping of Skylake-U, present on Core i3-61xxU / i5-62xxU / i5-63xxU / i7-65xxU / i7-66xxU, where XX is *not* 00). These processors report pf=0x40. OTOH, we don't have reports of success for them, either. Status for the Skylake-U/Y D-1 Pentiums and Celerons is unknown, and I cannot even find boot logs for any of those in the 'net. Given what is known about this issue so far, I will upload a new version of the intel-microcode package *removing* the update for signature 0x406e3 (for all pf combinations). This change will be reverted when Intel issues a new revision of the 0x406e3 microcode. -- "One disk to rule them all, One disk to find them. One disk to bring them all and in the darkness grind them. In the Land of Redmond where the shadows lie." -- The Silicon Valley Tarot Henrique Holschuh