[PATCH v3 0/5] Making microcode late-load robust
From: Ashok Raj
Date: Wed Aug 17 2022 - 01:12:05 EST
Hi Boris
I thought v2 worked, but there were some glaring errors in them.
I added more trace and this did go through and finish. More testing would
help.
Let me know how you want to handle 1-3. patch4,5 needs more review and some
help testing on AMD as well.
v2: https://lore.kernel.org/lkml/20220816043754.3258815-1-ashok.raj@xxxxxxxxx/
v1: https://lore.kernel.org/lkml/20220813223825.3164861-1-ashok.raj@xxxxxxxxx/
Changes since v2
- Dropped Documentation, queued for tip
https://lore.kernel.org/lkml/166063601497.401.9527776956724011207.tip-bot2@tip-bot2/
Patches: 1-3
No Changes since v1
Patch4:
x2apic doesn't handle NMI vector via the self IPI MSR, it can only
send Fixed mode interrupts. Enhance send_IPI_self() to comprehend
and use a function that will deliver to the NMI.
Patch5:
- Modified when the NMI was triggered. Moved it closer to when
__reload_late() is running.
- Used the new apic->send_IPI_self() support helper.
- Added an atomic to check if the primary has completed the load.
- Triggering the NMI and when its taken determines how long the primary
CPU needs to wait. For Now i have no timeout, but will check on couple
more platforms before coming with a number, so it can bail out if
secondaries don't show up in a reasonable amount of time.
Ashok Raj (5):
x86/microcode/intel: Check against CPU signature before saving
microcode
x86/microcode/intel: Allow a late-load only if a min rev is specified
x86/microcode: Avoid any chance of MCE's during microcode update
x86/x2apic: Support x2apic self IPI with NMI_VECTOR
x86/microcode: Place siblings in NMI loop while update in progress
arch/x86/include/asm/mce.h | 4 +
arch/x86/include/asm/microcode_intel.h | 4 +-
arch/x86/kernel/apic/x2apic_phys.c | 6 +-
arch/x86/kernel/cpu/mce/core.c | 9 +
arch/x86/kernel/cpu/microcode/core.c | 229 ++++++++++++++++++++++++-
arch/x86/kernel/cpu/microcode/intel.c | 34 +++-
6 files changed, 274 insertions(+), 12 deletions(-)
--
2.32.0