Skip to contentSkip to conversation

Page Table Isolation (PTI) v4.14 backporting base tree

GIT PULL Ingo Molnar 2 participants 2 messages Dec 17, 2017

lore
Share & tools
lore
#0 lore raw

Linus,

Please pull the latest WIP.x86-pti.base-for-linus git tree from:

git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git WIP.x86-pti.base-for-linus

# HEAD: 2aeb07365bcd489620f71390a7d2031cd4dfb83e x86/mm/kasan: Don't use vmemmap_populate() to initialize shadow

Note, this tree should be pulled after:

[RFC GIT PULL] Page Table Isolation (PTI) preparatory tree

This tree contains the v4.14 PTI backport preparatory tree, which consists of four merges of upstream trees and 7 cherry-picked commits, which the upcoming PTI work depends on:

e5d77a73f364: Merge commit 'upstream-x86-virt' into WIP.x86/mm 2ec077c186a0: Merge branch 'upstream-acpi-fixes' into WIP.x86/pti.base 650400b2ccb8: Merge branch 'upstream-x86-selftests' into WIP.x86/pti.base 0fd2e9c53d82: Merge commit 'upstream-x86-entry' into WIP.x86/mm

2aeb07365bcd: x86/mm/kasan: Don't use vmemmap_populate() to initialize shadow 3382290ed2d5: locking/barriers: Convert users of lockless_dereference() to READ_ONCE()
c2bc66082e10: locking/barriers: Add implicit smp_read_barrier_depends() to READ_ONCE()
ab95477e7cb3: bpf: fix build issues on um due to mising bpf_perf_event.h 2fe1bc1f501d: perf/x86: Enable free running PEBS for REGS_USER/INTR
f2dbad36c55e: x86: Make X86_BUG_FXSAVE_LEAK detectable in CPUID on AMD
a8b4db562e72: x86/cpufeature: Add User-Mode Instruction Prevention definitions

pulling this should generate a trivial conflict in arch/x86/kernel/cpu/Makefile, which after resolution should result in a no-op merge.

Thanks,

Ingo

------------------>
Andi Kleen (1):
perf/x86: Enable free running PEBS for REGS_USER/INTR

Andrey Ryabinin (1):
x86/mm/kasan: Don't use vmemmap_populate() to initialize shadow

Daniel Borkmann (1):
bpf: fix build issues on um due to mising bpf_perf_event.h

Ingo Molnar (4):
Merge commit 'upstream-x86-entry' into WIP.x86/mm Merge branch 'upstream-x86-selftests' into WIP.x86/pti.base Merge branch 'upstream-acpi-fixes' into WIP.x86/pti.base Merge commit 'upstream-x86-virt' into WIP.x86/mm

Ricardo Neri (1):
x86/cpufeature: Add User-Mode Instruction Prevention definitions

Rudolf Marek (1):
x86: Make X86_BUG_FXSAVE_LEAK detectable in CPUID on AMD

Will Deacon (2):
locking/barriers: Add implicit smp_read_barrier_depends() to READ_ONCE() locking/barriers: Convert users of lockless_dereference() to READ_ONCE()

[ no-op tree, no diffstat. ]

#1 lore raw

Ok, so I've pulled the two prep trees, since they are obviously safe and don't change any code.

I'm going to read through the actual real changes again one more time today, but as mentioned, I almost certainly won't actually pull them until after rc4. I'd like to have at least one calm rc here.

But as part of this all, _please_ make these prep pulls actually work out - even if you end up having more work in the actual x86 entry tree and the eventual PTI changes themselves, make sure they really are based on this, and contain no more games. I was already a bit dubious about the whole "let's cherry-pick and then merge to have a common point", I'm going to be _really_ annoyed if we then need even more odd games here.

So keep that common point as the common point, and no more backport games or odd back-merges.

Linus