On top of Thomas+Paolo's

Hey,

This patchset replaces Paolo's second patch but also adds the warnings that Linus wanted. Along with some extra knobs in case the MSR will have more things in the future.

Oh, and one bug-fix.

I am also working on using the VMX Load MSR list so that the L1D flush happens automatically on VMENTER.

Konrad Rzeszutek Wilk (5):
x86/bugs: Move the l1tf function and define pr_fmt properly x86/cpufeatures: Add detection of L1D cache flush support. x86/bugs: Read FLUSH_CMD MSR during boot.
x86/KVM: Warn user if KVM is loaded SMT and L1TF CPU bug being present. x86/KVM: Use L1 cache flush before VMENTER if available.

Paolo Bonzini (1):
kvm: x86: mitigation for L1 cache terminal fault vulnerabilities

 Documentation/admin-guide/kernel-parameters.txt |  4 ++
 arch/x86/include/asm/cpu.h                      |  4 ++
 arch/x86/include/asm/cpufeatures.h              |  1 +
 arch/x86/include/asm/kvm_host.h                 |  7 ++-
 arch/x86/include/asm/msr-index.h                |  6 ++
 arch/x86/kernel/cpu/bugs.c                      | 83 +++++++++++++++++--------
 arch/x86/kvm/mmu.c                              |  1 +
 arch/x86/kvm/svm.c                              |  3 +-
 arch/x86/kvm/vmx.c                              | 46 +++++++++++++-
 arch/x86/kvm/x86.c                              | 80 +++++++++++++++++++++++-
 kernel/cpu.c                                    |  1 +
 11 files changed, 205 insertions(+), 31 deletions(-)

Done.

Attached is the git bundle of the patches that were posted and also code to do L1D flush using the VMX load MSR list.

I can most certainly post it, but I hadn't run any performance numbers so not sure if that approach is even worth it or not.

That's a pretty neat trick.

Is there any way to just _view_ a bundle? Without the pain to load it into git and then figure out how to extract useful patches from that again? git-bundle doesn't seem to have that option.

Not that I know of and the Oracles of Gurgle do not know anything either.

Thanks,

tglx

I can post them all once I rebase on l1tf.v6 that Thomas just sent.

A bundle on its own doesn't even have all the information. It's not a collection of patches. It's literally an offline version of the git synchronization protocol, so it's delta-compressed against the repository etc, and a bundle on its own will not make sense.

So you have to do

git fetch bundle

to integrate it into the repository and give it all the context it needs to make sense. Then "gitk ..FETCH_HEAD" etc will work, and all the normal git tools to look at patches.

Linus