]> git.baikalelectronics.ru Git - kernel.git/commit
KVM: arm64: Allocate hyp vectors statically
authorWill Deacon <will@kernel.org>
Fri, 13 Nov 2020 11:38:44 +0000 (11:38 +0000)
committerMarc Zyngier <maz@kernel.org>
Mon, 16 Nov 2020 10:43:05 +0000 (10:43 +0000)
commit45f64ca21c57f0a8b03974281b69aee7989f6900
treeee08e27ab59bdf43f567d3edaa8e0ea5151d14a3
parent42133c28e4048629d3526c4db9f0540740c19ab0
KVM: arm64: Allocate hyp vectors statically

The EL2 vectors installed when a guest is running point at one of the
following configurations for a given CPU:

  - Straight at __kvm_hyp_vector
  - A trampoline containing an SMC sequence to mitigate Spectre-v2 and
    then a direct branch to __kvm_hyp_vector
  - A dynamically-allocated trampoline which has an indirect branch to
    __kvm_hyp_vector
  - A dynamically-allocated trampoline containing an SMC sequence to
    mitigate Spectre-v2 and then an indirect branch to __kvm_hyp_vector

The indirect branches mean that VA randomization at EL2 isn't trivially
bypassable using Spectre-v3a (where the vector base is readable by the
guest).

Rather than populate these vectors dynamically, configure everything
statically and use an enumerated type to identify the vector "slot"
corresponding to one of the configurations above. This both simplifies
the code, but also makes it much easier to implement at EL2 later on.

Signed-off-by: Will Deacon <will@kernel.org>
[maz: fixed double call to kvm_init_vector_slots() on nVHE]
Signed-off-by: Marc Zyngier <maz@kernel.org>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Quentin Perret <qperret@google.com>
Link: https://lore.kernel.org/r/20201113113847.21619-8-will@kernel.org
arch/arm64/include/asm/kvm_asm.h
arch/arm64/include/asm/spectre.h
arch/arm64/kernel/cpu_errata.c
arch/arm64/kernel/proton-pack.c
arch/arm64/kvm/arm.c
arch/arm64/kvm/hyp/Makefile
arch/arm64/kvm/hyp/hyp-entry.S
arch/arm64/kvm/hyp/smccc_wa.S [deleted file]
arch/arm64/kvm/va_layout.c