linux-kernel - [PATCH v2 0/6] KVM: arm64: emulation for CTR

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20240426104950.7382-1-sebott@redhat.com>
Date: Fri, 26 Apr 2024 12:49:44 +0200
From: Sebastian Ott <sebott@...hat.com>
To: linux-arm-kernel@...ts.infradead.org,
	kvmarm@...ts.linux.dev,
	linux-kernel@...r.kernel.org
Cc: Marc Zyngier <maz@...nel.org>,
	Oliver Upton <oliver.upton@...ux.dev>,
	James Morse <james.morse@....com>,
	Suzuki K Poulose <suzuki.poulose@....com>,
	Catalin Marinas <catalin.marinas@....com>,
	Will Deacon <will@...nel.org>
Subject: [PATCH v2 0/6] KVM: arm64: emulation for CTR_EL0

Hej folks,

I'm looking into supporting migration between 2 Ampere Altra (Max)
machines (using Neoverse-N1). They are almost identical regarding
their feature id register state except for CTR_EL0.DIC which is set
on one machine but not the other.

CTR_EL0 is currently marked as invariant and migrating a VM between
those 2 machines using qemu fails.

Changes RFC [0] -> V1 [1]:
 * store the emulated value per VM and not per VCPU
 * allow to change more values than just the DIC bit
 * only trap guest access to that reg when needed
 * make sure to not present the guest with an inconsistent register set
Changes V1 -> V2:
 * implemented Marc's suggestion for keeping registers consistent while
   not breaking userspace ABI / expectations (I hope correctly this time)
 * keep the shadowed value valid at all time
 * unify the code to setup traps

Note:
* in order to switch off CTR_EL0.IDC userspace first has to set up
  CLIDR_EL1 accordingly
* reading CCSIDR before and after changing the cache size in CTR_EL0
  could result in different values (but only if CCSIDR was not changed
  from the generated value)
* I'll prepare a testcase for the next revision

Thanks,
Sebastian

[0]: https://lore.kernel.org/all/20240318111636.10613-1-sebott@redhat.com/T/
[1]: https://lore.kernel.org/lkml/20240405120108.11844-1-sebott@redhat.com/T/

Sebastian Ott (6):
  KVM: arm64: change return value in arm64_check_features()
  KVM: arm64: unify trap setup code
  KVM: arm64: maintain per VM value for CTR_EL0
  KVM: arm64: add emulation for CTR_EL0 register
  KVM: arm64: show writable masks for feature registers
  KVM: arm64: rename functions for invariant sys regs

 arch/arm64/include/asm/kvm_emulate.h |  37 -----
 arch/arm64/include/asm/kvm_host.h    |   4 +-
 arch/arm64/kvm/arm.c                 |   3 +-
 arch/arm64/kvm/sys_regs.c            | 225 ++++++++++++++++++++-------
 4 files changed, 173 insertions(+), 96 deletions(-)

-- 
2.42.0