linux-kernel - [PATCH v2 00/11] KVM: x86: pvclock fixes and cleanups

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: <20250201013827.680235-1-seanjc@google.com>
Date: Fri, 31 Jan 2025 17:38:16 -0800
From: Sean Christopherson <seanjc@...gle.com>
To: Sean Christopherson <seanjc@...gle.com>, Paolo Bonzini <pbonzini@...hat.com>, 
	David Woodhouse <dwmw2@...radead.org>, Paul Durrant <paul@....org>
Cc: kvm@...r.kernel.org, linux-kernel@...r.kernel.org, 
	syzbot+352e553a86e0d75f5120@...kaller.appspotmail.com, 
	Paul Durrant <pdurrant@...zon.com>, David Woodhouse <dwmw@...zon.co.uk>, 
	Vitaly Kuznetsov <vkuznets@...hat.com>
Subject: [PATCH v2 00/11] KVM: x86: pvclock fixes and cleanups

Fix a lockdep splat in KVM's suspend notifier by simply removing a spurious
kvm->lock acquisition related to kvmclock, and then try to wrangle KVM's
pvclock handling into something approaching sanity (I made the mistake of
looking at how KVM handled PVCLOCK_GUEST_STOPPED).

The Xen changes are slightly better tested this time around, though given
how many bugs there were in v1, that isn't saying a whole lot.

Case in point, I had to shuffle the scariest patch (at first glance),
"KVM: x86/xen: Use guest's copy of pvclock when starting timer", to land
earlier in the series, because the existing usage of the cache pvclock's
"version" field in kvm_xen_start_timer() relies on the pvclock update to
persist _a_ version number to the cache.  Eww. :-(

v2:
 - Collect reviews. [Paul, Vitaly]
 - Account for the compat offset in xen_get_guest_pvclock().
 - Try both the compat and non-compat clocks in kvm_xen_start_timer().
 - Fix a horrendous uninitialized variable bug. [Paul]
 - Move the kvm_xen_start_timer() change earlier so that it wouldn't be
   transiently broken due to "version" always being '0'.
 - Shuffle patches so that the "set PVCLOCK_GUEST_STOPPED only for
   kvmclock" is fully isolated from "Don't bleed PVCLOCK_GUEST_STOPPED". [Paul]

v1: https://lore.kernel.org/all/20250118005552.2626804-1-seanjc@google.com

Sean Christopherson (11):
  KVM: x86: Don't take kvm->lock when iterating over vCPUs in suspend
    notifier
  KVM: x86: Eliminate "handling" of impossible errors during SUSPEND
  KVM: x86: Drop local pvclock_flags variable in kvm_guest_time_update()
  KVM: x86: Process "guest stopped request" once per guest time update
  KVM: x86/xen: Use guest's copy of pvclock when starting timer
  KVM: x86: Don't bleed PVCLOCK_GUEST_STOPPED across PV clocks
  KVM: x86: Set PVCLOCK_GUEST_STOPPED only for kvmclock, not for Xen PV
    clock
  KVM: x86: Pass reference pvclock as a param to
    kvm_setup_guest_pvclock()
  KVM: x86: Remove per-vCPU "cache" of its reference pvclock
  KVM: x86: Setup Hyper-V TSC page before Xen PV clocks (during clock
    update)
  KVM: x86: Override TSC_STABLE flag for Xen PV clocks in
    kvm_guest_time_update()

 arch/x86/include/asm/kvm_host.h |   3 +-
 arch/x86/kvm/x86.c              | 115 ++++++++++++++++----------------
 arch/x86/kvm/xen.c              |  69 +++++++++++++++++--
 3 files changed, 121 insertions(+), 66 deletions(-)


base-commit: eb723766b1030a23c38adf2348b7c3d1409d11f0
-- 
2.48.1.362.g079036d154-goog