| Message ID | 20200724173536.789982-1-oupton@google.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | kvm: x86: fix reversed timespec values in PV wall clock | expand |
On Fri, Jul 24, 2020 at 10:35 AM Oliver Upton <oupton@google.com> wrote: > > commit 8171cd68806b ("KVM: x86: use raw clock values consistently") > causes KVM to accidentally write seconds to the nanoseconds field (and > vice versa) in the KVM wall clock. Fix it by reversing this accidental > switch. Modulo the written nanoseconds value by NSEC_PER_SEC to correct > for the amount of time represented as seconds. > > Fixes: 8171cd68806b ("KVM: x86: use raw clock values consistently") > Cc: stable@vger.kernel.org > Reviewed-by: Jim Mattson <jmattson@google.com> > Reviewed-by: Peter Shier <pshier@google.com> > Signed-off-by: Oliver Upton <oupton@google.com> > --- > Parent commit: c34b26b98cac ("KVM: MIPS: clean up redundant 'kvm_run' parameters") > > arch/x86/kvm/x86.c | 5 +++-- > 1 file changed, 3 insertions(+), 2 deletions(-) > > diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c > index e27d3db7e43f..86228cc6b29e 100644 > --- a/arch/x86/kvm/x86.c > +++ b/arch/x86/kvm/x86.c > @@ -1809,8 +1809,9 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock) > */ > wall_nsec = ktime_get_real_ns() - get_kvmclock_ns(kvm); > > - wc.nsec = do_div(wall_nsec, 1000000000); > - wc.sec = (u32)wall_nsec; /* overflow in 2106 guest time */ > + /* overflow in 2106 guest time */ > + wc.sec = (u32)do_div(wall_nsec, NSEC_PER_SEC); > + wc.nsec = wall_nsec % NSEC_PER_SEC; > wc.version = version; > > kvm_write_guest(kvm, wall_clock, &wc, sizeof(wc)); > -- > 2.28.0.rc0.142.g3c755180ce-goog > Disregard, I crossed wires in my reading of do_div().
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index e27d3db7e43f..86228cc6b29e 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -1809,8 +1809,9 @@ static void kvm_write_wall_clock(struct kvm *kvm, gpa_t wall_clock) */ wall_nsec = ktime_get_real_ns() - get_kvmclock_ns(kvm); - wc.nsec = do_div(wall_nsec, 1000000000); - wc.sec = (u32)wall_nsec; /* overflow in 2106 guest time */ + /* overflow in 2106 guest time */ + wc.sec = (u32)do_div(wall_nsec, NSEC_PER_SEC); + wc.nsec = wall_nsec % NSEC_PER_SEC; wc.version = version; kvm_write_guest(kvm, wall_clock, &wc, sizeof(wc));
commit 8171cd68806b ("KVM: x86: use raw clock values consistently") causes KVM to accidentally write seconds to the nanoseconds field (and vice versa) in the KVM wall clock. Fix it by reversing this accidental switch. Modulo the written nanoseconds value by NSEC_PER_SEC to correct for the amount of time represented as seconds. Fixes: 8171cd68806b ("KVM: x86: use raw clock values consistently") Cc: stable@vger.kernel.org Reviewed-by: Jim Mattson <jmattson@google.com> Reviewed-by: Peter Shier <pshier@google.com> Signed-off-by: Oliver Upton <oupton@google.com> --- Parent commit: c34b26b98cac ("KVM: MIPS: clean up redundant 'kvm_run' parameters") arch/x86/kvm/x86.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-)