archive mirror
 help / color / mirror / Atom feed
From: Thomas Gleixner <>
To: LKML <>
Cc: Miklos Szeredi <>,, Vincenzo Frascino <>,
	Andy Lutomirski <>,
	Paolo Bonzini <>,
	Juergen Gross <>,
	Christophe Leroy <>,
Subject: [patch 3/3] x86/vdso: Unbreak paravirt VDSO clocks
Date: Sat, 06 Jun 2020 23:51:17 +0200	[thread overview]
Message-ID: <> (raw)
In-Reply-To: <>

The conversion of x86 VDSO to the generic clock mode storage broke the
paravirt and hyperv clocksource logic. These clock sources have their own
internal sequence counter to validate the clocksource at the point of
reading it. This is necessary because the hypervisor can invalidate the
clocksource asynchronously so a check during the VDSO data update is not
sufficient. If the internal check during read invalidates the clocksource
the read return U64_MAX. The original code checked this efficiently by
testing whether the result (casted to signed) is negative, i.e. bit 63 is
set. This was done that way because an extra indicator for the validity had
more overhead.

The conversion broke this check because the check was replaced by a check
for a valid VDSO clock mode.

The wreckage manifests itself when the paravirt clock is installed as a
valid VDSO clock and during runtime invalidated by the hypervisor,
e.g. after a host suspend/resume cycle. After the invalidation the read
function returns U64_MAX which is used as cycles and makes the clock jump
by ~2200 seconds, and become stale until the 2200 seconds have elapsed
where it starts to jump again. The period of this effect depends on the
shift/mult pair of the clocksource and the jumps and staleness are an
artifact of undefined but reproducible behaviour of math overflow.

Implement an x86 version of the new vdso_cycles_ok() inline which adds this
check back and a variant of vdso_clocksource_ok() which lets the compiler
optimize it out to avoid the extra conditional. That's suboptimal when the
system does not have a VDSO capable clocksource, but that's not the case
which is optimized for.

Reported-by: Miklos Szeredi <>
Fixes: 5d51bee725cc ("clocksource: Add common vdso clock mode storage")
Signed-off-by: Thomas Gleixner <>
 arch/x86/include/asm/vdso/gettimeofday.h |   18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

--- a/arch/x86/include/asm/vdso/gettimeofday.h
+++ b/arch/x86/include/asm/vdso/gettimeofday.h
@@ -271,6 +271,24 @@ static __always_inline const struct vdso
 	return __vdso_data;
+static inline bool arch_vdso_clocksource_ok(const struct vdso_data *vd)
+	return true;
+#define vdso_clocksource_ok arch_vdso_clocksource_ok
+ * Clocksource read value validation to handle PV and HyperV clocksources
+ * which can be invalidated asynchronously and indicate invalidation by
+ * returning U64_MAX, which can be effectively tested by checking for a
+ * negative value after casting it to s64.
+ */
+static inline bool arch_vdso_cycles_ok(u64 cycles)
+	return (s64)cycles >= 0;
+#define vdso_cycles_ok arch_vdso_cycles_ok
  * x86 specific delta calculation.

  parent reply	other threads:[~2020-06-07  9:36 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-06-06 21:51 [patch 0/3] vdso: Unbreak VDSO with PV and HyperV clocksources Thomas Gleixner
2020-06-06 21:51 ` [patch 1/3] clocksource: Remove obsolete ifdef Thomas Gleixner
2020-06-09 14:40   ` [tip: x86/urgent] " tip-bot2 for Thomas Gleixner
2020-06-06 21:51 ` [patch 2/3] lib/vdso: Provide sanity check for cycles (again) Thomas Gleixner
2020-06-09 14:40   ` [tip: x86/urgent] " tip-bot2 for Thomas Gleixner
2020-06-06 21:51 ` Thomas Gleixner [this message]
2020-06-09 14:40   ` [tip: x86/urgent] x86/vdso: Unbreak paravirt VDSO clocks tip-bot2 for Thomas Gleixner
2020-06-09 13:10 ` [patch 0/3] vdso: Unbreak VDSO with PV and HyperV clocksources Miklos Szeredi

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \ \ \ \ \ \ \ \ \ \ \ \
    --subject='Re: [patch 3/3] x86/vdso: Unbreak paravirt VDSO clocks' \

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).