All of lore.kernel.org
 help / color / mirror / Atom feed
From: Gerd Hoffmann <kraxel@redhat.com>
To: kvm-devel <kvm-devel@lists.sourceforge.net>,
	Glauber de Oliveira Costa <gcosta@redhat.com>,
	Avi Kivity <avi@qumranet.com>
Cc: Jeremy Fitzhardinge <jeremy@goop.org>
Subject: pv clock: kvm is incompatible with xen :-(
Date: Fri, 11 Apr 2008 09:33:21 +0200	[thread overview]
Message-ID: <47FF1441.6010801@redhat.com> (raw)

  Hi,

Tried to use kvmclock with xenner and noticed that the kvmclock
(MSR_KVM_SYSTEM_TIME msr) is incompatible with xen.

kvm guests do this to translate the tsc delta into nsecs:

   #define get_clock(cpu, field) per_cpu(hv_clock, cpu).field

   static inline u64 kvm_get_delta(u64 last_tsc)
   {
           int cpu = smp_processor_id();
           u64 delta = native_read_tsc() - last_tsc;
           return (delta * get_clock(cpu, tsc_to_system_mul)) >> 22;
   }

whereas xen guests do this (64bit version):

   static inline u64 scale_delta(u64 delta, u32 mul_frac, int shift)
   {
           u64 product;

           if (shift < 0)
                   delta >>= -shift;
           else
                   delta <<= shift;

           __asm__ (
                   "mul %%rdx ; shrd $32,%%rdx,%%rax"
                   : "=a" (product) : "0" (delta), "d" ((u64)mul_frac));
           return product;
   }

Note that xen does a 64bit multiply (of the 64bit delta and the 32bit
factor) yielding a 128bit result, then picking bits 32-95 for the 64bit
return value.  In contrast kvm does a simple 64bit multiply, which is
equivalent to using the lowest 64 bits.  Thus kvm is off by factor 2^32,
and that without even considering the ordering of two (shift + multiply)
operations and any rounding errors ...

cheers,
  Gerd

-- 
http://kraxel.fedorapeople.org/xenner/

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone

             reply	other threads:[~2008-04-11  7:33 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-04-11  7:33 Gerd Hoffmann [this message]
2008-04-11 12:06 ` pv clock: kvm is incompatible with xen :-( Avi Kivity
2008-04-11 13:44   ` Gerd Hoffmann
2008-04-11 15:01     ` Gerd Hoffmann
2008-04-11 19:02     ` Jeremy Fitzhardinge
2008-04-18 14:26       ` Gerd Hoffmann
2008-04-18 22:23         ` Jeremy Fitzhardinge
2008-04-21  7:31           ` Gerd Hoffmann
2008-04-21 11:46             ` Jeremy Fitzhardinge
2008-04-21 12:50               ` Gerd Hoffmann
2008-04-21 13:34                 ` Jeremy Fitzhardinge
2008-04-21 14:20                   ` Gerd Hoffmann
2008-04-22 17:54                 ` Glauber Costa
2008-04-23  6:03                   ` Gerd Hoffmann
2008-04-24 12:57                     ` Glauber Costa

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:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

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

  git send-email \
    --in-reply-to=47FF1441.6010801@redhat.com \
    --to=kraxel@redhat.com \
    --cc=avi@qumranet.com \
    --cc=gcosta@redhat.com \
    --cc=jeremy@goop.org \
    --cc=kvm-devel@lists.sourceforge.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.