[Qemu-devel] [PATCH v2] migration: Do not re-read the clock on pre_save in case of paused guest

Maxiwell S. Garcia posted 1 patch 1 week ago
Test docker-clang@ubuntu passed
Test asan passed
Test docker-mingw@fedora passed
Test FreeBSD passed
Test checkpatch passed
Test s390x passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20190711194702.26598-1-maxiwell@linux.ibm.com
Maintainers: David Gibson <david@gibson.dropbear.id.au>
hw/ppc/ppc.c         | 13 +++++++++----
target/ppc/cpu-qom.h |  1 +
2 files changed, 10 insertions(+), 4 deletions(-)

[Qemu-devel] [PATCH v2] migration: Do not re-read the clock on pre_save in case of paused guest

Posted by Maxiwell S. Garcia 1 week ago
Re-read the timebase before migrate was ported from x86 commit:
   6053a86fe7bd: kvmclock: reduce kvmclock difference on migration

The clock move makes the guest knows about the paused time between
the stop and migrate commands. This is an issue in an already-paused
VM because some side effects, like process stalls, could happen
after migration.

So, this patch checks the runstate of guest in the pre_save handler and
do not re-reads the timebase in case of paused state (cold migration).

Signed-off-by: Maxiwell S. Garcia <maxiwell@linux.ibm.com>
---
 hw/ppc/ppc.c         | 13 +++++++++----
 target/ppc/cpu-qom.h |  1 +
 2 files changed, 10 insertions(+), 4 deletions(-)

diff --git a/hw/ppc/ppc.c b/hw/ppc/ppc.c
index a9e508c496..8572e45274 100644
--- a/hw/ppc/ppc.c
+++ b/hw/ppc/ppc.c
@@ -1008,6 +1008,8 @@ static void timebase_save(PPCTimebase *tb)
      * there is no need to update it from KVM here
      */
     tb->guest_timebase = ticks + first_ppc_cpu->env.tb_env->tb_offset;
+
+    tb->runstate_paused = runstate_check(RUN_STATE_PAUSED);
 }
 
 static void timebase_load(PPCTimebase *tb)
@@ -1051,9 +1053,9 @@ void cpu_ppc_clock_vm_state_change(void *opaque, int running,
 }
 
 /*
- * When migrating, read the clock just before migration,
- * so that the guest clock counts during the events
- * between:
+ * When migrating a running guest, read the clock just
+ * before migration, so that the guest clock counts
+ * during the events between:
  *
  *  * vm_stop()
  *  *
@@ -1068,7 +1070,10 @@ static int timebase_pre_save(void *opaque)
 {
     PPCTimebase *tb = opaque;
 
-    timebase_save(tb);
+    /* guest_timebase won't be overridden in case of paused guest */
+    if (!tb->runstate_paused) {
+        timebase_save(tb);
+    }
 
     return 0;
 }
diff --git a/target/ppc/cpu-qom.h b/target/ppc/cpu-qom.h
index be9b4c30c3..5fbcdee9c9 100644
--- a/target/ppc/cpu-qom.h
+++ b/target/ppc/cpu-qom.h
@@ -201,6 +201,7 @@ typedef struct PowerPCCPUClass {
 typedef struct PPCTimebase {
     uint64_t guest_timebase;
     int64_t time_of_the_day_ns;
+    bool runstate_paused;
 } PPCTimebase;
 
 extern const struct VMStateDescription vmstate_ppc_timebase;
-- 
2.20.1


Re: [Qemu-devel] [PATCH v2] migration: Do not re-read the clock on pre_save in case of paused guest

Posted by David Gibson 1 week ago
On Thu, Jul 11, 2019 at 04:47:02PM -0300, Maxiwell S. Garcia wrote:
> Re-read the timebase before migrate was ported from x86 commit:
>    6053a86fe7bd: kvmclock: reduce kvmclock difference on migration
> 
> The clock move makes the guest knows about the paused time between
> the stop and migrate commands. This is an issue in an already-paused
> VM because some side effects, like process stalls, could happen
> after migration.
> 
> So, this patch checks the runstate of guest in the pre_save handler and
> do not re-reads the timebase in case of paused state (cold migration).
> 
> Signed-off-by: Maxiwell S. Garcia <maxiwell@linux.ibm.com>

I've applied this to ppc-for-4.2.  I think it probably is a correct
fix, but this could have subtle impacts on things that are mostly
working at the moment, so I don't want to risk it during hard freeze.

> ---
>  hw/ppc/ppc.c         | 13 +++++++++----
>  target/ppc/cpu-qom.h |  1 +
>  2 files changed, 10 insertions(+), 4 deletions(-)
> 
> diff --git a/hw/ppc/ppc.c b/hw/ppc/ppc.c
> index a9e508c496..8572e45274 100644
> --- a/hw/ppc/ppc.c
> +++ b/hw/ppc/ppc.c
> @@ -1008,6 +1008,8 @@ static void timebase_save(PPCTimebase *tb)
>       * there is no need to update it from KVM here
>       */
>      tb->guest_timebase = ticks + first_ppc_cpu->env.tb_env->tb_offset;
> +
> +    tb->runstate_paused = runstate_check(RUN_STATE_PAUSED);
>  }
>  
>  static void timebase_load(PPCTimebase *tb)
> @@ -1051,9 +1053,9 @@ void cpu_ppc_clock_vm_state_change(void *opaque, int running,
>  }
>  
>  /*
> - * When migrating, read the clock just before migration,
> - * so that the guest clock counts during the events
> - * between:
> + * When migrating a running guest, read the clock just
> + * before migration, so that the guest clock counts
> + * during the events between:
>   *
>   *  * vm_stop()
>   *  *
> @@ -1068,7 +1070,10 @@ static int timebase_pre_save(void *opaque)
>  {
>      PPCTimebase *tb = opaque;
>  
> -    timebase_save(tb);
> +    /* guest_timebase won't be overridden in case of paused guest */
> +    if (!tb->runstate_paused) {
> +        timebase_save(tb);
> +    }
>  
>      return 0;
>  }
> diff --git a/target/ppc/cpu-qom.h b/target/ppc/cpu-qom.h
> index be9b4c30c3..5fbcdee9c9 100644
> --- a/target/ppc/cpu-qom.h
> +++ b/target/ppc/cpu-qom.h
> @@ -201,6 +201,7 @@ typedef struct PowerPCCPUClass {
>  typedef struct PPCTimebase {
>      uint64_t guest_timebase;
>      int64_t time_of_the_day_ns;
> +    bool runstate_paused;
>  } PPCTimebase;
>  
>  extern const struct VMStateDescription vmstate_ppc_timebase;

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson