diff mbox

target/ppc: Fix backwards migration of msr_mask

Message ID 20180320022319.17199-1-david@gibson.dropbear.id.au (mailing list archive)
State New, archived
Headers show

Commit Message

David Gibson March 20, 2018, 2:23 a.m. UTC
21b786f "PowerPC: Add TS bits into msr_mask" added the transaction states
to msr_mask for recent POWER CPUs to allow correct migration of machines
that are in certain interim transactional memory states.

This was correct, but unfortunately breaks backwards of pseries-2.7 and
earlier machine types which (stupidly) transferred the msr_mask in the
migration stream and failed if it wasn't equal on each end.

This works around the problem by masking out the new MSR bits in the
compatibility code to send the msr_mask on old machine types.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
---
 target/ppc/machine.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

Comments

Laurent Vivier March 20, 2018, 8:08 a.m. UTC | #1
On 20/03/2018 03:23, David Gibson wrote:
> 21b786f "PowerPC: Add TS bits into msr_mask" added the transaction states
> to msr_mask for recent POWER CPUs to allow correct migration of machines
> that are in certain interim transactional memory states.
> 
> This was correct, but unfortunately breaks backwards of pseries-2.7 and
> earlier machine types which (stupidly) transferred the msr_mask in the
> migration stream and failed if it wasn't equal on each end.
> 
> This works around the problem by masking out the new MSR bits in the
> compatibility code to send the msr_mask on old machine types.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> ---
>  target/ppc/machine.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)

Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Greg Kurz March 20, 2018, 9:41 a.m. UTC | #2
On Tue, 20 Mar 2018 13:23:19 +1100
David Gibson <david@gibson.dropbear.id.au> wrote:

> 21b786f "PowerPC: Add TS bits into msr_mask" added the transaction states
> to msr_mask for recent POWER CPUs to allow correct migration of machines
> that are in certain interim transactional memory states.
> 
> This was correct, but unfortunately breaks backwards of pseries-2.7 and
> earlier machine types which (stupidly) transferred the msr_mask in the
> migration stream and failed if it wasn't equal on each end.
> 
> This works around the problem by masking out the new MSR bits in the
> compatibility code to send the msr_mask on old machine types.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> ---

Reviewed-by: Greg Kurz <groug@kaod.org>

and

Tested-by: Greg Kurz <groug@kaod.org>

>  target/ppc/machine.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/target/ppc/machine.c b/target/ppc/machine.c
> index e475206c6a..0634cdb295 100644
> --- a/target/ppc/machine.c
> +++ b/target/ppc/machine.c
> @@ -190,7 +190,15 @@ static int cpu_pre_save(void *opaque)
>  
>      /* Hacks for migration compatibility between 2.6, 2.7 & 2.8 */
>      if (cpu->pre_2_8_migration) {
> -        cpu->mig_msr_mask = env->msr_mask;
> +        /* Mask out bits that got added to msr_mask since the versions
> +         * which stupidly included it in the migration stream. */
> +        target_ulong metamask = 0
> +#if defined(TARGET_PPC64)
> +            | (1ULL << MSR_TS0)
> +            | (1ULL << MSR_TS1)
> +#endif
> +            ;
> +        cpu->mig_msr_mask = env->msr_mask & ~metamask;
>          cpu->mig_insns_flags = env->insns_flags & insns_compat_mask;
>          cpu->mig_insns_flags2 = env->insns_flags2 & insns_compat_mask2;
>          cpu->mig_nb_BATs = env->nb_BATs;
Lukáš Doktor March 20, 2018, 12:17 p.m. UTC | #3
Dne 20.3.2018 v 03:23 David Gibson napsal(a):
> 21b786f "PowerPC: Add TS bits into msr_mask" added the transaction states
> to msr_mask for recent POWER CPUs to allow correct migration of machines
> that are in certain interim transactional memory states.
> 
> This was correct, but unfortunately breaks backwards of pseries-2.7 and
> earlier machine types which (stupidly) transferred the msr_mask in the
> migration stream and failed if it wasn't equal on each end.
> 
> This works around the problem by masking out the new MSR bits in the
> compatibility code to send the msr_mask on old machine types.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> ---

Thanks, migration to 2.6.2 works again.

Tested-by: Lukáš Doktor <ldoktor@redhat.com> 

>  target/ppc/machine.c | 10 +++++++++-
>  1 file changed, 9 insertions(+), 1 deletion(-)
> 
> diff --git a/target/ppc/machine.c b/target/ppc/machine.c
> index e475206c6a..0634cdb295 100644
> --- a/target/ppc/machine.c
> +++ b/target/ppc/machine.c
> @@ -190,7 +190,15 @@ static int cpu_pre_save(void *opaque)
>  
>      /* Hacks for migration compatibility between 2.6, 2.7 & 2.8 */
>      if (cpu->pre_2_8_migration) {
> -        cpu->mig_msr_mask = env->msr_mask;
> +        /* Mask out bits that got added to msr_mask since the versions
> +         * which stupidly included it in the migration stream. */
> +        target_ulong metamask = 0
> +#if defined(TARGET_PPC64)
> +            | (1ULL << MSR_TS0)
> +            | (1ULL << MSR_TS1)
> +#endif
> +            ;
> +        cpu->mig_msr_mask = env->msr_mask & ~metamask;
>          cpu->mig_insns_flags = env->insns_flags & insns_compat_mask;
>          cpu->mig_insns_flags2 = env->insns_flags2 & insns_compat_mask2;
>          cpu->mig_nb_BATs = env->nb_BATs;
>
simon April 4, 2018, 3:27 a.m. UTC | #4
David,
On Tue, Mar 20, 2018 at 01:23:19PM +1100, David Gibson wrote:
> 21b786f "PowerPC: Add TS bits into msr_mask" added the transaction states
> to msr_mask for recent POWER CPUs to allow correct migration of machines
> that are in certain interim transactional memory states.
> 
> This was correct, but unfortunately breaks backwards of pseries-2.7 and
> earlier machine types which (stupidly) transferred the msr_mask in the
> migration stream and failed if it wasn't equal on each end.
> 
> This works around the problem by masking out the new MSR bits in the
> compatibility code to send the msr_mask on old machine types.
> 
> Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Sorry I didn't consider the backward compatibility in previous
commit. Thanks for fix that.

Thanks,
- Simon
diff mbox

Patch

diff --git a/target/ppc/machine.c b/target/ppc/machine.c
index e475206c6a..0634cdb295 100644
--- a/target/ppc/machine.c
+++ b/target/ppc/machine.c
@@ -190,7 +190,15 @@  static int cpu_pre_save(void *opaque)
 
     /* Hacks for migration compatibility between 2.6, 2.7 & 2.8 */
     if (cpu->pre_2_8_migration) {
-        cpu->mig_msr_mask = env->msr_mask;
+        /* Mask out bits that got added to msr_mask since the versions
+         * which stupidly included it in the migration stream. */
+        target_ulong metamask = 0
+#if defined(TARGET_PPC64)
+            | (1ULL << MSR_TS0)
+            | (1ULL << MSR_TS1)
+#endif
+            ;
+        cpu->mig_msr_mask = env->msr_mask & ~metamask;
         cpu->mig_insns_flags = env->insns_flags & insns_compat_mask;
         cpu->mig_insns_flags2 = env->insns_flags2 & insns_compat_mask2;
         cpu->mig_nb_BATs = env->nb_BATs;