Re: [Xen-devel] [PATCH] libxc: really tolerate empty PV records

2018-03-01 Thread Wei Liu
On Tue, Feb 27, 2018 at 07:51:22AM -0700, Jan Beulich wrote:
> Commit 119ee4d773 ("tools/libxc: Tolerate specific zero-content records
> in migration v2 streams") meant tolerate those, but failed to set rc
> accordingly.
> 
> Signed-off-by: Jan Beulich 

Reviewed-by: Wei Liu 

___
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [Xen-devel] [PATCH] libxc: really tolerate empty PV records

2018-02-27 Thread Andrew Cooper
On 27/02/18 15:00, Andrew Cooper wrote:
> On 27/02/18 14:51, Jan Beulich wrote:
>> ---
>> This is what is currently breaking save/restore on the 4.8 branch; I
>> haven't checked yet why the sending side produces an empty record there.
>> I wonder whether that's related to f1a0a8c3fe ("tools/libxc: Fix
>> restoration of PV MSRs after migrate") not having been backported yet.
>> Otoh the latter is missing on 4.7 too, yet there save/restore doesn't
>> fail.
> It will be 72efb1df6 "Avoid generating inappropriate zero-content
> records" which is missing.
>
> It triggers when there are a non-zero maximum number of MSRs, but zero
> MSRs to tend after Xen skips the empty MSRs.

send*

Sorry for the noise.

~Andrew

___
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

Re: [Xen-devel] [PATCH] libxc: really tolerate empty PV records

2018-02-27 Thread Andrew Cooper
On 27/02/18 14:51, Jan Beulich wrote:
> Commit 119ee4d773 ("tools/libxc: Tolerate specific zero-content records
> in migration v2 streams") meant tolerate those, but failed to set rc
> accordingly.
>
> Signed-off-by: Jan Beulich 

Reviewed-by: Andrew Cooper 

> ---
> This is what is currently breaking save/restore on the 4.8 branch; I
> haven't checked yet why the sending side produces an empty record there.
> I wonder whether that's related to f1a0a8c3fe ("tools/libxc: Fix
> restoration of PV MSRs after migrate") not having been backported yet.
> Otoh the latter is missing on 4.7 too, yet there save/restore doesn't
> fail.

It will be 72efb1df6 "Avoid generating inappropriate zero-content
records" which is missing.

It triggers when there are a non-zero maximum number of MSRs, but zero
MSRs to tend after Xen skips the empty MSRs.

~Andrew

>
> --- a/tools/libxc/xc_sr_restore_x86_pv.c
> +++ b/tools/libxc/xc_sr_restore_x86_pv.c
> @@ -770,6 +770,7 @@ static int handle_x86_pv_vcpu_blob(struc
>  {
>  DBGPRINTF("Skipping empty %s record for vcpu %u\n",
>rec_type_to_str(rec->type), vhdr->vcpu_id);
> +rc = 0;
>  goto out;
>  }
>  
>
>
>


___
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel

[Xen-devel] [PATCH] libxc: really tolerate empty PV records

2018-02-27 Thread Jan Beulich
Commit 119ee4d773 ("tools/libxc: Tolerate specific zero-content records
in migration v2 streams") meant tolerate those, but failed to set rc
accordingly.

Signed-off-by: Jan Beulich 
---
This is what is currently breaking save/restore on the 4.8 branch; I
haven't checked yet why the sending side produces an empty record there.
I wonder whether that's related to f1a0a8c3fe ("tools/libxc: Fix
restoration of PV MSRs after migrate") not having been backported yet.
Otoh the latter is missing on 4.7 too, yet there save/restore doesn't
fail.

--- a/tools/libxc/xc_sr_restore_x86_pv.c
+++ b/tools/libxc/xc_sr_restore_x86_pv.c
@@ -770,6 +770,7 @@ static int handle_x86_pv_vcpu_blob(struc
 {
 DBGPRINTF("Skipping empty %s record for vcpu %u\n",
   rec_type_to_str(rec->type), vhdr->vcpu_id);
+rc = 0;
 goto out;
 }
 




___
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel