RE: [Xen-ia64-devel] [PATCH] [RESEND]clean up vti code

["Xu, Anthony" <anthony.xu@xxxxxxxxx>]

Alex,
The serialization is not needed. Because itir is a parameter of itc,
there is implicit serialization before itc.
I removed a stop between itir and itc carelessly.

I will separate this patch, and send out later.

Thanks,
-Anthony 

>-----Original Message-----
>From: Alex Williamson [mailto:alex.williamson@xxxxxx]
>Sent: 2006?5?26? 23:55
>To: Xu, Anthony
>Cc: Zhang, Xiantao; xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
>Subject: RE: [Xen-ia64-devel] [PATCH] [RESEND]clean up vti code
>
>Hi Anthony,
>
>  A few comments.
>
>On Thu, 2006-05-25 at 20:02 +0800, Xu, Anthony wrote:
>> @@ -321,88 +285,82 @@ ENTRY(vmx_alt_itlb_miss)
>> -       ;;
>> -       mov cr.itir=r20
>> -       ;;
>> -       srlz.i
>> -       ;;
>> -       itc.i r19               // insert the TLB entry
>...
>> +    ;;
>> +    mov cr.itir=r20
>> +    itc.i r19          // insert the TLB entry
>...
>> -       ;;
>> -       mov cr.itir=r20
>> -       ;;
>> -       srlz.i
>> -       ;;
>> -(p7) itc.d r19         // insert the TLB entry
>...
>> +    ;;
>> +    mov cr.itir=r20
>> +(p7)itc.d r19          // insert the TLB entry
>
>   The loss of serialization between mv cr.itir and itc has me nervous.
>The assembler is nervous about it too:
>
>Warning: Use of 'itc.i' violates RAW dependency 'CR[ITIR]' (implied)
>Warning: Use of 'itc.d' may violate RAW dependency 'CR[ITIR]' (implied)
>
>Also, this patch has gotten very large, and I'm not sure it only
>includes cleanups anymore.  There seems to be lots of new code mixed in
>as well.  Could you separate simple formatting cleanup from code
>restructuring and modification?  I did verify that this version of the
>patch allows me to to boot domVTI again.  Thanks,
>
>       Alex
>
>--
>Alex Williamson                             HP Linux & Open Source Lab

_______________________________________________
Xen-ia64-devel mailing list
Xen-ia64-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-ia64-devel