Re: [PATCH] xen: arm: arm64: Fix memory cloberring issues during VFP save restore.

[George Dunlap <george.dunlap@eu.citrix.com>]

On 02/07/2014 11:39 AM, Ian Campbell wrote:
> On Fri, 2014-02-07 at 16:08 +0530, Pranavkumar Sawargaonkar wrote:
>> This patch addresses memory cloberring issue mentioed by Julien Grall
>> with my earlier patch -
>> Ref:
>> http://www.gossamer-threads.com/lists/xen/devel/316247
>>
>> Signed-off-by: Pranavkumar Sawargaonkar <pranavkumar@linaro.org>
>> Signed-off-by: Anup Patel <anup.patel@linaro.org>
>> ---
>>   xen/arch/arm/arm64/vfp.c |   70 ++++++++++++++++++++++++----------------------
>>   1 file changed, 36 insertions(+), 34 deletions(-)
>>
>> diff --git a/xen/arch/arm/arm64/vfp.c b/xen/arch/arm/arm64/vfp.c
>> index c09cf0c..62f56a3 100644
>> --- a/xen/arch/arm/arm64/vfp.c
>> +++ b/xen/arch/arm/arm64/vfp.c
>> @@ -8,23 +8,24 @@ void vfp_save_state(struct vcpu *v)
>>       if ( !cpu_has_fp )
>>           return;
>>   
>> -    asm volatile("stp q0, q1, [%0, #16 * 0]\n\t"
>> -                 "stp q2, q3, [%0, #16 * 2]\n\t"
>> -                 "stp q4, q5, [%0, #16 * 4]\n\t"
>> -                 "stp q6, q7, [%0, #16 * 6]\n\t"
>> -                 "stp q8, q9, [%0, #16 * 8]\n\t"
>> -                 "stp q10, q11, [%0, #16 * 10]\n\t"
>> -                 "stp q12, q13, [%0, #16 * 12]\n\t"
>> -                 "stp q14, q15, [%0, #16 * 14]\n\t"
>> -                 "stp q16, q17, [%0, #16 * 16]\n\t"
>> -                 "stp q18, q19, [%0, #16 * 18]\n\t"
>> -                 "stp q20, q21, [%0, #16 * 20]\n\t"
>> -                 "stp q22, q23, [%0, #16 * 22]\n\t"
>> -                 "stp q24, q25, [%0, #16 * 24]\n\t"
>> -                 "stp q26, q27, [%0, #16 * 26]\n\t"
>> -                 "stp q28, q29, [%0, #16 * 28]\n\t"
>> -                 "stp q30, q31, [%0, #16 * 30]\n\t"
>> -                 :: "r" ((char *)(&v->arch.vfp.fpregs)): "memory");
>> +    asm volatile("stp q0, q1, [%1, #16 * 0]\n\t"
>> +                 "stp q2, q3, [%1, #16 * 2]\n\t"
>> +                 "stp q4, q5, [%1, #16 * 4]\n\t"
>> +                 "stp q6, q7, [%1, #16 * 6]\n\t"
>> +                 "stp q8, q9, [%1, #16 * 8]\n\t"
>> +                 "stp q10, q11, [%1, #16 * 10]\n\t"
>> +                 "stp q12, q13, [%1, #16 * 12]\n\t"
>> +                 "stp q14, q15, [%1, #16 * 14]\n\t"
>> +                 "stp q16, q17, [%1, #16 * 16]\n\t"
>> +                 "stp q18, q19, [%1, #16 * 18]\n\t"
>> +                 "stp q20, q21, [%1, #16 * 20]\n\t"
>> +                 "stp q22, q23, [%1, #16 * 22]\n\t"
>> +                 "stp q24, q25, [%1, #16 * 24]\n\t"
>> +                 "stp q26, q27, [%1, #16 * 26]\n\t"
>> +                 "stp q28, q29, [%1, #16 * 28]\n\t"
>> +                 "stp q30, q31, [%1, #16 * 30]\n\t"
>> +                 :"=Q" (*v->arch.vfp.fpregs): "r" (v->arch.vfp.fpregs)
>> +                 : "memory");
> The point of this change was to be able to drop the memory clobbers.
>
> George, I'd like to take this in 4.4 if possible -- I wanted to get the
> baseline functionality fixed for 4.4 ASAP since it was quite a big hole
> which is why I committed without waiting for this respin.
>
> The issue is that the patch which was committed yesterday clobbers all
> of memory and not just the bits the inline asm touches.

Obviously there's not much point in releasing a version with a fix that 
doesn't work. :-)

Release-acked-by: George Dunlap <george.dunlap@eu.citrix.com>