Re: [Qemu-devel] [PATCH RFC 4/4] target-ppc: flush tlb from all the cpu

[Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>]

Benjamin Herrenschmidt <benh@kernel.crashing.org> writes:

> On Fri, 2016-09-02 at 12:02 +0530, Nikunj A Dadhania wrote:
>> Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
>> ---
>>  cputlb.c                | 15 +++++++++++++++
>>  include/exec/exec-all.h |  2 ++
>>  target-ppc/mmu-hash64.c |  2 +-
>>  3 files changed, 18 insertions(+), 1 deletion(-)
>> 
>> diff --git a/cputlb.c b/cputlb.c
>> index 64faf47..17ff58e 100644
>> --- a/cputlb.c
>> +++ b/cputlb.c
>> @@ -123,6 +123,21 @@ void tlb_flush(CPUState *cpu, int flush_global)
>>      }
>>  }
>>  
>> +static void tlb_flush_all_async_work(CPUState *cpu, void *opaque)
>> +{
>> +    tlb_flush_nocheck(cpu, GPOINTER_TO_INT(opaque));
>> +}
>> +
>> +void tlb_flush_all(CPUState *cpu, int flush_global)
>> +{
>> +    CPUState *c;
>> +
>> +    CPU_FOREACH(c) {
>> +        async_run_on_cpu(c, tlb_flush_all_async_work,
>> +                         GUINT_TO_POINTER(flush_global));
>> +    }
>> +}
>
> Hrm... this is asynchronous ?

Yes.

> It probably needs to be synchronous...

I see run_on_cpu() which seems suitable.

> We must provide a guarantee that no other processor can see the old
> translation when the tlb invalidation sequence completes. With the
> current lazy TLB flush, we already delay the invalidation until
> we hit that synchronization point so we need to be synchronous.


>> diff --git a/target-ppc/mmu-hash64.c b/target-ppc/mmu-hash64.c
>> index 8118143..d852c21 100644
>> --- a/target-ppc/mmu-hash64.c
>> +++ b/target-ppc/mmu-hash64.c
>> @@ -912,7 +912,7 @@ void ppc_hash64_tlb_flush_hpte(PowerPCCPU *cpu,
>>       * invalidate, and we still don't have a tlb_flush_mask(env, n,
>>       * mask) in QEMU, we just invalidate all TLBs
>>       */
>> -    tlb_flush(CPU(cpu), 1);
>> +    tlb_flush_all(CPU(cpu), 1);
>>  }
>>  
>>  void ppc_hash64_update_rmls(CPUPPCState *env)

Regards,
Nikunj