Re: [Qemu-devel] [PATCH -V3 1/4] target-ppc: Update slb array with correct index values.

[Alexander Graf]

On 26.08.2013, at 08:46, Aneesh Kumar K.V wrote:

> Alexander Graf <address@hidden> writes:
> 
>> On 23.08.2013, at 06:20, Aneesh Kumar K.V wrote:
>> 
>>> From: "Aneesh Kumar K.V" <address@hidden>
>>> 
>>> Without this, a value of rb=0 and rs=0 results in replacing the 0th
>>> index. This can be observed when using gdb remote debugging support.
>>> 
>>> (gdb) x/10i do_fork
>>> 0xc000000000085330 <do_fork>:        Cannot access memory at address 
>>> 0xc000000000085330
>>> (gdb)
>>> 
>>> This is because when we do the slb sync via kvm_cpu_synchronize_state,
>>> we overwrite the slb entry (0th entry) for 0xc000000000085330
>>> 
>>> Signed-off-by: Aneesh Kumar K.V <address@hidden>
>>> ---
>>> target-ppc/kvm.c | 20 ++++++++++++++++++--
>>> 1 file changed, 18 insertions(+), 2 deletions(-)
>>> 
>>> diff --git a/target-ppc/kvm.c b/target-ppc/kvm.c
>>> index 30a870e..6878af2 100644
>>> --- a/target-ppc/kvm.c
>>> +++ b/target-ppc/kvm.c
>>> @@ -1033,9 +1033,25 @@ int kvm_arch_get_registers(CPUState *cs)
>>> 
>>>    /* Sync SLB */
>>> #ifdef TARGET_PPC64
>>> +        /*
>>> +         * KVM_GET_SREGS doesn't return slb entry with slot information
>>> +         * same as index. The ioctl zero fills the array and update only
>>> +         * upto slb_max entries. We cannot depend on the slot value
>>> +         * in the slbe field for update, because a zero slbe value would
>>> +         * result in us wrongly updating the 0th index. Instead we zero 
>>> fill
>>> +         * the env->slb array first so that we mark all entries invalid and
>>> +         * update with only valid SLB entries.
>> 
>> Still too negative. How about something like this:
>> 
>> /*
>> * The packed SLB array we get from KVM only contains information
>> *  about valid entries. So we flush our internal copy to get rid of stale
>> *  ones, then put all valid SLB entries back in.
>> */
> 
> updated
> 
>> 
>>> +         */
>>> +        memset(env->slb, 0, 64 * sizeof(ppc_slb_t));
>> 
>> Can't we use ARRAY_SIZE here and below?
> 
> I was thinking we want to be explicit there saying we are zeroing out all
> the 64 entries. I could do sizeof(env->slb).

Yup, that works too. In the loop below s/64/ARRAY_SIZE(env->slb)/ should work 
too I think.


Alex