Re: [PATCH 1/3] target/arm: keep translation start level unsigned

[Richard Henderson]

On 12/30/20 2:10 PM, Richard Henderson wrote:
> On 12/18/20 6:33 AM, remi.denis.courmont@huawei.com wrote:
>> From: Rémi Denis-Courmont <remi.denis.courmont@huawei.com>
>>
>> Signed-off-by: Rémi Denis-Courmont <remi.denis.courmont@huawei.com>
>> ---
>>  target/arm/helper.c | 14 ++++++--------
>>  1 file changed, 6 insertions(+), 8 deletions(-)
> 
> The patch does more than what is described above.
> 
>>
>> diff --git a/target/arm/helper.c b/target/arm/helper.c
>> index df195c314c..b927e53ab0 100644
>> --- a/target/arm/helper.c
>> +++ b/target/arm/helper.c
>> @@ -10821,17 +10821,12 @@ do_fault:
>>   * Returns true if the suggested S2 translation parameters are OK and
>>   * false otherwise.
>>   */
>> -static bool check_s2_mmu_setup(ARMCPU *cpu, bool is_aa64, int level,
>> +static bool check_s2_mmu_setup(ARMCPU *cpu, bool is_aa64, uint32_t level,
>>                                 int inputsize, int stride)
>>  {
>>      const int grainsize = stride + 3;
>>      int startsizecheck;
>>  
>> -    /* Negative levels are never allowed.  */
>> -    if (level < 0) {
>> -        return false;
>> -    }
>> -
> 
> I would expect this to be the only hunk from the patch description.  Probably
> changing this negative check to a >= 3 check.

Having read the next patch, I think you should drop this type change.

>> @@ -11203,7 +11201,7 @@ static bool get_phys_addr_lpae(CPUARMState *env, 
>> uint64_t address,
>>  
>>          if (!aarch64 || stride == 9) {
>>              /* AArch32 or 4KB pages */
>> -            startlevel = 2 - sl0;
>> +            startlevel = (2 - sl0) & 3;

This hunk belongs with the next patch, implementing TTST, and should be
conditional.  I.e.

    if (stride == 9) {
        startlevel = 2 - sl0;
        if (aarch64 &&
            cpu_isar_feature(aa64_st, env_archcpu(env)) {
            startlevel &= 3;
        }
    ...


r~