Re: [PATCH v3 2/2] target/m68k: consolidate physical translation offset into get_physical_address()

[Mark Cave-Ayland]

On 30/06/2020 22:20, Laurent Vivier wrote:

> Le 30/06/2020 à 13:27, Mark Cave-Ayland a écrit :
>> Since all callers to get_physical_address() now apply the same page offset to
>> the translation result, move the logic into get_physical_address() itself to
>> avoid duplication.
>>
>> Suggested-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
>> ---
>>  target/m68k/helper.c | 18 +++++++-----------
>>  1 file changed, 7 insertions(+), 11 deletions(-)
>>
>> diff --git a/target/m68k/helper.c b/target/m68k/helper.c
>> index 631eab7774..71c2376910 100644
>> --- a/target/m68k/helper.c
>> +++ b/target/m68k/helper.c
>> @@ -643,7 +643,7 @@ static int get_physical_address(CPUM68KState *env, 
>> hwaddr *physical,
>>                  /* Transparent Translation Register bit */
>>                  env->mmu.mmusr = M68K_MMU_T_040 | M68K_MMU_R_040;
>>              }
>> -            *physical = address & TARGET_PAGE_MASK;
>> +            *physical = address;
>>              *page_size = TARGET_PAGE_SIZE;
>>              return 0;
>>          }
>> @@ -771,7 +771,8 @@ static int get_physical_address(CPUM68KState *env, 
>> hwaddr *physical,
>>      }
>>      *page_size = 1 << page_bits;
>>      page_mask = ~(*page_size - 1);
>> -    *physical = next & page_mask;
>> +    address &= TARGET_PAGE_MASK;
> 
> I don't think you need TARGET_PAGE_MASK here:
> - TARGET_PAGE_MASK is 4096
> - page_mask is either 4096 or 8192

Ah yes, of course - that will get handled fine by the statement below.

>> +    *physical = (next & page_mask) + (address & (*page_size - 1));
>>  
>>      if (access_type & ACCESS_PTEST) {
>>          env->mmu.mmusr |= next & M68K_MMU_SR_MASK_040;
>> @@ -826,8 +827,6 @@ hwaddr m68k_cpu_get_phys_page_debug(CPUState *cs, vaddr 
>> addr)
>>          return -1;
>>      }
>>  
>> -    addr &= TARGET_PAGE_MASK;
>> -    phys_addr += addr & (page_size - 1);
>>      return phys_addr;
>>  }
>>  
>> @@ -891,10 +890,8 @@ bool m68k_cpu_tlb_fill(CPUState *cs, vaddr address, int 
>> size,
>>      ret = get_physical_address(&cpu->env, &physical, &prot,
>>                                 address, access_type, &page_size);
>>      if (likely(ret == 0)) {
>> -        address &= TARGET_PAGE_MASK;
>> -        physical += address & (page_size - 1);
>> -        tlb_set_page(cs, address, physical,
>> -                     prot, mmu_idx, TARGET_PAGE_SIZE);
>> +        tlb_set_page(cs, address & TARGET_PAGE_MASK,
>> +                     physical & TARGET_PAGE_MASK, prot, mmu_idx, page_size);
> 
> I had a look to tl_set_page() to see how it manages the entry when the
> addresses are not aligned to page_size, and it calls
> tlb_set_page_with_attrs() where we have a comment:
> 
> /* Add a new TLB entry. At most one entry for a given virtual address
>  * is permitted. Only a single TARGET_PAGE_SIZE region is mapped, the
>  * supplied size is only used by tlb_flush_page.
> ...
> 
> So I think it's correct to use TARGET_PAGE_MASK and page_size.

Indeed, it certainly agrees with the documentation and what the majority of the
callers are doing - I'm just surprised that tlb_set_page_with_attrs() doesn't
assert() if any bits below TARGET_PAGE_MASK are set, rather than masking both 
vaddr
and paddr itself. But I'm happy with this part anyhow.


ATB,

Mark.