Re: [Qemu-devel] [PATCH v5 09/15] cputlb: Access MemoryRegion with MemOp

[Richard Henderson]

On 7/25/19 11:46 PM, address@hidden wrote:
> No-op MEMOP_SIZE and SIZE_MEMOP macros allows us to later easily
> convert memory_region_dispatch_{read|write} paramter "unsigned size"
> into a size+sign+endianness encoded "MemOp op".
> 
> Being a no-op macro, this patch does not introduce any logical change.
> 
> Signed-off-by: Tony Nguyen <address@hidden>
> ---
>  accel/tcg/cputlb.c | 21 ++++++++++-----------
>  1 file changed, 10 insertions(+), 11 deletions(-)
> 
> diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
> index 523be4c..5d88cec 100644
> --- a/accel/tcg/cputlb.c
> +++ b/accel/tcg/cputlb.c
> @@ -881,7 +881,7 @@ static void tlb_fill(CPUState *cpu, target_ulong addr, 
> int size,
> 
>  static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
>                           int mmu_idx, target_ulong addr, uintptr_t retaddr,
> -                         MMUAccessType access_type, int size)
> +                         MMUAccessType access_type, MemOp op)

As I mentioned for patch 2, don't change this now, wait until after patch 10.

> -    r = memory_region_dispatch_read(mr, mr_offset,
> -                                    &val, size, iotlbentry->attrs);
> +    r = memory_region_dispatch_read(mr, mr_offset, &val, op, 
> iotlbentry->attrs);

So size_memop here,

> -        cpu_transaction_failed(cpu, physaddr, addr, size, access_type,
> +        cpu_transaction_failed(cpu, physaddr, addr, MEMOP_SIZE(op), 
> access_type,
>                                 mmu_idx, iotlbentry->attrs, r, retaddr);

but no memop_size here.

>  static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
>                        int mmu_idx, uint64_t val, target_ulong addr,
> -                      uintptr_t retaddr, int size)
> +                      uintptr_t retaddr, MemOp op)

Likewise.

>          res = io_readx(env, &env_tlb(env)->d[mmu_idx].iotlb[index],
> -                       mmu_idx, addr, retaddr, access_type, size);
> +                       mmu_idx, addr, retaddr, access_type, 
> SIZE_MEMOP(size));

And when you do come back to change the types after patch 10, at the top of the
function:

-    unsigned a_bits = get_alignment_bits(get_memop(oi));
+    MemOp op = get_memop(oi);
+    unsigned a_bits = get_alignment_bits(op);

and then pass along op directly.  Which will fix some of the weirdness in patch 
11.


r~