Re: [PATCH-for-9.1 v2 1/2] target/mips: Pass page table entry size in bytes to get_pte()

[Richard Henderson]

On 8/13/24 20:18, Philippe Mathieu-Daudé wrote:
> In order to simplify a bit, pass the PTE size in
> bytes rather than bits.
>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
> ---
>   target/mips/tcg/sysemu/tlb_helper.c | 16 ++++++++--------
>   1 file changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/target/mips/tcg/sysemu/tlb_helper.c 
> b/target/mips/tcg/sysemu/tlb_helper.c
> index 3ba6d369a6..a8caf3ade8 100644
> --- a/target/mips/tcg/sysemu/tlb_helper.c
> +++ b/target/mips/tcg/sysemu/tlb_helper.c
> @@ -592,13 +592,13 @@ static void raise_mmu_exception(CPUMIPSState *env, 
> target_ulong address,
>    * resulting in a TLB or XTLB Refill exception.
>    */
>   
> -static bool get_pte(CPUMIPSState *env, uint64_t vaddr, int entry_size,
> +static bool get_pte(CPUMIPSState *env, uint64_t vaddr, unsigned entry_bytes,
>           uint64_t *pte)
>   {
> -    if ((vaddr & ((entry_size >> 3) - 1)) != 0) {
> +    if ((vaddr & (entry_bytes - 1)) != 0) {
>           return false;
>       }
> -    if (entry_size == 64) {
> +    if (entry_bytes == 8) {
>           *pte = cpu_ldq_code(env, vaddr);
>       } else {
>           *pte = cpu_ldl_code(env, vaddr);

Considering the next patch, where you need to make the MemOpIdx,
why not pass in the size as MemOp?

> @@ -630,8 +630,8 @@ static int walk_directory(CPUMIPSState *env, uint64_t 
> *vaddr,
>       int psn = (env->CP0_PWCtl >> CP0PC_PSN) & 0x3F;
>       int hugepg = (env->CP0_PWCtl >> CP0PC_HUGEPG) & 0x1;
>       int pf_ptew = (env->CP0_PWField >> CP0PF_PTEW) & 0x3F;
> -    uint32_t direntry_size = 1 << (directory_shift + 3);
> -    uint32_t leafentry_size = 1 << (leaf_shift + 3);
> +    uint32_t direntry_size = 1 << directory_shift;
> +    uint32_t leafentry_size = 1 << leaf_shift;

This would then be just directory_shift/leaf_shift unchanged.


r~