Re: [Qemu-devel] [PATCH 04/10] cputlb: Split large page tracking per mmu_idx

[Richard Henderson]

On 10/27/18 1:16 AM, Emilio G. Cota wrote:
> On Tue, Oct 23, 2018 at 08:02:47 +0100, Richard Henderson wrote:
>> +static void tlb_flush_page_locked(CPUArchState *env, int midx,
>> +                                  target_ulong addr)
>> +{
>> +    target_ulong lp_addr = env->tlb_d[midx].large_page_addr;
>> +    target_ulong lp_mask = env->tlb_d[midx].large_page_mask;
>> +
>> +    /* Check if we need to flush due to large pages.  */
>> +    if ((addr & lp_mask) == lp_addr) {
>> +        tlb_debug("forcing full flush midx %d ("
>> +                  TARGET_FMT_lx "/" TARGET_FMT_lx ")\n",
>> +                  midx, lp_addr, lp_mask);
>> +        tlb_flush_one_mmuidx_locked(env, midx);
>> +    } else {
>> +        int pidx = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
>> +        tlb_flush_entry_locked(&env->tlb_table[midx][pidx], addr);
>> +        tlb_flush_vtlb_page_locked(env, midx, addr);
> 
> Just noticed that we should use tlb_entry here, e.g.:
> 
>      } else {
> -        int pidx = (addr >> TARGET_PAGE_BITS) & (CPU_TLB_SIZE - 1);
> -        tlb_flush_entry_locked(&env->tlb_table[midx][pidx], addr);
> +        CPUTLBEntry *entry = tlb_entry(env, midx, addr);
> +
> +        tlb_flush_entry_locked(entry, addr);
>          tlb_flush_vtlb_page_locked(env, midx, addr);
>      }

Fixed, thanks.

r~