Re: [PATCH v4 6/7] xen/riscv: page table handling

[Jan Beulich <jbeulich@xxxxxxxx>]

On 20.08.2024 16:42, oleksii.kurochko@xxxxxxxxx wrote:
> On Tue, 2024-08-20 at 15:47 +0200, Jan Beulich wrote:
>> On 20.08.2024 15:18, oleksii.kurochko@xxxxxxxxx wrote:
>>> On Tue, 2024-08-13 at 12:31 +0200, Jan Beulich wrote:
>>>> From all I can determine we also get here when making brand new
>>>> entries.
>>>> What am I overlooking?
>>> Yes, but in this case an intermediate page tables should be
>>> read_only,
>>> so alloc_only will be true and it will be allowed to create new
>>> intermediate page table.
>>
>> Hmm, so instead of "read-only" do you maybe mean page tables are not
>> supposed to be modified? There's a difference here: When they're
>> read-only, you can't write to them (or a fault will result). Whereas
>> when in principle they can be modified, there still may be a rule
>> saying "in this case they shouldn't be altered".
> 
> There is such rule which checks that page tables aren't supposed to be
> modified ( so that is why they are read-only ):

Hmm, you're saying "read-only" again in reply to me explaining that this
isn't the correct term here. I find this increasingly confusing.

Jan

> ```
>     /* Sanity check when modifying an entry. */
>     if ( (flags & PTE_VALID) && mfn_eq(mfn, INVALID_MFN) )
>     {
>       ...
> 
>         /* We don't allow modifying a table entry */
>         if ( pte_is_table(entry) )
>         {
>             printk("Modifying a table entry is not allowed.\n");
>             return false;
>         }
> ```
> 
> ~ Oleksii