Re: [for 4.22 v5 17/18] xen/riscv: add support of page lookup by GFN

[Jan Beulich <jbeulich@xxxxxxxx>]

On 17.11.2025 16:52, Oleksii Kurochko wrote:
> On 11/10/25 5:46 PM, Jan Beulich wrote:
>> On 20.10.2025 17:58, Oleksii Kurochko wrote:
>>> +static bool check_outside_boundary(gfn_t gfn, gfn_t boundary, bool 
>>> is_lower,
>>> +                                   unsigned int *level_out)
>>> +{
>>> +    unsigned int level;
>>> +
>>> +    if ( is_lower ? gfn_x(gfn) < gfn_x(boundary)
>>> +                  : gfn_x(gfn) > gfn_x(boundary) )
>>> +    {
>>> +        unsigned long mask = 0;
>>> +
>>> +        for ( level = P2M_ROOT_LEVEL; level; level-- )
>>> +        {
>>> +            unsigned long masked_gfn;
>>> +
>>> +            mask |= PFN_DOWN(P2M_LEVEL_MASK(level));
>>> +            masked_gfn = gfn_x(gfn) & mask;
>>> +
>>> +            if ( is_lower ? masked_gfn < gfn_x(boundary)
>>> +                          : masked_gfn > gfn_x(boundary) )
>>> +            {
>>> +                *level_out = level;
>> For this to be correct in the is_lower case, don't you need to fill the
>> bottom bits of masked_gfn with all 1s, rather than with all 0s? Otherwise
>> the tail of the range may be above boundary.
> 
> I think that I didn't get what you mean by "the range" here and so I can't 
> understand
> what is "the tail of the range".
> Could you please clarify?

By applying "mask" you effectively produce a range (with "gfn" somewhere in
the middle). For the level (which you return to the caller) to be correct,
the entire range must be matching "gfn" in being below or above of the
boundary. My impression is that this isn't the case when is_lower is true.

Jan