Re: [PATCH RESEND 6/9] selftests/mm: mremap_test: replace random data with deterministic pattern

[David Hildenbrand (Arm)]

On 10/1/26 15:16, Sarthak Sharma wrote:
> 
> 
> On 10/1/26 5:30 PM, David Hildenbrand (Arm) wrote:
>> On 9/24/26 07:00, Sarthak Sharma wrote:
>>> mremap_test uses a random data stream to detect corruption after remap
>>> operations. This requires seed handling and byte by byte validation,
>>> which is inefficient.
>>>
>>> Replace it with a deterministic pattern where every word in a page
>>> contains its one-based page index. Use memcpy() and memcmp() to
>>> initialize and validate the mappings. Remove the pattern seed and
>>> its command line option. Also update the comment diagrams to
>>> reflect the new deterministic pattern.
>>>
>>> Suggested-by: David Hildenbrand (Arm) <david@xxxxxxxxxx>
>>> Signed-off-by: Sarthak Sharma <sarthak.sharma@xxxxxxx>
>>> ---
>>
>> How will this patch change with the change in threshold handling? Does it make
>> sense to reshuffle the patches?
> 
> From an intermediate patch POV, both ways of doing should be identical:
> 
> a) Remove threshold first, then remove randomization:
> 	i) Adjust random buffer size according to what the test requires
> 	   and implement the start, mid end pages checking
> 	ii) Replace rand approach with fixed pattern
> 
> b) Remove randomization first, then remove threshold:
> 	i) Adjust fixed pattern size according to threshold
> 	ii) Implement the new logic of start, mid and end pages checking
> 	    and adjust pattern size
> 
> But if we plan to remove perf and timing infrastructure before (which I
> plan to do in v2), yes removing threshold first would be neater, since
> it will keep both the removal patches together. Please let me know if I
> am missing something.

No, looks good. I'll take a look at the patches again in a v2.

-- 
Cheers,

David