[PATCH v2 2/2] drm/ttm: Increase pool shrinker batch target

[Tvrtko Ursulin]

On 03/06/2025 13:57, Christian König wrote:
> On 6/3/25 13:27, Tvrtko Ursulin wrote:
>> The default core shrink target of 128 pages (SHRINK_BATCH) is quite low
>> relative to how cheap TTM pool shrinking is, and how the free pages are
>> distributed in page order pools.
>>
>> We can make the target a bit more aggressive by making it roughly the
>> average number of pages across all pools, freeing more of the cached
>> pages every time shrinker core invokes our callback.
>>
>> Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin at igalia.com>
>> Cc: Christian König <christian.koenig at amd.com>
>> Cc: Thomas Hellström <thomas.hellstrom at linux.intel.com>
>> ---
>>   drivers/gpu/drm/ttm/ttm_pool.c | 9 ++++++++-
>>   1 file changed, 8 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/gpu/drm/ttm/ttm_pool.c b/drivers/gpu/drm/ttm/ttm_pool.c
>> index c060c90b89c0..e671812789ea 100644
>> --- a/drivers/gpu/drm/ttm/ttm_pool.c
>> +++ b/drivers/gpu/drm/ttm/ttm_pool.c
>> @@ -1265,10 +1265,16 @@ int ttm_pool_debugfs(struct ttm_pool *pool, struct seq_file *m)
>>   }
>>   EXPORT_SYMBOL(ttm_pool_debugfs);
>>   
>> +/* Free average pool number of pages.  */
>> +#define TTM_SHRINKER_BATCH ((1 << (MAX_PAGE_ORDER / 2)) * NR_PAGE_ORDERS)
> 
> To be honest this feels random. But I can't come up with a better idea either and it still looks better than the default as far as I can see.
> 
> So feel free to add Reviewed-by: Christian König <christian.koenig at amd.com> to the series.

Thanks!

> Should I push it to drm-misc-next or do you now have commit rights?

I should be able to, think I've tested it already once.

Regards,

Tvrtko

>> +
>>   /* Test the shrinker functions and dump the result */
>>   static int ttm_pool_debugfs_shrink_show(struct seq_file *m, void *data)
>>   {
>> -	struct shrink_control sc = { .gfp_mask = GFP_NOFS };
>> +	struct shrink_control sc = {
>> +		.gfp_mask = GFP_NOFS,
>> +		.nr_to_scan = TTM_SHRINKER_BATCH,
>> +	};
>>   
>>   	fs_reclaim_acquire(GFP_KERNEL);
>>   	seq_printf(m, "%lu/%lu\n", ttm_pool_shrinker_count(mm_shrinker, &sc),
>> @@ -1326,6 +1332,7 @@ int ttm_pool_mgr_init(unsigned long num_pages)
>>   
>>   	mm_shrinker->count_objects = ttm_pool_shrinker_count;
>>   	mm_shrinker->scan_objects = ttm_pool_shrinker_scan;
>> +	mm_shrinker->batch = TTM_SHRINKER_BATCH;
>>   	mm_shrinker->seeks = 1;
>>   
>>   	shrinker_register(mm_shrinker);
>