On Thu, 2025-01-16 at 16:45 +0800, Gao Xiang wrote: > > On 2025/1/16 16:24, Chen Linxuan wrote: > > On Thu, 2025-01-16 at 15:51 +0800, Gao Xiang wrote: > > > Hi Linxuan, > > > > > > On 2025/1/16 15:20, Chen Linxuan wrote: > > > > Comments in file include/linux/shrinker.h says that > > > > `count_objects` of `struct shrinker` should return SHRINK_EMPTY > > > > when there are no objects to free. > > > > > > > > > If there are no objects to free, it should return SHRINK_EMPTY, > > > > > while 0 is returned in cases of the number of freeable items cannot > > > > > be determined or shrinker should skip this cache for this time > > > > > (e.g., their number is below shrinkable limit). > > > > > > Thanks for the patch! > > > > > > Yeah, it seems that is the case. Yet it'd better to document > > > what the impact if 0 is returned here if you know more.. > > > > Sorry, I have no idea about that. > > I guess it has no difference if the shrinker is not memcg-aware, > see: > https://lore.kernel.org/r/153063070859.1818.11870882950920963480.stgit@localhost.localdomain > > But I'm fine to use SHRINK_EMPTY since it's clearly documented. > > So could you resend a patch to address my suggestion?
v2 patch has been sent. > > Thanks, > Gao Xiang > >
