On 12/17/22 11:58, Yafang Shao wrote: > On 64bit system, page extension is for debugging purpose only currently, > because of its overhead, in particular the memory overhead. > > Once a page_ext is enabled, at least it will take 0.2% of the total > memory because of the page_ext flags, no matter this page_ext uses it or > not. Currently this page_ext flags is only used for page_owner on 64bit > system. So it doesn't make sense to allocate this flags for all page_ext > by default. We'd better move it into page_owner's structure, then when > someone wants to introduce a new page_ext which may be memory-overhead > sensitive, it will save this unneeded overhead. > > On 32bit system, there's page_idle running on production envrionment, > which also uses this page_ext flags. So it will take another 0.2% of > total memory if the user enable both page_idle and page_owner after this > change, but considering page_owner is for debugging purpose only, the > memory overhead in this case won't be a problem. > > So, let split the page_ext flags. Hi, FYI I think Pasha's solution should work to avoid the waste even without the split: https://lore.kernel.org/all/20230113154253.92480-1-pasha.tatashin@xxxxxxxxxx/ > Yafang Shao (2): > mm: page_owner: split page_owner's flag from the comm flags > mm: page_idle: split 32bit page_idle's flag from the common flags > > include/linux/page_ext.h | 14 +------------- > include/linux/page_idle.h | 39 +++++++++++++++++++++++++++++++++------ > mm/page_ext.c | 10 ---------- > mm/page_idle.c | 12 ++++++++++++ > mm/page_owner.c | 36 ++++++++++++++++++++++-------------- > 5 files changed, 68 insertions(+), 43 deletions(-) >
