On Tue, Apr 26, 2022 at 12:18:39PM +0530, Anshuman Khandual wrote: > > > On 4/25/22 19:46, HORIGUCHI NAOYA(堀口 直也) wrote: > > On Sat, Apr 23, 2022 at 11:16:20AM +0800, Yu Xu wrote: > >> For this patch v2, > >> > >> Yang, David, Anshuman and Naoya all agree to fix the bug in > >> split_huge_page_to_list(), i.e., to replace the BUG to > >> WARN + returning -EBUSY for huge_zero_page. > >> > >> However, this patch v2 has just been merged into linux mainline. > >> Sorry for that I didn't hurry up. > >> > >> So now, should we continue sending v3, or should we end this > >> thread, and start a successive patch for optimization? > > > > No problem, you can start a new thread for additional patches. > > Maybe it's better to have separate patches (starting with one > > reverting v2 and then one doing v3). > > May be also pull in the following patch which adds MF_MSG_HUGE_ZERO > in the same series ? > > https://lore.kernel.org/linux-mm/20220425080306.1771480-1-anshuman.khandual@xxxxxxx/ I don't think so. I think that Yu's v3 would solely solve the reported problem, so any check for huge_zero_page in memory_failire() should not be needed. (But correct me if I miss something ...) Thanks, Naoya Horiguchi
