On 2022/5/12 15:28, David Hildenbrand wrote: >>>>> >>>>> Once the problematic DIMM would actually get unplugged, the memory block devices >>>>> would get removed as well. So when hotplugging a new DIMM in the same >>>>> location, we could online that memory again. >>>> >>>> What about PG_hwpoison flags? struct pages are also freed and reallocated >>>> in the actual DIMM replacement? >>> >>> Once memory is offline, the memmap is stale and is no longer >>> trustworthy. It gets reinitialize during memory onlining -- so any >>> previous PG_hwpoison is overridden at least there. In some setups, we >>> even poison the whole memmap via page_init_poison() during memory offlining. >>> >>> Apart from that, we should be freeing the memmap in all relevant cases >>> when removing memory. I remember there are a couple of corner cases, but >>> we don't really have to care about that. >> >> OK, so there seems no need to manipulate struct pages for hwpoison in >> all relevant cases. > > Right. When offlining a memory block, all we have to do is remember if > we stumbled over a hwpoisoned page and rememebr that inside the memory > block. Rejecting to online is then easy. BTW: How should we deal with the below race window: CPU A CPU B CPU C accessing page while hold page refcnt memory_failure happened on page offline_pages page can be offlined due to page refcnt is ignored when PG_hwpoison is set can still access page struct... Any in use page (with page refcnt incremented) might be offlined while its content, e.g. flags, private ..., can still be accessed if the above race happened. Is this possible? Or am I miss something? Any suggestion to fix it? I can't figure out a way yet. :( Thanks a lot! >
