On Sun, Jul 18, 2021 at 12:30:31PM +0800, Qi Zheng wrote: > Some malloc libraries(e.g. jemalloc or tcmalloc) usually > allocate the amount of VAs by mmap() and do not unmap > those VAs. They will use madvise(MADV_DONTNEED) to free > physical memory if they want. But the page tables do not > be freed by madvise(), so it can produce many page tables > when the process touches an enormous virtual address space. > > The following figures are a memory usage snapshot of one > process which actually happened on our server: > > VIRT: 55t > RES: 590g > VmPTE: 110g > > As we can see, the PTE page tables size is 110g, while the > RES is 590g. In theory, the process only need 1.2g PTE page > tables to map those physical memory. The reason why PTE page > tables occupy a lot of memory is that madvise(MADV_DONTNEED) > only empty the PTE and free physical memory but doesn't free > the PTE page table pages. So we can free those empty PTE page > tables to save memory. In the above cases, we can save memory > about 108g(best case). And the larger the difference between > the size of VIRT and RES, the more memory we save. > > In this patch series, we add a pte_refcount field to the > struct page of page table to track how many users of PTE page > table. Similar to the mechanism of page refcount, the user of > PTE page table should hold a refcount to it before accessing. > The PTE page table page will be freed when the last refcount > is dropped. The patch is very hard to review. Could you split up introduction of the new API in the separate patch? With a proper documentation of the API. Why pte_refcount is atomic? Looks like you do everything under pmd_lock(). Do I miss something? And performance numbers should be included. I don't expect pmd_lock() in all hotpaths to scale well. -- Kirill A. Shutemov
