tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

mm: fix accounting of memmap pages

For !CONFIG_SPARSEMEM_VMEMMAP, memmap page accounting is currently done
upfront in sparse_buffer_init(). However, sparse_buffer_alloc() may
return NULL in failure scenario.

Also, memmap pages may be allocated either from the memblock allocator
during early boot or from the buddy allocator. When removed via
arch_remove_memory(), accounting of memmap pages must reflect the original
allocation source.

To ensure correctness:
* Account memmap pages after successful allocation in sparse_init_nid()
and section_activate().
* Account memmap pages in section_deactivate() based on allocation
source.

Link: https://lkml.kernel.org/r/20250807183545.1424509-1-sumanthk@linux.ibm.com
Fixes: 15995a352474 ("mm: report per-page metadata information")
Signed-off-by: Sumanth Korikkar <sumanthk@linux.ibm.com>
Suggested-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Wei Yang <richard.weiyang@gmail.com>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

authored by

Sumanth Korikkar and committed by
Andrew Morton c3576889 9f68eaba

+9 -11
-5
mm/sparse-vmemmap.c
··· 578 578 if (r < 0) 579 579 return NULL; 580 580 581 - if (system_state == SYSTEM_BOOTING) 582 - memmap_boot_pages_add(DIV_ROUND_UP(end - start, PAGE_SIZE)); 583 - else 584 - memmap_pages_add(DIV_ROUND_UP(end - start, PAGE_SIZE)); 585 - 586 581 return pfn_to_page(pfn); 587 582 } 588 583
+9 -6
mm/sparse.c
··· 454 454 */ 455 455 sparsemap_buf = memmap_alloc(size, section_map_size(), addr, nid, true); 456 456 sparsemap_buf_end = sparsemap_buf + size; 457 - #ifndef CONFIG_SPARSEMEM_VMEMMAP 458 - memmap_boot_pages_add(DIV_ROUND_UP(size, PAGE_SIZE)); 459 - #endif 460 457 } 461 458 462 459 static void __init sparse_buffer_fini(void) ··· 564 567 sparse_buffer_fini(); 565 568 goto failed; 566 569 } 570 + memmap_boot_pages_add(DIV_ROUND_UP(PAGES_PER_SECTION * sizeof(struct page), 571 + PAGE_SIZE)); 567 572 sparse_init_early_section(nid, map, pnum, 0); 568 573 } 569 574 } ··· 679 680 unsigned long start = (unsigned long) pfn_to_page(pfn); 680 681 unsigned long end = start + nr_pages * sizeof(struct page); 681 682 682 - memmap_pages_add(-1L * (DIV_ROUND_UP(end - start, PAGE_SIZE))); 683 683 vmemmap_free(start, end, altmap); 684 684 } 685 685 static void free_map_bootmem(struct page *memmap) ··· 854 856 * The memmap of early sections is always fully populated. See 855 857 * section_activate() and pfn_valid() . 856 858 */ 857 - if (!section_is_early) 859 + if (!section_is_early) { 860 + memmap_pages_add(-1L * (DIV_ROUND_UP(nr_pages * sizeof(struct page), PAGE_SIZE))); 858 861 depopulate_section_memmap(pfn, nr_pages, altmap); 859 - else if (memmap) 862 + } else if (memmap) { 863 + memmap_boot_pages_add(-1L * (DIV_ROUND_UP(nr_pages * sizeof(struct page), 864 + PAGE_SIZE))); 860 865 free_map_bootmem(memmap); 866 + } 861 867 862 868 if (empty) 863 869 ms->section_mem_map = (unsigned long)NULL; ··· 906 904 section_deactivate(pfn, nr_pages, altmap); 907 905 return ERR_PTR(-ENOMEM); 908 906 } 907 + memmap_pages_add(DIV_ROUND_UP(nr_pages * sizeof(struct page), PAGE_SIZE)); 909 908 910 909 return memmap; 911 910 }