From f88ce2c84a341f44a7d00bc10868714bc4751f7e Mon Sep 17 00:00:00 2001 From: David Woodhouse Date: Wed, 23 Apr 2025 14:33:37 +0100 Subject: [PATCH] mm: introduce for_each_valid_pfn() and use it from reserve_bootmem_region() MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Patch series "mm: Introduce for_each_valid_pfn()", v4. There are cases where a naïve loop over a PFN range, calling pfn_valid() on each one, is horribly inefficient. Ruihan Li reported the case where memmap_init() iterates all the way from zero to a potentially large value of ARCH_PFN_OFFSET, and we at Amazon found the reserve_bootmem_region() one as it affects hypervisor live update. Others are more cosmetic. By introducing a for_each_valid_pfn() helper it can optimise away a lot of pointless calls to pfn_valid(), skipping immediately to the next valid PFN and also skipping *all* checks within a valid (sub)region according to the granularity of the memory model in use. This patch (of 7) Especially since commit 9092d4f7a1f8 ("memblock: update initialization of reserved pages"), the reserve_bootmem_region() function can spend a significant amount of time iterating over every 4KiB PFN in a range, calling pfn_valid() on each one, and ultimately doing absolutely nothing. On a platform used for virtualization, with large NOMAP regions that eventually get used for guest RAM, this leads to a significant increase in steal time experienced during kexec for a live update. Introduce for_each_valid_pfn() and use it from reserve_bootmem_region(). This implementation is precisely the same naïve loop that the functio used to have, but subsequent commits will provide optimised versions for FLATMEM and SPARSEMEM, and this version will remain for those architectures which provide their own pfn_valid() implementation, until/unless they also provide a matching for_each_valid_pfn(). Link: https://lkml.kernel.org/r/20250423133821.789413-1-dwmw2@infradead.org Link: https://lkml.kernel.org/r/20250423133821.789413-2-dwmw2@infradead.org Signed-off-by: David Woodhouse Reviewed-by: Mike Rapoport (Microsoft) Acked-by: David Hildenbrand Cc: Anshuman Khandual Cc: Ard Biesheuvel Cc: Catalin Marinas Cc: Marc Rutland Cc: Marc Zyngier Cc: Ruihan Li Cc: Will Deacon Cc: Lorenzo Stoakes Signed-off-by: Andrew Morton --- include/linux/mmzone.h | 10 ++++++++++ mm/mm_init.c | 23 ++++++++++------------- 2 files changed, 20 insertions(+), 13 deletions(-) diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 6ccec1bf2896f..230a29c2ed1a9 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -2177,6 +2177,16 @@ void sparse_init(void); #define subsection_map_init(_pfn, _nr_pages) do {} while (0) #endif /* CONFIG_SPARSEMEM */ +/* + * Fallback case for when the architecture provides its own pfn_valid() but + * not a corresponding for_each_valid_pfn(). + */ +#ifndef for_each_valid_pfn +#define for_each_valid_pfn(_pfn, _start_pfn, _end_pfn) \ + for ((_pfn) = (_start_pfn); (_pfn) < (_end_pfn); (_pfn)++) \ + if (pfn_valid(_pfn)) +#endif + #endif /* !__GENERATING_BOUNDS.H */ #endif /* !__ASSEMBLY__ */ #endif /* _LINUX_MMZONE_H */ diff --git a/mm/mm_init.c b/mm/mm_init.c index b35006d9d49d0..7191703a5820c 100644 --- a/mm/mm_init.c +++ b/mm/mm_init.c @@ -783,22 +783,19 @@ void __meminit init_deferred_page(unsigned long pfn, int nid) void __meminit reserve_bootmem_region(phys_addr_t start, phys_addr_t end, int nid) { - unsigned long start_pfn = PFN_DOWN(start); - unsigned long end_pfn = PFN_UP(end); + unsigned long pfn; - for (; start_pfn < end_pfn; start_pfn++) { - if (pfn_valid(start_pfn)) { - struct page *page = pfn_to_page(start_pfn); + for_each_valid_pfn(pfn, PFN_DOWN(start), PFN_UP(end)) { + struct page *page = pfn_to_page(pfn); - __init_deferred_page(start_pfn, nid); + __init_deferred_page(pfn, nid); - /* - * no need for atomic set_bit because the struct - * page is not visible yet so nobody should - * access it yet. - */ - __SetPageReserved(page); - } + /* + * no need for atomic set_bit because the struct + * page is not visible yet so nobody should + * access it yet. + */ + __SetPageReserved(page); } } -- 2.39.5