On 01.10.21 08:38, Anshuman Khandual wrote:
On 9/29/21 12:21 AM, Chris Goldsworthy wrote:From: Sudarshan Rajagopalan <quic_sudaraja@xxxxxxxxxxx> After new memory blocks have been hotplugged, max_pfn and max_low_pfn needs updating to reflect on new PFNs being hot added to system. Without this patch, debug-related functions that use max_pfn such as get_max_dump_pfn() or read_page_owner() will not work with any page in memory that is hot-added after boot. Fixes: 4ab215061554 ("arm64: Add memory hotplug support") Signed-off-by: Sudarshan Rajagopalan <quic_sudaraja@xxxxxxxxxxx> Signed-off-by: Chris Goldsworthy <quic_cgoldswo@xxxxxxxxxxx> Acked-by: David Hildenbrand <david@xxxxxxxxxx> Cc: Florian Fainelli <f.fainelli@xxxxxxxxx> Cc: Georgi Djakov <quic_c_gdjako@xxxxxxxxxxx> --- arch/arm64/mm/mmu.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index cfd9deb..fd85b51 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -1499,6 +1499,11 @@ int arch_add_memory(int nid, u64 start, u64 size, if (ret) __remove_pgd_mapping(swapper_pg_dir, __phys_to_virt(start), size); + else { + max_pfn = PFN_UP(start + size); + max_low_pfn = max_pfn; + }Do these variables get updated on *all* platforms that support memory hotplug via an arch_add_memory() ? If not, dont they all face similar issues as well ? Why not just update these in generic hotplug instead , after looking into arch_add_memory() return code.
s390x sets them to the possible maximum (based on the direct mapping size) in init code.
Other archs like x86-64 have to update other parameters. So I guess it just made sense to let the archs handle updating these 2 variables.
-- Thanks, David / dhildenb
