Re: [PATCH RFC 2/2] mm/memory_hotplug: handle memblocks only with CONFIG_ARCH_KEEP_MEMBLOCK

[Michal Hocko]

On Thu 16-04-20 12:47:07, David Hildenbrand wrote:
> The comment in add_memory_resource() is stale: hotadd_new_pgdat() will
> no longer call get_pfn_range_for_nid(), as a hotadded pgdat will simply
> span no pages at all, until memory is moved to the zone/node via
> move_pfn_range_to_zone() - e.g., when onlining memory blocks.
> 
> The only archs that care about memblocks for hotplugged memory (either
> for iterating over all system RAM or testing for memory validity) are
> arm64, s390x, and powerpc - due to CONFIG_ARCH_KEEP_MEMBLOCK. Without
> CONFIG_ARCH_KEEP_MEMBLOCK, we can simply stop messing with memblocks.

OK, makes sense to me.

> For s390x, it seems to be fairly easy to avoid CONFIG_ARCH_KEEP_MEMBLOCK.
> arm64 could rework most code (esp., pfn_valid(), valid_phys_addr_range()
> and kexec_file_load()) to not require memblocks for hotplugged
> memory. E.g., as hotplugged memory has no holes and can be identified
> using !early_section(), arm64's variant of pfn_valid() could be reworked
> fairly easily to not require memblocks for hotadded memory. powerpc might
> be more involed.

I haven't checked these architectures but is the information really
useful for this patch?

> Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> Cc: Michal Hocko <mhocko@xxxxxxxxxx>
> Cc: Baoquan He <bhe@xxxxxxxxxx>
> Cc: Oscar Salvador <osalvador@xxxxxxx>
> Cc: Pankaj Gupta <pankaj.gupta.linux@xxxxxxxxx>
> Cc: Mike Rapoport <rppt@xxxxxxxxxxxxx>
> Cc: Anshuman Khandual <anshuman.khandual@xxxxxxx>
> Signed-off-by: David Hildenbrand <david@xxxxxxxxxx>

Acked-by: Michal Hocko <mhocko@xxxxxxxx>

with a minor nit

> -	/*
> -	 * Add new range to memblock so that when hotadd_new_pgdat() is called
> -	 * to allocate new pgdat, get_pfn_range_for_nid() will be able to find
> -	 * this new range and calculate total pages correctly.  The range will
> -	 * be removed at hot-remove time.
> -	 */
> +#ifdef CONFIG_ARCH_KEEP_MEMBLOCK

	if (IS_ENABLED(CONFIG_ARCH_KEEP_MEMBLOCK)

would be slightly nicer. This should work for all the ifedefs in this
patch.

>  	memblock_add_node(start, size, nid);
> +#endif
>  
>  	ret = __try_online_node(nid, false);
>  	if (ret < 0)
> @@ -1075,7 +1071,9 @@ int __ref add_memory_resource(int nid, struct resource *res)
>  	/* rollback pgdat allocation and others */
>  	if (new_node)
>  		rollback_node_hotadd(nid);
> +#ifdef CONFIG_ARCH_KEEP_MEMBLOCK
>  	memblock_remove(start, size);
> +#endif
>  	mem_hotplug_done();
>  	return ret;
>  }
> @@ -1751,8 +1749,11 @@ static int __ref try_remove_memory(int nid, u64 start, u64 size)
>  	mem_hotplug_begin();
>  
>  	arch_remove_memory(nid, start, size, NULL);
> +
> +#ifdef CONFIG_ARCH_KEEP_MEMBLOCK
>  	memblock_free(start, size);
>  	memblock_remove(start, size);
> +#endif
>  	__release_memory_resource(start, size);
>  
>  	try_offline_node(nid);
> -- 
> 2.25.1

-- 
Michal Hocko
SUSE Labs