Re: [PATCH] vmemmap, memory_hotplug: fallback to base pages for vmmap

[Michal Hocko <mhocko@kernel.org>]

Ohh, scratch that. The patch is bogus. I have completely missed that
vmemmap_populate_hugepages already falls back to
vmemmap_populate_basepages. I have to revisit the bug report I have
received to see what happened apart from the allocation warning. Maybe
we just want to silent that warning.

Sorry about the noise!

On Tue 11-07-17 15:42:04, Michal Hocko wrote:
> From: Michal Hocko <mhocko@suse.com>
> 
> vmemmap_populate uses huge pages if the CPU supports them which is good
> and usually what we want. vmemmap_alloc_block will use the bootmem
> allocator in the early initialization so the allocation will most likely
> succeed. This is not the case for the memory hotplug though. Such an
> allocation can easily fail under memory pressure. Especially so when the
> kernel memory is restricted with movable_node parameter.
> 
> There is no real reason to fail the vmemmap_populate when
> vmemmap_populate_hugepages fails though. We can still fallback to
> vmemmap_populate_basepages and use base pages. The performance will not
> be optimal but this is much better than failing the memory hot add.
> 
> Signed-off-by: Michal Hocko <mhocko@suse.com>
> ---
>  arch/x86/mm/init_64.c | 7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
> index 136422d7d539..e6e3c755b9cb 100644
> --- a/arch/x86/mm/init_64.c
> +++ b/arch/x86/mm/init_64.c
> @@ -1401,15 +1401,16 @@ static int __meminit vmemmap_populate_hugepages(unsigned long start,
>  int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
>  {
>  	struct vmem_altmap *altmap = to_vmem_altmap(start);
> -	int err;
> +	int err = -ENOMEM;
>  
>  	if (boot_cpu_has(X86_FEATURE_PSE))
>  		err = vmemmap_populate_hugepages(start, end, node, altmap);
>  	else if (altmap) {
>  		pr_err_once("%s: no cpu support for altmap allocations\n",
>  				__func__);
> -		err = -ENOMEM;
> -	} else
> +		return err;
> +	}
> +	if (err)
>  		err = vmemmap_populate_basepages(start, end, node);
>  	if (!err)
>  		sync_global_pgds(start, end - 1);
> -- 
> 2.11.0
> 

-- 
Michal Hocko
SUSE Labs

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>