Re: Weirdness in __alloc_bootmem_node_high

[Tejun Heo]

On Tue, Apr 17, 2012 at 05:55:02PM +0200, Michal Hocko wrote:
> Hi,
> I just come across the following condition in __alloc_bootmem_node_high
> which I have hard times to understand. I guess it is a bug and we need
> something like the following. But, to be honest, I have no idea why we
> care about those 128MB above MAX_DMA32_PFN.
> ---
>  mm/bootmem.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/bootmem.c b/mm/bootmem.c
> index 0131170..5adb072 100644
> --- a/mm/bootmem.c
> +++ b/mm/bootmem.c
> @@ -737,7 +737,7 @@ void * __init __alloc_bootmem_node_high(pg_data_t *pgdat, unsigned long size,
>  	/* update goal according ...MAX_DMA32_PFN */
>  	end_pfn = pgdat->node_start_pfn + pgdat->node_spanned_pages;
>  
> -	if (end_pfn > MAX_DMA32_PFN + (128 >> (20 - PAGE_SHIFT)) &&
> +	if (end_pfn > MAX_DMA32_PFN + (128 << (20 - PAGE_SHIFT)) &&
>  	    (goal >> PAGE_SHIFT) < MAX_DMA32_PFN) {
>  		void *ptr;
>  		unsigned long new_goal;

Regardless of x86 not using it, this is a bug fix and this code path
seems to be used by mips at least.  Michal, can you please post proper
signed-off patch?  The code is simply trying to use memory above DMA32
limit if there seems to be enough space (128M) to avoid unnecessarily
using DMA32 memory.

Thanks.

-- 
tejun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/