Re: [patch v2] hugetlb: correct page offset index for sharing pmd

From: Hillf Danton
Date: Thu Aug 16 2012 - 08:40:23 EST


On Thu, Aug 16, 2012 at 7:03 AM, Andrew Morton
<akpm@xxxxxxxxxxxxxxxxxxxx> wrote:

> Don't be too concerned about the size of a change - it's the end result
> which matters. If a larger patch results in a better end result, then
> do the larger patch.

Hi Andrew,

This work was triggered by the fact that huge_pmd_share mismatches
unmap_ref_private. But it does match hugetlb_vmtruncate_list.

Plus RADIX_INDEX and HEAP_INDEX are defined, and used when inserting
vma into prio tree.
===
/*
* The following macros are used for implementing prio_tree for i_mmap
*/

#define RADIX_INDEX(vma) ((vma)->vm_pgoff)
#define VMA_SIZE(vma) (((vma)->vm_end - (vma)->vm_start) >> PAGE_SHIFT)
/* avoid overflow */
#define HEAP_INDEX(vma) ((vma)->vm_pgoff + (VMA_SIZE(vma) - 1))
===

Thus it is incorrect to use huge pgoff in searching vma in prio tree, and
I have to withdraw this work.

Thanks,
Hillf
--
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/