Re: [PATCH v2] hugetlb: fix race condition in hugetlb_fault()

[Hillf Danton]

On Sat, Mar 31, 2012 at 4:07 AM, Chris Metcalf <cmetcalf@xxxxxxxxxx> wrote:
> The race is as follows. ÂSuppose a multi-threaded task forks a new
> process, thus bumping up the ref count on all the pages. ÂWhile the fork
> is occurring (and thus we have marked all the PTEs as read-only), another
> thread in the original process tries to write to a huge page, taking an
> access violation from the write-protect and calling hugetlb_cow(). ÂNow,
> suppose the fork() fails. ÂIt will undo the COW and decrement the ref
> count on the pages, so the ref count on the huge page drops back to 1.
> Meanwhile hugetlb_cow() also decrements the ref count by one on the
> original page, since the original address space doesn't need it any more,
> having copied a new page to replace the original page. ÂThis leaves the
> ref count at zero, and when we call unlock_page(), we panic.
>
> The solution is to take an extra reference to the page while we are
> holding the lock on it.
>
If the following chart matches the above description,

===
	fork on CPU A				fault on CPU B
	=============				==============
	...
	down_write(&parent->mmap_sem);
	down_write_nested(&child->mmap_sem);
	...
	while duplicating vmas
		if error
			break;
	...
	up_write(&child->mmap_sem);
	up_write(&parent->mmap_sem);		...
						down_read(&parent->mmap_sem);
						...
						lock_page(page);
						handle COW
						page_mapcount(old_page) == 2
						alloc and prepare new_page
	...
	handle error
	page_remove_rmap(page);
	put_page(page);
	...
						fold new_page into pte
						page_remove_rmap(page);
						put_page(page);
						...
				oops ==>	unlock_page(page);
						up_read(&parent->mmap_sem);
===

would you please spin with description refreshed?

> Cc: stable@xxxxxxxxxx

Let Andrew do the stable work, ok?

Best Regards
-hd

> Signed-off-by: Chris Metcalf <cmetcalf@xxxxxxxxxx>
> ---
> This change incorporates Hillf Danton's suggestion to just unconditionally
> get and put the page around the region of code in question.
>
> Âmm/hugetlb.c | Â Â2 ++
> Â1 files changed, 2 insertions(+), 0 deletions(-)
>
> diff --git a/mm/hugetlb.c b/mm/hugetlb.c
> index 1871753..5f53d6b 100644
> --- a/mm/hugetlb.c
> +++ b/mm/hugetlb.c
> @@ -2701,6 +2701,7 @@ int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
> Â Â Â Â * so no worry about deadlock.
> Â Â Â Â */
> Â Â Â Âpage = pte_page(entry);
> + Â Â Â get_page(page);
> Â Â Â Âif (page != pagecache_page)
> Â Â Â Â Â Â Â Âlock_page(page);
>
> @@ -2732,6 +2733,7 @@ out_page_table_lock:
> Â Â Â Â}
> Â Â Â Âif (page != pagecache_page)
> Â Â Â Â Â Â Â Âunlock_page(page);
> + Â Â Â put_page(page);
>
> Âout_mutex:
> Â Â Â Âmutex_unlock(&hugetlb_instantiation_mutex);
> --
> 1.6.5.2
>
--
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/