mm/dynamic_hugetlb: replace spin_lock with mutex_lock and fix kabi broken

hulk inclusion
category: bugfix
bugzilla: https://gitee.com/openeuler/kernel/issues/I6MH03


CVE: NA

--------------------------------

When memory is fragmented, update_reserve_pages() may call migrate_pages()
to collect continuous memory. This function can sleep, so we should use
mutex lock instead of spin lock. Use KABI_EXTEND to fix kabi broken.

Fixes: 0c06a1c0 ("mm/dynamic_hugetlb: add interface to configure the count of hugepages")
Signed-off-by: Liu Shixin <liushixin2@huawei.com>