On Wed 17-10-18 21:10:22, Mike Kravetz wrote:
Some test systems were experiencing negative huge page reserve counts and incorrect file block counts. This was traced to /proc/sys/vm/drop_caches removing clean pages from hugetlbfs file pagecaches. When non-hugetlbfs explicit code removes the pages, the appropriate accounting is not performed.
This can be recreated as follows: fallocate -l 2M /dev/hugepages/foo echo 1 > /proc/sys/vm/drop_caches fallocate -l 2M /dev/hugepages/foo grep -i huge /proc/meminfo AnonHugePages: 0 kB ShmemHugePages: 0 kB HugePages_Total: 2048 HugePages_Free: 2047 HugePages_Rsvd: 18446744073709551615 HugePages_Surp: 0 Hugepagesize: 2048 kB Hugetlb: 4194304 kB ls -lsh /dev/hugepages/foo 4.0M -rw-r--r--. 1 root root 2.0M Oct 17 20:05 /dev/hugepages/foo
To address this issue, dirty pages as they are added to pagecache. This can easily be reproduced with fallocate as shown above. Read faulted pages will eventually end up being marked dirty. But there is a window where they are clean and could be impacted by code such as drop_caches. So, just dirty them all as they are added to the pagecache.
In addition, it makes little sense to even try to drop hugetlbfs pagecache pages, so disable calls to these filesystems in drop_caches code.
Fixes: 70c3547e36f5 ("hugetlbfs: add hugetlbfs_fallocate()") Cc: stable@vger.kernel.org Signed-off-by: Mike Kravetz mike.kravetz@oracle.com
I do agree with others that HUGETLBFS_MAGIC check in drop_pagecache_sb is wrong in principal. I am not even sure we want to special case memory backed filesystems. What if we ever implement MADV_FREE on fs? Should those pages be dropped? My first idea take would be yes.
Acked-by: Michal Hocko mhocko@suse.com to the set_page_dirty dirty part.
Although I am wondering why you haven't covered only the fallocate path wrt Fixes tag. In other words, do we need the same treatment for the page fault path? We do not set dirty bit on page there as well. We rely on the dirty bit in pte and only for writable mappings. I have hard time to see why we have been safe there as well. So maybe it is your Fixes: tag which is not entirely correct, or I am simply missing the fault path.