On Thu, Apr 22, 2021 at 11:04:11AM +0200, Jan Kara wrote:

Yes, I understand that. What I was more asking about is: Does it really
matter we leave those buffer heads and journal heads unreclaimed. I
understand it could be triggering premature OOM in theory but is it a
problem in practice? Was there some observed practical case for which this
was added or was it just added due to the theoretical concern?

I was doing some research, and found the mail thread which inspired
bdev_try_to_free_page():

https://lore.kernel.org/linux-ext4/20081202200647.72cc5807.toshi.okajima@jp.fujitsu.com/ (local)

From what I can tell Toshi Okajima did have a test workload which
would trigger blkdev_releasepage().  He didn't specify it in the mail
thread as near as I can tell, but he did use it to test the page.
Thinking about it, it shouldn't be hard to trigger it via something like:

   find /mnt -print0 | xargs -0 touch

in a memory contrained box with a large file system attached (a
bookshelf NAS scenario).  Under the right circumstances, I'm pretty
sure a premature OOM could be demonstrated.

					- Ted