Hello,

On Tue, Mar 03, 2020 at 03:03:58PM +0100, Michael Stapelberg wrote:

Here’s a /proc/<pid>/stack from when the issue is happening:

[<0>] balance_dirty_pages_ratelimited+0x2ca/0x3b0
[<0>] __handle_mm_fault+0xe6e/0x1280
[<0>] handle_mm_fault+0xbe/0x1d0
[<0>] __do_page_fault+0x249/0x4f0
[<0>] page_fault+0x1e/0x30

How can I obtain the numbers for the next step?

Yes, that's dirty throttling alright. Hopefully, the
balance_dirty_pages tracepoint which can be enabled from under
/sys/kernel/debug/tracing/events/writeback/balance_dirty_pages/ should
tell us why bdp thinks it needs throttling and then we can go from
there. Unfortunately, I'm rather preoccupied and afraid I don't have a
lot of bandwidth to work on it myself for the coming weeks.

Thanks.

-- 
tejun