2013-09-06 05:16:50

by Weijie Yang

[permalink] [raw]
Subject: [PATCH v2 0/4] mm/zswap bugfix: memory leaks and other problems

This patch series fix a few bugs in zswap based on Linux-3.11.

v1 --> v2
- free memory in zswap_frontswap_invalidate_area (in patch 1)
- fix whitespace corruption (line wrapping)

Corresponding mail thread: https://lkml.org/lkml/2013/8/18/59

These issues fixed/optimized are:

1. memory leaks when re-swapon

2. memory leaks when invalidate and reclaim occur concurrently

3. avoid unnecessary page scanning

4. use GFP_NOIO instead of GFP_KERNEL to avoid zswap store and reclaim
functions called recursively

Issues discussed in that mail thread NOT fixed as it happens rarely or
not a big problem:

1. a "theoretical race condition" when reclaim page
When a handle alloced from zbud, zbud considers this handle is used
validly by upper(zswap) and can be a candidate for reclaim. But zswap has
to initialize it such as setting swapentry and adding it to rbtree.
so there is a race condition, such as:
thread 0: obtain handle x from zbud_alloc
thread 1: zbud_reclaim_page is called
thread 1: callback zswap_writeback_entry to reclaim handle x
thread 1: get swpentry from handle x (it is random value now)
thread 1: bad thing may happen
thread 0: initialize handle x with swapentry

2. frontswap_map bitmap not cleared after zswap reclaim
Frontswap uses frontswap_map bitmap to track page in "backend" implementation,
when zswap reclaim a page, the corresponding bitmap record is not cleared.

mm/zswap.c | 34 +++++++++++++++++++++++-----------
1 file changed, 23 insertions(+), 11 deletions(-)