Re: [PATCH v3] mm/swap: fix race when skipping swapcache
From: Kairui Song
Date: Sun Feb 18 2024 - 03:29:32 EST
On Sat, Feb 17, 2024 at 2:02 AM David Hildenbrand <david@xxxxxxxxxx> wrote:
> On 16.02.24 17:53, David Hildenbrand wrote:
> > On 16.02.24 10:51, Kairui Song wrote:
> >> From: Kairui Song <kasong@xxxxxxxxxxx>
> >>
> >> When skipping swapcache for SWP_SYNCHRONOUS_IO, if two or more threads
> >> swapin the same entry at the same time, they get different pages (A, B).
> >> Before one thread (T0) finishes the swapin and installs page (A)
> >> to the PTE, another thread (T1) could finish swapin of page (B),
> >> swap_free the entry, then swap out the possibly modified page
> >> reusing the same entry. It breaks the pte_same check in (T0) because
> >> PTE value is unchanged, causing ABA problem. Thread (T0) will
> >> install a stalled page (A) into the PTE and cause data corruption.
> >>
> >> One possible callstack is like this:
> >>
> >> CPU0 CPU1
> >> ---- ----
> >> do_swap_page() do_swap_page() with same entry
> >> <direct swapin path> <direct swapin path>
> >> <alloc page A> <alloc page B>
> >> swap_read_folio() <- read to page A swap_read_folio() <- read to page B
> >> <slow on later locks or interrupt> <finished swapin first>
> >> ... set_pte_at()
> >> swap_free() <- entry is free
> >> <write to page B, now page A stalled>
> >> <swap out page B to same swap entry>
> >> pte_same() <- Check pass, PTE seems
> >> unchanged, but page A
> >> is stalled!
> >> swap_free() <- page B content lost!
> >> set_pte_at() <- staled page A installed!
> >>
> >> And besides, for ZRAM, swap_free() allows the swap device to discard
> >> the entry content, so even if page (B) is not modified, if
> >> swap_read_folio() on CPU0 happens later than swap_free() on CPU1,
> >> it may also cause data loss.
> >>
> >> To fix this, reuse swapcache_prepare which will pin the swap entry using
> >> the cache flag, and allow only one thread to pin it. Release the pin
> >> after PT unlocked. Racers will simply wait since it's a rare and very
> >> short event. A schedule() call is added to avoid wasting too much CPU
> >> or adding too much noise to perf statistics
> >>
> >> Other methods like increasing the swap count don't seem to be a good
> >> idea after some tests, that will cause racers to fall back to use the
> >> swap cache again. Parallel swapin using different methods leads to
> >> a much more complex scenario.
> >>
> >> Reproducer:
> >>
> >> This race issue can be triggered easily using a well constructed
> >> reproducer and patched brd (with a delay in read path) [1]:
> >>
> >> With latest 6.8 mainline, race caused data loss can be observed easily:
> >> $ gcc -g -lpthread test-thread-swap-race.c && ./a.out
> >> Polulating 32MB of memory region...
> >> Keep swapping out...
> >> Starting round 0...
> >> Spawning 65536 workers...
> >> 32746 workers spawned, wait for done...
> >> Round 0: Error on 0x5aa00, expected 32746, got 32743, 3 data loss!
> >> Round 0: Error on 0x395200, expected 32746, got 32743, 3 data loss!
> >> Round 0: Error on 0x3fd000, expected 32746, got 32737, 9 data loss!
> >> Round 0 Failed, 15 data loss!
> >>
> >> This reproducer spawns multiple threads sharing the same memory region
> >> using a small swap device. Every two threads updates mapped pages one by
> >> one in opposite direction trying to create a race, with one dedicated
> >> thread keep swapping out the data out using madvise.
> >>
> >> The reproducer created a reproduce rate of about once every 5 minutes,
> >> so the race should be totally possible in production.
> >>
> >> After this patch, I ran the reproducer for over a few hundred rounds
> >> and no data loss observed.
> >>
> >> Performance overhead is minimal, microbenchmark swapin 10G from 32G
> >> zram:
> >>
> >> Before: 10934698 us
> >> After: 11157121 us
> >> Non-direct: 13155355 us (Dropping SWP_SYNCHRONOUS_IO flag)
> >>
> >> Fixes: 0bcac06f27d7 ("mm, swap: skip swapcache for swapin of synchronous device")
> >> Link: https://github.com/ryncsn/emm-test-project/tree/master/swap-stress-race [1]
> >> Reported-by: "Huang, Ying" <ying.huang@xxxxxxxxx>
> >> Closes: https://lore.kernel.org/lkml/87bk92gqpx.fsf_-_@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/
> >> Signed-off-by: Kairui Song <kasong@xxxxxxxxxxx>
> >> Cc: stable@xxxxxxxxxxxxxxx
> >>
> >> ---
> >> Update from V2:
> >> - Add a schedule() if raced to prevent repeated page faults wasting CPU
> >> and add noise to perf statistics.
> >> - Use a bool to state the special case instead of reusing existing
> >> variables fixing error handling [Minchan Kim].
> >>
> >> V2: https://lore.kernel.org/all/20240206182559.32264-1-ryncsn@xxxxxxxxx/
> >>
> >> Update from V1:
> >> - Add some words on ZRAM case, it will discard swap content on swap_free so the race window is a bit different but cure is the same. [Barry Song]
> >> - Update comments make it cleaner [Huang, Ying]
> >> - Add a function place holder to fix CONFIG_SWAP=n built [SeongJae Park]
> >> - Update the commit message and summary, refer to SWP_SYNCHRONOUS_IO instead of "direct swapin path" [Yu Zhao]
> >> - Update commit message.
> >> - Collect Review and Acks.
> >>
> >> V1: https://lore.kernel.org/all/20240205110959.4021-1-ryncsn@xxxxxxxxx/
> >>
> >> include/linux/swap.h | 5 +++++
> >> mm/memory.c | 20 ++++++++++++++++++++
> >> mm/swap.h | 5 +++++
> >> mm/swapfile.c | 13 +++++++++++++
> >> 4 files changed, 43 insertions(+)
> >>
> >> diff --git a/include/linux/swap.h b/include/linux/swap.h
> >> index 4db00ddad261..8d28f6091a32 100644
> >> --- a/include/linux/swap.h
> >> +++ b/include/linux/swap.h
> >> @@ -549,6 +549,11 @@ static inline int swap_duplicate(swp_entry_t swp)
> >> return 0;
> >> }
> >>
> >> +static inline int swapcache_prepare(swp_entry_t swp)
> >> +{
> >> + return 0;
> >> +}
> >> +
> >> static inline void swap_free(swp_entry_t swp)
> >> {
> >> }
> >> diff --git a/mm/memory.c b/mm/memory.c
> >> index 7e1f4849463a..7059230d0a54 100644
> >> --- a/mm/memory.c
> >> +++ b/mm/memory.c
> >> @@ -3799,6 +3799,7 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >> struct page *page;
> >> struct swap_info_struct *si = NULL;
> >> rmap_t rmap_flags = RMAP_NONE;
> >> + bool need_clear_cache = false;
> >> bool exclusive = false;
> >> swp_entry_t entry;
> >> pte_t pte;
> >> @@ -3867,6 +3868,20 @@ vm_fault_t do_swap_page(struct vm_fault *vmf)
> >> if (!folio) {
> >> if (data_race(si->flags & SWP_SYNCHRONOUS_IO) &&
> >> __swap_count(entry) == 1) {
> >> + /*
> >> + * Prevent parallel swapin from proceeding with
> >> + * the cache flag. Otherwise, another thread may
> >> + * finish swapin first, free the entry, and swapout
> >> + * reusing the same entry. It's undetectable as
> >> + * pte_same() returns true due to entry reuse.
> >> + */
> >> + if (swapcache_prepare(entry)) {
> >> + /* Relax a bit to prevent rapid repeated page faults */
> >> + schedule();
> >> + goto out;
> >> + }
> >> + need_clear_cache = true;
> >> +
Hi David
Thanks for the review! I saw you added more replies so I'll just post
reply on your last mail.
> >
> > I took a closer look at __read_swap_cache_async() and it essentially
> > does something similar.
> >
> > Instead of returning, it keeps retrying until it finds that
> > swapcache_prepare() fails for another reason than -EEXISTS (e.g.,
> > freed concurrently) or it finds the entry in the swapcache.
> >
> > So if you would succeed here on a freed+reused swap entry,
> > __read_swap_cache_async() would simply retry.
> >
> > It spells that out:
> >
> > /*
> > * We might race against __delete_from_swap_cache(), and
> > * stumble across a swap_map entry whose SWAP_HAS_CACHE
> > * has not yet been cleared. Or race against another
> > * __read_swap_cache_async(), which has set SWAP_HAS_CACHE
> > * in swap_map, but not yet added its folio to swap cache.
> > */
> >
> > Whereby we could not race against this code here as well where we
> > speculatively set SWAP_HAS_CACHE and might never add something to the swap
> > cache.
> >
> >
> > I'd probably avoid the wrong returns and do something even closer to
> > __read_swap_cache_async().
> >
> > while (true) {
> > /*
> > * Fake that we are trying to insert a page into the swapcache, to
> > * serialize against concurrent threads wanting to do the same.
> > * [more from your description]
> > */
> > ret = swapcache_prepare(entry);
> > if (likely(!ret)
> > /*
> > * Move forward with swapin, we'll recheck if the PTE hasn't
> > * changed later.
> > */
> > break;
> > else if (ret != -EEXIST)
> > goto out;
> >
> > /*
> > * See __read_swap_cache_async(). We might either have raced against
> > * another thread, or the entry could have been freed and reused in the
> > * meantime. Make sure that the PTE did not change, to detect freeing.
> > */
> > vmf->pte = pte_offset_map_lock(vma->vm_mm, vmf->pmd,
> > vmf->address, &vmf->ptl);
> > if (!vmf->pte || !pte_same(ptep_get(vmf->pte), vmf->orig_pte))
> > goto unlock;
> >
> >
> > schedule();
> > }
It was discussed earlier about looping in the page fault. One issue is
about swap entry may stuck in swapcache for a long time, so at lease
an extra cache loop up is need. To be safe we need to implement a
similar loop like the one in mm/swap_state.c, which I doubt is
necessary...
> >
> > I was skeptical about the schedule(), but __read_swap_cache_async() does it
> > already because there is no better way to wait for the event to happen.
> >
> > With something like above you would no longer depend on the speed of schedule() to
> > determine how often you would retry the fault, which would likely make sense.
> >
> > I do wonder about the schedule() vs. schedule_timeout_uninterruptible(), though.
> > No expert on that area, do you have any idea?
> >
schedule_timeout_uninterruptible seems more reasonable here from its
name (delay a bit to wait). My idea here is that SWP_SYNCHRONOUS_IO
devices are supposed to be super fast, so usually a second try will
just work (when tested with a less stressed test case and that seems
to be always true), and the race itself is rare enough to be ignore
for 7 years.
But when system is really stressed (eg. the reproducer I provided), it
may take longer to finish (SWP_SYNCHRONOUS_IO devices are CPU bound).
So a schedule() can help to avoid one task from looping page fault,
for better statistic and CPU usage.
Previous test results:
https://lore.kernel.org/lkml/CAMgjq7BvTJmxrWQOJvkLt4g_jnvmx07NdU63sGeRMGde4Ov=gA@xxxxxxxxxxxxxx/
It showed schedule() works fine here.