[PATCH] mm/migrate: fix race between lock page and clear PG_Isolated

[Matthew Wilcox]

On Tue, Mar 15, 2022 at 04:45:13PM +0100, David Hildenbrand wrote:
> On 15.03.22 05:21, Andrew Morton wrote:
> > On Tue, 15 Mar 2022 11:05:15 +0800 Andrew Yang <andrew.yang at mediatek.com> wrote:
> > 
> >> When memory is tight, system may start to compact memory for large
> >> continuous memory demands. If one process tries to lock a memory page
> >> that is being locked and isolated for compaction, it may wait a long time
> >> or even forever. This is because compaction will perform non-atomic
> >> PG_Isolated clear while holding page lock, this may overwrite PG_waiters
> >> set by the process that can't obtain the page lock and add itself to the
> >> waiting queue to wait for the lock to be unlocked.
> >>
> >> CPU1                            CPU2
> >> lock_page(page); (successful)
> >>                                 lock_page(); (failed)
> >> __ClearPageIsolated(page);      SetPageWaiters(page) (may be overwritten)
> >> unlock_page(page);
> >>
> >> The solution is to not perform non-atomic operation on page flags while
> >> holding page lock.
> > 
> > Sure, the non-atomic bitop optimization is really risky and I suspect
> > we reach for it too often.  Or at least without really clearly
> > demonstrating that it is safe, and documenting our assumptions.
> 
> I agree. IIRC, non-atomic variants are mostly only safe while the
> refcount is 0. Everything else is just absolutely fragile.

We could add an assertion ... I just tried this:

+++ b/include/linux/page-flags.h
@@ -342,14 +342,16 @@ static __always_inline                                                    \
 void __folio_set_##lname(struct folio *folio)                          \
 { __set_bit(PG_##lname, folio_flags(folio, FOLIO_##policy)); }         \
 static __always_inline void __SetPage##uname(struct page *page)                \
-{ __set_bit(PG_##lname, &policy(page, 1)->flags); }
+{ VM_BUG_ON_PGFLAGS(atomic_read(&policy(page, 1)->_refcount), page);   \
+  __set_bit(PG_##lname, &policy(page, 1)->flags); }

 #define __CLEARPAGEFLAG(uname, lname, policy)                          \
 static __always_inline                                                 \
 void __folio_clear_##lname(struct folio *folio)                                \
 { __clear_bit(PG_##lname, folio_flags(folio, FOLIO_##policy)); }       \
 static __always_inline void __ClearPage##uname(struct page *page)      \
-{ __clear_bit(PG_##lname, &policy(page, 1)->flags); }
+{ VM_BUG_ON_PGFLAGS(atomic_read(&policy(page, 1)->_refcount), page);   \
+  __clear_bit(PG_##lname, &policy(page, 1)->flags); }

 #define TESTSETFLAG(uname, lname, policy)                              \
 static __always_inline                                                 \

... but it dies _really_ early:

(gdb) bt
#0  0xffffffff820055e5 in native_halt ()
    at ../arch/x86/include/asm/irqflags.h:57
#1  halt () at ../arch/x86/include/asm/irqflags.h:98
#2  early_fixup_exception (regs=regs at entry=0xffffffff81e03cf8,
    trapnr=trapnr at entry=6) at ../arch/x86/mm/extable.c:283
#3  0xffffffff81ff243c in do_early_exception (regs=0xffffffff81e03cf8,
    trapnr=6) at ../arch/x86/kernel/head64.c:419
#4  0xffffffff81ff214f in early_idt_handler_common ()
    at ../arch/x86/kernel/head_64.S:417
#5  0x0000000000000000 in ?? ()

and honestly, I'm not sure how to debug something that goes wrong this
early.  Maybe I need to make that start warning 5 seconds after boot
or only if we're not in pid 1, or something ...