Re: BUG FIX?: mm->rss is modified in some places without holding the page_table_lock

["Theodore Y. Ts'o" <tytso@MIT.EDU>]

      Given that we don't have a 64-bit atomic_t type, what do people
      think of Davej's patch?  (attached, below)

   Broken, in 9 out of 10 places where he adds page_table_lock
   acquisitions, this lock is already held --> instant deadlock.

   This report is complicated by the fact that people were forgetting
   that vmlist_*_{lock,unlock}(mm) was actually just spin_{lock,unlock}
   on mm->page_table_lock.  I fixed that already by removing the dumb
   vmlist locking macros which were causing all of this confusion.

Are you saying that the original bug report may not actually be a
problem?  Is ms->rss actually protected in _all_ of the right places, but
people got confused because of the syntactic sugar?

						- Ted
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux.eu.org/Linux-MM/