Messages in this thread | | | Subject | Re: [PATCH] hugetlbfs: lockdep annotate root inode properly | From | Peter Zijlstra <> | Date | Thu, 08 Mar 2012 23:46:20 +0100 |
| |
On Thu, 2012-03-08 at 23:44 +0100, Peter Zijlstra wrote: > On Thu, 2012-03-08 at 21:44 +0000, Al Viro wrote: > > I suspect that they right thing would be to have a way to set explicit > > nesting rules, not tied to speficic call trace. > > See might_lock() / might_lock_read(), these are used to implement > might_fault(), which is used to annotate paths that could -- but rarely > do -- fault.
This will of course result in a specific trace, but if you do it early enough the trace points to your setup function, which can contain a comment explaining things.
| |