Re: [RFC PATCH] rt/aio: fix rcu garbage collection might_sleep() splat

[Mike Galbraith <umgwanakikbuti@gmail.com>]

On Mon, 2014-06-09 at 05:17 +0200, Mike Galbraith wrote: 
> On Mon, 2014-06-09 at 10:08 +0800, Lai Jiangshan wrote: 
> > Hi, rt-people
> > 
> > I don't think it is the correct direction.
> 
> Yup, it's a band-aid, ergo RFC.

Making aio play by the rules is safe.  Another option is to bend the
rules up a little.  With the below, ltp aio-stress testcases haven't yet
griped or exploded either.. which somehow isn't quite as comforting as
"is safe" :)

---
 fs/aio.c |    6 ++++++
 1 file changed, 6 insertions(+)

--- a/fs/aio.c
+++ b/fs/aio.c
@@ -509,7 +509,9 @@ static void free_ioctx_reqs(struct percp
 	struct kioctx *ctx = container_of(ref, struct kioctx, reqs);
 
 	INIT_WORK(&ctx->free_work, free_ioctx);
+	preempt_enable_rt();
 	schedule_work(&ctx->free_work);
+	preempt_disable_rt();
 }
 
 /*
@@ -522,7 +524,9 @@ static void free_ioctx_users(struct perc
 	struct kioctx *ctx = container_of(ref, struct kioctx, users);
 	struct kiocb *req;
 
+	preempt_enable_rt();
 	spin_lock_irq(&ctx->ctx_lock);
+	local_irq_disable_rt();
 
 	while (!list_empty(&ctx->active_reqs)) {
 		req = list_first_entry(&ctx->active_reqs,
@@ -536,6 +540,8 @@ static void free_ioctx_users(struct perc
 
 	percpu_ref_kill(&ctx->reqs);
 	percpu_ref_put(&ctx->reqs);
+	preempt_disable_rt();
+	local_irq_enable_rt();
 }
 
 static int ioctx_add_table(struct kioctx *ctx, struct mm_struct *mm)