Commit 1d86e29b authored by Corey Minyard's avatar Corey Minyard
Browse files

ipmi: Fix a memory ordering issue



From a locking point of view it is safe to check waiting_msg without
a lock, but there is a memory ordering issue that causes it to
possibly not be set right when viewed from another processor.  We are
already claiming a lock right after that, move the check to inside
the lock to enforce the memory ordering.
Signed-off-by: default avatarCorey Minyard <cminyard@mvista.com>
parent d6c5dc18
......@@ -932,9 +932,6 @@ static void sender(void *send_info,
enum si_sm_result result;
unsigned long flags;
BUG_ON(smi_info->waiting_msg);
smi_info->waiting_msg = msg;
debug_timestamp("Enqueue");
if (smi_info->run_to_completion) {
......@@ -942,7 +939,7 @@ static void sender(void *send_info,
* If we are running to completion, start it and run
* transactions until everything is clear.
*/
smi_info->curr_msg = smi_info->waiting_msg;
smi_info->curr_msg = msg;
smi_info->waiting_msg = NULL;
/*
......@@ -960,6 +957,15 @@ static void sender(void *send_info,
}
spin_lock_irqsave(&smi_info->si_lock, flags);
/*
* The following two lines don't need to be under the lock for
* the lock's sake, but they do need SMP memory barriers to
* avoid getting things out of order. We are already claiming
* the lock, anyway, so just do it under the lock to avoid the
* ordering problem.
*/
BUG_ON(smi_info->waiting_msg);
smi_info->waiting_msg = msg;
check_start_timer_thread(smi_info);
spin_unlock_irqrestore(&smi_info->si_lock, flags);
}
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment