----- On Nov 16, 2017, at 11:18 AM, Peter Zijlstra peterz-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org wrote:

On Tue, Nov 14, 2017 at 03:03:51PM -0500, Mathieu Desnoyers wrote:

quoted

@@ -977,6 +978,13 @@ struct task_struct {
 	unsigned long			numa_pages_migrated;
 #endif /* CONFIG_NUMA_BALANCING */
 
+#ifdef CONFIG_RSEQ
+	struct rseq __user *rseq;
+	u32 rseq_len;
+	u32 rseq_sig;
+	bool rseq_preempt, rseq_signal, rseq_migrate;

No bool please. Use something that has a defined size in ILP32/LP64.
_Bool makes it absolutely impossible to speculate on structure layout
across architectures.

I should as well make all those a bitmask within a "u32 rseq_event_mask" then,
sounds fair ?

Thanks,

Mathieu

quoted

+#endif
+
 	struct tlbflush_unmap_batch	tlb_ubc;
 
 	struct rcu_head			rcu;

-- 
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com