[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [gpsd-dev] SHM broken
From: |
Hal Murray |
Subject: |
Re: [gpsd-dev] SHM broken |
Date: |
Tue, 24 Mar 2015 13:44:57 -0700 |
> Do you actually see working memory barriers on non-C11 compiles?
Good question.
gpsd.h says:
#define HAVE_STDATOMIC_H 1
I can't find anything that looks right in objdump output.
Looks like a bug in compiler.h
It's testing STD_ATOMIC_H
stdatomic.h is setting _STDATOMIC_H
This is one system. I'll go see what happens on others.
diff --git a/compiler.h b/compiler.h
index 4fde42d..53902dc 100644
--- a/compiler.h
+++ b/compiler.h
@@ -81,7 +81,7 @@ static /address@hidden@*/ inline void memory_barrier(void)
/* prevent instruction reordering across any call to this function */
{
#ifndef S_SPLINT_S
-#ifdef STD_ATOMIC_H
+#ifdef _STDATOMIC_H
#ifndef __COVERITY__
atomic_thread_fence(memory_order_seq_cst);
#endif /* __COVERITY__ */
Works now.
---
objdump -S doesn't show any source.
Is there something funny I'm doing?
It works in other contexts.
--
These are my opinions. I hate spam.