提交 757be67f 编写于 作者: R Ralf Baechle

MIPS: Octeon: Fix warning.

Cong Ding <dinggnu@gmail.com> reports correctly that the variable dummy
is being used without initialization.  That said, I can't reproduce this
warning with GCC 4.7.1.  However, since the variable dummy servces no
real purpose, I'm going for a different fix.  This fix
includes https://patchwork.linux-mips.org/patch/4801/ plus Geert's
suggestion to use ACCESS_ONCE().
Signed-off-by: NRalf Baechle <ralf@linux-mips.org>
上级 4ea494b5
......@@ -30,6 +30,7 @@
* measurement, and debugging facilities.
*/
#include <linux/compiler.h>
#include <linux/irqflags.h>
#include <asm/octeon/cvmx.h>
#include <asm/octeon/cvmx-l2c.h>
......@@ -285,22 +286,22 @@ uint64_t cvmx_l2c_read_perf(uint32_t counter)
*/
static void fault_in(uint64_t addr, int len)
{
volatile char *ptr;
volatile char dummy;
char *ptr;
/*
* Adjust addr and length so we get all cache lines even for
* small ranges spanning two cache lines.
*/
len += addr & CVMX_CACHE_LINE_MASK;
addr &= ~CVMX_CACHE_LINE_MASK;
ptr = (volatile char *)cvmx_phys_to_ptr(addr);
ptr = cvmx_phys_to_ptr(addr);
/*
* Invalidate L1 cache to make sure all loads result in data
* being in L2.
*/
CVMX_DCACHE_INVALIDATE;
while (len > 0) {
dummy += *ptr;
ACCESS_ONCE(*ptr);
len -= CVMX_CACHE_LINE_SIZE;
ptr += CVMX_CACHE_LINE_SIZE;
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册