2017-08-23 14:40:28

by Arnd Bergmann

[permalink] [raw]
Subject: [PATCH] clk: sunxi: fix uninitialized access

gcc-8 reports an uninitialized variable access in a code path
that we would see with incorrect DTB input:

drivers/clk/sunxi/clk-sun8i-bus-gates.c: In function 'sun8i_h3_bus_gates_init':
drivers/clk/sunxi/clk-sun8i-bus-gates.c:85:27: error: 'clk_parent' may be used uninitialized in this function [-Werror=maybe-uninitialized]

This works around by skipping invalid input and printing a warning
instead if it ever happens. The problem was apparently part of the
initiali driver submission, but older compilers don't notice it.

Fixes: ab6e23a4e388 ("clk: sunxi: Add H3 clocks support")
Signed-off-by: Arnd Bergmann <[email protected]>
---
drivers/clk/sunxi/clk-sun8i-bus-gates.c | 4 ++++
1 file changed, 4 insertions(+)

diff --git a/drivers/clk/sunxi/clk-sun8i-bus-gates.c b/drivers/clk/sunxi/clk-sun8i-bus-gates.c
index 63fdb790df29..bee305bdddbe 100644
--- a/drivers/clk/sunxi/clk-sun8i-bus-gates.c
+++ b/drivers/clk/sunxi/clk-sun8i-bus-gates.c
@@ -78,6 +78,10 @@ static void __init sun8i_h3_bus_gates_init(struct device_node *node)
clk_parent = APB1;
else if (index >= 96 && index <= 127)
clk_parent = APB2;
+ else {
+ WARN_ON(true);
+ continue;
+ }

clk_reg = reg + 4 * (index / 32);
clk_bit = index % 32;
--
2.9.0


2017-08-24 08:50:39

by Maxime Ripard

[permalink] [raw]
Subject: Re: [PATCH] clk: sunxi: fix uninitialized access

On Wed, Aug 23, 2017 at 04:39:27PM +0200, Arnd Bergmann wrote:
> gcc-8 reports an uninitialized variable access in a code path
> that we would see with incorrect DTB input:
>
> drivers/clk/sunxi/clk-sun8i-bus-gates.c: In function 'sun8i_h3_bus_gates_init':
> drivers/clk/sunxi/clk-sun8i-bus-gates.c:85:27: error: 'clk_parent' may be used uninitialized in this function [-Werror=maybe-uninitialized]
>
> This works around by skipping invalid input and printing a warning
> instead if it ever happens. The problem was apparently part of the
> initiali driver submission, but older compilers don't notice it.
>
> Fixes: ab6e23a4e388 ("clk: sunxi: Add H3 clocks support")
> Signed-off-by: Arnd Bergmann <[email protected]>

That's weird it stayed for that long without any compiler noticing.

Acked-by: Maxime Ripard <[email protected]>

Mike, Stephen, can you pick it up as a fix? We don't have anything
else coming your way.

Thanks!
Maxime

--
Maxime Ripard, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com


Attachments:
(No filename) (1.03 kB)
signature.asc (801.00 B)
Download all attachments

2017-08-31 17:57:03

by Stephen Boyd

[permalink] [raw]
Subject: Re: [PATCH] clk: sunxi: fix uninitialized access

On 08/23, Arnd Bergmann wrote:
> gcc-8 reports an uninitialized variable access in a code path
> that we would see with incorrect DTB input:
>
> drivers/clk/sunxi/clk-sun8i-bus-gates.c: In function 'sun8i_h3_bus_gates_init':
> drivers/clk/sunxi/clk-sun8i-bus-gates.c:85:27: error: 'clk_parent' may be used uninitialized in this function [-Werror=maybe-uninitialized]
>
> This works around by skipping invalid input and printing a warning
> instead if it ever happens. The problem was apparently part of the
> initiali driver submission, but older compilers don't notice it.
>
> Fixes: ab6e23a4e388 ("clk: sunxi: Add H3 clocks support")
> Signed-off-by: Arnd Bergmann <[email protected]>
> ---

Applied to clk-next

--
Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum,
a Linux Foundation Collaborative Project