[RESEND PATCH] drivers: base: power: Optimize array out-of-bounds access logic

[RESEND PATCH] drivers: base: power: Optimize array out-of-bounds access logic

[xueqin Luo]

The code previously used snprintf to format a string into a buffer and
manually checked for potential buffer overflows by comparing the returned
length with the buffer size. This approach introduced unnecessary
complexity and was prone to subtle errors.

Replaced snprintf with scnprintf, which directly returns the actual number
of characters written to the buffer (excluding the null terminator). This
change eliminates the need for manual overflow checks and simplifies the
buffer offset and size adjustment logic.

Signed-off-by: xueqin Luo <luoxueqin@kylinos.cn>
---
 drivers/base/power/trace.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/drivers/base/power/trace.c b/drivers/base/power/trace.c
index cd6e559648b2..d8da7195bb00 100644
--- a/drivers/base/power/trace.c
+++ b/drivers/base/power/trace.c
@@ -238,10 +238,8 @@ int show_trace_dev_match(char *buf, size_t size)
 		unsigned int hash = hash_string(DEVSEED, dev_name(dev),
 						DEVHASH);
 		if (hash == value) {
-			int len = snprintf(buf, size, "%s\n",
+			int len = scnprintf(buf, size, "%s\n",
 					    dev_driver_string(dev));
-			if (len > size)
-				len = size;
 			buf += len;
 			ret += len;
 			size -= len;
-- 
2.34.1

Re: [RESEND PATCH] drivers: base: power: Optimize array out-of-bounds access logic

[Greg KH]

On Mon, Dec 02, 2024 at 11:23:06AM +0800, xueqin Luo wrote:
> The code previously used snprintf to format a string into a buffer and
> manually checked for potential buffer overflows by comparing the returned
> length with the buffer size. This approach introduced unnecessary
> complexity and was prone to subtle errors.

What errors are in the original code here?  Is it incorrect?

> Replaced snprintf with scnprintf, which directly returns the actual number
> of characters written to the buffer (excluding the null terminator). This
> change eliminates the need for manual overflow checks and simplifies the
> buffer offset and size adjustment logic.

Your lines should be wrapped at 72 columns, right?

> 
> Signed-off-by: xueqin Luo <luoxueqin@kylinos.cn>

Why is this a resend?  What was wrong with the first version?

> ---
>  drivers/base/power/trace.c | 4 +---
>  1 file changed, 1 insertion(+), 3 deletions(-)
> 
> diff --git a/drivers/base/power/trace.c b/drivers/base/power/trace.c
> index cd6e559648b2..d8da7195bb00 100644
> --- a/drivers/base/power/trace.c
> +++ b/drivers/base/power/trace.c
> @@ -238,10 +238,8 @@ int show_trace_dev_match(char *buf, size_t size)
>  		unsigned int hash = hash_string(DEVSEED, dev_name(dev),
>  						DEVHASH);
>  		if (hash == value) {
> -			int len = snprintf(buf, size, "%s\n",
> +			int len = scnprintf(buf, size, "%s\n",
>  					    dev_driver_string(dev));
> -			if (len > size)
> -				len = size;

How was this tested?  I think if code could just be cleaned up
automatically like this, it would have already, right?

thanks,

greg k-h